Supplementary Appendix

Supplementary Appendix This appendix has been provided by the authors to give readers additional information about their work. Supplement to: Jänne PA, Yang JC-H, Kim D-W, et al. AZD9291 in EGFR inhibitor resistant non small-cell lung cancer. N Engl J Med 215;372:1689-99. DOI: 1.156/NEJMoa1411817

SUPPLEMENTARY APPENDIX Table of contents Page Supplementary Methods: Study Oversight 3 Supplementary Methods: Patients 3 Supplementary Methods: Study Design 4 Supplementary Methods: Expansion Cohort Size 5 Supplementary Methods: Dose-Limiting Toxicities 6 Supplementary Methods: Study Assessments Supplementary Methods: Central T79M testing 8 Supplementary Results: Patients 9 Supplementary Results: Pharmacokinetics 9 Supplementary Results: Pneumonitis-Like Events 9 Supplementary Results: Efficacy by Prior EGFR-TKI Therapy 1 Supplementary Results: Efficacy by EGFR Mutation Type 1 Figure S1. Chemical Structure of AZD9291 12 Figure S2. Overall Study Schema. 13 Figure S3. Pharmacokinetics following Single Dose and Multiple Dosing. 14 Figure S4. Response Rate Broken Down by 1 or >1 Prior EGFR-TKI and Centrally Tested T79M Status 15 Figure S5. Response Rate Broken Down by Immediate versus No Immediate Prior EGFR-TKI and Centrally Tested T79M Status 16 Figure S6. Response Rate Broken Down by EGFR Mutation Type L858R versus Exon 19 Deletion and T79M Status. 17 1

Figure S7. Duration of Treatment and Response by Centrally Tested T79M Status. 18 A) Patients with a detectable EGFR T79M. B) Patients without a detectable EGFR T79M. Table S1. Jackman Criteria for Acquired Resistance to EGFR TKIs in Lung Cancer. 2 Table S2. Response Evaluation Criteria in Solid Tumors (Version 1.1), Used to Assess Overall Visit Response. 21 Table S3. Prior Cytotoxic Chemotherapy at Baseline. 23 Table S4. All-Causality Serious Adverse Events 24 Table S5. Summary of Pharmacokinetic Parameters. 25 Table S6. Objective Response Rate by Dose. 26 Table S7. Best Objective Response in Patients with EGFR Mutations Other than Exon 19 Deletion and L858R 27 Table S8. Best Objective Response in Patients with No Sensitizing EGFR Mutations Detected 28 2

SUPPLEMENTARY METHODS Study Oversight An agreement was in place between the study sponsor and the authors, which established the authors rights to publish the study and have access to the data. The study sponsor was permitted a period of 3 days for review of the proposed final manuscript to allow for filing of any relevant patent applications. Responsibility for opinions, conclusion, and interpretation of the data lies with the authors. Patients In the dose escalation cohorts, pre-treatment EGFR T79M testing was optional, whilst in the dose expansion cohorts, a new tumor biopsy was required after disease progression on the most recent regimen. Testing for EGFR T79M was performed in a central laboratory or in a local laboratory followed by confirmation in a central laboratory using the cobas EGFR Mutation Test (Roche Molecular Systems). For the first-line expansion cohort (to be reported elsewhere), central confirmation that the tumor had a known EGFR-TKI-sensitizing mutation was required and that patients had received no prior therapy for advanced diseases. Other key eligibility criteria included an age of 18 years or older (patients from Japan aged at least 2 years), a World Health Organization (WHO) performance status of or 1, measurable disease, and adequate end organ function. Patients with stable, untreated, or treated asymptomatic central nervous system metastases were eligible. Patients were excluded if they had: received EGFR-TKI treatment within 8 days (or approximately 5x half-life) of the first dose of study treatment, received any other therapy for advanced disease within 14 days of first dose 3

