Study of MEDI4736 (Durvalumab) With or Without Tremelimumab Versus Standard of Care Chemotherapy in Urothelial Cancer

• ClinicalTrials.gov Identifier: NCT02516241
• Recruitment Status: Active, not recruiting
• First Posted: August 5, 2015
• Results First Posted: May 13, 2021
• Last Update Posted: May 25, 2023
• Sponsor: AstraZeneca
• Information provided by (Responsible Party): AstraZeneca

• Study Type: Interventional
• Study Design: Allocation: Randomized; Intervention Model: Parallel Assignment; Masking: None (Open Label); Primary Purpose: Treatment
• Condition: Urothelial Cancer
• Interventions: Drug: MEDI4736 (Durvalumab); Drug: Tremelimumab; Drug: Cisplatin; Drug: Carboplatin; Drug: Gemcitabine
• Enrollment: 1126

Participant Flow

Recruitment Details
Pre-assignment Details

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Period Title: Overall Study
Started	342	346	344
Received Treatment	340	345	313
Completed	0	0	0
Not Completed	342	346	344
Reason Not Completed
Death	255	262	270
Lost to Follow-up	1	1	3
Withdrawal by Subject	2	7	12
participants ongoing at data cut-off	84	76	59

Baseline Characteristics

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care	Total
Arm/Group Description		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment	Total of all reporting groups
Overall Number of Baseline Participants		342	346	344	1032
Baseline Analysis Population Description		Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.
Age, Continuous; Mean (Standard Deviation); Unit of measure: Years
Age (years)	Number Analyzed	342 participants	346 participants	344 participants	1032 participants
		66.4 (9.60)	66.3 (9.9)	67.0 (9.32)	66.5 (9.60)
Age, Customized; Measure Type: Count of Participants; Unit of measure: Participants
Age Group (Years)	Number Analyzed	342 participants	346 participants	344 participants	1032 participants
	<50	19 5.6%	22 6.4%	14 4.1%	55 5.3%
	>= 50 - < 65	118 34.5%	115 33.2%	119 34.6%	352 34.1%
	>= 65 - < 75	132 38.6%	141 40.8%	137 39.8%	410 39.7%
	>= 75	73 21.3%	68 19.7%	74 21.5%	215 20.8%
Sex: Female, Male; Measure Type: Count of Participants; Unit of measure: Participants
Sex	Number Analyzed	342 participants	346 participants	344 participants	1032 participants
	Female	86 25.1%	97 28.0%	70 20.3%	253 24.5%
	Male	256 74.9%	249 72.0%	274 79.7%	779 75.5%
Ethnicity (NIH/OMB); Measure Type: Count of Participants; Unit of measure: Participants
Ethnicity	Number Analyzed	342 participants	346 participants	344 participants	1032 participants
	Hispanic or Latino	15 4.4%	14 4.0%	17 4.9%	46 4.5%
	Not Hispanic or Latino	320 93.6%	329 95.1%	322 93.6%	971 94.1%
	Unknown or Not Reported	7 2.0%	3 0.9%	5 1.5%	15 1.5%
Race/Ethnicity, Customized; Measure Type: Count of Participants; Unit of measure: Participants
Race	Number Analyzed	342 participants	346 participants	344 participants	1032 participants
	White	253 74.0%	278 80.3%	260 75.6%	791 76.6%
	Black or African American	3 0.9%	3 0.9%	0 0.0%	6 0.6%
	Asian	72 21.1%	60 17.3%	76 22.1%	208 20.2%
	Native Hawaiian or other Pacific islander	0 0.0%	0 0.0%	0 0.0%	0 0.0%
	American Indian or Alaska Native	0 0.0%	0 0.0%	0 0.0%	0 0.0%
	Other	13 3.8%	4 1.2%	8 2.3%	25 2.4%
	Unknown or Not Reported	1 0.3%	1 0.3%	0 0.0%	2 0.2%

Outcome Measures

1. Primary Outcome

Title	To Assess the Efficacy of Durvalumab + Tremelimumab Combination Therapy Versus SoC in Terms of OS in Full Analysis Set
Description	The OS was defined as the time from the date of randomization until death due to any cause (ie, date of death or censoring - date of randomization + 1). Any participant not known to have died at the time of analysis were censored based on the last recorded date on which the participant was known to be alive. Median OS was calculated using the Kaplan-Meier technique.
Time Frame	From randomization date until death due to any cause, assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	344
Median (95% Confidence Interval); Unit of Measure: Months
	15.1; (13.1 to 18.0)	12.1; (10.9 to 14)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0751
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.85
	Confidence Interval	(2-Sided) 98.66%; 0.688 to 1.063
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

2. Primary Outcome

Title	To Assess the Efficacy of Durvalumab Monotherapy Versus SoC in Terms of OS in PD-L1-High Analysis Set
Description	The OS was defined as the time from the date of randomization until death due to any cause (ie, date of death or censoring - date of randomization + 1). Any participant not known to have died at the time of analysis were censored based on the last recorded date on which the participant was known to be alive. Median OS was calculated using the Kaplan-Meier technique.
Time Frame	From randomization date until death due to any cause, assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	209	207
Median (95% Confidence Interval); Unit of Measure: Months
	14.4; (10.4 to 17.3)	12.1; (10.4 to 15.0)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3039
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.89
	Confidence Interval	(2-Sided) 96.99%; 0.695 to 1.139
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

3. Secondary Outcome

Title	OS, Full Analysis Set - Durvalumab Monotherapy vs SoC
Description	The OS was defined as the time from the date of randomization until death due to any cause (ie, date of death or censoring - date of randomization + 1). Any participant not known to have died at the time of analysis were censored based on the last recorded date on which the participant was known to be alive. Median OS was calculated using the Kaplan-Meier technique.
Time Frame	From randomization date until death due to any cause, assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	346	344
Median (95% Confidence Interval); Unit of Measure: Months
	13.2; (10.3 to 15)	12.1; (10.9 to 14)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.8637
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.99
	Confidence Interval	(2-Sided) 95%; 0.830 to 1.169
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

