Impact of EGFR-TKI Treatment on the Tumor Immune Microenvironment in EGFR Mutation–Positive Non–Small Cell Lung Cancer

Lung cancer is the leading cause of cancer-related death worldwide (1). Recent advances in the development of molecularly targeted drugs such as EGFR tyrosine kinase inhibitors (TKI) and immune-checkpoint inhibitors (ICI), including programmed cell death-1 (PD-1) and programmed cell death–ligand 1 (PD-L1) inhibitors, have greatly improved therapeutic outcome in non–small cell lung cancer (NSCLC).

The tumor immune microenvironment (TME) as characterized by indicators such as the expression of PD-L1 in tumor cells and the tumor nonsynonymous mutation burden (TMB) is an emerging predictor of ICI response in NSCLC (2, 3). ICIs were recently found to be of limited benefit for patients with NSCLC harboring EGFR mutations (4, 5). Furthermore, recent studies have shown that EGFR mutation–positive NSCLC has a low TMB (6) and a high frequency of inactive tumor-infiltrating lymphocytes (TIL; ref. 7). Nevertheless, our previous study suggested that patients with EGFR mutation–positive tumors might receive benefit from PD-1 inhibitor treatment after acquiring resistance to conventional EGFR-TKIs if they are negative for the T790M TKI resistance–conferring mutation of EGFR and have a favorable TME for ICI therapy such as a high PD-L1 expression level and a high TMB (8). Although changes in the TME induced by EGFR-TKI treatment are likely to be an important determinant of the efficacy of subsequent ICI therapy, only a few previous studies have focused on such changes (5, 9).

We have now evaluated changes in the TME for patients with EGFR mutation–positive NSCLC between before and after the development of resistance to EGFR-TKI treatment.

We reviewed the medical records of all patients with EGFR mutation–positive advanced or recurrent NSCLC treated at Kindai University Hospital, Kanagawa Cardiovascular and Respiratory Center, Kishiwada City Hospital, and Kurashiki Central Hospital between February 2012 and May 2018. EGFR mutations were detected with commercial assays such as the cobas EGFR Mutation Test (Roche Molecular Diagnostics). The current study was performed according to the Declaration of Helsinki and protocols approved by the Institutional Review Board (IRB) of each hospital. All patients provided written informed consent, where applicable, or such informed consent was waived by IRB-approved protocols for aggregate deidentified data analysis. We identified 138 patients who had baseline tissue specimens available and who underwent rebiopsy after disease progression during EGFR-TKI therapy including that with osimertinib. Patients who were diagnosed on the basis of cytology only or for whom the amount of residual tumor tissue specimens was insufficient for analysis were excluded (Supplementary Figure S1).

Data regarding clinicopathologic features and treatment history were extracted from medical records. Data were updated as of 31 May, 2018. Responses were assessed according to RECIST v1.1. Progression-free survival (PFS) was measured from treatment initiation to clinical or radiographic progression or death from any cause. Overall survival (OS) was measured from treatment initiation to death from any cause. Patients without documented clinical or radiographic disease progression or who were still alive were censored on the date of last follow-up.

Tumor histology was classified according to WHO criteria. IHC was performed with mAbs to PD-L1 (kit with clone 22C3 and clone 28–8, Agilent Technologies), to CD8 (clone C8/144B, Agilent Technologies), to FOXP3 (clone 236A/E7, Abcam), and to CD73 (clone D7F9A, Cell Signaling Technology) and with the use of an automated stainer (Dako). The stained slides were evaluated by board-certified pathologists who were blinded to clinical outcome.

PD-L1 immunostaining was optimized with human placenta and tonsil as positive controls. We defined PD-L1 positivity as membranous staining of tumor cells at any intensity with cutoffs of ≥1% and ≥50% (<1% defined as negative, 1 to 49% as low positive, and ≥50% as strong positive) of tumor cells in a section that included at least 100 evaluable tumor cells (10). The percentage of PD-L1–positive tumor-infiltrating immune cells occupying the tumor was also recorded with cutoffs of ≥1%, ≥5%, and ≥10% (11, 12). Tumor CD73 expression was graded on the basis of a semiquantitative H-score, which was calculated by multiplying the cytomembrane staining intensity (0, absent; 1, weak; 2, moderate; 3, strong) by the percentage of stained cells (0%–100%) to yield a value of 0 to 300 (13).