of study treatment, a prior history of interstitial lung disease, unresolved toxicities from prior therapy, or clinically significant cardiac conduction abnormalities. Patients were also excluded if they had any of the following cardiac criteria: mean resting corrected QT interval (QTc) > 47 msec obtained from three electrocardiograms (ECGs), using the screening clinic ECG machine derived QTc value; any clinically important abnormalities in rhythm, conduction, or morphology of resting ECG, e.g., complete left bundle branch block, third degree heart block, second degree heart block, PR interval >25 msec; any factors that increase the risk of QTc prolongation or risk of arrhythmic events such as heart failure, hypokalemia, congenital long QT syndrome, family history of long QT syndrome, or unexplained sudden death under 4 years of age in first degree relatives or any concomitant medication known to prolong the QT interval. Study Design Up to six patients were enrolled in each dose escalation cohort to ensure a minimum of three and a maximum of six safety evaluable patients, using the rolling six approach. 27 Following completion of cycle 1 of each escalation cohort, a safety review committee, minimally consisting of Principal Investigators from each site, the study and global safety physicians, reviewed the safety and pharmacokinetic data and made a decision regarding subsequent dose escalation or expansion. Each subsequent dose cohort represented a 1% increase from the prior dose with the exception of the final dose escalation from 16 mg once daily to 24 mg once daily. Expansion cohorts were initiated following identification of a clinical signal of efficacy in the 2 mg escalation cohort. Expansion cohorts were opened at a particular dose only after that dose had been determined to be sufficiently well tolerated in the escalation cohort to permit dose 4

escalation to the next dose level. Patients were evenly allocated into all appropriate open expansion cohorts once they had been confirmed eligible for the study. As expansion cohorts were opened after an escalation cohort completed, and enrolment into the expansion cohorts was rapid, it was uncommon for more than two cohorts to be open at any one time for a molecular subset (T79M+ or T79M-). Following a protocol amendment the study included the extension cohort of an additional 175 patients with the same entry criteria as for the expansion cohorts, who are receiving a dose of 8 mg AZD9291 tablet formulation (Fig. S1). Primary and secondary objectives include evaluation of safety, tolerability, and efficacy (ORR) of AZD9291, assessed by adverse events, vital signs, laboratory variables, physical examination, electrocardiogram, ophthalmic examinations, and RECIST v1.1. The study amendment that included this additional extension cohort indicated that the study was now a phase 1/2 study. The findings from this cohort will be reported elsewhere when complete. Expansion Cohort Size (Protocol Extract) Cohort expansion will be in patients whose NSCLC has progressed following therapy with an EGFR-TKI agent and whose tumor status is T79M+. Data from these cohorts will provide a preliminary assessment of anti-tumor activity based on ORR (providing an acceptable falsenegative risk of concluding that there is no activity if the true response rate is at least 3%). A conclusion of no evidence of activity will be reached if no objective RECIST responses (confirmed PR or CR) are observed. If the true response rate is 3%, the chance of observing no responses in 3 evaluable patients is <1%. 5

In the case that anti-tumor activity, in the form of responses, is observed, 3 evaluable patients will provide reasonable confidence of estimating what the true response rate would be in this population as well as in the second-line T79M- population and a first-line population of EGFR-TKI treatment-naïve patients with EGFRm+ advanced NSCLC. Confidence intervals will be constructed (using the Clopper-Pearson interval) around the response rates observed in each population to enable decisions to be made about the likely success of future studies in each of the populations. For example: 17% ORR (5/3 responses), 8% CI, 8% to 29%; 3% ORR (9/3 responses), 8% CI, 19% to 43%; 5% ORR (15/3 responses), 8% CI, 37% to 63%; 7% ORR (21/3 responses), 8% CI, 57% to 81%. In the event that there is little evidence of anti-tumor activity being observed within the cohort expansions, individual cohorts may be closed prior to 3 evaluable patients being recruited. This will avoid exposing patients to a potentially ineffective therapy, providing the evidence of no anti-tumor activity is strong enough at that time. For example, if eight evaluable patients are recruited to a cohort, and no responses are observed, it may be appropriate to conclude no evidence of activity because if the true response was 3%, the chance of observing no responses in eight evaluable patients is <6%. Dose-Limiting Toxicities A dose-limiting toxicity was defined as any toxicity not attributable to the disease or diseaserelated processes under investigation, which occurs from the first dose of study treatment (Day 1, Cycle ) up to the last day of Cycle 1 (21 days after start of multiple dosing) in the dose 6

