Programmed death-ligand 1 expression and T790M status in EGFR-mutant non-small cell lung cancer

Accepted Manuscript Title: Programmed death-ligand 1 expression and T790M status in EGFR-mutant non-small cell lung cancer Authors: Akito Hata, Nobuyuki Katakami, Shigeki Nanjo, Chiyuki Okuda, Reiko Kaji, Katsuhiro Masago, Shiro Fujita, Hiroshi Yoshida, Kota Zama, Yukihiro Imai, Yukio Hirata PII: DOI: Reference:

S0169-5002(17)30401-4 http://dx.doi.org/doi:10.1016/j.lungcan.2017.07.022 LUNG 5419

To appear in:

Lung Cancer

Received date: Revised date: Accepted date:

17-4-2017 14-6-2017 17-7-2017

Please cite this article as: Hata Akito, Katakami Nobuyuki, Nanjo Shigeki, Okuda Chiyuki, Kaji Reiko, Masago Katsuhiro, Fujita Shiro, Yoshida Hiroshi, Zama Kota, Imai Yukihiro, Hirata Yukio.Programmed death-ligand 1 expression and T790M status in EGFR-mutant non-small cell lung cancer.Lung Cancer http://dx.doi.org/10.1016/j.lungcan.2017.07.022 This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

PD-L1 expression and T790M status

Article type: Original Research Title: Programmed death-ligand 1 expression and T790M status in EGFR-mutant non-small cell lung cancer Running title: PD-L1 expression and T790M Authors: Akito Hata, MDa; Nobuyuki Katakami, MD, PhDa; Shigeki Nanjo, MD, PhDa; Chiyuki Okuda, MDa; Reiko Kaji, MDa; Katsuhiro Masago, MD, PhDa; Shiro Fujita, MD, PhDa; Hiroshi Yoshidab; Kota Zama, PhDb; Yukihiro Imai, MD, PhDc, and Yukio Hirata, MD, PhDa Affiliations: aDivision of integrated oncology, Institute of Biomedical Research and Innovation Hospital, Kobe, Japan; bDepartment of contract research for clinical pathology, GeneticLab Co. Ltd., Sapporo, Japan; and vDepartment of clinical pathology, Kobe City Medical Center, General Hospital, Kobe, Japan Corresponding author: Dr. Akito Hata, Division of integrated oncology, Institute of Biomedical Research and Innovation Hospital, 2-2, Minatojima-minamimachi, Chuo-ku, Kobe, 650-0047, Japan. Tel.: +81-78-304-5200. E-mail: [email protected].

1

PD-L1 expression and T790M status

Highlights   

T790M+ status was correlated to lower PD-L1 expression. PD-L1 expression seems to be dynamic and affected by EGFR-TKI treatment. PD-L1 expression might have a prognostic value and interaction with T790M.

ABSTRACT Background: Differential biology and prognosis between T790M+ and T790Mpopulations imply immunological differences also. Methods: We retrospectively analyzed programmed death-ligand 1 (PD-L1) expression and T790M status in rebiopsied samples of epidermal growth factor receptor (EGFR)-mutant non-small cell lung cancer (NSCLC). PD-L1 immunohistochemistry was performed using the SP142 antibody for tumour cell (TC) and tumour-infiltrating immune cell (IC) and the 28-8 antibody for TC. PD-L1+ was defined as TC or IC ≥1%. Results: We investigated 67 available rebiopsied histologic samples in 47 patients. Using the SP142, prevalence of PD-L1 any+, moderate+, and strong+ in T790M+ vs. T790Msamples were 31% vs. 61%, 8% vs. 15%, and 0% vs. 2%, respectively, representing PD-L1+ prevalence of T790M+ samples was significantly lower than that of T790M(p=0.0149). Prevalence of any TC+/IC+ in T790M+ vs. T790M- samples were TC: 31% vs. 51% (p=0.0997) and IC: 8% vs. 27% (p=0.0536), respectively. Using the 28-8, median percentage of PD-L1+ in T790M+ samples was 1.9 (range, 0-27.2), whereas T790M- was 4.1 (range, 0-89.8) (p=0.0801). Prevalence of PD-L1+ ≥1%, ≥5%, and ≥10% in T790M+ vs. T790M- samples were 77% vs. 83% (p=0.5476), 31% vs. 49% (p=0.1419), and 12% vs. 27% (p=0.1213), respectively. In 9 of 11 patients receiving multiple rebiopsies, T790M and/or PD-L1 expression revealed temporal dynamism. Survival curves according to 2

PD-L1 expression and T790M status

PD-L1 expression/T790M status suggested better prognosis in PD-L1-/T790M+ population. Conclusions: T790M+ status was correlated to lower PD-L1 expression. PD-L1 expression might have a prognostic value and interaction with T790M mutation in EGFR-mutant NSCLC. (245 words)

Keywords: programmed death-ligand 1, T790M, epidermal growth factor receptor mutation, non-small cell lung cancer, rebiopsy