The number of TILs was evaluated at an absolute magnification of ×200 (0.0625 mm²/field). At least one and a maximum of five scanned fields of tumor regions were randomly chosen for each TIL count. TILs were counted by one pathologist, and the density of TILs in the tumor was calculated by dividing the number of TILs by the sum of the area (mm²) of the viewed fields (8, 13). TILs were defined as cells positive for CD8, FOXP3, or CD73 regardless of staining intensity. Cytology specimens and fragmented biopsy specimens were excluded from quantitative TIL assessment because of the lack of an adequate stromal component for analysis. For TIL analysis, all fields were evaluated as intratumoral regions.

DNA was extracted from formalin-fixed paraffin-embedded tumor specimens with the use of a QIAamp DNA Micro Kit (Qiagen). Samples for which the concentration of extracted DNA was <20 ng/mL were excluded. Whole-exome capture libraries were constructed with the use of Agilent Sure-Select Human All Exon v5.0. Samples with a library concentration of >20 ng/μL were used for further analysis. Enriched exome libraries were sequenced with the HiSeq 2500 platform (Illumina), yielding an average of 115 million reads (11.6 Gb). Nonsynonymous mutations were identified with the CLC whole-exome pipeline (ver. CLC 5.0) followed by additional annotation for SIFT, polyphen2, and GERP as described previously (8). Samples with >20% C→T transitions, potentially resulting from DNA damage, were excluded from further analysis. For comparison with previous findings (14), TMB was calculated as the total number of nonsynonymous single-nucleotide or insertion–deletion mutations divided by the coding region captured in the latest version of the Memorial Sloan Kettering–Integrated Mutation Profiling of Actionable Cancer Targets (MSK-IMPACT) panel of 468 genes (1.22 Mb; ref. 9). Only those bases with sufficient coverage in the sequencing run and with an allele frequency of ≥10% were included in the calculation of TMB.

Fisher exact test and the Mann–Whitney U test were applied to compare categorical or continuous variables, respectively. Bowker test of symmetry was used for categorical comparisons of PD-L1 expression. Paired data were compared with the Wilcoxon signed rank test. Differences in PFS or OS curves constructed by the Kaplan–Meier method were assessed with the log-rank test, and the Cox proportional hazard regression model was adopted to determine HRs. All P values were based on a two-sided hypothesis, and a P value of <0.05 was considered statistically significant. Given the exploratory nature of the study, multiplicity adjustment was not performed. Statistical analysis was performed with JMP software version 10.0.2 (SAS Institute, Cary, NC).

A total of 138 patients with EGFR mutation–positive NSCLC who had been treated with EGFR-TKIs and who had both baseline and postprogression tumor samples available was identified (Supplementary Figure S1). Of these 138 patients, paired analysis of pre–EGFR-TKI treatment and postprogression samples was possible for 134 individuals in the case of PD-L1 expression and CD8⁺ TILs, 133 individuals in the case of FOXP3⁺ TILs, 70 individuals in the case of CD73⁺ TILs, and 15 individuals in the case of TMB. Clinicopathologic characteristics of the patients included in the study are shown in Table 1. A total of 28 patients (20%) received the ICIs nivolumab or pembrolizumab intravenously after the development of EGFR-TKI resistance.

