Clinical Trial Results:
A phase II baseline versus treatment study to determine the efficacy of raltegravir (ISENTRESS) in preventing progression of relapsing remitting multiple sclerosis as determined by gadoliniumenhanced MRI
Summary


EudraCT number 
201200484761 
Trial protocol 
GB 
Global end of trial date 
10 Jun 2015

Results information


Results version number 
v1(current) 
This version publication date 
25 Jun 2016

First version publication date 
25 Jun 2016

Other versions 

Summary report(s) 
INSPIRE end of study report 
Trial Information
Subject Disposition
Baseline Characteristics
End Points
Adverse Events
More Information
Subject Disposition
Baseline Characteristics
End Points
Adverse Events
More Information


Trial identification


Sponsor protocol code 
008717QM


Additional study identifiers


ISRCTN number 
  
US NCT number 
NCT02104661  
WHO universal trial number (UTN) 
  
Sponsors


Sponsor organisation name 
Queen Mary University of London


Sponsor organisation address 
5 Walden Street, London, United Kingdom, E12EF


Public contact 
Prof Gavin Giovannoni, Queen Mary University of London, +44 02078822579, g.giovannoni@qmul.ac.uk


Scientific contact 
Prof Gavin Giovannoni, Queen Mary University of London, +44 02078822579, g.giovannoni@qmul.ac.uk


Sponsor organisation name 
Queen Mary University of London


Sponsor organisation address 
5 Walden Street, London, United Kingdom, E12EF


Public contact 
Prof Gavin Giovannoni
Neuroscience and Trauma Centre
Blizard Institute 4 Newark St London E12AT, Queen Mary University of London
Blizard Institute 4 Newark St London E12AT, +44 020 7882 8954, g.giovannoni@qmul.ac.uk


Scientific contact 
Prof Gavin Giovannoni
Neuroscience and Trauma Centre
Blizard Institute 4 Newark St London E12AT, Queen Mary University of London
Blizard Institute 4 Newark St London E12AT, +44 020 7882 8954, g.giovannoni@qmul.ac.uk


Paediatric regulatory details


Is trial part of an agreed paediatric investigation plan (PIP) 
No


Does article 45 of REGULATION (EC) No 1901/2006 apply to this trial? 
No


Does article 46 of REGULATION (EC) No 1901/2006 apply to this trial? 
No


Results analysis stage


Analysis stage 
Final


Date of interim/final analysis 
12 Oct 2015


Is this the analysis of the primary completion data? 
Yes


Primary completion date 
10 Sep 2014


Global end of trial reached? 
Yes


Global end of trial date 
10 Jun 2015


Was the trial ended prematurely? 
No


General information about the trial


Main objective of the trial 
The primary objective is to assess whether treatment with raltegravir in patients with active MS has the effect of reducing the total number or rate of development of new or recurrent Gdenhanced lesions on brain MRI over the period of treatment, compared to baseline.


Protection of trial subjects 
All participants provided written informed consent before any study specific assessments were performed. Participants were given ample time for consideration before consenting to take part. Participants were made aware of their right to withdraw from the study at any time for any reason. The investigator also had the right to withdraw participants from the study. The total time on the study for enrolled participants was six months, which was considered to be an ethically acceptable timeframe for patients who are in the early stages of RRMS as this is the time limit before they meet Association of British Neurologists (ABN) criteria for currently licensed disease modifying treatment.


Background therapy 
Not applicable  
Evidence for comparator 
Not applicable for this open label single arm study  
Actual start date of recruitment 
01 Feb 2013


Long term followup planned 
No


Independent data monitoring committee (IDMC) involvement? 
Yes


Population of trial subjects


Number of subjects enrolled per country 

Country: Number of subjects enrolled 
United Kingdom: 31


Worldwide total number of subjects 
31


EEA total number of subjects 
31


Number of subjects enrolled per age group 

In utero 
0


Preterm newborn  gestational age < 37 wk 
0


Newborns (027 days) 
0


Infants and toddlers (28 days23 months) 
0


Children (211 years) 
0


Adolescents (1217 years) 
0


Adults (1864 years) 
31


From 65 to 84 years 
0


85 years and over 
0



Recruitment


Recruitment details 
All participants were recruited at the Clinical Research Centre of the Royal London Hospital and were drawn prevalently from the catchment area of greater London. Participants were also referred to the site by six Patient Identification Centres (PICs). Recruitment into the study started in May 2013. The last patient was screened in June 2014.  
Preassignment


Screening details 
A total of 31 participants were screened, of these 8 had no evidence of Gd enhancing lesions in their baseline MRI and were screen failed. Of the 23 participants who were recruited into the study 3 were withdrawn prior to starting the treatment phase; one at the request of the participant and the remaining two due to MS relapse.  
Preassignment period milestones


Number of subjects started 
31  
Intermediate milestone: Number of subjects 
Observation period: 23


Number of subjects completed 
20  
Preassignment subject noncompletion reasons


Reason: Number of subjects 
No gdenhancing lesions in MRI: 8  
Reason: Number of subjects 
MS relapse: 2  
Reason: Number of subjects 
Consent withdrawn by subject: 1  
Period 1


Period 1 title 
Treatment period (overall period)


Is this the baseline period? 
Yes  
Allocation method 
Not applicable


Blinding used 
Not blinded  
Blinding implementation details 
Not applicable for this open label single arm trial.


Arms


Arm title

Treatment  
Arm description 
Single arm open label  
Arm type 
Experimental  
Investigational medicinal product name 
Raltegravir


Investigational medicinal product code 

Other name 
Isentress


Pharmaceutical forms 
Tablet


Routes of administration 
Oral use


Dosage and administration details 
400mg twice a day administrated as the potassium salt in a film coated tablet




Notes [1]  The number of subjects reported to be in the baseline period are not the same as the worldwide number enrolled in the trial. It is expected that these numbers will be the same. Justification: Of the 31 subjects screened, 20 received treatment. 


Baseline characteristics reporting groups


Reporting group title 
Treatment period


Reporting group description 
  


Subject analysis sets


Subject analysis set title 
Per protocol


Subject analysis set type 
Per protocol  
Subject analysis set description 
4 subjects were excluded from per protocol (PP) analysis due to concomitant medications (2 has steroids/immunosuppressants at screening and 2 had proton pump inhibitors during the study).
PP 12 females (75%), 4 males (25%). Mean age baseline 41.62yrs (31.1552.99); mean height 168.91cm (156.0180.0); mean weight 77.9Kg (51.9108.3); mean EDSS 2.25 (0.03.5); mean no. relapses past year 1.44 (13)


Subject analysis set title 
ITT


Subject analysis set type 
Intentiontotreat  
Subject analysis set description 
All 20 subjects who completed the study were includedin the ITT analysis.
ITT 14 females (70%), 6 males (30%). Mean age baseline 40.73yrs (31.1552.99); mean height 169.04cm (156.0183.0); mean weight 78.73Kg (51.9109.7); mean EDSS 2.4 (0.04.0); mean no. relapses past year 1.5 (13)


Subject analysis set title 
Flexible per protocol


Subject analysis set type 
Per protocol  
Subject analysis set description 
Flexible PP sample n=18. Two subjects were excluded for all visits, one subject had visits seven and eight excluded, and one subject had just visit eight excluded.