1. Introduction Lung cancer is the leading cause of cancer deaths worldwide. Non-small cell lung cancer (NSCLC) accounts for approximately 80% of lung cancers, and the majority are already unresectable and metastatic upon their initial diagnosis. Cytotoxic chemotherapies such as platinum-based regimens were once the primary therapeutic option for metastatic NSCLC, but their advancement has reached a plateau. Molecular-targeted therapies have been recently developed, and they have provided a remarkable benefit to patients harboring specific genetic alterations such as epidermal growth factor receptor (EGFR) gene mutations or anaplastic lymphoma kinase (ALK) gene fusions [1-3]. Efficacies of up-front EGFR- and ALK-tyrosine kinase inhibitors (TKIs) have been established for patients harboring these genetic alterations in prospective randomized phase III trials comparing

3

PD-L1 expression and T790M status

platinum doublets, and the median progression-free survivals (PFSs) are approximately 12 months [4-5]. Despite an initial dramatic response, most patients receiving these TKIs finally acquire resistance. Several acquired resistant mechanisms to EGFR-TKIs have been identified [6-9], and the "gatekeeper" EGFR mutation, a threonine-to-methionine substitution at amino acid position 790 in exon 20 (T790M), is the most common mechanism and accounts for more than half of acquired resistance to EGFR-TKIs [9-10]. Therefore, it is clinically important to develop more effective therapies for patients with T790M. Indeed, third-generation EGFR-TKIs have been developed, and have demonstrated their remarkable efficacies for patients with T790M [11-13]. After spread of these third-generation EGFR-TKIs, a T790M status-based treatment selection becomes essential in clinical practice for patients after acquired resistance to EGFR-TKIs. On the other hand, current advancement of immunotherapies is evolving. Among them, anti-programmed death-1 (PD-1)/programmed death-ligand 1 (PD-L1) antibodies have demonstrated their notable efficacies in pretreated NSCLC. Anti-PD-1 antibodies, such as nivolumab and pembrolizumab, have shown survival benefit in pretreated patients with NSCLC after failure of platinum doublet chemotherapies, in randomized phase III trials compared to docetaxel monotherapy [14-17]. Based on results of these trials, anti-PD-1 antibody monotherapies have become standard treatments for pretreated NSCLC. In cases responding to such immunotherapies, durable response is expected over 1-2 years, much longer than common cytotoxic agents [14-17]. Unfortunately, the response rate and PFS of these immunotherapies are generally 10-20% and 2-3 months, respectively, and

4

PD-L1 expression and T790M status

relatively many patients obtain no response and experience early progression. Besides, several studies demonstrated a possible poorer efficacy of anti-PD-1 antibodies for patients with EGFR mutations [15-17]. However, certain patients with EGFR mutations do respond to anti-PD-1 antibodies. Immunotherapies could play an important role as a salvage therapy after acquired resistance to EGFR-TKIs. Thus, development of reliable predictive markers for immunotherapies is needed for patients with EGFR mutations. Several predictive markers for anti-PD-1/PD-L1 antibodies have been developed. Among them, PD-L1 expression is the most widely investigated predictive marker for many types of cancers. Some studies for NSCLC have demonstrated correlations between PD-L1 expression and response to anti-PD-1/PD-L1 antibodies [15-19]. Moreover, several reports have mentioned that PD-L1 expression has not only a predictive value but also a prognostic value [20,21]. T790M mutation is also considered to be a prognostic marker as well as a predictive marker for third-generation EGFR-TKIs. T790M+ patients seem to have better prognosis than T790M- patients after acquired resistance to EGFR-TKIs [10,22]. Differential biology and prognosis between T790M+ and T790M- populations imply immunological differences also. Based on the above background, we hypothesized that PD-L1 expression is distinct according to T790M status. The aim of this study was to investigate correlation between PD-L1 expression and T790M status in EGFR-mutant NSCLC after acquired resistance to EGFR-TKIs.

5

PD-L1 expression and T790M status

2. Methods 2.1. Samples and patients We retrospectively screened electronic medical records of patients with NSCLC who had received rebiopsy in our institute. Among them, we focused on only samples with EGFR-mutant NSCLC after acquired resistance (complete/partial response or stable disease ≥ 6 months on previous EGFR-TKI therapy) to EGFR-TKIs [23], after exclusion of samples with ALK-fusion and EGFR/ALK wild-type NSCLC. Except for unsuccessful rebiopsied cases, we examined whether each sample contained cancer cells enough to perform PD-L1 immunohistochemistry (IHC). After confirmation of cancer cell sufficiency, PD-L1 IHC was carried out. The study was approved by the institutional review board, and complied with the Declaration of Helsinki.

2.2. Rebiopsy Rebiopsies were performed on various sites using various procedures. Computed tomography-guided biopsy (using percutaneous core needle), transbronchial biopsy with flexible bronchoscopy, or ultrasound-guided biopsy (using percutaneous core needle) were used as rebiopsy procedures after obtaining appropriate written informed consent. Some patients underwent multiple (sites and/or times) rebiopsies. We investigated dynamism of PD-L1 expression and T790M status by examination of these samples. We also examined whether EGFR-TKI treatment had affected PD-L1 expression and T790M status.