We first investigated changes in PD-L1 expression in tumor cells between before EGFR-TKI treatment and the development of drug resistance. PD-L1 expression was significantly changed after EGFR-TKI treatment (n = 134, P = 0.0007; Fig. 1A), and the proportion of patients with PD-L1 strong positive tumors increased from 14% at baseline to 28% after EGFR-TKI treatment (n = 134, P = 0.0010; Fig. 1B; Supplementary Table S1). A similar significant increase in the proportion of PD-L1 strong positive tumors was apparent for individuals negative for the T790M TKI resistance–conferring mutation at progression (n = 70, P = 0.0125) but not for those positive for this mutation (n = 64, P = 0.0620; Fig. 1C). In addition, the proportion of PD-L1 strong positive tumors postprogression tended to be greater for T790M-negative patients than for T790M-positive patients (P = 0.0565; Supplementary Table S1). The proportion of PD-L1 strong positive tumors was also significantly increased or tended to be increased after EGFR-TKI treatment regardless of smoking status or the type of EGFR-activating mutation (Supplementary Fig. S2A and S2B). We also investigated the change in PD-L1 expression by tumor-infiltrating immune cells between baseline and the development of EGFR-TKI resistance, which was evaluable in 101 cases. Although no significant change was detected (n = 101, P = 0.1393), the proportion of patients with a PD-L1 expression level in tumor-infiltrating immune cells of ≥10% increased from 11% at baseline to 25% after EGFR-TKI treatment (Supplementary Fig. S2C).

Twenty-seven patients with paired samples assayed for PD-L1 expression in tumor cells were subsequently treated with antibodies to PD-1. We evaluated PFS for ICI treatment according to PD-L1 expression in tumor cells in both pre– and post–EGFR-TKI treatment samples to examine which result was more related to ICI efficacy. Consistent with our previous findings (8), the median PFS was longer for patients with PD-L1 strong positive tumors than for those with a PD-L1 expression level of <50% (negative or low positive) with regard to the post–EGFR-TKI treatment samples (Fig. 1D). A significantly better PFS was also apparent for patients with PD-L1 strong positive tumors with regard to the pre–EGFR-TKI treatment samples (Supplementary Fig. S2D). A similar trend was observed for OS relative to the initiation of ICI treatment (Supplementary Fig. S3A and S3B). Two of five patients whose PD-L1 expression level changed from negative or low positive at baseline to strong positive after the development of EGFR-TKI resistance showed a PFS for ICI treatment of >6 months (Fig. 1E). We also found that patients negative for T790M and strongly positive for PD-L1 expression in post–EGFR-TKI treatment samples showed the most favorable PFS and OS outcomes (Supplementary Fig. S3C and S3D). In addition, ICI treatment showed a durable response with regard to PFS in patients with a PD-L1 expression level in tumor-infiltrating immune cells in the post–EGFR-TKI treatment sample of ≥10% (Supplementary Fig. S4A and S4B), consistent with previous findings (12). Although recent studies have suggested that EGFR-mutated tumors differ in their response to ICI treatment according to the type of mutation (15, 16), we detected no significant difference in PFS or OS between patients with exon-19 deletions and those with the L858R point mutation (Supplementary Fig. S5A and S5B).

We next evaluated CD8⁺ and FOXP3⁺ TIL densities with paired tissue samples obtained before and after the development of EGFR-TKI resistance. The densities of both CD8⁺ and FOXP3⁺ TILs were significantly lower after EGFR-TKI treatment compared with before (median of 292.8→224.0/mm², P = 0.0274, and of 249.6→150.4/mm², P < 0.0001, respectively; Fig. 2A and B; Supplementary Table S1).

In the case of specimens obtained after EGFR-TKI treatment, CD8⁺ TIL density was significantly higher in PD-L1 strong positive tumors than in PD-L1 negative or low positive tumors (median of 337.6 vs. 172.8/mm², P < 0.0001), whereas there was no such difference between the corresponding pretreatment samples (Fig. 2C; Supplementary Table S2). In the case of both pre– and post–EGFR-TKI treatment samples, FOXP3⁺ TIL density was significantly higher in PD-L1 strong positive tumors than in PD-L1 negative or low positive tumors (median of 403.2 vs. 203.2/mm², P = 0.0015, and of 316.8 vs. 89.6/mm², P = 0.0001, respectively; Fig. 2D; Supplementary Table S2).

