ORR was defined as the percentage of patients with at least one visit response of confirmed complete response (CR) or partial response (PR). CR was defined as disappearance of all target lesions (TLs) since baseline. Any pathological lymph nodes selected as TLs must have had reduction in short axis to <10 millimeters (mm). PR was defined as at least a 30% decrease in the sum of diameters of TLs, taking as reference the baseline sum of diameters. A confirmed response meant that a response of CR/PR was recorded at 1 visit and confirmed by repeat imaging, preferably at the next regularly scheduled imaging visit and not less than 4 weeks after the visit when response was first observed with no evidence of progression between the initial and CR/PR confirmation visits. Results are reported as percentage of patients with a confirmed response and percentage of patients with confirmed or unconfirmed responses (i.e., including single visit responses).

The FAS included all randomised patients.

PFS was defined as the time from the date of randomisation until the date of objective disease progression or death (by any cause in the absence of progression) regardless of whether the patient had withdrawn from randomised therapy or had received another anti-cancer therapy prior to progression. Results are reported as median time from randomisation to PFS, calculated using the Kaplan-Meier technique.

The FAS included all randomised patients.

PFS was defined as the time from the date of randomisation until the date of objective disease progression or death (by any cause in the absence of progression) regardless of whether the patient had withdrawn from randomised therapy or had received another anti-cancer therapy prior to progression. PFS rates were calculated using Kaplan-Meier estimates of the cumulative probability of PFS. The PFS rate at 3 months and 6 months was equivalent to the percentage of patients with PFS after 3 months and 6 months, respectively.

The FAS included all randomised patients.

OS was defined as the time from the date of randomisation until death due to any cause. Results are reported as median OS, calculated using the Kaplan-Meier technique.

The FAS included all randomised patients.

OS was defined as the time from the date of randomisation until death due to any cause. OS rates were calculated using Kaplan-Meier estimates of the cumulative probability of survival at each indicated time period. The OS rate at 6 months and 12 months was equivalent to the percentage of patients with OS after 6 months and 12 months, respectively.

The FAS included all randomised patients.

BoR was calculated based on the overall visit responses from each RECIST assessment. It was the best response a patient had following date of first dosing but prior to starting any subsequent cancer therapy and prior to RECIST progression or last evaluable assessment in absence of RECIST progression. Categorisation of BoR was based on RECIST using the following response categories: CR and PR for status of 'Response'; Stable Disease (SD) ≥6 weeks, Progressive Disease (PD) and Not Evaluable (NE) for status of 'Non-response'. SD was defined as neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD. PD was defined as at least a 20% increase in the sum of diameters of TLs taking as reference the smallest sum on study. NE was only relevant if any of the TLs were not assessed or not evaluable or had a lesion intervention at this visit. Results are reported as number of patients with BoR for each of the indicated categories.

The FAS included all randomised patients.

DCR at 3 months was defined as the percentage of patients who have a BoR of CR or PR in the first 3 months or who have demonstrated SD for a minimum interval of 13 weeks following the start of treatment. DCR at 6 months was defined as the percentage of patients who have a BoR of CR or PR in the first 6 months or who have demonstrated SD for a minimum interval of 26 weeks following the start of treatment. DCR at 12 months was defined as the percentage of patients who have a BoR of CR or PR in the first 12 months or who have demonstrated SD for a minimum interval of 52 weeks following the start of treatment. Results are reported as the percentage of patients with disease control for each of the indicated categories.

The FAS included all randomised patients.

To evaluate PK, blood samples were collected pre- and post-dose and durvalumab (MEDI4736) concentrations in serum were determined. On Day 1 of Cycles 1, 4 and 7 (Weeks 0, 12 and 24), PK samples were collected pre-dose (within 60 minutes prior to treatment with any IP) and post-dose at the end of infusion (within 10 minutes of end of infusion of durvalumab and within 10 minutes of end of infusion of tremelimumab [for patients receiving durvalumab plus tremelimumab]). On Day 1 of Cycle 2 (Week 4), PK samples were collected pre-dose (within 60 minutes prior to treatment with any IP) only. The 3-month follow-up sample for durvalumab was relative to the respective last dose. Results are reported as mean pre- or post-dose durvalumab concentrations as indicated by the individual categories (1 cycle=4 weeks). Samples below lower limit of quantification were treated as missing in the analyses.

The PK analysis set included all patients who received at least 1 dose of IP, and had PK sampling data post-dose without important deviations or events to affect PK (n=64). Only patients with data available at the timepoints of testing were included in the analyses.

To evaluate PK, blood samples were collected pre- and post-dose and tremelimumab concentrations in serum were determined. On Day 1 of Cycles 1 and 4 (Weeks 0 and 12), PK samples were collected pre-dose (within 60 minutes prior to treatment with any IP) and post-dose at the end of infusion (within 10 minutes of end of infusion of tremelimumab). On Day 1 of Cycle 2 (Week 4), PK samples were collected pre-dose (within 60 minutes prior to treatment with any IP) only. The 3-month follow-up sample for tremelimumab was relative to the respective last dose. Results are reported as mean pre- or post-dose tremelimumab concentrations as indicated by the individual categories (1 cycle=4 weeks). Samples below lower limit of quantification were treated as missing in the analyses.

The PK analysis set included all patients who received at least 1 dose of IP, and had PK sampling data post-dose without important deviations or events to affect PK (n=64). Only patients with data available at the timepoints of testing were included in the analyses.

ADA prevalence was the proportion of the ADA evaluable set who had an ADA positive result at any point in time, baseline or post-baseline. ADA incidence (treatment-emergent ADA) was the sum of both treatment-induced (post-baseline ADA positive only) and treatment-boosted ADA positive patients as a proportion of the evaluable patient population. Treatment-boosted ADA was defined as baseline positive ADA titre boosted to a 4-fold or higher level following IP administration. Persistently positive was defined as positive at ≥2 post-baseline assessments (with ≥16 weeks between first and last positive) or positive at last post-baseline assessment. Transiently positive was defined as at least 1 post-baseline ADA positive assessment and not fulfilling the conditions of persistently positive. Results are reported as number of patients with detectable anti-durvalumab antibodies satisfying each of the indicated categories. Note: 'positive' is denoted by 'pos' in some category titles.

The ADA evaluable set included all patients in the safety analysis set (i.e. those who had received any IP) who had non-missing baseline ADA data and at least 1 non-missing post-baseline ADA result.

ADA prevalence was the proportion of the ADA evaluable set who had an ADA positive result at any point in time, baseline or post-baseline. ADA incidence (treatment-emergent ADA) was the sum of both treatment-induced (post-baseline ADA positive only) and treatment-boosted ADA positive patients as a proportion of the evaluable patient population. Treatment-boosted ADA was defined as baseline positive ADA titre boosted to a 4-fold or higher level following IP administration. Persistently positive was defined as positive at ≥2 post-baseline assessments (with ≥16 weeks between first and last positive) or positive at last post-baseline assessment. Transiently positive was defined as at least 1 post-baseline ADA positive assessment and not fulfilling the conditions of persistently positive. Results are reported as number of patients with detectable anti- tremelimumab antibodies satisfying each of the indicated categories. Note: 'positive' is denoted by 'pos' in some category titles.

The ADA evaluable set included all patients in the safety analysis set (i.e. those who had received any IP) who had non-missing baseline ADA data and at least 1 non-missing post-baseline ADA result.