4. Secondary Outcome

Title	OS, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC
Description	The OS was defined as the time from the date of randomization until death due to any cause (ie, date of death or censoring - date of randomization + 1). Any participant not known to have died at the time of analysis were censored based on the last recorded date on which the participant was known to be alive. Median OS was calculated using the Kaplan-Meier technique.
Time Frame	From randomization date until death due to any cause, assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	207
Median (95% Confidence Interval); Unit of Measure: Months
	17.9; (14.8 to 24.2)	12.1; (10.4 to 15.0)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0091
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.74
	Confidence Interval	(2-Sided) 95%; 0.589 to 0.928
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

5. Secondary Outcome

Title	OS, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	The OS was defined as the time from the date of randomization until death due to any cause (ie, date of death or censoring - date of randomization + 1). Any participant not known to have died at the time of analysis were censored based on the last recorded date on which the participant was known to be alive. Median OS was calculated using the Kaplan-Meier technique.
Time Frame	From randomization date until death due to any cause, assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Median (95% Confidence Interval); Unit of Measure: Months
	11.8; (8.9 to 15.8)	10.9; (8.0 to 14.8)	12.2; (10.4 to 14)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.7599
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.04
	Confidence Interval	(2-Sided) 95%; 0.799 to 1.359
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.4424
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.90
	Confidence Interval	(2-Sided) 95%; 0.692 to 1.175
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

6. Secondary Outcome

Title	Alive at 24 Months (OS24), Full Analysis Set
Description	Alive at 24 months (OS24) is defined as the Kaplan-Meier estimate of OS at 24 months.
Time Frame	From randomization date until death due to any cause, assessed up to 24 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	39.0; (33.8 to 44.2)	31.5; (26.6 to 36.4)	29.0; (24.2 to 34.0)

7. Secondary Outcome

Title	Alive at 24 Months (OS24), PD-L1-High Analysis Set
Description	Alive at 24 months (OS24) is defined as the Kaplan-Meier estimate of OS at 24 months.
Time Frame	From randomization date until death due to any cause, assessed up to 24 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	43.7; (36.8 to 50.3)	36.0; (29.5 to 42.6)	29.3; (23.1 to 35.7)

8. Secondary Outcome

Title	Alive at 24 Months (OS24), PD-L1-Low/Negative Analysis Set
Description	Alive at 24 months (OS24) is defined as the Kaplan-Meier estimate of OS at 24 months.
Time Frame	From randomization date until death due to any cause, assessed up to 24 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	32.1; (24.5 to 40.0)	24.5; (17.6 to 32.0)	28.6; (21.2 to 36.4)

9. Secondary Outcome

Title	PFS, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Progression free survival (PFS) (per RECIST 1.1, as assessed by investigator) was defined as the time from the date of randomization until the date of objective disease progression or death (by any cause in the absence of progression) regardless of whether the patient withdraws from randomized therapy or receives another anticancer therapy prior to progression. Median PFS was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Median (95% Confidence Interval); Unit of Measure: Months
	3.7; (3.4 to 3.8)	2.3; (1.9 to 3.5)	6.7; (5.7 to 7.3)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2579
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.10
	Confidence Interval	(2-Sided) 95%; 0.931 to 1.304
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0008
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.33
	Confidence Interval	(2-Sided) 95%; 1.124 to 1.567
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

10. Secondary Outcome

Title	PFS, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Monotherapy vs SoC
Description	Progression free survival (PFS) (per RECIST 1.1, as assessed by investigator) was defined as the time from the date of randomization until the date of objective disease progression or death (by any cause in the absence of progression) regardless of whether the patient withdraws from randomized therapy or receives another anticancer therapy prior to progression. Median PFS was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Median (95% Confidence Interval); Unit of Measure: Months
	4.1; (3.6 to 5.7)	2.4; (1.9 to 3.7)	5.8; (5.6 to 7.2)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2395
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.88
	Confidence Interval	(2-Sided) 95%; 0.705 to 1.091
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2431
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.14
	Confidence Interval	(2-Sided) 95%; 0.917 to 1.406
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

11. Secondary Outcome

Title	PFS, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	Progression free survival (PFS) (per RECIST 1.1, as assessed by investigator) was defined as the time from the date of randomization until the date of objective disease progression or death (by any cause in the absence of progression) regardless of whether the patient withdraws from randomized therapy or receives another anticancer therapy prior to progression. Median PFS was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Median (95% Confidence Interval); Unit of Measure: Months
	2.0; (1.9 to 3.6)	2.0; (1.9 to 3.5)	7.2; (5.7 to 7.5)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0014
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.53
	Confidence Interval	(2-Sided) 95%; 1.177 to 1.990
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.6236
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.94
	Confidence Interval	(2-Sided) 95%; 0.729 to 1.209
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

12. Secondary Outcome

Title	Alive and Progression-free at 12 Months (APF12), Full Analysis Set
Description	Alive and progression-free at 12 months (APF12) was defined as the Kaplan-Meier estimate of PFS (per RECIST 1.1 as assessed by investigator) at 12 months.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	21.4; (17.2 to 26.0)	16.8; (13.1 to 21.1)	15.3; (11.4 to 19.7)

13. Secondary Outcome

Title	Alive and Progression-free at 12 Months (APF12), PD-L1-High Analysis Set
Description	Alive and progression-free at 12 months (APF12) was defined as the Kaplan-Meier estimate of PFS (per RECIST 1.1 as assessed by investigator) at 12 months.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	25.6; (19.7 to 31.8)	21.2; (15.9 to 27.0)	15.0; (10.2 to 20.7)