With regard to the change in TIL density according to PD-L1 expression status, both CD8⁺ and FOXP3⁺ TIL counts were significantly lower after EGFR-TKI treatment than before in PD-L1 negative or low positive tumors (median of 281.6→172.8/mm², P = 0.0016, and of 203.2→89.6/mm², P < 0.0001, respectively), whereas CD8⁺ TIL density essentially did not change (median of 312.0→337.6/mm², P = 0.4014) and FOXP3⁺ TIL density also showed a nonsignificant change (median of 403.2→316.8/mm², P = 0.1544) between before and after EGFR-TKI treatment in PD-L1 strong positive tumors (Fig. 2C and D; Supplementary Table S2). Together, these findings suggested that any improvement in the TME in terms of CD8⁺ cell infiltration induced by EGFR-TKI treatment may occur only in PD-L1 strong positive tumors, despite the higher FOXP3⁺ TIL density in such tumors both before and after treatment compared with PD-L1 negative or low positive tumors.

We next evaluated CD73 expression in both tumor cells and TILs to explore why ICI efficacy is unsatisfactory in some patients with a favorable TME as characterized by a high PD-L1 expression level with CD8⁺ cell infiltration. CD73 is a mediator of immune suppression (17), and its expression in tumor cells was found to be high in EGFR mutation–positive NSCLC (18). CD73⁺ TIL density was significantly higher in tumors with a high FOXP3⁺ TIL density (with a cutoff of the median value for all study patients) than in those with a low FOXP3⁺ TIL density after EGFR-TKI treatment (median of 9.6 vs. 3.2/mm², P = 0.0318; Supplementary Fig. S6), consistent with a previous finding that CD73 is expressed on Treg cells (19). Unexpectedly, CD73⁺ TIL density was not significantly changed after EGFR-TKI treatment compared with baseline (median of 9.6→6.4/mm², P = 0.0653; Fig. 2E; Supplementary Table S1). Also, evaluation of CD73 expression in tumor cells by determination of H-score values revealed no significant change between before and after EGFR-TKI treatment (median of 120→90, P = 0.9012; Fig. 2F). Of note, however, expression of CD73 in tumor cells at the time of disease progression during EGFR-TKI treatment was significantly higher for PD-L1 strong positive tumors than for PD-L1 negative or low positive tumors (median of 270 vs. 40, P = 0.0004). In addition, a significant increase in CD73 expression was observed after EGFR-TKI treatment compared with baseline in PD-L1 strong positive tumor (median of 140→270, P = 0.0155; Fig. 2G; Supplementary Table S2).

Representative IHC data and CT images for a patient who showed marked changes in the TME including an increase in PD-L1 expression level in tumor cells (5%–85%) after the development of EGFR-TKI resistance but who showed an unsatisfactory response to subsequent pembrolizumab treatment (PFS of 1.9 months) are shown in Fig. 3. The densities of CD8⁺ and FOXP3⁺ as well as the expression of CD73 in tumor cells changed from negative or low at baseline to high after the development of EGFR-TKI resistance, suggestive of possible immunosuppressive effects of regulatory T (Treg) cells and CD73 expression in tumor cells.

We determined TMB for paired tissue samples collected before EGFR-TKI treatment and at the time of resistance development. TMB tended to be higher after EGFR-TKI treatment than before for the 15 patients with paired samples available (median of 3.3→4.1 mutations/Mbp, P = 0.0508; Fig. 4A). Given that we found that, among patients with PD-L1 strong positive tumors after EGFR-TKI treatment, those who were T790M negative at the time of resistance development showed a more durable PFS response during subsequent ICI treatment than did those who were T790M positive (Supplementary Fig. S3C), we speculated that this difference in PFS might be attributable to a difference in TMB. However, we did not detect a significant difference in TMB according to T790M status at the time of EGFR-TKI resistance development for tumor samples obtained either before (median of 3.1 vs. 3.2 mutations/Mbp for T790M positive or negative, respectively, n = 17, P = 0.4430; Fig. 4B) or after (median of 4.1 vs. 4.1 mutations/Mbp, n = 41, P = 0.3866; Fig. 4C) EGFR-TKI treatment.

