Adverse events were summarized by maximum reported Common Terminology Criteria for Adverse Event (CTCAE) grade, version 5.0.

Grade 1 (Mild): asymptomatic or mild symptoms; clinical or diagnostic observations only; intervention not indicated.

Grade 2 (Moderate): minimal, local or noninvasive intervention indicated; limiting age-appropriate instrumental activities of daily living.

Grade 3 (Severe or medically significant but not immediately life-threatening): hospitalization or prolongation of hospitalization indicated; disabling; limiting self-care activities of daily living.

Grade 4 (Life-threatening consequences): urgent intervention indicated.

Grade 5: Death related to AE.

Includes adverse events with onset date on or after the date of first dose and up to and including 28 days following discontinuation of treatment but prior to the start of a new anti-cancer therapy.

Participants from the Safety Analysis Set, which includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed).

Progression-free survival (PFS) using Investigator assessment as defined by RECIST 1.1. Median progression free survival (months) calculated using the Kaplan-Meier method.

Progression-free survival (PFS) is defined as the time from 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 anti-cancer therapy prior to progression. Patients who have not progressed or died at the time of analysis will be censored at the time of the latest date of assessment from their last evaluable RECIST assessment.

The primary efficacy analysis of the investigator-assessed progression-free survival will be performed when approximately 278 PFS events and at least 16 months of follow-up after Last subject in, has occurred in the 556 randomized patients.

Full Analysis Set, includes all randomized participants.

Sensitivity analysis for progression-free survival (PFS) by blinded independent central review (BICR) using Investigator assessment as defined by RECIST 1.1. Median progression free survival (months) calculated using the Kaplan-Meier method.

Progression-free survival (PFS) is defined as the time from 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 anti-cancer therapy prior to progression. Patients who have not progressed or died at the time of analysis will be censored at the time of the latest date of assessment from their last evaluable RECIST assessment.

The primary efficacy analysis of the investigator-assessed progression-free survival will be performed when approximately 278 PFS events and at least 16 months of follow-up after Last subject in, has occurred in the 556 randomized patients.

Full Analysis Set, includes all randomized participants.

Overall survival is defined as the time from the date of first dose until death due to any causes. Subjects not known to have died at the time of analysis are censored at the last recorded date on which the subject was known to be alive. Median overall survival calculated using the Kaplan-Meier method.

Per the protocol, the safety run-in treatment arms were combined and analyzed regardless of chemotherapy received for consistency with the randomized component.

Safety Analysis Set, includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed).

Duration of response (DoR) is defined as the time from the date of first documented response (which is subsequently confirmed) until date of documented progression or death in the absence of disease progression (i.e. date of progression free survival event or censoring - date of first response + 1). The end of response should coincide with the date of progression or death from any cause used for the progression free survival endpoint. The time of the initial response is defined as the latest of the dates contributing towards the first visit that was Complete response or Partial response that was subsequently confirmed.

Per the protocol, the safety run-in treatment arms were combined and analyzed regardless of chemotherapy received for consistency with the randomized component.

Participants from the Safety Analysis Set, which includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed), that had a confirmed response.

Confirmed objective response rate (per RECIST 1.1 using Investigator assessments) is defined as the number (%) of subjects with at least 1 visit response of Completed Response (CR) or Partial Response (PR), where each CR or PR must be subsequently confirmed at least 4 weeks after the visit when the response was first observed with no evidence of progression between the initial and CR/PR confirmation visit.

Confidence Interval is calculated using the exact Clopper-Pearson method.

Per the protocol, the safety run-in treatment arms were combined and analyzed regardless of chemotherapy received for consistency with the randomized component.

Safety Analysis Set, includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed).

Depth of Response (percent change from Baseline in tumor diameter) is defined as the relative percent change in the sum of the longest diameters of RECIST 1.1 target lesions (TL) at the nadir in the absence of new lesions or progression of non-target lesions (NTL) compared to baseline. The best percentage change in TL size is the maximum reduction from baseline or the minimum increase from baseline in the absence of a reduction. Tumor assessments of the chest and abdomen (including the entire liver and both adrenal glands) were performed using RECIST 1.1 by the investigator on images from CT (preferred) or MRI with IV contrast.

Per the protocol, the safety run-in treatment arms were combined and analyzed regardless of chemotherapy received for consistency with the randomized component.

Safety Analysis Set, includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed).

Disease control rate is defined as the percentage of subjects who have a best overall response of Complete response (CR) or Partial response (PR) or Stable disease (SD) by RECIST 1.1 as assessed by the Investigator. For participants with a best overall response of SD, a RECIST assessment of SD must have been observed at least 6 weeks minus 1 week (at least 35 study days) following the first dose.

Per the protocol, the safety run-in treatment arms were combined and analyzed regardless of chemotherapy received for consistency with the randomized component.

Safety Analysis Set, includes all participants in the Safety Run-In Treatment Arms who received at least 1 dose of study treatment (osimertinib, cisplatin, carboplatin, or pemetrexed).