14. Secondary Outcome

Title	Alive and Progression-free at 12 Months (APF12), PD-L1-Low/Negative Analysis Set
Description	Alive and progression-free at 12 months (APF12) was defined as the Kaplan-Meier estimate of PFS (per RECIST 1.1 as assessed by investigator) at 12 months.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Measure Type: Number; Number (95% Confidence Interval); Unit of Measure: percentage of participants
	15.2; (9.6 to 21.8)	9.7; (5.4 to 15.7)	15.6; (9.6 to 22.9)

15. Secondary Outcome

Title	PFS2, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Time from randomization to second progression or death (PFS2) was defined as the time from the date of randomization to the earliest of the progression event subsequent to first subsequent therapy or death (ie, date of PFS2 event or censoring - date of randomization +1). Median PFS2 was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until first confirmed disease progression, disease then assessed per local practice until 2nd progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Median (95% Confidence Interval); Unit of Measure: Months
	14.6; (11.3 to 17.8)	11.9; (8.9 to 14.6)	11.5; (10.3 to 12.8)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0477
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.83
	Confidence Interval	(2-Sided) 95%; 0.683 to 0.998
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.7885
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.97
	Confidence Interval	(2-Sided) 95%; 0.809 to 1.175
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

16. Secondary Outcome

Title	PFS2, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Monotherapy vs SoC
Description	Time from randomization to second progression or death (PFS2) was defined as the time from the date of randomization to the earliest of the progression event subsequent to first subsequent therapy or death (ie, date of PFS2 event or censoring - date of randomization +1). Median PFS2 was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until first confirmed disease progression, disease then assessed per local practice until 2nd progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Median (95% Confidence Interval); Unit of Measure: Months
	17.2; (12.1 to 24.8)	13.4; (9.7 to 17.5)	11.3; (8.9 to 14.3)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0053
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.70
	Confidence Interval	(2-Sided) 95%; 0.549 to 0.901
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1336
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.83
	Confidence Interval	(2-Sided) 95%; 0.651 to 1.059
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

17. Secondary Outcome

Title	PFS2, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	Time from randomization to second progression or death (PFS2) was defined as the time from the date of randomization to the earliest of the progression event subsequent to first subsequent therapy or death (ie, date of PFS2 event or censoring - date of randomization +1). Median PFS2 was calculated using the Kaplan-Meier technique.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until first confirmed disease progression, disease then assessed per local practice until 2nd progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Median (95% Confidence Interval); Unit of Measure: Months
	10.7; (7.5 to 16.6)	9.4; (7.0 to 13.7)	11.6; (10.3 to 13.4)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.8148
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	1.04
	Confidence Interval	(2-Sided) 95%; 0.772 to 1.391
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2684
	Comments	The 2-sided p-value was calculated using a stratified log-rank test adjusting for visceral metastases and cisplatin eligibility status.
	Method	Log Rank
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Hazard Ratio (HR)
	Estimated Value	0.85
	Confidence Interval	(2-Sided) 95%; 0.636 to 1.134
	Estimation Comments	The HR and confidence interval (CI) were calculated using a stratified Cox proportional hazards model, adjusting for visceral metastases and cisplatin eligibility status with ties handled by the Breslow approach.

18. Secondary Outcome

Title	Objective Response Rate (ORR), Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Objective response rate ORR (per RECIST 1.1 as assessed by investigator) is defined as the number (%) of patients with at least 1 visit response of CR or PR.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Measure Type: Number; Unit of Measure: percentage of participants
	36.3	25.7	49.1

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0005
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.58
	Confidence Interval	(2-Sided) 95%; 0.422 to 0.788
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.35
	Confidence Interval	(2-Sided) 95%; 0.253 to 0.485
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

19. Secondary Outcome

Title	Objective Response Rate (ORR), PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Monotherapy vs SoC
Description	Objective response rate ORR (per RECIST 1.1 as assessed by investigator) is defined as the number (%) of patients with at least 1 visit response of CR or PR.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Measure Type: Number; Unit of Measure: percentage of participants
	46.8	27.8	48.3

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.7708
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.94
	Confidence Interval	(2-Sided) 95%; 0.639 to 1.394
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.41
	Confidence Interval	(2-Sided) 95%; 0.268 to 0.610
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

20. Secondary Outcome

Title	Objective Response Rate (ORR), PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	Objective response rate ORR (per RECIST 1.1 as assessed by investigator) is defined as the number (%) of patients with at least 1 visit response of CR or PR.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Measure Type: Number; Unit of Measure: percentage of participants
	20.4	22.6	50.4

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.24
	Confidence Interval	(2-Sided) 95%; 0.135 to 0.406
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.6195
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.86
	Confidence Interval	(2-Sided) 95%; 0.469 to 1.566
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

21. Secondary Outcome

Title	Objective Response Rate (ORR) Based on BICR Assessment According to RECIST 1.1 - Responses Are Confirmed - Durvalumab Cisplatin Ineligible Population
Description	Objective response rate ORR (per RECIST 1.1 as assessed by investigator) is defined as the number (%) of patients with at least 1 visit response of CR or PR. unconfirmed responses are excluded.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (01NOV2017, a maximum of 3 years).

Outcome Measure Data

Analysis Population Description

Durvalumab Cisplatin Ineligible Population include all patients who had received D monotherapy and were not eligible for cisplatin treatment at baseline (per eCRF).