End points reporting groups


Reporting group title 
Treatment


Reporting group description 
Single arm open label  
Subject analysis set title 
Per protocol


Subject analysis set type 
Per protocol  
Subject analysis set description 
4 subjects were excluded from per protocol (PP) analysis due to concomitant medications (2 has steroids/immunosuppressants at screening and 2 had proton pump inhibitors during the study).
PP 12 females (75%), 4 males (25%). Mean age baseline 41.62yrs (31.1552.99); mean height 168.91cm (156.0180.0); mean weight 77.9Kg (51.9108.3); mean EDSS 2.25 (0.03.5); mean no. relapses past year 1.44 (13)


Subject analysis set title 
ITT


Subject analysis set type 
Intentiontotreat  
Subject analysis set description 
All 20 subjects who completed the study were includedin the ITT analysis.
ITT 14 females (70%), 6 males (30%). Mean age baseline 40.73yrs (31.1552.99); mean height 169.04cm (156.0183.0); mean weight 78.73Kg (51.9109.7); mean EDSS 2.4 (0.04.0); mean no. relapses past year 1.5 (13)


Subject analysis set title 
Flexible per protocol


Subject analysis set type 
Per protocol  
Subject analysis set description 
Flexible PP sample n=18. Two subjects were excluded for all visits, one subject had visits seven and eight excluded, and one subject had just visit eight excluded.



End point title 
Total T1 Gdenhancing lesions in brain MRI scans  
End point description 
Withinpatient average no. new Gdenh lesions observed on serial T1weighted brain MRI scans. Counts of new or recurrent Gdenh lesions appearing on brain T1weighted MRI. These counts are available at each of visits 2 to 8 in the majority of patients.
ITT n=20
‘flexible’ PP n=18. 2 patients were excluded for all visits, 1 patient had just visits 7 and 8 excluded, and 1 had visit 8 excluded.
PP n=16, 4 patients had all their visits excluded.
There were missing counts for 1patient at visit 3, 1 at visit 5, and missing counts for 2 at visit 6.
Lesion outcomes provide no statistical evidence consistent with an effect of raltegravir.
This is not because of lack of power: changes were not only nonsignificant statistically, but also generally clinically small (including both small reductions and small increases).
For the most substantial change, a reduction in persisting T1 Gd lesions in the PP sample, the decrease before intervention was greater than that afterwards


End point type 
Primary


End point timeframe 
Visit two (enrolment) is excluded from analysis, which covers visits three, four and five "before", and six, seven and eight "after" medication was first dispensed.




Statistical analysis title 
Mixed effect Poisson regression model  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator, adjusting for the (log) enrolment visit value


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other ^{[1]}  
Pvalue 
= 0.681 ^{[2]}  
Method 
Regression, Linear  
Parameter type 
Rate ratio  
Point estimate 
1.04


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.85  
upper limit 
1.29  
Notes [1]  The analyses compare the mean rate after vs before. The potential gradients of change over the threemonth before vs after periods were also compared. For all three samples above there was no significant change in the after vs before gradients of monthly lesion accrual: Pvalues were respectively P=0.659, 0.429 and 0.463 for ITT, flexible PP and PP. [2]  Est rate ratio after vs before: 1.04 (95% CI .85, 1.29); represents 4% nonsignificant incr lesions/month in after period, weighted ITT rate ratio 1.03 Simple nonparametric Wilcoxon sign rank test withinpatient changes nonsignificant, P=0.646 

Statistical analysis title 
Mixed effect Poisson regression model  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator, adjusting for the (log) enrolment visit value


Comparison groups 
Treatment v Flexible per protocol


Number of subjects included in analysis 
38


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.714 ^{[3]}  
Method 
Regression, Linear  
Parameter type 
rate ratio  
Point estimate 
0.95


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.75  
upper limit 
1.22  
Notes [3]  5% nonsignificant decrease in rate. This is in close agreement with the summary weighted rate ratio of 0.92. The pvalue from the simple nonparametric Wilcoxon sign rank test of changes is nonsignificant, P=0.183. 

Statistical analysis title 
Mixed effect Poisson regression model  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator, adjusting for the (log) enrolment visit value


Comparison groups 
Treatment v Per protocol


Number of subjects included in analysis 
36


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.577 ^{[4]}  
Method 
Regression, Linear  
Parameter type 
rate ratio  
Point estimate 
0.93


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.72  
upper limit 
1.2  
Notes [4]  Nonsignificant 7% decrease. This is similar to the summary weighted rate ratio of 0.92. The pvalue from the simple nonparametric Wilcoxon sign rank test of changes is also nonsignificant, P=0.197. 


End point title 
Persisting T1 Gdenhancing brain MRI lesions  
End point description 

End point type 
Primary


End point timeframe 
Visits 3, 4 and 5 (before) and 6, 7 and 8 (after medication was first dispensed).




Statistical analysis title 
Poisson regression  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator.


Comparison groups 
Treatment v Per protocol


Number of subjects included in analysis 
36


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.16 ^{[5]}  
Method 
Regression, Linear  
Parameter type 
Slope  
Point estimate 
0.78


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.55  
upper limit 
1.1  
Notes [5]  The rate ratio was estimated as 0.78 (95% CI:0.55, 1.10) P=0.160, a nonsignificant reduction. There was no significant change in gradient in the PP (0.762). In this PP sample the rate of reduction after was slightly lower than that before. 

Statistical analysis title 
Poisson regression  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.652 ^{[6]}  
Method 
Regression, Linear  
Parameter type 
Slope  
Point estimate 
0.93


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.69  
upper limit 
1.26  
Notes [6]  The rate ratio was estimated as 0.93 (95% CI .69, 1.26), P=0.652, a slight and nonsignificant reduction in persisting lesions. 


End point title 
New T1 Gd enhancing brain MRI lesions  
End point description 
Rates in each 3months period.


End point type 
Primary


End point timeframe 
Before (visits 3, 4, 5) vs After (visits 6, 7, 8).




Statistical analysis title 
Poisson regression model  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator estimates the rate ratio after vs before as 1.16 (95% CI 0.87, 1.55), P=0.314; this represents a slight and nonsignificant increase in new lesions per month in the ‘after’ period. The above analyses compare the mean rate after vs before. The gradients of change over the threemonth before vs after periods were also compared.


Comparison groups 
Treatment v Per protocol


Number of subjects included in analysis 
36


Analysis specification 
Prespecified


Analysis type 
other ^{[7]}  
Pvalue 
= 0.456  
Method 
Regression, Linear  
Confidence interval 

Notes [7]  There was no significant change in the after vs before gradients of monthly lesion accrual in the PP (P=0.137). 

Statistical analysis title 
Poisson regression model  
Statistical analysis description 
A mixed effect Poisson regression model with before/after indicator estimates the rate ratio after vs before as 1.16 (95% CI 0.87, 1.55), P=0.314; this represents a slight and nonsignificant increase in new lesions per month in the ‘after’ period. This analysis compares the mean rate after vs before. The gradients of change over the threemonth before vs after periods were also compared.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.314 ^{[8]}  
Method 
Regression, Linear  
Parameter type 
Slope  
Point estimate 
1.16


Confidence interval 

level 
95%  
sides 
2sided


lower limit 
0.87  
upper limit 
1.55  
Notes [8]  There was no significant change in the after vs before gradients of monthly lesion accrual in the ITT (P=0.562) analysis. 