escalation cohorts and which includes, despite optimal therapeutic intervention: hematological toxicity CTCAE (version 4) grade 4 present for more than 4 days; non-hematological toxicity CTCAE grade 3 including: infection (including febrile neutropenia), confirmed QTc prolongation (>5 msec absolute or >6 msec above baseline); any other toxicity that: is greater than that at baseline, is clinically significant and/or unacceptable, and is judged to be a doselimiting toxicity by the Safety Review Committee, that is a protocol defined stopping criteria (i.e. confirmed corneal ulceration), results in a disruption of dosing schedule of more than 7 days. Study Assessments Single-dose pharmacokinetic analyses were based on the data from patients in the dose escalation phase, while multiple-dose analysis was based on all patients who received capsules. Plasma concentrations of AZD9291 and the potentially active metabolites AZ514 and AZ755 were measured. Pharmacokinetic concentration data are included from bioanalytical runs, at the time of data cut-off, that passed the acceptance criteria for samples from patients dosed with the capsule. Non-compartmental pharmacokinetic parameters were calculated and summarized in Phoenix WinNonlin using nominal sampling times. Thus preliminary pharmacokinetic data are presented that are draft, unvalidated, and subject to change. The relatively flat pharmacokinetic profile observed means that the errors introduced by using nominal times should be relatively small. Final full pharmacokinetic data will be reported separately. Initially multiple-dose pharmacokinetic assessment was performed on Day 8 of dosing but as the longer pharmacokinetic profile emerged, the steady state assessments were adjusted to Day 22 of multiple dosing. 7

Central T79M Testing Central testing was performed using the cobas EGFR Mutation Test, a real-time polymerase chain reaction (PCR) test for the qualitative detection of mutations in exons 18, 19, 2, and 21 of the EGFR gene in DNA derived from formalin-fixed paraffin-embedded human non-small cell lung cancer tumor tissue. The CE-IVD version of the test, which includes detection of L858R, Ex19del, T79M, Exon 2ins, S768I, and G719X, was used. The central test result was used for final assignment of patients to the T79M positive or negative cohorts. Due to the nature of the samples received from local testing sites/laboratories, adjustments were made to the cobas EGFR Mutation Test Instructions For Use in order to maximize the test success rate and minimize the test reporting time. This was to ensure that a central test result was obtained from as many patients recruited into the study (often based on a local test result) as possible. Instead of using 1 2 sections of 5 μm thickness, up to 1 sections were used. In addition to using biopsy/resected tissue, testing in AURA also allowed fixed cytology samples. In order to maximize assay success rates, the quantity of tissue permissible for extraction was based on assessment of tissue area based on the hematoxylin and eosin stain review. 8