Arm/Group Title	Monotherapy - PD-L1 High	Monotherapy - PD-L1 Low/Negative	Monotherapy - Total
Arm/Group Description:	MEDI4736 (Durvalumab) - PD-L1 high	MEDI4736 (Durvalumab) - PD-L1 low/negative	MEDI4736 (Durvalumab)
Overall Number of Participants Analyzed	87	56	143
Measure Type: Number; Unit of Measure: percentage of participants
	23.0	17.9	21.0

22. Secondary Outcome

Title	Disease Control Rate (DCR) at 6 Months, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Disease control rate (DCR) at 6 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 24 weeks (-7 days, i.e., 161 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 6 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Measure Type: Number; Unit of Measure: percentage of participants
	41.5	31.8	55.5

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0002
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.56
	Confidence Interval	(2-Sided) 95%; 0.410 to 0.759
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.37
	Confidence Interval	(2-Sided) 95%; 0.267 to 0.500
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

23. Secondary Outcome

Title	Disease Control Rate (DCR) at 6 Months, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Monotherapy vs SoC
Description	Disease control rate (DCR) at 6 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 24 weeks (-7 days, i.e., 161 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 6 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Measure Type: Number; Unit of Measure: percentage of participants
	49.8	34.4	53.1

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.4959
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.87
	Confidence Interval	(2-Sided) 95%; 0.590 to 1.291
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.46
	Confidence Interval	(2-Sided) 95%; 0.305 to 0.679
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

24. Secondary Outcome

Title	Disease Control Rate (DCR) at 6 Months, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	Disease control rate (DCR) at 6 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 24 weeks (-7 days, i.e., 161 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 6 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Measure Type: Number; Unit of Measure: percentage of participants
	29.2	27.7	59.1

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.27
	Confidence Interval	(2-Sided) 95%; 0.160 to 0.448
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.8074
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.07
	Confidence Interval	(2-Sided) 95%; 0.623 to 1.836
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

25. Secondary Outcome

Title	Disease Control Rate (DCR) at 12 Months, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Disease control rate (DCR) at 12 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 48 weeks (-7 days, i.e., 329 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020)

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Measure Type: Number; Unit of Measure: percentage of participants
	38.0	28.0	50.6

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0008
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.59
	Confidence Interval	(2-Sided) 95%; 0.432 to 0.802
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.37
	Confidence Interval	(2-Sided) 95%; 0.270 to 0.512
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.

26. Secondary Outcome

Title	Disease Control Rate (DCR) at 12 Months, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Monotherapy vs SoC
Description	Disease control rate (DCR) at 12 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 48 weeks (-7 days, i.e., 329 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Measure Type: Number; Unit of Measure: percentage of participants
	47.3	30.6	48.8

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.7717
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.94
	Confidence Interval	(2-Sided) 95%; 0.639 to 1.394
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.46
	Confidence Interval	(2-Sided) 95%; 0.303 to 0.682
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

27. Secondary Outcome

Title	Disease Control Rate (DCR) at 12 Months, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy Versus SoC and Durvalumab + Tremelimumab Combination Therapy Versus Durvalumab Monotherapy
Description	Disease control rate (DCR) at 12 months is defined as the percentage of patients who have a best objective response (BoR) of CR or PR, or who have demonstrated SD for a minimum interval of 48 weeks (-7 days, i.e., 329 days), following the start of study treatment.
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to 12 months or the data cut-off date (27JAN2020).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Measure Type: Number; Unit of Measure: percentage of participants
	24.1	24.1	53.3

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	[Not Specified]
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	<0.0001
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.26
	Confidence Interval	(2-Sided) 95%; 0.151 to 0.439
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Monotherapy
	Comments	monotherapy as reference.
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.9692
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	0.99
	Confidence Interval	(2-Sided) 95%; 0.556 to 1.759
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status.

28. Secondary Outcome

Title	Duration of Response (DoR), Full Analysis Set
Description	Duration of response (DoR) (per RECIST 1.1 as assessed by investigator) was defined as the time from the date of first documented response until the first date of documented progression or death in the absence of disease progression (i.e. date of PFS event or censoring - date of first response + 1).
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	342	346	344
Median (Inter-Quartile Range); Unit of Measure: Months
	11.1 [1]; (4.3 to NA)	9.3 [1]; (5.4 to NA)	5.7; (3.7 to 9.3)

[1]

upper limit is not reached

29. Secondary Outcome

Title	Duration of Response (DoR), PD-L1-High Analysis Set
Description	Duration of response (DoR) (per RECIST 1.1 as assessed by investigator) was defined as the time from the date of first documented response until the first date of documented progression or death in the absence of disease progression (i.e. date of PFS event or censoring - date of first response + 1).
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	205	209	207
Median (Inter-Quartile Range); Unit of Measure: Months
	10.0 [1]; (3.8 to NA)	18.5 [1]; (5.6 to NA)	5.8; (3.7 to 11.5)

[1]

upper limit is not reached.

30. Secondary Outcome

Title	Duration of Response (DoR), PD-L1-Low/Negative Analysis Set
Description	Duration of response (DoR) (per RECIST 1.1 as assessed by investigator) was defined as the time from the date of first documented response until the first date of documented progression or death in the absence of disease progression (i.e. date of PFS event or censoring - date of first response + 1).
Time Frame	Tumour scans performed at baseline then every 8 weeks since randomization until confirmed disease progression. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title	Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:	MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed	137	137	137
Median (Inter-Quartile Range); Unit of Measure: Months
	12.9 [1]; (5.5 to NA)	5.6 [1]; (3.8 to NA)	5.7; (3.9 to 8.2)

[1]

upper limit is not reached.

31. Secondary Outcome

Title	Serum Concentrations of Durvalumab, Pharmacokinetic Analysis Set
Description	Blood samples were collected to determine the serum concentration of durvalumab.
Time Frame	Pre-dose and within 1 hour after end of infusion at Week 0, 12 and 24, pre-dose at week 4, and at follow-up Month 3.