End point title 
New or enlarging T2 weighted lesions on brain MRI  
End point description 
For T2 lesions (where the only ‘after’ observation is visit 8, so no comparison is possible for the two ‘flexible PP’ patients), only ITT and PP are given. Withinpatient ratios averaged on log scale before back transforming, substituting 0.1 for zero counts. Weighted mean ratios pooled on log scale by weighting for lesion counts.


End point type 
Secondary


End point timeframe 
Counts of new or enlarging T2weighted lesions at visits five (before) and eight (after).




Statistical analysis title 
Wilcoxon sign rank test  
Comparison groups 
Treatment v Per protocol


Number of subjects included in analysis 
36


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.462 ^{[9]}  
Method 
Wilcoxon (MannWhitney)  
Confidence interval 

Notes [9]  Nonsignificant: Wilcoxon sign rank test of the withinpatient changes was P= 0.462 for PP analysis. 

Statistical analysis title 
Wilcoxon sign rank test  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.472 ^{[10]}  
Method 
Wilcoxon (MannWhitney)  
Confidence interval 

Notes [10]  Nonsignificant: Wilcoxon sign rank test of the withinpatient changes was P=0.472 for ITT analysis. 


End point title 
Inflammatory markers  
End point description 
Statistical analysis performed on three inflammatory markers:
Interleukin8 (IL8), or chemokine (CXC motif) ligand 8, CXCL8, is a chemokine produced by macrophages and other cell types. IL8 secretion is increased by oxidant stress, which thereby cause the recruitment of inflammatory cells and induces a further increase in oxidant stress mediators, making it a key parameter in localized inflammation. Reported in pg/mL Unit.
Serum CD163 (a soluble form of the receptor exists in plasma, commonly denoted sCD163. It is generated by ectodomain shedding of the membrane bound receptor. sCD163 is upregulated in a large range of inflammatory diseases). Reported in ng/mL Unit.
Human Creactive protein (HCRP), CRP is used mainly as a marker of inflammation.
For HCRP, three measurements above the measureable threshold were assigned values of 10000 ng/ml. The largest measurable HCRP value in the dataset is 9492.25. Reported in ng/mL Unit.


End point type 
Secondary


End point timeframe 
Changes in mean at visits 3, 4, and 5 (before) and 6, 7, and 8 (after).




Statistical analysis title 
IL8 One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.1 ^{[11]}  
Method 
One sample ttest  
Confidence interval 

Notes [11]  Non significant 

Statistical analysis title 
IL8 Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.47 ^{[12]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [12]  Non significant 

Statistical analysis title 
IL8 Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.193 ^{[13]}  
Method 
Change in gradient  
Confidence interval 

Notes [13]  Non significant 

Statistical analysis title 
CD163 one sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.041 ^{[14]}  
Method 
One sample ttest  
Confidence interval 

Notes [14]  Significant decline changes to a significant positive gradient. The gradient change is significant in both ITT and PP. 

Statistical analysis title 
CD163 Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.04 ^{[15]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [15]  Significant decline changes to a significant positive gradient. The gradient change is significant in both ITT and PP. 

Statistical analysis title 
CD163 Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.018 ^{[16]}  
Method 
Change in gradient  
Confidence interval 

Notes [16]  Significant. 

Statistical analysis title 
HCRP one sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.069 ^{[17]}  
Method 
One sample ttest  
Confidence interval 

Notes [17]  Significant overall increase not credibly attributable to intervention, since the gradient before, though nonsignificant, is too similar to the gradient after intervention. 

Statistical analysis title 
HCRP Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.04 ^{[18]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [18]  Significant overall increase not credibly attributable to intervention, since the gradient before, though nonsignificant, is too similar to the gradient after intervention 

Statistical analysis title 
HCRP Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.944 ^{[19]}  
Method 
Change in gradient  
Confidence interval 

Notes [19]  Non significant 


End point title 
Retroviral activity  
End point description 
To date, there has been no definitive evidence to link HERVs as the cause of immunemediated disease; however, HERV elements have been found in sera of people with a range of diseases such as type 1 diabetes, rheumatoid arthritis and SLE but not in control populations. The evidence suggesting a postulated link between HERVs and MS has been accumulating.
evidence is summarized to demonstrate that HERVH and HERVW are epidemiologically linked to patients with relapsing remitting MS. Further evidence was recently published by Perron et al. (4) that also links MS to a HERV which Perron calls multiple sclerosis associated retroviral element (MSRV).
Raltegravir effect in relation to MS is not known, it may act by inhibiting HERVs, possibly in a similar mode of action that Raltegravir inhibits HIV replication.


End point type 
Secondary


End point timeframe 
Changes in mean after  before




Notes [20]  HERV_W_a; HERV_W_m; HERV_W_n n=18 HERV_H_b; HERV_H_c; HERV_H_k n=17 HERV_W_i n=13 HERV_H_j n=15 

Statistical analysis title 
One sample ttest MSRV_a  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.551 ^{[21]}  
Method 
One sample ttest  
Confidence interval 

Notes [21]  Non significant. 

Statistical analysis title 
One sample ttest MSRV_b  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.484 ^{[22]}  
Method 
One sample ttest  
Confidence interval 

Notes [22]  Non significant. 

Statistical analysis title 
One sample ttest HERV_c  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.835 ^{[23]}  
Method 
One sample ttest  
Confidence interval 

Notes [23]  Non significant. 

Statistical analysis title 
One sample ttest HERV_d  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.928  
Method 
One sample ttest  
Confidence interval 

Statistical analysis title 
One sample ttest HERV_e  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.563 ^{[24]}  
Method 
One sample ttest  
Confidence interval 

Notes [24]  Non significant. 

Statistical analysis title 
One sample ttest HERV_f  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.408 ^{[25]}  
Method 
One sample ttest  
Confidence interval 

Notes [25]  Non significant. 

Statistical analysis title 
One sample ttest HERV_W_a  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.012 ^{[26]}  
Method 
One sample ttest  
Confidence interval 

Notes [26]  There is a significant drop in mean from before to after; however, there is a negative gradient of decline throughout the trial period. Therefore the after vs before drop cannot reasonably be attributed to the intervention. 

Statistical analysis title 
One sample ttest HERV_H_b  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.372 ^{[27]}  
Method 
One sample ttest  
Confidence interval 

Notes [27]  Non significant. 

Statistical analysis title 
One sample ttest HERV_H_c  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.177 ^{[28]}  
Method 
One sample ttest  
Confidence interval 

Notes [28]  Non significant. 

Statistical analysis title 
One sample ttest HERV_H_d  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.113 ^{[29]}  
Method 
One sample ttest  
Confidence interval 

Notes [29]  Non significant. 

Statistical analysis title 
One sample ttest HERV_W_e  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.907 ^{[30]}  
Method 
One sample ttest  
Confidence interval 

Notes [30]  Non significant. 

Statistical analysis title 
One sample ttest HERV_W_f  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.632 ^{[31]}  
Method 
One sample ttest  
Confidence interval 

Notes [31]  Non significant. 