6

PD-L1 expression and T790M status

2.3. EGFR mutational analysis We isolated tumor DNA from each specimen, and analyzed EGFR sensitive mutations and T790M mutation using highly sensitive assays: the peptide nucleic acid-locked nucleic acid PCR clamp method or the cycleave method [24,25].

2.4. PD-L1 immunohistochemistry Paraffin-embedded tumor tissue was sectioned at a thickness of 4 μm, and the sections were then pasted on coated glass slides for PD-L1 IHC. PD-L1 IHC was performed using SP142 antibody for staining of tumour cells (TC) and tumour-infiltrating immune cells (IC). Slides were stained with a Ventana GX automated system. Antigen retrieval was performed in Cell conditioning 1. The primary antibody of PD-L1 (clone: SP142) was diluted at 1:25, and incubated for 32 min at room temperature. The antibody was detected with Ventana Amplification Kit and Ventana ultraView Universal DAB Detection Kit. Digital image was captured using Aperio Scanscope AT Turbo slide scanner under 20x objective magnification. Scoring of PD-L1 was performed using digital image analysis software, namely Aperio membrane v9 and Aperio Genie Classifier. TC score was defined as percentage of PD-L1 expressing tumour cells: TC3≥50%, TC2≥5% to <50%, TC1≥1% to <5%, and TC0<1%; and IC score as percentage of tumor area: IC3≥10%, IC2≥5% to <10%, IC1≥1% to <5%, and IC0<1%. These scores were based on previous studies using SP142 [17-19]. We defined TC1/2/3 or IC1/2/3 as PD-L1+, TC1/2 or IC1/2 as PD-L1 moderate+, and TC3 or IC3 as PD-L1 strong+. PD-L1 IHC was additionally performed using 28-8 antibody for TC. Slides were

7

PD-L1 expression and T790M status

stained with Dako Autostainer Link48. Antigen retrieval was performed in Target Retrieval Solution Low pH. PD-L1 antibody (clone: 28-8) was diluted at 1:600, and incubated for 45 min at room temperature. The antibody was detected with Rabbit (LINKER) and EnVision FLEX/HRP. Digital image was captured using Aperio Scanscope AT Turbo slide scanner under 20x objective magnification. Scoring of PD-L1 was performed using digital image analysis software, namely Aperio membrane v9 and Aperio Genie Classifier. PD-L1 expression ≥1%, ≥5%, and ≥10% were investigated as cut-off values. We defined ≥1% as PD-L1+, ≥5% as PD-L1 moderate+, and ≥10% as PD-L1 strong+.

2.5. Survival analyses according to PD-L1 expression and T790M status We investigated overall survival (OS) according to PD-L1 expression (PD-L1+ vs. PD-L1-) by SP142 and T790M status (T790M+ vs. T790M-) at first rebiopsy. OS was defined as time from initial diagnosis of stage IV/recurrence after surgery to death.

2.6. Statistical analyses Chi-square test was done to compare PD-L1 positivity at various cut-offs between T790M+ and T790M- samples. To compare percentage of PD-L1+ between T790M+ and T790M- samples with 28-8, we used the Wilcoxon rank sum test. OS curves and 95% confidence intervals (CI) were estimated by the Kaplan-Meier method. OS was compared using the log-rank test, according to PD-L1 expression and/or T790M status. A P-value less than 0.05 was considered significant. The statistical analyses were performed using JMP 12 (SAS Institute, Inc., Cary, NC, USA).

8

PD-L1 expression and T790M status

3. Results 3.1. Sample and patient profile Between January 2010 and November 2016, 121 rebiopsies to obtain histologic tissue samples were done in 89 patients with NSCLC. Eleven rebiopsies were unsuccessful and failed to obtain malignant tissue samples. Three ALK-fusion positive and 22 EGFR/ALK wild-type samples were excluded, and a total 85 tissue samples of EGFR-mutant NSCLC were examined. Eleven samples were unavailable, and 7 samples contained few enough cancer cells to perform PD-L1 IHC. Finally, we investigated 67 rebiopsied histologic samples in 47 patients (Fig. 1). Characteristics of final 47 patients are shown as the following: median age, 68 (range, 47-84); male/female, 22/25; never/ever smoker, 18/29; adeno/squamous cell/large cell carcinoma, 45/1/1; exon 19 (deletion)/exon 21 (L858R)/exon21 (L861Q), 19/26/2; T790M+/-, 17/30; times of rebiopsy 1/2/3/4, 36/5/4/2. Except T790M mutation, other acquired resistant molecular mechanisms (e.g., MET) were not examined. No findings of small cell transformation and epithelial mesenchymal transition were confirmed. Rebiopsies were performed using: 35 (52%) computed tomography guided biopsy (CTGB); 20 (30%) transbronchial biopsy (TBB); and 12 (18%) ultrasound guided biopsy (USGB). CTGB was adopted to 32 lung and 3 pleural lesions. All 20 TBBs were done to lung lesions. USGB was utilized to: 4 lung; 4 liver; 2 rib; 1 cervical lymph node; and 1 supraclavicular lymph node lesions. Thirty-six (77%) patients received rebiopsy once, and

9

PD-L1 expression and T790M status

11 (23%) underwent more than twice.