Outcome Measure Data

Analysis Population Description

Pharmacokinetic analysis set included all patients who received at least 1 dose of durvalumab or tremelimumab per the protocol and had at least one post dose evaluable PK data of durvalumab or tremelimumab

Arm/Group Title		Combination Therapy	Monotherapy
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)
Overall Number of Participants Analyzed		339	343
Mean (Standard Deviation); Unit of Measure: μg/mL
Week 0 - Predose	Number Analyzed	329 participants	339 participants
		NA [1] (NA)	NA [1] (NA)
Week 0 - Postdose	Number Analyzed	328 participants	334 participants
		511 (338)	484 (162)
Week 4 - Predose	Number Analyzed	285 participants	300 participants
		75.3 (84.4)	78.9 (56.5)
Week 12 - Predose	Number Analyzed	198 participants	195 participants
		125 (97.7)	144 (82.5)
Week 12 - Postdose	Number Analyzed	184 participants	186 participants
		594 (298)	576 (233)
Week 24 - Predose	Number Analyzed	127 participants	119 participants
		150 (94.4)	163 (82.5)
Week 24 - Postdose	Number Analyzed	126 participants	116 participants
		604 (249)	600 (244)
Follow-up Month 3	Number Analyzed	77 participants	66 participants
		23.3 (50)	19.9 (18.4)

[1]

week 0 pre-dose serum concentrations are below the level of detection

32. Secondary Outcome

Title	Serum Concentrations of Tremelimumab, Pharmacokinetic Analysis Set
Description	Blood samples were collected to determine the serum concentration of tremelimumab.
Time Frame	Pre-dose and within 1 hour after end of infusion at Week 0, 12 and 24, pre-dose at week 4, and at follow-up Month 3.

Outcome Measure Data

Analysis Population Description

Pharmacokinetic analysis set included all patients who received at least 1 dose of durvalumab or tremelimumab per the protocol and had at least one post dose evaluable PK data of durvalumab or tremelimumab

Arm/Group Title		Combination Therapy
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab
Overall Number of Participants Analyzed		339
Mean (Standard Deviation); Unit of Measure: μg/mL
Week 0 - Predose	Number Analyzed	332 participants
		NA [1] (NA)
Week 0 - Postdose	Number Analyzed	332 participants
		24.0 (11.5)
Week 4 - Predose	Number Analyzed	283 participants
		3.75 (3.18)
Week 12 - Predose	Number Analyzed	197 participants
		5.43 (4.01)
Week 12 - Postdose	Number Analyzed	182 participants
		28.1 (13.2)
Follow-up Month 3	Number Analyzed	63 participants
		0.943 (1.40)

[1]

week 0 pre-dose serum concentrations are below the level of detection

33. Secondary Outcome

Title	Number of Participants With Anti-Drug Antibody (ADA) Response to Durvalumab, Safety Analysis Set - ADA Evaluable Patients
Description	Serum Samples will be measured for the presence of ADAs and ADA-neutralizing antibodies for Durvalumab using validated assays. Tiered analysis will be performed to include screening, confirmatory, and titer assay components, and positive-negative cut points previously statistically determined from drug-naïve validation samples will be used. Persistently positive is defined as having at least 2 post-baseline ADA positive measurements with at least 16 weeks (112 days) between the first and last positive measurements, or an ADA positive result at the last available assessment. Transiently positive is defined as having at least one post-baseline ADA positive measurement and not fulfilling the conditions for persistently positive. The category may include patients meeting these criteria who are ADA positive at baseline.
Time Frame	At week 0, 4, 12 and 24, and at follow-up Month 3.

Outcome Measure Data

Analysis Population Description

ADA evaluable patients is defined as patients in the safety analysis set who have a non-missing baseline ADA and at least one non-missing post-baseline result

Arm/Group Title		Combination Therapy	Monotherapy
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)
Overall Number of Participants Analyzed		340	345
Measure Type: Count of Participants; Unit of Measure: Participants
ADA evaluable patients	Number Analyzed	340 participants	345 participants
		293 86.2%	302 87.5%
ADA positive at any visit (ADA Prevalence)	Number Analyzed	293 participants	302 participants
		37 12.6%	31 10.3%
Treatment-emergent ADA positive (ADA Incidence)	Number Analyzed	293 participants	302 participants
		28 9.6%	11 3.6%
Treatment-boosted ADA	Number Analyzed	293 participants	302 participants
		0 0.0%	1 0.3%
Treatment-induced ADA (Positive Post-baseline only)	Number Analyzed	293 participants	302 participants
		28 9.6%	10 3.3%
ADA Positive at Baseline only	Number Analyzed	293 participants	302 participants
		8 2.7%	18 6.0%
ADA Positive Post-baseline and Positive at Baseline	Number Analyzed	293 participants	302 participants
		1 0.3%	3 1.0%
Persistently Positive	Number Analyzed	293 participants	302 participants
		15 5.1%	8 2.6%
Transiently Positive	Number Analyzed	293 participants	302 participants
		14 4.8%	5 1.7%
nAb Positive at any visit	Number Analyzed	293 participants	302 participants
		3 1.0%	2 0.7%

34. Secondary Outcome

Title	Number of Participants With Anti-Drug Antibody (ADA) Response to Tremelimumab, Safety Analysis Set - ADA Evaluable Patients
Description	Serum Samples will be measured for the presence of ADAs and ADA-neutralizing antibodies for Tremelimumab using validated assays. Tiered analysis will be performed to include screening, confirmatory, and titer assay components, and positive-negative cut points previously statistically determined from drug-naïve validation samples will be used. Persistently positive is defined as having at least 2 post-baseline ADA positive measurements with at least 16 weeks (112 days) between the first and last positive measurements, or an ADA positive result at the last available assessment. Transiently positive is defined as having at least one post-baseline ADA positive measurement and not fulfilling the conditions for persistently positive. The category may include patients meeting these criteria who are ADA positive at baseline.
Time Frame	At week 0, 4, 12 and at follow-up Month 3.