Statistical analysis title 
One sample ttest PROP_B_g  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.148 ^{[32]}  
Method 
One sample ttest  
Confidence interval 

Notes [32]  Non significant. 

Statistical analysis title 
One sample ttest PROP_mon_h  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.053  
Method 
One sample ttest  
Confidence interval 

Statistical analysis title 
One sample ttest HERV_W_i  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.561 ^{[33]}  
Method 
One sample ttest  
Confidence interval 

Notes [33]  Non significant. 

Statistical analysis title 
One sample ttest HERV_H_j  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.029 ^{[34]}  
Method 
One sample ttest  
Confidence interval 

Notes [34]  Non significant. 

Statistical analysis title 
One sample ttest HERV_H_k  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.266 ^{[35]}  
Method 
One sample ttest  
Confidence interval 

Notes [35]  Non significant. 

Statistical analysis title 
One sample ttest HERV_H_l  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.522 ^{[36]}  
Method 
One sample ttest  
Confidence interval 

Notes [36]  Non significant. 

Statistical analysis title 
One sample ttest HERV_W_m  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.035 ^{[37]}  
Method 
One sample ttest  
Confidence interval 

Notes [37]  Non significant. 

Statistical analysis title 
One sample ttest HERV_W_n  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.158 ^{[38]}  
Method 
One sample ttest  
Confidence interval 

Notes [38]  Non significant. 

Statistical analysis title 
Mixed model MSRV_a  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.364 ^{[39]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [39]  Non significant. 

Statistical analysis title 
Mixed model MSRV_b  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.433 ^{[40]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [40]  Non significant. 

Statistical analysis title 
Mixed model HERV_c  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.775 ^{[41]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [41]  Non significant. 

Statistical analysis title 
Mixed model HERV_d  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.609 ^{[42]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [42]  Non significant. 

Statistical analysis title 
Mixed model HERV_e  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.781 ^{[43]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [43]  Non significant. 

Statistical analysis title 
Mixed model HERV_f  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.324 ^{[44]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [44]  Non significant. 

Statistical analysis title 
Mixed model HERV_W_a  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.008  
Method 
Mixed models analysis  
Confidence interval 

Statistical analysis title 
Mixed model HERV_H_b  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.307 ^{[45]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [45]  Non significant. 

Statistical analysis title 
Mixed model HERV_H_c  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.212 ^{[46]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [46]  Non significant. 

Statistical analysis title 
Mixed model HERV_H_d  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.203 ^{[47]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [47]  Non significant. 

Statistical analysis title 
Mixed model HERV_W_e  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.235 ^{[48]}  
Method 
Change in gradient  
Confidence interval 

Notes [48]  Non significant. 

Statistical analysis title 
Mixed model HERV_W_f  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.627 ^{[49]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [49]  Non significant. 

Statistical analysis title 
Mixed model PROP_B_g  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.298 ^{[50]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [50]  Non significant. 

Statistical analysis title 
Mixed model PROP_mon_h  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.16 ^{[51]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [51]  Non significant. 

Statistical analysis title 
Mixed model HERV_W_i  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.93 ^{[52]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [52]  Non significant. 

Statistical analysis title 
Mixed model HERV_H_j  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.008  
Method 
Mixed models analysis  
Confidence interval 

Statistical analysis title 
Mixed model HERV_H_k  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.642  
Method 
Mixed models analysis  
Confidence interval 

Statistical analysis title 
Mixed model HERV_H_l  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.807  
Method 
Mixed models analysis  
Confidence interval 

Statistical analysis title 
Mixed model HERV_W_m  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.005  
Method 
Mixed models analysis  
Confidence interval 

Statistical analysis title 
Mixed model HERV_W_n  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.259 ^{[53]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [53]  Non significant. 

Statistical analysis title 
Change in gradient MSRV_a  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.488 ^{[54]}  
Method 
Change in gradient  
Confidence interval 

Notes [54]  Non significant. 

Statistical analysis title 
Change in gradient MSRV_b  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.691 ^{[55]}  
Method 
Change in gradient  
Confidence interval 

Notes [55]  Non significant. 

Statistical analysis title 
Change in gradient HERV_c  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.172 ^{[56]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [56]  Non significant. 

Statistical analysis title 
Change in gradient HERV_d  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.802 ^{[57]}  
Method 
Change in gradient  
Confidence interval 

Notes [57]  Non significant. 

Statistical analysis title 
Change in gradient HERV_e  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.621 ^{[58]}  
Method 
Change in gradient  
Confidence interval 

Notes [58]  Non significant. 

Statistical analysis title 
Change in gradient HERV_f  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.533 ^{[59]}  
Method 
Change in gradient  
Confidence interval 

Notes [59]  Non significant. 

Statistical analysis title 
Change in gradient HERV_W_a  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.438 ^{[60]}  
Method 
Change in gradient  
Confidence interval 

Notes [60]  Non significant. 

Statistical analysis title 
Change in gradient HERV_H_b  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.012 ^{[61]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [61]  Gradient change significant P=0.012. Gradient before borderline significantly negative. After is significantly positive. Decline in values in period before intervention appears to change significantly after v 5 into an increase in values over time. 

Statistical analysis title 
Change in gradient HERV_H_c  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.197 ^{[62]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [62]  Non significant. 

Statistical analysis title 
Change in gradient HERV_H_d  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[63]}  
Pvalue 
= 0.01 ^{[64]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [63]  The change in gradient is consistent with intervention effect. However, interpretation dependent on biological plausibility because of possibility of Type I error. [64]  Borderline significant negative gradient before changes to significant positive after v 5. Decline in values appears to change significantly after intervention into increase in values over time. 

Statistical analysis title 
Change in gradient HERV_W_e  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.008 ^{[65]}  
Method 
Change in gradient  
Confidence interval 

Notes [65]  Significantly negative decline before visit 5 changes to a nonsignificant positive gradient after; the change in gradients is significant P=0.008. 

Statistical analysis title 
Change in gradient HERV_W_f  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.195 ^{[66]}  
Method 
Change in gradient  
Confidence interval 

Notes [66]  Non significant. 

Statistical analysis title 
Change in gradient PROP_B_g  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.405 ^{[67]}  
Method 
Change in gradient  
Confidence interval 

Notes [67]  Non significant. 

Statistical analysis title 
Change in gradient PROP_mon_h  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.072 ^{[68]}  
Method 
Change in gradient  
Confidence interval 

Notes [68]  Nonsignificant positive gradient changes to borderline significant negative gradient; change in gradient borderline significant consistent with intervention effect. Interpretation dependent on biological plausibility due to possibility Type 1 error 

Statistical analysis title 
Change in gradient HERV_W_i  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.552 ^{[69]}  
Method 
Change in gradient  
Confidence interval 

Notes [69]  Nonsignificant. 

Statistical analysis title 
Change in gradient HERV_H_i  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.054 ^{[70]}  
Method 
Change in gradient  
Confidence interval 

Notes [70]  A nonsignifiant negative gradient before intervention changes to a significantly positive gradient after; the change is borderline signififcant P=0.054. 

Statistical analysis title 
Change in gradient HERV_H_k  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.802 ^{[71]}  
Method 
Change in gradient  
Confidence interval 

Notes [71]  Nonsignificant. 