3.2. Comparison of PD-L1 expression between T790M+ and T790M- samples using SP142 Prevalence of PD-L1 any+, moderate+, and strong+ in T790M+ vs. T790Msamples were 31% vs. 61%, 8% vs. 15%, and 0% vs. 2%, respectively (Fig. 2). Prevalence of PD-L1+ in T790M+ samples was significantly lower than that of T790M- (p=0.0149). Prevalence of any TC+ in T790M+ vs. T790M- samples were 31% vs. 51%, respectively (p=0.0997). That of any IC+ in T790M+ vs. T790M- samples were 8% vs. 27%, respectively (p=0.0536). Number of TC3/2/1/0 and IC3/2/1/0 in T790M+ samples were 0/2/6/18 and 0/0/2/24, respectively (2 TC2/IC0, 2 TC1/IC1, 4 TC1/IC0, and 18 TC0/IC0). Those of TC3/2/1/0 and IC3/2/1/0 in T790M- samples were 1/5/15/20 and 1/0/10/30, respectively (1 TC3/IC3, 1 TC2/IC1, 4 TC2/IC0, 5 TC1/IC1, 10 TC1/IC0, 4 TC0/IC1, and 16 TC0/IC0). Fig. 2 shows overlap between PD-L1 expression on TC/IC.

3.3. Comparison of PD-L1 expression between T790M+ and T790M- samples using 28-8 Median percentage of PD-L1+ in T790M+ samples was 1.9 (range, 0-27.2), whereas T790M- was 4.1 (range, 0-89.8) (p=0.0801 by Wilcoxon rank sum test) (Fig. 3A). Prevalence of PD-L1+ ≥1%, ≥5%, ≥10%, ≥20%, and ≥50% in T790M+ vs. T790Msamples were 77% (20/26) vs. 83% (34/41) (p=0.5476), 31% (8/26) vs. 49% (20/41) (p=0.1419), 12% (3/26) vs. 27% (11/41) (p=0.1213), 4% (1/26) vs. 15% (6/41) (p=0.1336), and 4% (1/26) vs. 7% (3/41) (p=0.5476), respectively (Fig. 3B).

10

PD-L1 expression and T790M status

3.4. Concordance of PD-L1-positivity between SP142 and 28-8 antibodies We examined concordance of PD-L1-positivity between SP142 and 28-8 antibodies. The concordance rate was 58% (39 of 67 samples). Details were: SP142+/28-8+, 31 (46%) samples; SP142-/28-8-, 8 (12%) samples; SP142-/28-8+, 26 (39%) samples; and SP142+/28-8-, 2 (3%) samples.

3.5. PD-L1 expression of representative samples Fig. 4 shows PD-L1 expression of representative samples: sample A (T790M+), TC0/IC0 by SP142 (Fig. 4A1) and 0% by 28-8 (Fig. 4A2); sample B (T790M+), TC2/IC0 by SP142 (Fig. 4B1) and 6% by 28-8 (Fig. 4B2); and sample C (T790M-), TC3/IC3 by SP142 (Fig. 4C1) and 90% by 28-8 (Fig. 4C2).

3.6. Dynamism of PD-L1 expression and T790M status affected by EGFR-TKI treatment Eleven patients received multiple rebiopsies (4 patients also underwent rebiopsies to different sites). Table 1 shows PD-L1/T790M dynamism at plural rebiopsies and EGFR-TKI exposure (“on” or “off”) before each rebiopsy. PD-L1 expression by SP142 had changed in 7 of these 11 patients (case #1-7). PD-L1 status changed from PD-L1- to PD-L1+ after EGFR-TKI-free interval in 6 patients (case #1-4, 6, and 7), and from PD-L1+ to PD-L1- after EGFR-TKI exposure in 3 patients (case #1, 2 and 4). In 7 of these 11 patients, T790M status had changed from T790M+ to T790M- after EGFR-TKI-free interval (case #1-5, 8 and 9). Among 3 of these 7, PD-L1 expression also

11

PD-L1 expression and T790M status

changed from PD-L1- to PD-L1+, inversely to a T790M status changed from T790M+ to T790M- (case #1, 3 and 4).

3.7. Survival analyses according to PD-L1 expression and/or T790M status Survival analyses were performed according to PD-L1 expression (by SP142) and/or T790M status at first rebiopsy. Median OS of PD-L1+ (n=24) vs. PD-L1- (n=23) were 43.8 (95% CI, 32.2-99.5) months vs. 82.5 (95% CI, 48.0-undeterminable) months, respectively (p=0.0503) (Fig. 5A). Median OS of T790M+ (n=16) vs. T790M- (n=31) were 78.7 (95% CI, 37.2-undeterminable) months vs. 68.2 (95% CI, 40.5-100.0) months, respectively (p=0.5039) (Fig. 5B). Survival analysis of PD-L1 expression/T790M status was also performed: median OS of PD-L1-/T790M+ (n=11), not reached (95% CI, 22.0-undeterminable) months; PD-L1-/T790M- (n=13), 82.5 (95% CI, 31.5-undeterminable) months; PD-L1+/T790M- (n=18), 48.4 (95% CI, 24.5-99.5) months; and PD-L1+/T790M+ (n=5), 38.1 (95% CI, 33.0-undeterminable) months, respectively (Fig. 5C).