Outcome Measure Data

Analysis Population Description

ADA evaluable patients is defined as patients in the safety analysis set who have a non-missing baseline ADA and at least one non-missing post-baseline result

Arm/Group Title		Combination Therapy
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab
Overall Number of Participants Analyzed		340
Measure Type: Count of Participants; Unit of Measure: Participants
ADA evaluable patients	Number Analyzed	340 participants
		292 85.9%
ADA positive at any visit (ADA Prevalence)	Number Analyzed	292 participants
		64 21.9%
Treatment-emergent ADA positive (ADA Incidence)	Number Analyzed	292 participants
		54 18.5%
Treatment-boosted ADA	Number Analyzed	292 participants
		1 0.3%
Treatment-induced ADA (Positive Post-baseline only)	Number Analyzed	292 participants
		53 18.2%
ADA Positive at Baseline only	Number Analyzed	292 participants
		8 2.7%
ADA Positive Post-baseline and Positive at Baseline	Number Analyzed	292 participants
		3 1.0%
Persistently Positive	Number Analyzed	292 participants
		31 10.6%
Transiently Positive	Number Analyzed	292 participants
		25 8.6%
nAb Positive at any visit	Number Analyzed	292 participants
		50 17.1%

35. Secondary Outcome

Title	Change From Baseline in FACT-BL (Derived NFBlSI-18 Score, FACT-BL TOI, and FACT-BL Total Score) by MMRM Analysis, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	the change from baseline in the following total/index scores will be evaluated as secondary endpoints: FACT-BL TOI (refer to as TOI), FACT-BL Total score, and NFBlSI-18 score. All the 5 subscales (PWB (0-28), FWB (0-28), EWB (0-24), SWB (0-28), and BlCS(0-48)) are summed as the FACT-BL total score (range 0-156), while the sum of PWB, FWB and BICS constitutes the FACT-BL TOI (range 0-104). NFBlSI-18 (range 0-72) is based on the scores of 16 items (GP4, C2, BL1, GP3, GE6, GE1, C6, BL5, GF5, GP2, GP1, GP6, C3, GP5, GF3, GF7) and 2 extra items "I feel weak all overall" and "I feel light-headed (dizzy)". The range of each item is 0-4. Higher score represent worse outcome.
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		342	346	344
Mean (Standard Error); Unit of Measure: Scores on a scale
NFBLSI - 18 Score (Average overall visits)	Number Analyzed	220 participants	204 participants	189 participants
		-3.7 (0.70)	-3.1 (0.76)	-5.2 (0.82)
FACT-BL TOI (Average overall visits)	Number Analyzed	220 participants	204 participants	189 participants
		-5.5 (0.87)	-3.9 (0.95)	-7.2 (1.02)
FACT-BL Total score (Average overall visits)	Number Analyzed	220 participants	204 participants	189 participants
		-7.2 (1.19)	-4.7 (1.30)	-8.8 (1.39)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1617
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	1.5
	Confidence Interval	(2-Sided) 95%; -0.6 to 3.59
	Estimation Comments	[Not Specified]

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0578
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	2.1
	Confidence Interval	(2-Sided) 95%; -0.07 to 4.29
	Estimation Comments	[Not Specified]

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2003
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	1.7
	Confidence Interval	(2-Sided) 95%; -0.91 to 4.32
	Estimation Comments	[Not Specified]

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0178
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	3.3
	Confidence Interval	(2-Sided) 95%; 0.57 to 6.00
	Estimation Comments	[Not Specified]

Statistical Analysis 5

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3765
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	1.6
	Confidence Interval	(2-Sided) 95%; -1.96 to 5.17
	Estimation Comments	[Not Specified]

Statistical Analysis 6

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0300
	Comments	MMRM model included patient, treatment, cisplatin eligibility, PD-L1 status and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	4.1
	Confidence Interval	(2-Sided) 95%; 0.40 to 7.81
	Estimation Comments	[Not Specified]

36. Secondary Outcome

Title	Change From Baseline in FACT-BL (Derived NFBlSI-18 Score, FACT-BL TOI, and FACT-BL Total Score) by MMRM Analysis, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	the change from baseline in the following total/index scores will be evaluated as secondary endpoints: FACT-BL TOI (refer to as TOI), FACT-BL Total score, and NFBlSI-18 score. All the 5 subscales (PWB (0-28), FWB (0-28), EWB (0-24), SWB (0-28), and BlCS(0-48)) are summed as the FACT-BL total score (range 0-156), while the sum of PWB, FWB and BICS constitutes the FACT-BL TOI (range 0-104). NFBlSI-18 (range 0-72) is based on the scores of 16 items (GP4, C2, BL1, GP3, GE6, GE1, C6, BL5, GF5, GP2, GP1, GP6, C3, GP5, GF3, GF7) and 2 extra items "I feel weak all overall" and "I feel light-headed (dizzy)". The range of each item is 0-4. Higher score represent worse
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		205	209	207
Mean (Standard Error); Unit of Measure: Scores on a scale
NFBLSI - 18 Score (Average overall visits)	Number Analyzed	129 participants	131 participants	119 participants
		-3.9 (0.85)	-2.0 (0.87)	-5.7 (0.95)
FACT-BL TOI (Average overall visits)	Number Analyzed	129 participants	131 participants	119 participants
		-5.8 (1.09)	-2.6 (1.11)	-8.0 (1.22)
FACT-BL Total score (Average overall visits)	Number Analyzed	129 participants	131 participants	119 participants
		-8.0 (1.47)	-2.7 (1.51)	-10.1 (1.65)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1373
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	1.9
	Confidence Interval	(2-Sided) 95%; -0.6 to 4.36
	Estimation Comments	[Not Specified]

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0034
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	3.8
	Confidence Interval	(2-Sided) 95%; 1.26 to 6.27
	Estimation Comments	[Not Specified]

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1774
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	2.2
	Confidence Interval	(2-Sided) 95%; -1.00 to 5.39
	Estimation Comments	[Not Specified]

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0011
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	5.4
	Confidence Interval	(2-Sided) 95%; 2.19 to 8.65
	Estimation Comments	[Not Specified]