SUPPLEMENTARY RESULTS Patients In 22 patients (1%), a central test result was not available due to: unavailability of the tumor specimen for testing (6 of 22), insufficient remaining tumor in specimen submitted for testing (8 of 22), or failure of the test (4 of 22). The remaining four samples were in the testing process at the time of data cut-off. Pharmacokinetics AZD9291 had a median (min max) time to C max (t max ) of 6 (3 24) hours and was eliminated with apparent mean (min max) half-life of 55 (29.6 145) hours. Based on this half-life, predictable accumulation of AZD9291 was observed after multiple dosing, with steady state reached within 22 days. This resulted in a relatively flat pharmacokinetic profile on daily dosing with limited difference between C max and minimum plasma concentration (C min ). At the 8 mg dose, the geometric mean (min max) AUC tau on day 22 was 1212 (368 388) nm.h and geometric mean (min max) C max was 635.4 (167 211) nm. AUC of the metabolites AZ514 and AZ755 was approximately 1% of the exposure of AZD9291 after multiple dosing. Visual inspection suggests no obvious differences in pharmacokinetic exposure in Asian versus non- Asian patients. Pneumonitis-Like Events In the six cases (four patients at 16 mg and two at 8 mg) of potential pneumonitis-like events, the median time to onset was 41 days (range 14 to 84 days). All six patients received steroid treatment and five of the six received concomitant antibiotic therapy. AZD9291 was 9

discontinued by all six patients around the time of the event: one due to patient decision, two due to disease progression, and three due to the adverse event. Efficacy by Prior EGFR-TKI Therapy We evaluated whether response rate was influenced by prior EGFR-TKI therapy and presence of detectable EGFR T79M. The response rate to AZD9291 in patients with a detectable EGFR T79M who had received only one prior therapy was 57% (33/58; 95% CI, 43% to 7%) and 66% (45/68; 95% CI, 54% to 77%) in patients who had received more than one prior EGFR-TKI treatment (Fig. S3). In patients with no detectable EGFR T79M, the response rates were: only one prior EGFR-TKI, 21% (7/34; 95% CI, 9% to 38%); more than one prior EGFR-TKI, 22% (6/27; 95% CI, 9% to 42%). In patients with a detectable EGFR T79M whose immediate prior therapy, defined as the most recent treatment regimen prior to study entry, was an EGFR-TKI, the objective response rate was 56% (46/82; 95% CI, 45% to 67%) and was 74% (32/43; 95% CI, 59% to 86%) if this treatment was not an EGFR-TKI (Fig. S4). In patients with no detectable EGFR T79M, the objective response rate was lower (11%, 4/36; 95% CI, 3% to 26%) if an EGFR-TKI was the immediate prior treatment compared to if this treatment was not an EGFR- TKI (36%, 9/25; 95% CI, 18% to 57%) (Fig. S4). Efficacy by EGFR mutation type We evaluated whether response rate was influenced by type of EGFR mutation and presence of detectable EGFR T79M. The response rate to AZD9291 in patients with a detectable EGFR T79M and an exon 19 deletion was 64% (49/77; 95% CI, 52% to 74%) and 57% (24/42; 95% CI, 41% to 72%) in patients with an L858R mutation (Fig. S5). In patients with no detectable 1

EGFR T79M, the response rates were: exon 19 deletion, 3% (8/27; 95% CI, 14% to 5%); L858R, 11% (2/19; 95% CI, 1% to 33%). Responses for patients with other EGFR mutations (G719X, S768I, exon 2 insertions) are shown in Table S4 and for patients with no sensitizing EGFR mutations in Table S5. 11

Figure S1. Chemical Structure of AZD9291. 21 12

Figure S2. Overall Study Schema. Study design for the dose escalation/expansion (Phase 1 component) and dose extension (Phase 2 component of the AZD9291 first time in patients ascending dose study (AURA; NCT182632). * Prior therapy not permissible in this cohort, n = 3. Paired biopsy cohort patients with T79M+ tumors, n = 12 evaluable. Not selected by T79M status, US only, n = 12. cobas EGFR Mutation Test (Roche Molecular Systems). 13

Figure S3. Pharmacokinetics Following Single Dose and Multiple Dosing. A B Geometric mean and individual AZD9291 plasma concentration time profiles. Panel A. After single administration of 8 mg capsule. Panel B. After 22 days of multiple once daily dosing of 8 mg capsule. Data are preliminary, unvalidated, and subject to change. 14