Finally, we investigated nine patients who underwent rebiopsy after progression on osimertinib (Table 2). Among the eight individuals for whom paired samples from before and after osimertinib treatment were evaluable, PD-L1 expression was increased in two (25%). No trend for changes in CD8⁺, FOXP3⁺, and CD73⁺ TIL densities between before and after osimertinib treatment were apparent, whereas five of the seven patients for whom TMB was evaluated in paired pre- and posttreatment samples showed a decrease in this parameter.

We have here retrospectively assessed whether EGFR-TKI therapy might improve the TME of EGFR mutation–positive NSCLC by analysis of paired tissue samples collected before and after such treatment. As far as we aware, our study is the first to evaluate the impact of EGFR-TKI therapy on the TME by comprehensive analysis of PD-L1 expression, TMB, as well as CD8⁺, FOXP3⁺, and CD73⁺ TIL densities in such a relatively large number of patients. We found that targeted therapy was associated with a significant increase in PD-L1 expression in tumor cells in postprogression tumor samples compared with those obtained at baseline, especially in the case of T790M-negative patients, consistent with previous findings that T790M-negative status was associated with higher PD-L1 expression (8, 20). In addition, our results suggested that EGFR-TKI treatment also might lead to an increase in PD-L1 expression in immune cells. Furthermore, patients who showed an increase in PD-L1 expression in tumor cells tended to manifest a durable response to subsequent ICI treatment, suggestive of a favorable impact of EGFR-TKI therapy on the TME with regard to ICI efficacy.

Little information has been available regarding whether PD-L1 expression changes in association with the development of EGFR-TKI resistance in patients with EGFR-mutated NSCLC. Studies have suggested that inhibition of the EGFR–MEK–ERK signaling pathway results in increased tumor cell expression of MHC class I molecules (MHC-I), which are important for the presentation of antigens to CD8⁺ cytotoxic T cells (21, 22). Such upregulation of MHC-I might thus be expected to increase the availability of recognizable antigens and to promote the antitumor activity of CD8⁺ TILs. Although there was a significant decrease in CD8⁺ TIL density between before and after EGFR-TKI treatment in our study patients, CD8⁺ TIL density was significantly higher in PD-L1 strong positive tumors than in PD-L1 negative or low positive tumors after EGFR-TKI treatment, consistent with the previous finding that PD-L1 expression in cancer cells was induced by IFNγ released by CD8⁺ TILs (23).

FOXP3⁺ TIL density was increased in PD-L1 strong positive tumor specimens before and after EGFR-TKI treatment compared with corresponding PD-L1 negative or low positive tumor specimens. Although antibodies to PD-1 have shown efficacy in patients with EGFR mutation–positive NSCLC, especially in those with tumors strongly positive for PD-L1 (median PFS of 7.1 months in the current study), the results are unsatisfactory compared with the efficacy of these agents in patients with tumors wild type for EGFR (24). Immunosuppression mediated via the adaptive induction of Treg cells might contribute to this lower efficacy of PD-1 blockade therapy in EGFR-mutated NSCLC, even that strongly positive for PD-L1. Live and apoptotic Treg cells express the ectoenzymes CD39 and CD73 that convert extracellular ATP to adenosine (19), the latter of which mediates immunosuppression via interaction with the A2A receptor (17). Actually, a significant association between FOXP3 and CD73 expression in TILs was observed, but we did not detect a significant difference in the density of CD73⁺ TILs between before and after progression with EGFR-TKI treatment. Interestingly, however, we observed a higher CD73 expression in PD-L1 strong positive tumor cells after EGFR-TKI treatment compared with baseline. Tumor-derived CD73 was previously shown to suppress T cell and NK cell–mediated antitumor immunity in preclinical models (25). Therefore, the expression of CD73 in tumor cells might contribute in part to this insufficient efficacy of ICIs. The relevance of CD73 expression in both EGFR-mutated tumor cells and TILs to the efficacy of PD-1 blockade therapy remains unclear, with further investigation of the conversion of ATP to adenosine by CD39 and CD73 in EGFR mutation–positive NSCLC being warranted. Moreover, given that several clinical trials of antibodies that target CD73 or the A2A receptor are ongoing in patients with cancer (26), such agents are also candidates for combination therapy with ICIs that target the PD-1–PD-L1 system in selected populations such as patients with EGFR-mutated NSCLC strongly positive for PD-L1.