Statistical Analysis 5

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3494
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	2.1
	Confidence Interval	(2-Sided) 95%; -2.27 to 6.38
	Estimation Comments	[Not Specified]

Statistical Analysis 6

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0011
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	7.3
	Confidence Interval	(2-Sided) 95%; 2.96 to 11.71
	Estimation Comments	[Not Specified]

37. Secondary Outcome

Title	Change From Baseline in FACT-BL (Derived NFBlSI-18 Score, FACT-BL TOI, and FACT-BL Total Score) by MMRM Analysis, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	the change from baseline in the following total/index scores will be evaluated as secondary endpoints: FACT-BL TOI (refer to as TOI), FACT-BL Total score, and NFBlSI-18 score. All the 5 subscales (PWB (0-28), FWB (0-28), EWB (0-24), SWB (0-28), and BlCS(0-48)) are summed as the FACT-BL total score (range 0-156), while the sum of PWB, FWB and BICS constitutes the FACT-BL TOI (range 0-104). NFBlSI-18 (range 0-72) is based on the scores of 16 items (GP4, C2, BL1, GP3, GE6, GE1, C6, BL5, GF5, GP2, GP1, GP6, C3, GP5, GF3, GF7) and 2 extra items "I feel weak all overall" and "I feel light-headed (dizzy)". The range of each item is 0-4. Higher score represent worse
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		137	137	137
Mean (Standard Error); Unit of Measure: Scores on a scale
NFBLSI - 18 Score (Average overall visits)	Number Analyzed	91 participants	73 participants	69 participants
		-3.4 (1.01)	-3.8 (1.18)	-2.0 (1.16)
FACT-BL TOI (Average overall visits)	Number Analyzed	91 participants	73 participants	69 participants
		-4.9 (1.18)	-5.0 (1.39)	-3.0 (1.36)
FACT-BL Total score (Average overall visits)	Number Analyzed	91 participants	73 participants	69 participants
		-5.7 (1.62)	-6.5 (1.90)	-3.2 (1.86)

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3865
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-1.3
	Confidence Interval	(2-Sided) 95%; -4.35 to 1.69
	Estimation Comments	[Not Specified]

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	NFBLSI- 18 score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2840
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-1.8
	Confidence Interval	(2-Sided) 95%; -5.05 to 1.49
	Estimation Comments	[Not Specified]

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2970
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-1.9
	Confidence Interval	(2-Sided) 95%; -5.41 to 1.66
	Estimation Comments	[Not Specified]

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL TOI (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3026
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-2.0
	Confidence Interval	(2-Sided) 95%; -5.85 to 1.83
	Estimation Comments	[Not Specified]

Statistical Analysis 5

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.3168
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-2.5
	Confidence Interval	(2-Sided) 95%; -7.29 to 2.38
	Estimation Comments	[Not Specified]

Statistical Analysis 6

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	FACT-BL Total score (Average overall visits)
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2179
	Comments	MMRM model included patient, treatment, cisplatin eligibility and visceral metastasis, visit and treatment by visit interaction as explanatory variables, and the appropriate baseline FACT-BL value as a covariate.
	Method	Mixed Models Analysis
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Mean Difference (Final Values)
	Estimated Value	-3.3
	Confidence Interval	(2-Sided) 95%; -8.51 to 1.96
	Estimation Comments	[Not Specified]

38. Secondary Outcome

Title	Improvement in Fatigue and Deterioration in Pain Per FACT-BL, Full Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Fatigue will be based on the question of "I have a lack of energy" and pain will be based on the question of "I have pain", according to GP1 and GP4 in PWB, respectively. Improvement in fatigue is defined at least 1 point improvement from baseline using GP1 of FACT-BL. Deterioration in pain is defined as at least 1 point deterioration from baseline using GP4 of FACT-BL.
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

Full analysis set (FAS) included all randomized participants analyzed on an intent-to-treat (ITT) basis.

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		342	346	344
Measure Type: Count of Participants; Unit of Measure: Participants
Patients with improvement in fatigue	Number Analyzed	141 participants	129 participants	115 participants
		38 27.0%	37 28.7%	24 20.9%
Patient with deterioration in pain	Number Analyzed	265 participants	270 participants	226 participants
		142 53.6%	130 48.1%	83 36.7%

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2419
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.4
	Confidence Interval	(2-Sided) 95%; 0.8 to 2.6
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1553
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.5
	Confidence Interval	(2-Sided) 95%; 0.9 to 2.8
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0003
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	2
	Confidence Interval	(2-Sided) 95%; 1.4 to 2.8
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0148
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.6
	Confidence Interval	(2-Sided) 95%; 1.1 to 2.3
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for PD-L1 status, visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

39. Secondary Outcome

Title	Improvement in Fatigue and Deterioration in Pain Per FACT-BL, PD-L1-High Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Fatigue will be based on the question of "I have a lack of energy" and pain will be based on the question of "I have pain", according to GP1 and GP4 in PWB, respectively. Improvement in fatigue is defined at least 1 point improvement from baseline using GP1 of FACT-BL. Deterioration in pain is defined as at least 1 point deterioration from baseline using GP4 of FACT-BL.
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-High analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-High as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

≥25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control OR ≥25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		205	209	207
Measure Type: Count of Participants; Unit of Measure: Participants
Patients with improvement in fatigue	Number Analyzed	86 participants	83 participants	74 participants
		23 26.7%	24 28.9%	14 18.9%
Patient with deterioration in pain	Number Analyzed	156 participants	164 participants	145 participants
		78 50.0%	71 43.3%	51 35.2%

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.2473
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.6
	Confidence Interval	(2-Sided) 95%; 0.7 to 3.4
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0148
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.7
	Confidence Interval	(2-Sided) 95%; 0.8 to 3.8
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0090
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.9
	Confidence Interval	(2-Sided) 95%; 1.2 to 3.0
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.1510
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.4
	Confidence Interval	(2-Sided) 95%; 0.9 to 2.3
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

40. Secondary Outcome

Title	Improvement in Fatigue and Deterioration in Pain Per FACT-BL, PD-L1-Low/Negative Analysis Set -Durvalumab + Tremelimumab Combination Therapy vs SoC and Durvalumab Mono Therapy vs SoC
Description	Fatigue will be based on the question of "I have a lack of energy" and pain will be based on the question of "I have pain", according to GP1 and GP4 in PWB, respectively. Improvement in fatigue is defined at least 1 point improvement from baseline using GP1 of FACT-BL. Deterioration in pain is defined as at least 1 point deterioration from baseline using GP4 of FACT-BL.
Time Frame	At baseline then every 8 weeks until second progression/death, whichever comes first. Assessed up to the data cut-off date (27JAN2020, a maximum of 5 years).