Statistical analysis title 
Change in gradient HERV_H_l  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.314 ^{[72]}  
Method 
Change in gradient  
Confidence interval 

Notes [72]  Nonsignificant. 

Statistical analysis title 
Change in gradient HERV_W_m  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.178 ^{[73]}  
Method 
Change in gradient  
Confidence interval 

Notes [73]  Nonsignificant. 

Statistical analysis title 
Change in gradient HERV_W_n  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.047 ^{[74]}  
Method 
Change in gradient  
Confidence interval 

Notes [74]  Nonsignificant decline before visit 5 becomes a borderline significant increase after; the change in gradient is significant. Biological plausibility important, both for change displayed and for negative correlation with Gd T1 lesions. 


End point title 
EDSS Clinical responses (disability data)  
End point description 
Expanded Disability Status Scale score at screening between 06.0 inclusive for trial eligibility. Disability measures summaries at baseline, before and after.


End point type 
Secondary


End point timeframe 
EDSS performed at visits 1, 2, 4, 6 and 8. Summaries of disability measured at baseline, before and after.




Statistical analysis title 
Onesample ttest  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other ^{[75]}  
Pvalue 
= 0.179  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [75]  Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant. 

Statistical analysis title 
Mixed model  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other ^{[76]}  
Pvalue 
= 0.13  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [76]  Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8. 


End point title 
MSFC Clinical responses (disability data)  
End point description 
MSFC (the standardly derived composite score from 9hole peg test (9HPT), timed walk and PASAT scores); higher scores indicate less disability


End point type 
Secondary


End point timeframe 
Changes in mean after  before (treatment)




Statistical analysis title 
Onesample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.002 ^{[77]}  
Method 
One sample ttest  
Confidence interval 

Notes [77]  Both change and change in gradient significant. 

Statistical analysis title 
Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
< 0.001 ^{[78]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [78]  Both change and change in gradient significant. 


End point title 
9HPT speed Clinical responses (disability data)  
End point description 

End point type 
Secondary


End point timeframe 
Changes in mean after  before




Statistical analysis title 
One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
< 0.001 ^{[79]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [79]  Highly significant improvement. There are statistically significant improvements in the 9HPT, but the rate of improvement slowed after intervention. This is not consistent with an effect of intervention. 

Statistical analysis title 
Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
< 0.001 ^{[80]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [80]  There are statistically significant improvements in the 9HPT, but the rate of improvement slowed after intervention. This is not consistent with an effect of intervention. 

Statistical analysis title 
Change in gradient  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.217 ^{[81]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [81]  No significant gradient change. Rate of improvement substantially greater before than after intervention 


End point title 
25 foot walk speed Clinical responses (disability data)  
End point description 

End point type 
Secondary


End point timeframe 
Changes in mean after  before




Statistical analysis title 
One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.738 ^{[82]}  
Method 
One sample ttest  
Confidence interval 

Notes [82]  Non significant. 

Statistical analysis title 
Mixed model analysis  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.669 ^{[83]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [83]  Non significant. 

Statistical analysis title 
Change in gradient  
Statistical analysis description 
Change in gradient’: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.908 ^{[84]}  
Method 
change in gradient  
Confidence interval 

Notes [84]  Non significant 


End point title 
PASAT Clinical responses (disability data)  
End point description 

End point type 
Secondary


End point timeframe 
Changes in mean after  before




Statistical analysis title 
One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.001 ^{[85]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [85]  Highly significant improvement, but rate of improvement substantially greater before than after intervention 

Statistical analysis title 
Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
< 0.001 ^{[86]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [86]  Highly significant improvement, but rate of improvement substantially greater before than after intervention. 

Statistical analysis title 
Change in gradient  
Statistical analysis description 
Change in gradient’: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
other  
Pvalue 
= 0.004 ^{[87]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

sides 
2sided


lower limit 
  
upper limit 
  
Notes [87]  There is a significant gradient change, but rate of improvement substantially greater before than after intervention 


End point title 
Quality of life measures  
End point description 
Quality of life measures: patient reported outcomes (PROs) including 3 questionnaires: Health Status Questionnaire (SF36), Multiple Sclerosis Impact Scale (MSIS29), Multiple Sclerosis Walking Scale (MSWS12) and 2 Visual Analogue Scales (VAS): Patient Fatigue Assessment (PFA) and Patient Pain Assessment (PPA). The SF36 generates 8 subscale scores for Physical Functional Scale (PF), RolePhysical Scale (RP), Bodily Pain Scale (BP), General Health scale (GH), Vitality Scale (VT), Social Functioning Scale (SF), Role Emotional Scale (RE) and Mental Health Scale (MH). Higher scores indicate better health/quality of life on all eight SF36 measures, and worse health/quality of life on the last four measures.


End point type 
Secondary


End point timeframe 
Quality of life measures baseline, before and after summeries. Patient reported outcomes completed at visits 2 (baseline), 3, 4 and 5 (before) and 6, 7 and 8 (after).




Statistical analysis title 
SF36 PF One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.713 ^{[88]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [88]  Non significant 

Statistical analysis title 
SF36 PF Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.685 ^{[89]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [89]  Non significant 

Statistical analysis title 
SF36 PF Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[90]}  
Pvalue 
= 0.063 ^{[91]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [90]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [91]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 RP One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.198 ^{[92]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [92]  Non significant. 

Statistical analysis title 
SF36 RP Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[93]}  
Pvalue 
= 0.034 ^{[94]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [93]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [94]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 RP Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.873 ^{[95]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [95]  Non significant. 

Statistical analysis title 
SF36 BP One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.14 ^{[96]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [96]  Non significant 

Statistical analysis title 
SF36 BP Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[97]}  
Pvalue 
= 0.024 ^{[98]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [97]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [98]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 BP Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[99]}  
Pvalue 
= 0.026 ^{[100]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [99]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [100]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 GH One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[101]}  
Pvalue 
= 0.019 ^{[102]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [101]  Because of the subjective nature of these measures compared to the disability data, they are susceptible to placebo effects, in particular, a positive response both to being in a trial and to receiving an unblinded intervention. Therefore, an improvement, in the period after intervention, though consistent with an effect of the drug administered at intervention, cannot credibly be attributed to the drug. [102]  Significant improvements in wellbeing/slowing of deterioration after compared to before intervention. Although consistent with effect of raltegravir, this is most credibly attributable to placebo effect of patients receiving unblinded intervention. 

Statistical analysis title 
SF36 GH Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[103]}  
Pvalue 
= 0.001 ^{[104]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [103]  Because of the subjective nature of these measures compared to the disability data, they are susceptible to placebo effects, in particular, a positive response both to being in a trial and to receiving an unblinded intervention. Therefore, an improvement, in the period after intervention, though consistent with an effect of the drug administered at intervention, cannot credibly be attributed to the drug. [104]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 GH Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[105]}  
Pvalue 
= 0.022 ^{[106]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [105]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [106]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 VT One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.14 ^{[107]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [107]  Non significant. 

Statistical analysis title 
SF36 VT Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[108]}  
Pvalue 
= 0.011 ^{[109]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [108]  Because of the subjective nature of these measures compared to the disability data, they are susceptible to placebo effects, in particular, a positive response both to being in a trial and to receiving an unblinded intervention. Therefore, an improvement, in the period after intervention, though consistent with an effect of the drug administered at intervention, cannot credibly be attributed to the drug. [109]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 VT Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[110]}  
Pvalue 
= 0.004 ^{[111]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [110]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [111]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 SF One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.567 ^{[112]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [112]  Non significant. 