4. Discussion To the best of our knowledge, this is the first report to investigate correlation between PD-L1 expression and T790M status in EGFR-mutant NSCLC after acquired resistance to EGFR-TKIs. Our study has demonstrated that PD-L1 expression of T790M+ samples was significantly lower than that of T790M-. Moreover, more than two-thirds of

12

PD-L1 expression and T790M status

T790M+ samples showed PD-L1- by SP142, and no T790M+ samples exhibited strong PD-L1+. These results suggest a possible poorer efficacy of anti-PD-1/PD-L1 antibodies for T790M+ population. Third-generation EGFR-TKIs will be preferentially administered in T790M+ population, whereas priority of anti-PD-1/PD-L1 antibodies may be lower in T790M+ populations after acquired resistance to EGFR-TKIs. Meanwhile, approximately 60% of T790M- samples were PD-L1+ by SP142, some of which exhibited strong PD-L1+. These results suggest a potential better efficacy of anti-PD-1/PD-L1 antibodies for T790M- population. Notably, only 2 (8%) of 26 T790M+ samples were IC+, whereas 11 (27%) of 41 T790M- were conformed as IC+ (P=0.0536). Herbst el al. demonstrated that higher IC score was associated with greater efficacy of anti-PD-L1 antibody than TC score [16]. Our study suggested higher IC score in T790Msamples, and tumour-infiltrating immune cells could greater affect immunotherapies in T790M populations. Since third-generation EGFR-TKIs are not indicated in T790Mpopulation, immunotherapies using anti-PD-1/PD-L1 antibodies could be important therapeutic options for T790M- population. Interestingly, some cases receiving multiple rebiopsies showed dynamism in PD-L1 expression. PD-L1 expression in some cases changed from PD-L1- to PD-L1+, inversely to a T790M status change from T790M+ to T790M- after EGFR-TKI-free interval. Similarly, PD-L1+ status changed to PD-L1- after EGFR-TKI exposure in other cases. These results suggest that PD-L1 expression is affected by EGFR-TKIs, as T790M status is influenced by EGFR-TKI exposure [9,26]. Preclinical data has suggested that PD-L1 expression could be suppressed by EGFR-TKIs, and no synergism exists between third-generation

13

PD-L1 expression and T790M status

EGFR-TKI and anti-PD-1 antibody [27]. Gainor et al. have reported spatiotemporal heterogeneity of PD-L1 expression [28]. They demonstrated PD-L1 expression of some cases varied after EGFR-TKI treatment, and could be heterogeneous in a single tumor. Azuma et al. have showed that PD-L1 expression was regulated by EGFR-TKI treatment [21]. Akbay et al have also demonstrated that PD-L1 expression was reduced by EGFR-TKIs in NSCLC cell lines harboring EGFR activating mutations [29]. Similar to T790M mutation, PD-L1 expression may be affected by EGFR-TKI treatment in EGFR-mutant NSCLC. Therefore, rebiopsied fresh samples might be desirable to examine PD-L1 expression as a predictive marker for immunotherapies. Our study has revealed that PD-L1 expression could have a prognostic value in EGFR-mutant NSCLC. Several studies also showed better prognosis of PD-L1- population than PD-L1+ population in NSCLC patients [20,21]. T790M is known as a prognostic marker after acquired resistance to EGFR-TKIs [10,22]. According to multiple rebiopsy results, PD-L1 expression and T790M synchronously changed, affected by EGFR-TKI treatments. Generally, T790M+ cancer cells have an indolent nature, but T790M+ patients do not always have better prognosis. Our Fig. 5C shows a possible better prognosis in PD-L1-/T790M+ population. Correlation between PD-L1 expression and T790M suggest their interaction, and better prognosis could be achieved by both biological and immunological aspects. Our study includes several limitations. First, it is retrospective and small sample size. Although our study has demonstrated statistical significances of lower PD-L1 expression in T790M+ samples than that in T790M-, prognostic distinction between