Overall survival is defined as the time from the date of randomization until death due to any cause. Subjects not known to have died at the time of analysis are censored at the last recorded date on which the subject was known to be alive. Median overall Survival calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

Landmark Overall Survival at 1, 2, and 3 years looks at the percent of patients alive at 1, 2 and 3 year time points.

Overall survival at 36 months not included to data cut off prior to 36-month timepoint.

Overall survival percentage calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

Objective Response Rate (ORR) (per Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) using Investigator assessments) is defined as the number (%) of patients with at least 1 visit response of Complete Response or Partial Response. Data obtained up until progression, or the last evaluable assessment in the absence of progression, was included in the assessment of Objective Response Rate.

The investigator-assessed ORR was summarized with a logistic regression stratified by race (Chinese/Asian vs. Non-Chinese/Asian vs. Non-Asian), WHO performance status (0 vs. 1), and method used for tissue testing (central vs. local).

Full Analysis Set, includes all randomized participants.

The duration of response is defined as the time from the date of first documented response until date of documented progression or death in the absence of disease progression. The end of response should coincide with the date of progression or death from any cause used for the progression-free survival endpoint. The time of the initial response is defined as the latest of the dates contributing towards the first response of Partial response or Complete response.

Median values of the duration of response, along with two-sided 95% CI in each treatment group were computed using the Kaplan-Meier method.

Participants from the Full Analysis Set, which includes all randomized participants, that had a response.

Depth of Response (percent change from Baseline in tumor diameter) is defined as the relative percent change in the sum of the longest diameters of RECIST 1.1 target lesions (TL) at the nadir in the absence of new lesions or progression of non-target lesions (NTL) compared to baseline. The best percentage change in TL size is the maximum reduction from baseline or the minimum increase from baseline in the absence of a reduction. Tumor assessments of the chest and abdomen (including the entire liver and both adrenal glands) were performed using RECIST 1.1 by the investigator on images from CT (preferred) or MRI with IV contrast.

Full Analysis Set, includes all randomized participants.

Disease control rate (DCR) is defined as the percentage of subjects who have a best overall response of Complete Response (CR) or Partial Response (PR) or Stable Disease (SD) by Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST 1.1) as assessed by the Investigator. For participants with a best overall response of SD, a RECIST assessment of SD must have been observed at least 6 weeks minus 1 week (at least 35 study days) following the randomization.

The adjusted disease control rate was calculated using a logistic regression stratified by race (Chinese/Asian vs. Non-Chinese/Asian vs. Non-Asian), WHO performance status (0 vs. 1), and method used for tissue testing (central vs. local).

Full Analysis Set, includes all randomized participants.

Progression Free Survival 2 is defined as the time from the date of randomization to the earliest of the progression events subsequent to that used for the primary Progression Free Survival (PFS), or death in absence of a first or second progression. The second progression event must have occurred after subsequent treatment administered after the initial progression free survival event. Any participant that was lost to follow-up, withdrew consent or discontinued for other reasons at the time of the analysis were censored at the last evaluable progression assessment. Median second progression free survival (months) calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

Time to first subsequent therapy (TFST) or death is defined as the time from the date of randomization to the earlier of the date of anti- cancer therapy start date following IP discontinuation or death. Any patient not known to have had a subsequent therapy or not known to have died at the time of the analysis were censored at the last known time to have not received subsequent therapy; i.e., the last follow-up visit where this was confirmed. Median time to first subsequent therapy or death calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

Time to second subsequent therapy (TSST) or death is defined as the time from the date of randomization to the earlier of the date of second subsequent anti-cancer therapy start date following IP discontinuation or death. Any patient not known to have died at the time of the analysis and not known to have had a second subsequent therapy will be censored at the last known time to have not received second subsequent therapy, i.e., the last follow-up visit where this was confirmed. If a patient terminated the study for reason other than death before second subsequent therapy, these patients will be censored at the earliest of their last known to be alive and termination dates. Median time to second subsequent therapy or death calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

QLQ-C30 has 30 questions and scores range from 0-100 after a linear transformation. Questions are combined to produce symptom scales, individual symptom items, functional scales, and global health status (GHS)/quality of life (QoL).

Positive change from baseline scores on the GHS/QoL and functioning scales indicate improvement on health status/function, and negative change scores on symptom scales/items represent less symptom severity/improvement on symptom status.

The score values calculated by averaging across patients overall mean across all visits. The analysis was performed using a MMRM analysis on the change from baseline in the score at each visit, including subject (random effect), treatment, visit (fixed effect & repeated measure) and treatment by visit interaction as explanatory variables, with the baseline score as a covariate along with the baseline score by assessment interaction.

Full Analysis Set, includes all randomized participants.

The EORTC QLQ-C30 consists of 30 questions that are combined to produce 3 symptom scales (fatigue, pain, and nausea/vomiting), 5 individual symptom items (dyspnea, insomnia, appetite loss, constipation, and diarrhea), 5 functional scales (physical, role, cognitive, emotional, and social), and a global measure of health status/QoL. Scores range from 0-100 after a linear transformation.