NSCLC harboring EGFR mutations has been shown to have a lower TMB compared with tumors without oncogenic drivers (6), but TMB was recently found to be significantly increased after EGFR-TKI treatment (9). Another recent study of a large cohort of patients with EGFR-mutated NSCLC with longitudinal cell-free DNA samples and clinical context showed that EGFR-TKI treatment resulted in enrichment of cooccurring genetic alterations (27). We have now also shown with paired tissue samples that TMB tended to be higher at the time of development of EGFR-TKI resistance compared with baseline, consistent with these previous findings. However, we were not able to investigate the relation between TMB and the efficacy of antibodies to PD-1 because of the lack of a sufficient number of tissue samples from treated cases. On the other hand, we did not detect a significant difference in TMB with regard to T790M status that was shown in previous study (9). As the number of patients was small in both studies, however, further investigation with a large number of patients is necessary to resolve this issue.

There are several limitations to our study. First, it was retrospective in nature. Second, a substantial number of patients received cytotoxic chemotherapy between the termination of EGFR-TKI treatment and rebiopsy. Third, despite the recent adoption of osimertinib in the first-line setting (28, 29), we focused mainly on those patients initially treated with conventional EGFR-TKIs. Further studies with more patients are needed to clarify the change in the TME between before and after osimertinib treatment. Another important limitation of our study relates to the heterogeneity of PD-L1 expression. Previous evaluations of PD-L1 expression in primary tumors and their metastases have shown that PD-L1 expression varies among tumor sites in the majority of cases (30–32). In the current study, pre- and post-TKI samples for evaluation of PD-L1 expression in tumor cells were obtained from the same anatomic lesions in only 65 of the 134 cases (49%). In addition, tissue samples for IHC analysis of PD-L1 are intrinsically nonhomogeneous (30), with the result that any change in PD-L1 expression observed might have been due to such heterogeneity.

In conclusion, we have shown that there were significant changes in the TME, including PD-L1 expression in tumor cells and TIL densities, between baseline and the time of development of EGFR-TKI resistance. The observed increase in PD-L1 expression suggests that EGFR-TKI therapy might confer a promising TME for subsequent ICI treatment. Nevertheless, unsatisfactory results with PD-1 inhibitors were obtained even in patients with tumors that express PD-L1 at a high level, possibly as a result of immunosuppressive factors such as an increase in Treg cell density and CD73 expression in tumor cells. Future investigations of the efficacy of combination immune therapy to overcome such immunosuppressive factors are warranted.