14

PD-L1 expression and T790M status

PD-L1+ and PD-L1- populations was insignificant, and marginal because of small sample size. Furthermore, our study lacks information on the therapies patients were receiving at the time of progression according to PD-L1 expression/T790M status. Subsequent therapies might have affected patients’ prognoses. Second is the selection of antibody for PD-L1 IHC and cut-off for PD-L1 positivity. Our study adopted the SP142 and 28-8 antibodies. Both antibodies have demonstrated a trend of lower PD-L1 expression in T790M+ samples, but their positivity were inconsistent in some samples. The concordance rate was 58%, and notably, 39% of samples revealed SP142-/28-8+. Four anti PD-L1 antibodies (SP142, 28-8, 22c3, and SP263) are clinically used for PD-L1 IHC, but PD-L1 IHC is not globally standardized [30]. Each IHC antibody has been developed simultaneously with each anti-PD-1/PD-L1 therapeutic antibody (atezolizumab, nivolumab, pembrolizumab, and durvalumab). A comparison data of staining degree has been reported (Blueprint PD-L1 IHC Assay Comparison Project [31]), and which found consistently fewer stained tumor cells by SP142, compared with other antibodies, similar to our results. However, neither comparison data with clinical efficacies exist, nor actual relative merits are known. In order to translate basic data for clinical practice, one of these four antibodies for PD-L1 IHC should be used in clinical studies to investigate PD-L1 expression. Our adopted cut-offs by the SP142 and 28-8 were based on evidenced clinical trials which showed correlation between PD-L1 expression and clinical efficacy of anti-PD-1/PD-L1 immunotherapies [14-19,30]. Although optimal cut off for PD-L1 positivity is yet to be determined, we are sure that our adopted cut offs are appropriate. Third, dynamism of PD-L1 expression might have affected survival analyses. We performed survival analyses according to PD-L1

15

PD-L1 expression and T790M status

expression and T790M status at first rebiopsy. EGFR-TKI might have affected PD-L1 expression and T790M status in our study. After first EGFR-TKI treatment, EGFR-TKI exposure is long and sufficient, and PD-L1 expression and T790M status at first rebiopsy properly reflect biological and immunological nature of cancer cells after acquired resistance to EGFR-TKIs. Our study investigated PD-L1 expression as a reliable and popular predictive marker for anti-PD-1/PD-L1 immunotherapies. However, PD-L1+ tumors do not always respond to these immunotherapies, whereas a few PD-L1- tumors respond to them. Whether PD-L1 expression is a definitive predictive marker in clinical practice remains controversial [32]. We should carefully interpret study results of PD-L1 expression, and further clinical evidence is necessary to better utilize PD-L1 expression as predictive marker in clinical practice. In conclusion, our study has demonstrated that T790M+ status was correlated to lower PD-L1 expression. Conversely, T790M- status was associated with higher PD-L1 expression, suggesting a potential efficacy of anti-PD-1/PD-L1 immunotherapies for T790M- population. PD-L1 expression seems to be dynamic and affected by EGFR-TKI treatment. PD-L1 expression might have a prognostic value in EGFR-mutant NSCLC. Further studies are warranted to investigate the detailed interaction between PD-L1 expression and T790M status.

Founding: The study was partially supported by cancer research prize for encouragement of hyogo prefecture health promotion association and research assistance funds from Foundation for Biomedical Research and Innovation. 16

PD-L1 expression and T790M status

Contribution: Study concept, AH and NK; Study design, AH and NK; Data acquisition, AH and NK; Quality control of data and algorithms; AH and NK; Data analysis and interpretation, AH, SN, HY, KZ, and YI; Statistical analysis, AH; Manuscript preparation, AH, NK and YH; Manuscript editing, AH, NK, and YH; Manuscript review, All authors.

Conflicts of interest statement Akito Hata received lecture fee from Chugai, Astra Zeneca, Boeringer Ingelheim, and Eli Lilly. Dr. Katakami received grants from Astra Zeneca, Eisai, Ono, Kyowa Kirin, Shionogi, Daiichi-Sankyo, Taiho, Chugai, Eli Lilly, Boeringer Ingelheim, and Merck Serono, and payment for lectures from Dainippon Sumitomo, Chugai, Boeringer Ingelheim, Astra Zeneca, Eli Lilly, Taiho, Janssen, Novartis, Pfizer, Ono, and Daiichi-Sankyo. The other authors declare no conflicts of interest.

Acknowledgement The study was partially supported by cancer research prize for encouragement of hyogo prefecture health promotion association and research assistance funds from Foundation for Biomedical Research and Innovation. We thank Mr. David Martin for writing support.

References 1.

Lynch TJ, Bell DW, Sordella R, et al. Activating mutations in the epidermal growth factor receptor underlying responsiveness of non-small-cell lung cancer to gefitinib. N Engl J Med 2004;350:2129-39.

2.

Paez JG, Janne PA, Lee JC, et al. EGFR mutations in lung cancer: correlation with clinical response to gefitinib therapy. Science 2004;304:1497-500.

3.

Soda M, Choi YL, Enomoto M, et al. Identification of the transforming EML4-ALK fusion gene in non-small-cell lung cancer. Nature 2007;448:561-6.

4.

Lee CK, Wu YL, Ding PN, et al. Impact of Specific Epidermal Growth Factor Receptor (EGFR) Mutations and Clinical Characteristics on Outcomes After Treatment With EGFR Tyrosine Kinase

17

PD-L1 expression and T790M status

Inhibitors Versus Chemotherapy in EGFR-Mutant Lung Cancer: A Meta-Analysis. J Clin Oncol 2015;33:1958-65. 5.

Solomon BJ, Mok T, Kim DW, et al. First-line crizotinib versus chemotherapy in ALK-positive lung cancer. N Engl J Med 2014;371:2167-77.

6.

Pao W, Miller VA, Politi KA, et al. Acquired resistance of lung adenocarcinomas to gefitinib or erlotinib is associated with a second mutation in the EGFR kinase domain. PLoS Med 2005;2:e73.

