Amendment 1: Inclusion criteria: amendment to 2 barrier methods to satisfy the requirement of adequate measures to avoid pregnancy.

Amendment 2: The allowance of prior docetaxel treatment in the first line setting was removed. The fourth stratification factor was amended to “prior taxane therapy (yes vs. no)”.

Amendment 3: EORTC QLQ-C30 evaluation after protocol treatment discontinuation amended to avoid patients who went off study therapy to have additional visits to obtain quality of life questionnaires. Language pertaining to tumor evaluation scanning options added to standardize the method of tumor measurement among the patients. Language pertaining to SAE reporting and paclitaxel overdose was added to clarify the details of data collection for medical monitoring and pharmacovigilance purposes. Clarification of the acceptable number of unstained slides in combination with cores of tumor tissue, if tumor blocks were not available. Language was changed to remove inclusion of patient initials from tissue samples collected during the study. The roles and responsibilities of the DSMB were clarified.

Amendment 4: Inclusion criteria; changed to 6 months after the last dose of paclitaxel and 30 days after the final dose of napabucasin/placebo to satisfy the SmPC recommendations for paclitaxel use. Accordingly, adequate contraception was redefined to include 6 months and 30 days of abstinence following the final dose of paclitaxel and napabucasin/placebo, respectively. Addition of language to report pregnancies occurring up to 6 months following the last dose of paclitaxel and to clarify a definition of true abstinence. Clarification of language to allow the Investigator to have unilateral right to unblind the patient in circumstances of medical emergency requiring unblinding. Addition of the recommendation to hold the bowel regimen for a specified period of time unless absence of bowel movement was noted in the first 2 days of protocol; this was to avoid the development of diarrhoea while on study drug. For the management of napabucasin-related AEs, dose holiday was extended to 3 days, based on the phase II trial of napabucasin in combination with paclitaxel.

Amendment 5: The duration of treatment was amended to avoid imbalance in paclitaxel treatment between patients treated with active agent napabucasin and patients treated with placebo. The period of required contraception was expanded to 90 days after the final dose of napabucasin/placebo for male patients in response to the French regulatory authority. It was clarified that if pregnancies occurred before 90 days after the final dose for male patients, the event was to be reported to the Sponsor. The following changes/clarifications were made, based on the SmPC: severe hepatic impairment and history of severe hypersensitivity to paclitaxel or to any of the excipients, including macrogolglycerol ricinoleate were included in the exclusion criteria; clarification that paclitaxel is a tetratogen, embryotoxic and a mutagen with mandatory contraception to 6 months following the last dose of paclitaxel; a cautionary note that male patients should seek advice on cryopreservation of sperm before paclitaxel treatment; inclusion of a cautionary note pertaining to the co-administration of paclitaxel and/or napabucasin with specific cytochrome inhibitors and inducers. Clarification that immediate unblinding without the need to prior sponsor notification was permissible in instances where emergent unblinding was necessary for patient safety, as per the UK regulatory agency. Serial blood sampling reduced based on the recommendations of the preclinical team.

Amendment 6: The total sample size was increased from 680 to 700 to accommodate the 5% expected patient yearly dropout rate. The background information about Ramucirumab’s approval by the Food and Drug Administration was included. It was clarified in the inclusion criteria that all 3 Siewert types of GEJ adenocarcinoma would be eligible. Treatment failure on first line therapy was expanded to include patients progressing within ≤ 6 months of last dose of therapy. Also, the criterion was further clarified. The number of slides of archival tumor sample was amended. Uncontrolled intercurrent illness definition was clarified. Prior cancer history exclusion criterion was clarified to define species of other prior malignant conditions. Instructions for protocol-defined administration of paclitaxel were specified. Management of severe hematologic AEs was amended to include the Medical Monitor. Patient compliance was clarified as measurement against drug dose prescribed by the investigator. In Appendix I, data collection following protocol treatment discontinuation was modified. The timing of blood collection was amended.

Amendment 7: The statistical calculations for the interim analysis were changed to correct an error made in the previous calculation. The biomarker-positive definition was revised to specify patients with p-STAT3 positivity on IHC staining of archival tissue. Safety profile text was moved to the end of list to main consistency with secondary and exploratory objectives in rest of protocol. Information in pre-medication recommendation was updated in line with current Investigator Brochure. For sensitivity analysis of the primary endpoint, factors used in the Cox proportional hazards model were modified to prevent treatment effect bias. In the correlative study section, analysis of duration of response was removed and the complete set of stratification factors was included for consistency with data collected. The procedure for unblinding was clarified. Regarding Appendix VI, changes were made as a clarification to define specifics of other prior malignant conditions; modified was done to state that unblinding would not be allowed for informational purposes or to permit participation in other clinical trials. Additionally, the necessity of specifying a reason for unblinding in the process of the unblinding request was stipulated. This change was made to clarify the procedures for unblinding study participants.

Amendment 8: The study was amended following the results of the planned interim analysis at 380 events to allow unblinding of the study and continued treatment as the discretion of the investigator. Additionally, the study is planned to terminate by 15 September 2017.