Statistical analysis title 
SF36 SF Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.414 ^{[113]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [113]  Non significant. 

Statistical analysis title 
SF36 SF Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[114]}  
Pvalue 
= 0.039 ^{[115]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [114]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [115]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
SF36 RE One sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.252 ^{[116]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [116]  Non significant. 

Statistical analysis title 
SF36 RE Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.096 ^{[117]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [117]  Non significant. 

Statistical analysis title 
SF36 RE Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.768 ^{[118]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [118]  Non significant. 

Statistical analysis title 
PFA One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.665 ^{[119]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [119]  Non significant. 

Statistical analysis title 
PFA Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.683 ^{[120]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [120]  Non significant. 

Statistical analysis title 
PFA Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[121]}  
Pvalue 
= 0.057 ^{[122]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [121]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [122]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
PPA One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.17 ^{[123]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [123]  Non significant. 

Statistical analysis title 
PPA Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.144 ^{[124]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [124]  Non significant. 

Statistical analysis title 
PPA Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[125]}  
Pvalue 
= 0.01 ^{[126]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [125]  Because of subjective nature these measures are susceptible to placebo effects, in particular, a positive response to being in a trial and receiving an unblinded intervention. Improvement after intervention cannot credibly be attibuted to the drug. [126]  Statistically significant improvements in the period after compared to before intervention. Although consistent with an effect of raltegravir, this is most credibly attributable to the placebo effect of patients receiving an unblinded intervention. 

Statistical analysis title 
MSIS One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.477 ^{[127]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [127]  Non significant. 

Statistical analysis title 
MSIS Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.22 ^{[128]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [128]  Non significant. 

Statistical analysis title 
MSIS Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.196 ^{[129]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [129]  Non significant. 

Statistical analysis title 
MSWS One sample ttest  
Statistical analysis description 
Onesample ttest’: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.817 ^{[130]}  
Method 
One sample ttest  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [130]  Non significant. 

Statistical analysis title 
MSWS Mixed model  
Statistical analysis description 
Mixed model’: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.772 ^{[131]}  
Method 
Mixed models analysis  
Parameter type 
Mean difference (net)  
Confidence interval 

Notes [131]  Non significant. 

Statistical analysis title 
MSWS Change in gradient  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.12 ^{[132]}  
Method 
Change in gradient  
Parameter type 
Slope  
Confidence interval 

Notes [132]  Non significant. 


End point title 
EBV copy number in saliva  
End point description 
Epstein–Barr virus (EBV), also called human herpesvirus 4 (HHV4), infection is associated with with a higher risk of certain autoimmune diseases, such as Multiple Sclerosis. In particular, people who have had glandular fever, the symptomatic EBV infection , have a higher risk to develop MS.
EBV may be found in the saliva of someone who has had glandular fever for several months after their symptoms pass, and most people may continue to have the virus in their saliva on and off for years. Studies of dynamics of virus shedding in healthy carriers demonstrate that EBV shedding into saliva is constant.


End point type 
Secondary


End point timeframe 
Changes in mean after  before.




Statistical analysis title 
EBV one sample ttest  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
33


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.226 ^{[133]}  
Method 
One sample ttest  
Confidence interval 

Notes [133]  Non significant. 

Statistical analysis title 
EBV Mixed model  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 2 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
33


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.616 ^{[134]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [134]  Non significant. 

Statistical analysis title 
EBV change in gradient  
Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
33


Analysis specification 
Prespecified


Analysis type 
^{[135]}  
Pvalue 
= 0.219 ^{[136]}  
Method 
Change in gradient  
Confidence interval 

Notes [135]  Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero). [136]  Non significant. 


End point title 
Laboratory safety data  
End point description 
Laboratory safety outcomes. Assessments collected at screening were used to determe study eligibility only. Safety assessments were collected at all visits.
Severity of abnormal results was evaluated by the investigator as mild, moderate or severe. Lab findings which the investigator felt were clinically significant based on the Laboratory Guidelines were recorded as adverse events. The relationship of the adverse event to the administration of the study drug was also determined by the investigator.


End point type 
Secondary


End point timeframe 
Changes in after  before (treatment).




Statistical analysis title 
One sample ttest Adjusted Calcium serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.025 ^{[137]}  
Method 
One sample ttest  
Confidence interval 

Notes [137]  There was a significant increase after vs before, 0.02 (P=0.025). However, there was no significant change in gradient P=0.406 

Statistical analysis title 
One sample ttest ALT  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.575 ^{[138]}  
Method 
One sample ttest  
Confidence interval 

Notes [138]  Non significant. 

Statistical analysis title 
One sample ttest AST  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.266 ^{[139]}  
Method 
One sample ttest  
Confidence interval 

Notes [139]  Non significant. 

Statistical analysis title 
One sample ttest basophil count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.883 ^{[140]}  
Method 
One sample ttest  
Confidence interval 

Notes [140]  Non significant. 

Statistical analysis title 
One sample ttest chloride serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.055 ^{[141]}  
Method 
One sample ttest  
Confidence interval 

Notes [141]  Borderline significant drop, 0.75 (P=0.055). However, there was no significant change in gradient, P=0.194. 

Statistical analysis title 
One sample ttest Cholesterol HDL ratio serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.369 ^{[142]}  
Method 
One sample ttest  
Confidence interval 

Notes [142]  Non significant. 

Statistical analysis title 
One sample ttest Cholesterol serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.004 ^{[143]}  
Method 
One sample ttest  
Confidence interval 

Notes [143]  There was a significant increase P=0.004. However, there was no significant change in gradient P=0.437. Faster increase before than after. 

Statistical analysis title 
One sample ttest Creatinine serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.132 ^{[144]}  
Method 
One sample ttest  
Confidence interval 

Notes [144]  Non significant. 

Statistical analysis title 
One sample ttest Eosinophil count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.855 ^{[145]}  
Method 
One sample ttest  
Confidence interval 

Notes [145]  Non significant. 

Statistical analysis title 
One sample ttest Estimated GFR  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.186 ^{[146]}  
Method 
One sample ttest  
Confidence interval 

Notes [146]  Non significant. 

Statistical analysis title 
One sample ttest GGT  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.789 ^{[147]}  
Method 
One sample ttest  
Confidence interval 

Notes [147]  Non significant. 

Statistical analysis title 
One sample ttest Glucose plasma  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.108 ^{[148]}  
Method 
One sample ttest  
Confidence interval 

Notes [148]  Non significant. 

Statistical analysis title 
One sample ttest Haemoglobin  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.308 ^{[149]}  
Method 
One sample ttest  
Confidence interval 

Notes [149]  Non significant. 

Statistical analysis title 
One sample ttest HDL Cholesterol serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.174 ^{[150]}  
Method 
One sample ttest  
Confidence interval 

Notes [150]  Non significant. 

Statistical analysis title 
One sample ttest LDL Cholesterol serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.027 ^{[151]}  
Method 
One sample ttest  
Confidence interval 

Notes [151]  There was a significant increase, but no significant change in gradient P=0.432. 