Figure S4. Response Rate Broken Down by 1 or >1 Prior EGFR-TKI and Centrally Tested T79M Status. T79M+ T79M- 7% 7% 6% 6% 5% 5% 4% 3% 62% 78/126 57% 33/58 66% 45/68 Overall Previous TKI regimens: 1 Previous TKI regimens: >1 4% 3% Overall Previous TKI regimens: 1 Previous TKI regimens: >1 2% 2% 1% 1% 21% 21% 22% 13/61 7/34 6/27 % % 15

Figure S5. Response Rate Broken Down by Immediate versus No Immediate Prior EGFR-TKI and Centrally Tested T79M Status. T79M+ T79M- 8% 8% 7% 7% 6% 6% 5% Overall 5% Overall 4% 3% 62% 78/125 56% 74% 32/43 Immediate TKI: Yes Immediate TKI: No 4% 3% Immediate TKI: Yes Immediate TKI: No 2% 46/82 2% 36% 1% % 1% % 21% 13/61 11% 4/36 9/25 16

Figure S6. Response Rate Broken Down by EGFR Mutation Type L858R versus Exon 19 Deletion and T79M Status T79M+ T79M- 7% 7% 6% 6% 5% 5% Overall Overall 4% 3% 61% 73/119 64% 49/77 57% Exon 19 deletion L858R 4% 3% Exon 19 deletion L858R 24/42 2% 2% 3% 1% 1% 22% 8/27 % % 1/46 11% 2/19 17

Figure S7. Duration of Treatment and Response by Centrally Tested T79M Status. A 18

B Panel A, patients with a detectable EGFR T79M. Panel B, patients without a detectable EGFR T79M. X represents the start of a confirmed response (complete or partial by Response Evaluation Criteria In Solid Tumors v1.1), O represents the end of a confirmed response. Arrows represent ongoing exposure. Dotted lines represent dose interruptions. 19

Table S1. Jackman Criteria for Acquired Resistance to EGFR TKIs in Lung Cancer. 22 1. Previously received treatment with a single-agent EGFR TKI (e.g. gefitinib or erlotinib) 2. Either of the following: A. A tumor that harbors an EGFR mutation known to be associated with drug sensitivity (ie, G719X, exon 19 deletion, L858R, L861Q) B. Objective clinical benefit from treatment with an EGFR TKI as defined by either: i. Documented partial or complete response (RECIST or WHO), or ii. Significant and durable ( 6 months) clinical benefit (stable disease as defined by RECIST or WHO) after initiation of gefitinib or erlotinib 3. Systemic progression of disease (RECIST or WHO) while on continuous treatment with gefitinib or erlotinib within the last 3 days 4. No intervening systemic therapy between cessation of gefitinib or erlotinib and initiation of new therapy 2

Table S2. Response Evaluation Criteria in Solid Tumors (Version 1.1), used to assess overall visit response. Overall visit response for target lesions (TLs) Complete Response (CR) Disappearance of all TLs since baseline. Any pathological lymph nodes selected as TLs must have a reduction in short axis to <1 mm. Partial Response (PR) Stable Disease (SD) Progressive Disease (PD) At least a 3% decrease in the sum of diameters of TLs, taking as reference the baseline sum of diameters. Neither sufficient shrinkage to qualify for PR nor sufficient increase to qualify for PD. At least a 2% increase in the sum of diameters of TLs, taking as reference the smallest sum on study (this includes the baseline sum if that is the smallest on study). In addition to the relative increase of 2%, the sum must also demonstrate an absolute increase of at least 5mm. Not Evaluable (NE) Only relevant if any of the TLs were not assessed or not evaluable or had a lesion intervention at this visit. Note: If the sum of diameters meets the progressive disease criteria, progressive disease overrides not evaluable as a TL response Overall visit response for non-target lesions (NTLs) Complete Response (CR) Disappearance of all NTLs since baseline. All lymph nodes must be non-pathological in size (<1 mm short axis). Non-CR/Non-PD Progressive Disease (PD) Persistence of one or more NTLs. Unequivocal progression of existing NTLs. Unequivocal progression may be due to an important progression in one lesion only or in several lesions. In all cases the progression MUST clinically significant for the physician to consider changing or stopping therapy. Not Evaluable (NE) Only relevant when one or some of the NTLs were not assessed and in the investigator s opinion they are not able to provide an evaluable overall NTL assessment at this visit. Note: For patients without TLs at baseline, this is relevant if any of the NTLs were not assessed at this visit and the progression criteria have not been met. New lesions Details of any new lesions will also be recorded with the date of assessment. The presence of one or more new lesions is assessed as progression. A lesion identified at a follow up assessment in an anatomical location that was not scanned at baseline is considered a new lesion and will 21