7.

Kobayashi S, Boggon TJ, Dayaram T, et al. EGFR mutation and resistance of non–small-cell lung cancer to gefitinib. N Engl J Med 2005;352:786-92.

8.

Engelman JA, Zejnullahu K, Mitsudomi T, et al. MET amplification leads to gefitinib resistance in lung cancer by activating ERBB3 signaling. Science 2007;316:1039-43.

9.

Sequist LV, Waltman BA, Dias-Santagata D, et al. Genotypic and histological evolution of lung cancers acquiring resistance to EGFR inhibitors. Sci Transl Med 2011;3:75ra26.

10. Yu HA, Arcila ME, Rekhtman N, et al. Analysis of tumor specimens at the time of acquired resistance to EGFR-TKI therapy in 155 patients with EGFR-mutant lung cancers. Clin Cancer Res 2013;19:2240-7. 11. Jänne PA, Yang JC, Kim DW, et al. AZD9291 in EGFR inhibitor-resistant non-small-cell lung cancer. N Engl J Med 2015;372:1689-99. 12. Sequist LV, Soria JC, Goldman JW, et al. Rociletinib in EGFR-mutated non-small-cell lung cancer. N Engl J Med 2015;372:1700-9. 13. Mok TS, Wu YL, Ahn ML, et al. Osimertinib or Platinum-Pemetrexed in EGFR T790M-Positive Lung Cancer. N Engl J Med 2017;376:629-40. 14. Brahmer J, Reckamp KL, Baas P, et al. Nivolumab versus Docetaxel in Advanced Squamous-Cell Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:123-35. 15. Borghaei H, Paz-Ares L, Horn L, et al. Nivolumab versus Docetaxel in Advanced Nonsquamous Non-Small-Cell Lung Cancer. N Engl J Med 2015;373:1627-39. 16. Herbst RS, Baas P, Kim DW, et al. Pembrolizumab versus docetaxel for previously treated, PD-L1-positive, advanced non-small-cell lung cancer (KEYNOTE-010): a randomised controlled trial.

18

PD-L1 expression and T790M status

Lancet 2015;387:1540-50. 17. Rittmeyer A, Barlesi F, Waterkamp D, et al. Atezolizumab versus docetaxel in patients with previously treated non-small-cell lung cancer (OAK): a phase 3, open-label, multicentre randomised controlled trial. Lancet 2017;389:255-65. 18. Herbst RS, Soria JC, Kowanetz M, et al. Predictive correlates of response to the anti-PD-L1 antibody MPDL3280A in cancer patients. Nature 2014;515:563-7. 19. Fehrenbacher L, Spira A, Ballinger M, et al. Atezolizumab versus docetaxel for patients with previously treated non-small-cell lung cancer (POPLAR): a multicentre, open-label, phase 2 randomised controlled trial. Lancet 2016;387:1837-46. 20. Zhong A, Xing Y, Pan X, et al. Prognostic value of programmed cell death-ligand 1 expression in patients with non-small-cell lung cancer: evidence from an updated meta-analysis. Onco Targets Ther 2015;8:3595-601. 21. Azuma K, Ota K, Kawahara A, et al. Association of PD-L1 overexpression with activating EGFR mutations in surgically resected nonsmall-cell lung cancer. Ann Oncol 2014;25:1935-40. 22. Hata A, Katakami N, Yoshioka H, et al. Rebiopsy of non-small cell lung cancer patients with acquired resistance to epidermal growth factor receptor-tyrosine kinase inhibitor: comparison between T790M mutation-positive and -negative populations. Cancer 2013;119:4325-32. 23. Jackman D, Pao W, Riely GJ, et al. Clinical definition of acquired resistance to epidermal growth factor receptor tyrosine kinase inhibitors in non-small-cell lung cancer. J Clin Oncol 2010;28:357-60. 24. Nagai Y, Miyazawa H, Huqun, et al. Genetic heterogeneity of the epidermal growth factor receptor in non-small cell lung cancer cell lines revealed by a rapid and sensitive detection system, the peptide nucleic acid-locked nucleic acid PCR clamp. Cancer Res 2005;65:7276-82. 25. Kosaka T, Yatabe Y, Endoh H, et al. Analysis of epidermal growth factor receptor gene mutation in patients with non-small cell lung cancer and acquired resistance to gefitinib. Clin Cancer Res 2006;12:5764-9. 26. Hata A, Katakami N, Yoshioka H, et al. Spatiotemporal T790M Heterogeneity in Individual Patients with