Statistical analysis title 
One sample ttest Lymphocyte count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.465 ^{[152]}  
Method 
One sample ttest  
Confidence interval 

Notes [152]  Non significant. 

Statistical analysis title 
One sample ttest Monocyte count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.958 ^{[153]}  
Method 
One sample ttest  
Confidence interval 

Notes [153]  Non significant. 

Statistical analysis title 
One sample ttest Neutrophil count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.291 ^{[154]}  
Method 
One sample ttest  
Confidence interval 

Notes [154]  Non significant. 

Statistical analysis title 
One sample ttest Platelet count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.04 ^{[155]}  
Method 
One sample ttest  
Confidence interval 

Notes [155]  There was asignificant increase P=0.040. However, there was no significant change in gradient P=0.352. 

Statistical analysis title 
One sample ttest Potassium serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.161 ^{[156]}  
Method 
One sample ttest  
Confidence interval 

Notes [156]  There was asignificant increase P=0.040. However, there was no significant change in gradient P=0.352. 

Statistical analysis title 
One sample ttest Sodium serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.692 ^{[157]}  
Method 
One sample ttest  
Confidence interval 

Notes [157]  Non significant. 

Statistical analysis title 
One sample ttest Total bilirubin serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.591 ^{[158]}  
Method 
One sample ttest  
Confidence interval 

Notes [158]  Non significant. 

Statistical analysis title 
One sample ttest Triglicerides serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.225 ^{[159]}  
Method 
One sample ttest  
Confidence interval 

Notes [159]  Non significant. 

Statistical analysis title 
One sample ttest Urea serum  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.183 ^{[160]}  
Method 
One sample ttest  
Confidence interval 

Notes [160]  Non significant. 

Statistical analysis title 
One sample ttest White blood count  
Statistical analysis description 
Onesample ttest: this uses one datapoint per patient, the withinpatient change in means, testing whether the mean of these changes is significant.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.184 ^{[161]}  
Method 
One sample ttest  
Confidence interval 

Notes [161]  Non significant. 

Statistical analysis title 
Mixed model Adjusted calcium serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.027 ^{[162]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [162]  There was a significant increase after vs before. However, there was no significant change in gradient. 

Statistical analysis title 
Mixed model ALT  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.643 ^{[163]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [163]  Non significant. 

Statistical analysis title 
Mixed model AST  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.265 ^{[164]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [164]  Non significant. 

Statistical analysis title 
Mixed model Basophil count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.954 ^{[165]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [165]  Non significant. 

Statistical analysis title 
Mixed model Chloride serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.032 ^{[166]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [166]  Borderline significant drop. However, there was no significant change in gradient 

Statistical analysis title 
Mixed model Cholesterol HDL ratio serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.441 ^{[167]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [167]  Borderline significant drop. 

Statistical analysis title 
Mixed model Cholesterol serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0 ^{[168]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [168]  There was a significant increase. However, there was no significant change in gradient. 

Statistical analysis title 
Mixed model Creatinine serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.097 ^{[169]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [169]  Non significant. 

Statistical analysis title 
Mixed model Eosinophil count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.711 ^{[170]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [170]  Non significant. 

Statistical analysis title 
Mixed model Estimated GFR  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.153 ^{[171]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [171]  Non significant. 

Statistical analysis title 
Mixed model GGT  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.794 ^{[172]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [172]  Non significant. 

Statistical analysis title 
Mixed model Glucose Plasma  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.196 ^{[173]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [173]  Non significant. 

Statistical analysis title 
Mixed model Haemoglobin  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.168 ^{[174]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [174]  Non significant. 

Statistical analysis title 
Mixed model HDL cholesterol serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.101 ^{[175]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [175]  Non significant. 

Statistical analysis title 
Mixed model LDL cholesterol serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.018 ^{[176]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [176]  There was a significant increase. However, there was no significant change in gradient. 

Statistical analysis title 
Mixed model Lymphocyte count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.304 ^{[177]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [177]  Non significant. 

Statistical analysis title 
Mixed model Monocyte count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.789 ^{[178]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [178]  Non significant. 

Statistical analysis title 
Mixed model Neutrophil count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.244 ^{[179]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [179]  Non significant. 

Statistical analysis title 
Mixed model Platelet count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.013 ^{[180]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [180]  There was a significant increase. However, there was no significant change in gradient. 

Statistical analysis title 
Mixed model Potassium serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.143 ^{[181]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [181]  Non significant. 

Statistical analysis title 
Mixed model Sodium serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.732 ^{[182]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [182]  Non significant. 

Statistical analysis title 
Mixed model Total bilirubin serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.57 ^{[183]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [183]  Non significant. 

Statistical analysis title 
Mixed model Triglicerides serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.133 ^{[184]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [184]  Non significant. 

Statistical analysis title 
Mixed model Urea serum  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.107 ^{[185]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [185]  Non significant. 

Statistical analysis title 
Mixed model White blood count  
Statistical analysis description 
Mixed model: this compares the change in means in a potentially more powerful analysis using a linear mixed model which uses all of a patient’s values, from visit 1 to visit 8.


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.154 ^{[186]}  
Method 
Mixed models analysis  
Confidence interval 

Notes [186]  Non significant. 

Statistical analysis title 
Change in gradient Adjusted calcium serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[187]}  
Pvalue 
= 0.406 ^{[188]}  
Method 
Change in gradient  
Confidence interval 

Notes [187]  Gradients reported are estimated rates of change in test units per month. [188]  No significant chnge gradient SE z Pvalue 95% Conf. Int before .0000267 .005385 0.00 0.996 .0105811 .0105277 after .0077586 .0053828 1.44 0.149 .0027916 .0183087 

Statistical analysis title 
Change in gradient ALT  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[189]}  
Pvalue 
= 0.179 ^{[190]}  
Method 
Change in gradient  
Confidence interval 

Notes [189]  Gradients reported are estimated rates of change in test units per month [190]  No signif change gradient SE z Pvalue 95% Conf. Int before .0000267 .005385 0.00 0.996 .0105811 .0105277 after .0077586 .0053828 1.44 0.149 .0027916 0183087 

Statistical analysis title 
Change in gradient AST  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.385 ^{[191]}  
Method 
Change in gradient  
Confidence interval 

Notes [191]  Non significant. 

Statistical analysis title 
Change in gradient Basophil count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.779 ^{[192]}  
Method 
Change in gradient  
Confidence interval 

Notes [192]  Non significant. 

Statistical analysis title 
Change in gradient Chloride serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.194 ^{[193]}  
Method 
Change in gradient  
Confidence interval 

Notes [193]  No signif change gradient SE z Pvalue 95% Conf. Int before .1363255 .2036459 0.67 0.503 .2628131 .5354641 after .3119152 .1915487 1.63 0.103 .6873437 .0635134 

Statistical analysis title 
Change in gradient Chlolesterol HDL ratio serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.928 ^{[194]}  
Method 
Change in gradient  
Confidence interval 

Notes [194]  Non significant. 