indicate disease progression. Overall visit response algorithm Target lesions Non-Target lesions New Lesions Overall response CR CR No CR CR NA No CR CR Non-CR/Non PD No PR CR NE No PR PR Non PD or NE No PR SD Non PD or NE No SD NA Non CR/Non PD No SD (Non CR/non PD) NE Non-PD or NE No NE NA NE No NE PD Any Yes or No PD Any PD Yes or No PD Any Any Yes PD 22

Table S3. Prior Cytotoxic Chemotherapy at Baseline. 2 mg N = 21 4 mg N = 58 Median number 2 2 of regimens ( 3) ( 9) (range) Number of regimens, n 1 2 3 4 5 6 7 8 9 Unknown 3 7 7 4 1 18 12 8 7 2 1 8 mg N = 9 2 ( 7) 16 18 29 17 2 4 2 1 1 16 mg N = 62 1 ( 8) 17 17 17 6 2 1 2 1 24 mg N = 21 2 ( 8) 5 5 5 3 2 1 Total N = 253 2 ( 9) 51 65 7 38 13 6 3 1 3 1 2 23

Table S4. All-Causality Serious Adverse Events Serious adverse event (SAE), 2 mg n (%) (N = 21) 4 mg (N = 58) 8 mg (N = 9) 16 mg (N = 63) 24 mg (N = 21) Total (N = 253) All SAEs 4 (19) 13 (22) 2 (22) 16 (25) 3 (14) 56 (22) SAEs occurring in >1 patient Pneumonia 1 (5) 3 (5) 2 (2) 6 (2) Pulmonary embolism 1 (5) 2 (3) 1 (1) 2 (3) 6 (2) Pleural effusion 1 (5) 1 (2) 1 (1) 1 (2) 4 (2) Diarrhea 1 (1) 1 (2) 2 (1) Dysphagia 1 (1) 1 (2) 2 (1) Ischemic stroke 1 (1) 1 (2) 2 (1) Pneumonitis 1 (1) 1 (2) 2 (1) Tumor pain 1 (1) 1 (2) 2 (1) Serious adverse events occurring in a single patient: bacteremia (8 mg), bacterial sepsis (4 mg), bronchitis (16 mg), lung infection (2 mg), mediastinitis (8 mg), pneumonia pneumococcal (8 mg), post-procedural infection (8 mg), pyelonephritis acute (4 mg), sepsis (4 mg), septic shock (8 mg), upper respiratory tract infection (16 mg), urinary tract infection (4 mg), viremia (16 mg), hepatic cancer (8 mg), uterine leiomyoma (16 mg), anemia (8 mg), pancytopenia (8 mg), hypoglycemia (8 mg), delusion (4 mg), depression (4 mg), insomnia (4 mg), brain injury (24 mg), convulsion (8 mg), headache (16 mg), presyncope (16 mg), blindness (4 mg), cardiac arrest (24 mg), cardiac failure acute (16 mg), myocardial infarction (4 mg), stress cardiomyopathy (16 mg), deep vein thrombosis (4 mg), embolism (8 mg), dyspnea (2 mg), interstitial lung disease (16 mg), lung disorder (8 mg), pneumothorax (8 mg), pulmonary artery thrombosis (4 mg), abdominal pain upper (16 mg), intestinal obstruction (16 mg), esophageal stenosis (8 mg), esophagitis (16 mg), nausea (2 mg), pancreatitis (24 mg), fatigue (2 mg), cholecystitis (8 mg), skin toxicity (8 mg), musculoskeletal chest pain (16 mg), renal failure (8 mg), renal impairment (8 mg), benign prostatic hyperplasia (4 mg), chest pain (8 mg), general physical health deterioration (8 mg), pyrexia (4 mg), alanine aminotransferase increased (2 mg), aspartate aminotransferase increased (2 mg), blood bilirubin increased (2 mg), overdose (16 mg), thoracic vertebral fracture (24 mg), uncoded terms (16 mg). 24

