1. Several changes were made to the inclusion criterion concerning subjects who were (1) resistant to or failed prior taxane and an anthracycline OR (2) resistant to or failed prior taxane and for whom further anthracycline therapy was not indicated 2. Clarification of Consent to Long-Term Follow-up as information on overall survival (a secondary study objective) must be collected for all subjects 3. Subjects in the Long-term Follow-up Period should have regular computed tomography (CT) scans/magnetic resonance images (MRIs) until disease progression is documented 4. The IVRS used in this study has a built-in 3-day time period for the screening phase 5. Participation in the pharmacokinetics part of the study is not mandatory 6. To allow collection of pharmacokinetic samples at the scheduled Day 15 visit of appropriate cycles rather than Day 14 to enhance subject convenience by eliminating the need for this additional Day-14 visit 7. Anti-cancer drugs or device therapy for the treatment of breast cancer was prohibited within 4 weeks (28 days) or 5 half-lives, whichever was longer, prior to randomization 8. Allowed for the use of hematopoietic growth factor (only cycles subsequent to Cycle 1) in study subjects at the discretion of the investigator 9. To emphasize that a subject may decline to participate in the optional genetic biomarker study and yet still participate in the study trial.

1. Clarifications on inclusion criteria regarding radiological evaluations, dose range of doxorubicin equivalents, eligibility following taxane and anthracycline therapy, discontinuation of prior chemotherapy, and pregnancy test 2. Clarifications on exclusion criteria regarding testing of subjects with unknown HER2 status, breast cancer types, subjects with brain metastases, baseline infections, prior radiation, excluded therapies, hemorrhage/bleeding events 3. To allow dose re-escalation of sorafenib/placebo only under the specific circumstances that are clearly defined in protocol 4. Included detailed guidelines on dose modifications, dose interruptions, and criteria for reintroduction of study medications in response to hematologic toxicities 5. To allow subjects for collection of pharmacokinetic samples, to attend their weekly clinic visit scheduled for Cycle 2/Day 15 to be conducted on Cycle 2/Day 14 6. To state that, if 10 slides are not available, fewer slides may be accepted for formalin-fixed paraffin embedded biopsy after discussion and agreement with the sponsor 7. In the interest of increasing the subject safety the protocol was modified to specify that subjects should undergo electrocardiograms (ECGs) on Day 1 of each cycle in addition to previously specified baseline, end-of-treatment, and as-required ECGs 8. The text on excluded concurrent therapies was clarified for consistency with the remainder of the protocol 9. The inclusion criterion regarding cancer-evaluation imaging at baseline was amended to allow a bone scan to be conducted up to 8 weeks before randomization to be used as the baseline scan 10. Screening study assessments should be performed within specified time periods relative to randomization 11. The instructions regarding HER2 status determination were changed in order to reflect the current practices regarding laboratory accreditation and usage of testing kits.

Amendment 6 modified inclusion criteria defining the specification for prior treatment with taxane and anthracycline therapy in subjects who relapsed over 12 months following the end of treatment and clarifications were made. 1. The inclusion criterion regarding baseline cancer evaluation imaging was amended to allow the baseline bone scan to have been conducted up to 12 weeks before randomization 2. Detailed guidelines on baseline criteria for laboratory evaluations were provided. 3. Re-screening was permitted in cases in which the subject’s eligibility for the study depended on the completion of further treatment, or to allow protocol compliant time lines for assessment or washout periods 4. Detailed guidelines on dose modifications, dose interruptions, and criteria for re-introduction of study medications in response to hematologic toxicities based on the CTCAE v4.0 grade of events

1. The original assumption of a 1-sided alpha of 0.025, power of 98% and a randomization ratio of 1:1 between treatments for 363 PFS events (for progression events based on central radiology review) to detect a 66.7% increase in PFS was amended to an assumption of a 1-sided alpha of 0.025, power of 92.8% and a randomization ratio of 1:1 between treatments for 250 PFS events (including progression events based on central radiology review and death, if death occurs before disease progression) to detect a 66.7% increase in PFS 2. The following statement: “assuming a median OS of 12 months for the control group, approximately 270 deaths would be expected at the time of the PFS analysis. For the final analysis of OS, 405 deaths are projected to occur by approximately 43.3 months after the first subject is randomized” was amended to “assuming a median of 12 months for the control group, approximately 270 deaths would be expected at the time of the PFS analysis. However, at the time approximately 250 PFS events are reached, if the observed number of deaths is less than approximately 270, the trial will continue and remain blinded until approximately 270 deaths are observed. In such a case, the analysis of PFS will include all PFS data up to the later data cutoff date for approximately 270 deaths. For the final analysis of OS, and based on actual events, 405 deaths are projected to occur by approximately 56.2 months after the first subject was randomized.” 3. DCR was added as a secondary end-point.