K. Haratani reports receiving commercial research grants from AstraZeneca K.K., and reports receiving speakers bureau honoraria from AstraZeneca K.K., Ono Pharmaceutical Co. Ltd., AS ONE Corp., Bristol-Myers Squibb Co. Ltd., Chugai Pharmaceutical Co. Ltd., MSD K.K., and Pfizer Japan Inc. H. Hayashi reports receiving commercial research grants from AstraZeneca, Boehringer Ingelheim Japan Inc., and Ono Pharmaceutical Co. Ltd., and reports receiving speakers bureau honoraria from AstraZeneca K.K., Ono Pharmaceutical Co. Ltd., Boehringer Ingelheim Japan Inc., Bristol-Myers Squibb Co. Ltd., Chugai Pharmaceutical Co. Ltd., Eli Lilly Japan K.K., MSD K.K., Pfizer Japan Inc., Shanghai Haihe Biopharma, and Taiho Pharmaceutical Co. Ltd. T. Yokoyama reports receiving speakers bureau honoraria from AstraZeneca, Chugai Pharmaceutical, Boehringer Ingelheim, and Ono Pharmaceutical. R. Kato reports receiving speakers bureau honoraria from Bristol-Myers Squibb Co. Ltd. J. Tanizaki reports receiving speakers bureau honoraria from Boehringer-Ingelheim Japan Inc., Eli Lilly Japan K.K., Bristol-Myers Squibb Co. Ltd., AstraZeneca K.K., MSD K.K., and Taiho Pharmaceutical Co. Ltd. K. Tanaka reports receiving speakers bureau honoraria from AstraZeneca, Merck Biopharma, and Bristol-Myers Squibb. M. Takeda reports receiving speakers bureau honoraria from Ono Pharmaceutical Co. Ltd., Chugai Pharmaceutical Co., Ltd., Novartis Pharma K.K., and Bristol-Myers Squibb Co. T. Ogura reports receiving commercial research grants from Boehringer Ingelheim Japan Inc., and The Mental Health Okamoto Memorial Foundation, and reports receiving speakers bureau honoraria from Boehringer Ingelheim Japan Inc. and Shionogi & Co. Ltd. K. Nakagawa reports receiving other commercial research support from MSD K.K., ICON Japan K.K., Takeda Pharmaceutical Co., Ltd., Eli Lilly Japan K.K., Bristol-Myers Squibb Company, Taiho Pharmaceutical Co. Ltd., PAREXEL International Corp., Ono Pharmaceutical Co. Ltd., IQIVIA Services Japan K.K., A2 Healthcare Corp., AbbVie, Chugai Pharmaceutical Co. Ltd., SymBio Pharmaceuticals Ltd., Merck Serono Co., EP-CRSU Co., Linical Co., Otsuka Pharmaceutical Co., EPS International Co., AstraZeneca K.K., Astellas Pharma, Novartis, Quintiles, CMIC Shift Zero, Eisai Co., Nippon Boehringer Ingelheim, Kissei Pharmaceutical, Pfizer Japan, Kyowa Hakko Kirin Co., EPS Corporation, Daiichi Sankyo Co., Bayer Yakuhin, inVentiv Health Japan, GRITSTONE ONCOLOGY Inc., Yakult Honsha Co., and Covance, and reports receiving speakers bureau honoraria from Astellas, AstraZeneca, MSD, Ono Pharmaceutical, Daiichi Sankyo, Taiho, Bristol-Myers Squibb, KYORIN, CareNet, Nichi-Iko, Hisamitsu, YODOSHA, Clinical Trial Co., MEDICUS SHUPPAN Publishers Co., Takeda, Chugai, Eli Lilly Japan, Nippon Boehringer Ingelheim, Novartis, Pfizer Japan, Nikkei Business Publications, Thermo Fisher Scientific, NANZANDO Co., Medical Review Co., YOMIURI TELECASTING CORPORATION, Renal Medical K.K., SymBio Pharmaceuticals Ltd., and AYUMI Pharmaceutical Corp. No potential conflicts of interest were disclosed by the other authors.

Conception and design: K. Isomoto, K. Haratani, H. Hayashi, K. Nakagawa

Development of methodology: K. Isomoto, K. Haratani, H. Hayashi, T. Niwa

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): K. Isomoto, K. Haratani, H. Hayashi, T. Niwa, T. Yokoyama, Y. Fukuda, R. Kato, K. Tanaka, M. Takeda, T. Ogura, T. Ishida

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K. Isomoto, K. Haratani, H. Hayashi, S. Shimizu, S. Tomida, Y. Chiba, R. Kato, J. Tanizaki, T. Ogura, K. Nakagawa

Writing, review, and/or revision of the manuscript: K. Isomoto, K. Haratani, H. Hayashi, R. Kato, K. Tanaka, T. Ogura

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): K. Isomoto, K. Haratani, H. Hayashi, T. Yokoyama, Y. Fukuda, J. Tanizaki

Study supervision: K. Haratani, H. Hayashi, A. Ito, K. Nakagawa

The authors thank Ai Yukumoto, Mami Kitano, Yume Shinkai, Michiko Kitano, and Haruka Sakamoto of Kindai University as well as Emi Honda and Misaki Sugiyama of Kanagawa Cardiovascular and Respiratory Center for their technical support. This study was supported by AstraZeneca K.K.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.