Table S5. Summary of Multiple-Dose AZD9291 Pharmacokinetic Parameters after Daily Administration of Capsules for 22 Days. 2 mg N = 7 4 mg N = 46 8 mg N = 9 16 mg N = 61 24 mg N = 14 AUC tau, nm.h 1965 [72.8] (873 499) 5636 [43.9] (24 141) 1212* [54.] (368 388) 1916 [61.6] (19 554) 2877 [48.1] (115 512) C max, nm 16.3 [72.5] (45.4 28) 35.1 [44.7] (128 87) 635.4 [55.4] (167 211) 16 [58.7] (132 293) 151 [45.7] (722 262) Data are preliminary, unvalidated, and subject to change. AUC, area under the concentration time curve; C max, maximum plasma concentration. * N=87. 25

Table S6. Objective Response Rate by Dose. Overall population ORR 52% (11/21) 95% CI, 3% to 74% T79M+ patients ORR 5% (5/1) 95% CI, 19% to 81% 2 mg 4 mg 8 mg 16 mg 24 mg Total 43% (25/58) 95% CI, 3% to 57% 59% (19/32) 95% CI, 41% to 76% 52% (4/77) 95% CI, 4% to 63% 7% (3/43) 95% CI, 54% to 83% 58% (36/62) 95% CI, 45% to 7% 61% (17/28) 95% CI, 41% to 79% 52% (11/21) 95% CI, 3% to 74% 5% (7/14) 95% CI, 23% to 77% 51% (123/239) 95% CI, 45% to 58% 61% (78/127) 95% CI, 52% to 7% T79M- patients ORR 67% (2/3) 95% CI, 9% to 99% 6% (1/17) 95% CI,.1% to 29% 17% (4/23) 95% CI, 5% to 39% 33% (6/18) 95% CI, 13% to 59% - 21% (13/61) 95% CI, 12% to 34% Patients are eligible for confirmed response if they have two post-baseline Response Evaluation Criteria In Solid Tumors (RECIST) assessments or patients who withdraw/die prior to the second RECIST assessment. 26

Table S7. Best Objective Response in Patients with EGFR Mutations Other than Exon 19 Deletion and L858R EGFR mutation Dose N Best objective response T79M+ patients G719X 8 mg 2 1 partial response 1 stable disease 16 mg 1 1 stable disease* G719X, S768I 16 mg 1 1 partial response T79M- patients G719X 4 mg 3 1 stable disease 2 progressive disease 16 mg 1 1 progressive disease G719X, S768I 8 mg 1 1 stable disease Exon 2 insertion 4 mg 1 1 progressive disease * Unconfirmed partial response. 27

Table S8. Best Objective Response in Patients with No Sensitizing EGFR Mutations Detected Dose N Best objective response T79M+ patients 8 mg 3 2 partial response 1 stable disease 24 mg 1 1 partial response T79M- patients 4 mg 4 1 partial response 3 progressive disease 8 mg 2 1 partial response 1 stable disease 16 mg 3 1 partial response 2 stable disease 28