Outcome Measure Data

Analysis Population Description

The PD-L1-Low/Negative analysis set will include the subset of patients in the FAS whose PD-L1 status is PD-L1-Low/negative as defined by an Immunohistochemistry (IHC) assay developed by Ventana as:

<25% tumor cell membrane positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control AND <25% tumor associated immune cell positivity for PD-L1 at any intensity above background staining as noted on the corresponding negative control.

Arm/Group Title		Combination Therapy	Monotherapy	Standard of Care
Arm/Group Description:		MEDI4736 (Durvalumab) + Tremelimumab	MEDI4736 (Durvalumab)	Standard of Care Chemotherapy Treatment
Overall Number of Participants Analyzed		137	137	137
Measure Type: Count of Participants; Unit of Measure: Participants
Patients with improvement in fatigue	Number Analyzed	55 participants	46 participants	41 participants
		15 27.3%	13 28.3%	10 24.4%
Patient with deterioration in pain	Number Analyzed	109 participants	106 participants	81 participants
		64 58.7%	59 55.7%	32 39.5%

Statistical Analysis 1

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.6763
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.2
	Confidence Interval	(2-Sided) 95%; 0.5 to 3.2
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 2

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patients with improvement in fatigue
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.6539
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.2
	Confidence Interval	(2-Sided) 95%; 0.5 to 3.3
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Statistical Analysis 3

Statistical Analysis Overview	Comparison Group Selection	Combination Therapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0092
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	2.2
	Confidence Interval	(2-Sided) 95%; 1.2 to 4.0
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood

Statistical Analysis 4

Statistical Analysis Overview	Comparison Group Selection	Monotherapy, Standard of Care
	Comments	Patient with deterioration in pain
	Type of Statistical Test	Superiority
	Comments	[Not Specified]
Statistical Test of Hypothesis	P-Value	0.0324
	Comments	P-value was based on twice the change in log-likelihood resulting from logistic regression model, adjusting for visceral metastases and cisplatin eligibility status.
	Method	Regression, Logistic
	Comments	[Not Specified]
Method of Estimation	Estimation Parameter	Odds Ratio (OR)
	Estimated Value	1.9
	Confidence Interval	(2-Sided) 95%; 1.1 to 3.5
	Estimation Comments	The OR and confidence interval (CI) were calculated using logistic regression, adjusting for visceral metastases and cisplatin eligibility status, with 95% CI calculated by profile likelihood.

Adverse Events

Time Frame	From first administration of study treatment up to 90 days after last dose of study medication or date of initiation of the first subsequent therapy, whichever occurs first, approximately 5 years. All-cause mortality, from screening up to data cut-off date (27JAN2020, approximately 5 years).
Adverse Event Reporting Description	[Not Specified]
Arm/Group Title	Combination Therapy		Monotherapy		Standard of Care
Arm/Group Description	MEDI4736 (Durvalumab) + Tremelimumab		MEDI4736 (Durvalumab)		Standard of Care Chemotherapy Treatment
All-Cause Mortality
	Combination Therapy		Monotherapy		Standard of Care
	Affected / at Risk (%)		Affected / at Risk (%)		Affected / at Risk (%)
Total	255/342 (74.56%)		263/346 (76.01%)		270/344 (78.49%)
Serious Adverse Events
	Combination Therapy		Monotherapy		Standard of Care
	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events	Affected / at Risk (%)	# Events
Total	179/340 (52.65%)		139/345 (40.29%)		125/313 (39.94%)
Blood and lymphatic system disorders
Anaemia † 1	2/340 (0.59%)	2	7/345 (2.03%)	9	9/313 (2.88%)	9
Bone marrow failure † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	1/313 (0.32%)	1
Febrile neutropenia † 1	1/340 (0.29%)	1	0/345 (0.00%)	0	4/313 (1.28%)	4
Haematotoxicity † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	3/313 (0.96%)	4
Leukopenia † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	3/313 (0.96%)	3
Neutropenia † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	1/313 (0.32%)	1
Pancytopenia † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	1/313 (0.32%)	1
Thrombocytopenia † 1	0/340 (0.00%)	0	0/345 (0.00%)	0	6/313 (1.92%)	6
Cardiac disorders
Acute coronary syndrome † 1	0/340 (0.00%)	0	1/345 (0.29%)	1	0/313 (0.00%)	0
Acute myocardial infarction † 1	0/340 (0.00%)	0	3/345 (0.87%)	4	0/313 (0.00%)	0
Angina pectoris † 1	0/340 (0.00%)	0	1/345 (0.29%)	1	1/313 (0.32%)	1
Atrial fibrillation † 1	2/340 (0.59%)	2	1/345 (0.29%)	1	2/313 (0.64%)	2
Cardiac arrest † 1	0/340 (0.00%)	0	3/345 (0.87%)	3	3/313 (0.96%)	3
Cardiac failure † 1	1/340 (0.29%)	1	0/345 (0.00%)	0	1/313 (0.32%)	1
Cardiac failure acute † 1	0/340 (0.00%)	0	1/345 (0.29%)	1	0/313 (0.00%)	0