Statistical analysis title 
Change in gradient Chlolesterol serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[195]}  
Pvalue 
= 0.437 ^{[196]}  
Method 
Change in gradient  
Confidence interval 

Notes [195]  Gradients reported are estimated rates of change in test units per month. [196]  No signif change gradient SE z Pvalue 95% Conf. Int before .0826927 .043773 1.89 0.059 .0031008 .1684862 after .0258226 .040227 0.64 0.521 .0530208 .1046661 

Statistical analysis title 
Change in gradient Creatinine serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[197]}  
Pvalue 
= 0.769 ^{[198]}  
Method 
Change in gradient  
Confidence interval 

Notes [197]  Gradients reported are estimated rates of change in test units per month. [198]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Eosinophil count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.768 ^{[199]}  
Method 
Change in gradient  
Confidence interval 

Notes [199]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Estimated GFR  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.874 ^{[200]}  
Method 
Change in gradient  
Confidence interval 

Notes [200]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Estimated GGT  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.345 ^{[201]}  
Method 
Change in gradient  
Confidence interval 

Notes [201]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Estimated Glucose plasma  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.072 ^{[202]}  
Method 
Change in gradient  
Confidence interval 

Notes [202]  There was a borderline significant gradient change P=0.072, from a nonsignificant increase before to a significant decline after. 

Statistical analysis title 
Change in gradient Haemoglobin  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.558 ^{[203]}  
Method 
Change in gradient  
Confidence interval 

Notes [203]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient HDL Cholesterol serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.843 ^{[204]}  
Method 
Change in gradient  
Confidence interval 

Notes [204]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient LDL Cholesterol serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.432 ^{[205]}  
Method 
Change in gradient  
Confidence interval 

Notes [205]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Lymphocyte count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.775 ^{[206]}  
Method 
Change in gradient  
Confidence interval 

Notes [206]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Monocyte count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.089 ^{[207]}  
Method 
Change in gradient  
Confidence interval 

Notes [207]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Neutrophil count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.899 ^{[208]}  
Method 
Change in gradient  
Confidence interval 

Notes [208]  There was no significant change in gradient. 

Statistical analysis title 
Change in gradient Platelet count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 
^{[209]}  
Pvalue 
= 0.352 ^{[210]}  
Method 
Change in gradient  
Confidence interval 

Notes [209]  Gradients reported are estimated rates of change in test units per month. [210]  No signif change gradient SE z Pvalue 95% Conf. Int before .274596 1.820794 0.15 0.880 3.843288 3.294096 after 2.592704 1.709353 1.52 0.129 .7575659 5.942974 

Statistical analysis title 
Change in gradient Potassium serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.221 ^{[211]}  
Method 
Change in gradient  
Confidence interval 

Notes [211]  There was no significant change in gradient 

Statistical analysis title 
Change in gradient Sodium serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.445 ^{[212]}  
Method 
Change in gradient  
Confidence interval 

Notes [212]  There was no significant change in gradient 

Statistical analysis title 
Change in gradient total bilirubin serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.123 ^{[213]}  
Method 
Change in gradient  
Confidence interval 

Notes [213]  There was no significant change in gradient 

Statistical analysis title 
Change in gradient Triglicerides serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.767 ^{[214]}  
Method 
Change in gradient  
Confidence interval 

Notes [214]  There was no significant change in gradient 

Statistical analysis title 
Change in gradient Urea serum  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.197 ^{[215]}  
Method 
Change in gradient  
Confidence interval 

Notes [215]  There was no significant change in gradient 

Statistical analysis title 
Change in gradient White blood count  
Statistical analysis description 
Change in gradient: this uses the linear mixed model, again using all of a patient’s measurements over the seven visits, to determine if the gradient of change after is significantly different from the gradient of change before. Note that the pvalue refers to the change in gradients, not to either of the two gradients, which may or may not be significant (ie significantly different from zero).


Comparison groups 
Treatment v ITT


Number of subjects included in analysis 
40


Analysis specification 
Prespecified


Analysis type 

Pvalue 
= 0.674 ^{[216]}  
Method 
Change in gradient  
Confidence interval 

Notes [216]  There was no significant change in gradient 


End point title 
Adverse events  
End point description 

End point type 
Secondary


End point timeframe 
After  before (treatment) changes




No statistical analyses for this end point 


Adverse events information


Timeframe for reporting adverse events 
Reporting of adverse events from screening


Adverse event reporting additional description 
Adverse events were recorded for all subjects screened (n=31) and include events for patients who did not receive IMP.


Assessment type 
Systematic  
Dictionary used for adverse event reporting


Dictionary name 
MedDRA  
Dictionary version 
19.0


Reporting groups


Reporting group title 
Nonserious adverse events


Reporting group description 
Adverse events were recorded from screening (n=31) and include subjects who did not receive IMP.  


Frequency threshold for reporting nonserious adverse events: 1%  



Substantial protocol amendments (globally) 

Were there any global substantial amendments to the protocol? Yes  
Date 
Amendment 

23 May 2013 
Approved documents:
• Protocol Version 5.0
• PIS/ICF Version 6.0
• Questionnaires Version 2.0
• Website Advertising Text Version 1.0
• Additional IMP Label
1. Updates to Patient Questionnaires
The following questionnaires remain in the study and patients will be asked to complete these at visits 28.
1. Health Status Questionnaire (SF36)
2. Multiple Sclerosis Impact Scale (MSIS29)
3. Multiple Sclerosis Walking Scale (MSWS12v2)
The following assessments replace patient assessments of pain and fatigue.
1. Patient Fatigue Assessment – Visual Analogue Scale
2. Patient Pain Assessment – Visual Analogue Scale
2. Reduction of EDSS Frequency
Frequency of EDSS assessments was reduced (every 2 months)as it was deemed unnecessary by the Chief Investigator to have this number of EDSS in the study.
3. Use of an Additional Pharmacy Label
Additional IMP label to be attached to the study IMP.
4. Pregnancy Tests before MRI
Clarification on pregnancy tests to be performed prior to MRI scans (standard of care but this information was not clearly outlined in the Protocol and Patient Information Sheet).
5. Advertising Materials
In order to advertise the study on the websites of patient support groups such as the MS Society and Shift MS to aid recruitment.


11 Nov 2014 
Approved document: Protocol Version 6.0
Summary of changes made to the protocol:
1. End of Trial Definition
The end of study definition was revised from ‘Last Patient Last Visit’ to ‘Last Patient Last Visit plus six months’.
2. Criteria for Premature Withdrawal
Both protocol sections 3.4 and sections 5.8 of the protocol indicated the reasons for
premature withdrawal from this study. However, section 3.4 did not include all reasons as given in protocol section 5.8.
Section 3.4 was updated with the reason, which was previously present in protocol section 5.8 but missing in protocol section 3.4 : Severe or disabling MS relapse needing IVMP and admission to hospital during the 3 months on treatment phase of the study.


13 Apr 2015 
Clarifications were made to the following sections of the protocol post last patient last visit:
1. Inclusion/Exclusion criteria section 3.3
2. Criteria for Premature Withdrawal section 3.4
3. Prior and concomitant therapies section 4.8
The amendment was approved by the MHRA but rejected by the ethics committee. Therefore it was felt appropriate to withdraw the amendment. MHRA were notified of this on 08/06/15.


Interruptions (globally) 

Were there any global interruptions to the trial? No  
Limitations and caveats 

Limitations of the trial such as small numbers of subjects analysed or technical problems leading to unreliable data.  
None 