19

PD-L1 expression and T790M status

EGFR-Mutant Non-Small-Cell Lung Cancer after Acquired Resistance to EGFR-TKI. J Thorac Oncol 2015;10:1553-9. 27. Chen N, Fang W, Zhan J, et al. Upregulation of PD-L1 by EGFR Activation Mediates the Immune Escape in EGFR-Driven NSCLC: Implication for Optional Immune Targeted Therapy for NSCLC Patients with EGFR Mutation. J Thorac Oncol 2015;10:910-23. 28. Gainor JF, Sequist LV, Shaw AT, et al. EGFR Mutations and ALK Rearrangements Are Associated with Low Response Rates to PD-1 Pathway Blockade in Non-Small Cell Lung Cancer: A Retrospective Analysis. Clin Cancer Res 2016;22:4585-93. 29. Akbay EA, Koyama S, Carretero J, et al. Activation of the PD-1 pathway contributes to immune escape in EGFR-driven lung tumors. Cancer Discov 2013;3:1355-63. 30. Shukuya T, Carbone DP. Predictive Markers for the Efficacy of Anti-PD-1/PD-L1 Antibodies in Lung Cancer. J Thorac Oncol 2016;11:976-88. 31. Hirsch FR, McElhinny A, Stanforth D, et al. PD-L1 Immunohistochemistry Assays for Lung Cancer: Results from Phase 1 of the Blueprint PD-L1 IHC Assay Comparison Project. J Thorac Oncol 2017;12:208-22. 32. Pennell NA. PD-L1 Testing and Lack of Benefit to Guide Treatment With Immune Checkpoint Inhibitors in Patients With Non-Small-Cell Lung Cancer. JAMA Oncol 2016 [Epub ahead of print]

Figure legends Fig. 1. Flow chart of final investigated samples and patients. Fig. 2. Criteria, overlap, and prevalence of T790M+ and T790M- samples on TC/IC score. Fig. 3. PD-L1 expression according to T790M status by the 28-8 antibody: (A), comparison by wilcoxon rank sum test; and (B), positivity according to each PD-L1 cut-off. Fig. 4. PD-L1 expression of representative samples. Fig. 5. Overall survival according to PD-L1 expression (A), T790M status (B), and PD-L1/T790M status expression (C).

20

PD-L1 expression and T790M status

21

PD-L1 expression and T790M status

Fig. 3. PD-L1 expression according to T790M status by the 28-8 antibody: (A), comparison by wilcoxon rank sum test; and (B), positivity according to each PD-L1 cut-off. (A)

p=0.0801

T790M

(B)

PD-L1, programmed death-ligand 1.

22

PD-L1 expression and T790M status

Fig. 4. PD-L1 expression of representative samples. Sample A (T790M+) A1, TC0/IC0 by SP142 A2, 0% by 28-8

Sample B (T790M+) B1, TC2/IC0 by SP142

Sample C (T790M-) C1, TC3/IC3 by SP142

B2, 6% by 28-8

C2, 90% by 28-8

PD-L1, programmed death-ligand 1; TC, tumour cells; IC, tumour-infiltrating immune cells.

23

PD-L1 expression and T790M status

Fig. 5. Overall survival according to PD-L1 expression (A), T790M status (B), and PD-L1 expression/T790M status (C).

Probability

Median OS of PD-L1-: 82.5 months

Median OS of PD-L1+: 43.8 months

(A)

Time (month)

Probability

Median OS of T790M+: 78.3 months

Median OS of T790M-: 68.2 months

(B)

Time (month)

24

PD-L1 expression and T790M status

Probability

PD-L1-/T790M+

PD-L1-/T790M-

PD-L1+/T790MPD-L1+/T790M+

(C)

Time (month)

OS, overall survival; PD-L1, programmed death-ligand 1. Table 1 Dynamism of PD-L1 expression and T790M status affected by EGFR-TKI treatment. Case #1 #2 #3 #4 #5 #6 #7 #8 #9 #10 #11

Dynamism

Location

TKI

PD-L1/T790M at 1st Rebiopsy PD-L1+/T790M+

TK I

PD-L1/T790M at 2nd Rebiopsy PD-L1-/T790M+

TKI

PD-L1/T790M at 3rd Rebiopsy PD-L1+/T790MPD-L1-/T790M+

Primary On On Off Metastasis Primary PD-L1+/T790MPD-L1/T790M On On Off Metastasis PD-L1-/T790M+ PD-L1+/T790M+ Primary PD-L1-/T790M+ PD-L1+/T790MPD-L1/T790M On On Off Metastasis PD-L1-/T790M+ PD-L1/T790M Primary On PD-L1-/T790M+ Off PD-L1-/T790M+ Off PD-L1+/T790MPD-L1/T790M Primary On PD-L1+/T790M+ Off PD-L1-/T790MPD-L1 Primary On PD-L1-/T790MOff PD-L1+/T790MPD-L1 Primary On PD-L1-/T790MOff PD-L1+/T790MT790M Primary On PD-L1-/T790M+ Off PD-L1-/T790MOn PD-L1-/T790M+ T790M Primary On PD-L1-/T790M+ Off PD-L1-/T790MPrimary PD-L1+/T790MNone On Off On Metastasis PD-L1+/T790MPD-L1+/T790MNone Primary On PD-L1-/T790M+ On PD-L1-/T790M+ PD-L1, programmed death-ligand 1; EGFR-TKI, epidermal growth factor receptor-tyrosine kinase inhibitor. PD-L1/T790M

25

TKI Off

On

PD-L1/T790M at 4th Rebiopsy PD-L1+/T790M-

PD-L1-/T790M+