Time to deterioration (TTD) is defined as the time from randomization until the date of the first clinically meaningful worsening (a change in the score from baseline of ≥10) that is confirmed at a subsequent assessment or death in the absence of a clinically meaningful symptom, function, or global health status/QoL worsening, regardless of whether the patient withdraws from study treatment or receives another anticancer therapy prior to symptom, function or GHS/QoL deterioration. The median TTD was calculated using the Kaplan-Meier method.

Full Analysis Set, includes all randomized participants.

EORTC QLQ-LC13 has 13 questions and scores range from 0-100 after a linear transformation. Questions assess cough, hemoptysis, dyspnea, site specific pain, sore mouth, dysphagia, peripheral neuropathy, and alopecia and pain medication. While the QLQ-LC13 includes more scales, only the Coughing, Pain in chest, and Dyspnea subscale scores were analyzed for this endpoint.

Negative change from baseline scores indicates less symptom severity, and thus improvement on health status.

The score values calculated by averaging across patients overall mean across all visits. The analysis was performed using a MMRM analysis on the change from baseline in the score at each visit, including subject (as a random effect), treatment, visit (as fixed effect and repeated measure) and treatment by visit interaction as explanatory variables, with the baseline score as a covariate along with the baseline score by assessment interaction.

Full Analysis Set, includes all randomized participants.

EORTC QLQ-LC13 has 13 questions and scores range from 0-100 after a linear transformation. Questions assess cough, hemoptysis, dyspnea, site specific pain, sore mouth, dysphagia, peripheral neuropathy, and alopecia and pain medication.

Time to deterioration (TTD) is defined as the time from randomization until the date of the first clinically meaningful worsening (a change in the score from baseline of ≥10) that is confirmed at a subsequent assessment or death in the absence of a clinically meaningful symptom, function, or global health status/QoL worsening, regardless of whether the patient withdraws from study treatment or receives another anticancer therapy prior to symptom, function or GHS/QoL deterioration. The median TTD was calculated using the Kaplan-Meier method.

Full Analysis Set, all randomized participants.

Comparing the local EGFR mutation test result used for patient selection with the retrospective central cobas® EGFR Mutation Test v2 results in participants with EXON 19 Deletion, excluding invalid results.

Since this endpoint uses pre-randomization data to compare results from a central vs. local lab, randomized component treatment arms were combined for analysis.

Participants with a valid local and central test result.

Comparing the local EGFR mutation test result used for patient selection with the retrospective central cobas® EGFR Mutation Test v2 results in participants with L858R, excluding invalid results.

Since this endpoint uses pre-randomization data to compare results from a central vs. local lab, randomized component treatment arms were combined for analysis.

Participants with a valid local and central test result.

PFS is defined as the time from randomization until an event. An event is defined as the date of objective disease progression or death (by any cause in the absence of progression), regardless of whether the subject withdraws from study treatment or receives another anti-cancer therapy prior to progression.

Subgroup analysis was performed using a Cox proportional hazards model including treatment, subgroup and a treatment-by subgroup interaction term.

Subgroup of Full Analysis Set, includes all randomized participants with Exon 19 Deletion EGFR mutation.

PFS is defined as the time from randomization until an event. An event is defined as the date of objective disease progression or death (by any cause in the absence of progression), regardless of whether the subject withdraws from study treatment or receives another anti-cancer therapy prior to progression.

Subgroup analysis was performed using a Cox proportional hazards model including treatment, subgroup and a treatment-by subgroup interaction term.

Subgroup of Full Analysis Set, includes all randomized participants with L858R EGFR mutation.

An analysis will be performed to assess whether the plasma concentration of osimertinib is affected when given with or without chemotherapy. Samples were collected pre-dose and 1-hour post-dose on day 22 and day 106; on day 43 samples were collected pre-dose and 1-, 2-, 4-, and 6-hours post-dose.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

An analysis will be performed to assess whether the plasma concentration of metabolite AZ5104 is affected when given with or without chemotherapy. Samples were collected pre-dose and 1-hour post-dose on day 22 and day 106; on day 43 samples were collected pre-dose and 1-, 2-, 4-, and 6-hours post-dose.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

Cmin,ss is the minimum plasma concentration of Osimertinib at steady state. Cmax,ss is the maximum plasma concentration of Osimertinib at steady state.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

Cmin,ss is the minimum plasma concentration of AZ5104 at steady state. Cmax,ss is the maximum plasma concentration of AZ5104 at steady state.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

AUCss is the area under the plasma concentration-time curve over a doing interval at a stead state.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

AUCss is the area under the plasma concentration-time curve over a doing interval at a stead state.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.

CLss/F is the apparent plasma clearance at steady state.

Pharmacokinetic Analysis Set, includes all randomized participants that received at least 1 dose of study treatment, have at least 1 measurable PK concentration and no missing doses 7 days prior to the PK sample, without any protocol deviation that affects PK, supported by the relevant date and time of this sample.