OS is defined for each participant as the time between randomization date and death due to any cause. The survival time for participants who had not died was censored at the last known alive date. Median and associated 2-sided 95% confidence intervals were calculated using the method of Brookmeyer and Crowley.

All randomized participants

PFS was defined as the time between randomization date and the date of progression or death, whichever occurred first. A participant who died without reported prior progression was considered to have progressed on the date of death. For a participant who underwent resection post randomization, PFS was censored on last tumor assessment date prior to resection. For those who remained alive and had not progressed, PFS was censored on last evaluable tumor assessment date. Participants who had not died and had no recorded post-baseline tumor assessment were censored at the day of randomization. For participants who had Progressive Disease (PD) prior to Week 12 and a subsequent assessment of Stable Disease (SD), Partial Response (PR), or Complete Response (CR), the date of PD following response was used in the analysis of PFS; otherwise these participants were censored on the date of their last tumor assessment. Median and 2-sided 95% CIs were calculated with Brookmeyer Crowley method.

All randomized participants.

BORR by treatment arm was defined as the total number of randomized participants in the arm whose BOR is CR or PR, divided by the total number of randomized participants in the arm. Any participant who was unevaluable for BOR, e.g. on account of missing or "not evaluable" assessments, was included in the denominator of the calculation (i.e. was considered a non-responder with respect to the BORR endpoint). 95% 2-sided exact confidence intervals were computed using the method of Clopper and Pearson.

All randomized participants

DCR by treatment arm was defined as the total number of randomized participants in the arm whose BOR is CR, PR or SD, divided by the total number of randomized participants in the arm. Any participant who was unevaluable for Disease Control (DC), (e.g. on account of missing or "not evaluable" assessments), was included in the denominator of the calculation (i.e. was considered a non-responder with respect to the DCR endpoint). 95% 2-sided exact confidence intervals were computed using the Clopper and Pearson method.

All randomized participants

Duration of response for participants whose BOR was CR or PR was defined as the time between the date measurement criteria were first met for overall response of PR or CR (whichever status was recorded first) and the date of disease progression or death (whichever occurred first). For participants who underwent tumor resection following response but prior to disease progression, duration of response was censored on the date of last evaluable tumor assessment prior to resection. For participants who had BOR of SD, PR or CR at Week 12, or a confirmed response of PR or CR before Week 12, the date of PD following thereafter (where available) was used in the analysis of duration of response. For those participants who remained alive and had not progressed following response, duration of response was censored on the date of last evaluable tumor assessment. Median and associated 2-sided 95% confidence intervals were calculated using the Brookmeyer Crowley method.

All randomized participants

Duration of stable disease was defined for participants whose BOR was SD as the time between when SD was first documented and the date of PD or death (whichever occurred first). For a participant who underwent tumor resection following Week 12 but prior to disease progression, duration of stable disease was censored on the date of the last evaluable tumor assessment prior to resection. For participants who had BOR of SD at Week 12, the date of PD following thereafter (where available) was used in the analysis of duration of stable disease. For participants with BOR of SD who had not subsequently progressed and who remained alive, duration of stable disease was censored on the date of last evaluable tumor assessment. Median and associated 2-sided 95% confidence intervals were calculated using the Brookmeyer and Crowley method.

All randomized participants

OS is defined for each participant as the time between randomization date and death due to any cause. The survival time for participants who had not died was censored at the last known alive date. Survival rates were calculated based on Kaplan-Meier estimation with log-log transformed confidence intervals. The survival rate at x year(s) is defined as the probability that a subject is alive at x year(s) following randomization.

All randomized participants

OS for each participant with brain metastases at baseline was measured as the time between randomization date and death due to any cause. The survival time for participants who had not died was censored at the last known alive date. Median OS, and associated 2-sided 95% confidence intervals were calculated using the Brookmeyer and Crowley method.

All randomized participants with brain metastases at baseline