Efficacy and safety of sintilimab in combination with chemotherapy in previously untreated advanced or metastatic nonsquamous or squamous NSCLC: two cohorts of an open-label, phase 1b study

Combining chemotherapy with immunotherapy improves the therapeutic outcome for first-line (1L) patients with advance nonsmall-cell lung cancer (NSCLC). Two cohorts of a phase 1b study (NCT02937116) aimed to evaluate the safety and efficacy of sintilimab, a PD-1 inhibitor, plus chemotherapy in 1L patients with nonsquamous and squamous NSCLC (nsqNSCLC/sqNSCLC); and to identify potential biomarkers for treatment response. Treatment-naïve patients with nsqNSCLC were enrolled and intravenously given sintilimab (200 mg), pemetrexed (500 mg/m2), and cisplatin (75 mg/m2), every 3 weeks (Q3W) for 4 cycles in cohort D. Treatment-naïve patients with sqNSCLC were enrolled and intravenously given sintilimab (200 mg), gemcitabine (1250 mg/m2), and cisplatin (75 mg/m2), Q3W, for 6 cycles in cohort E. The primary objective was to evaluate the safety and efficacy of the treatment. The additional objective was to explore biomarkers for the treatment efficacy. Twenty-one patients with nsqNSCLC, and 20 patients with sqNSCLC were enrolled in cohort D and cohort E, respectively. By the data cutoff (April 17, 2019), 8 (38.1%) patients in cohort D and 17 (85.0%) patients in cohort E experienced grade 3–4 adverse events. The median follow-up duration was 16.4 months (14.8–23.0) in cohort D and 15.9 months (11.7–17.7) in cohort E. The objective response rate was 68.4% (95% CI 43.4%, 87.4%) in cohort D and 64.7% (95% CI 38.3%, 85.8%) in cohort E. Neither PD-L1 expression nor tumor mutation burden value was significantly associated with an improved treatment response. Sintilimab plus chemotherapy exhibited manageable toxicity and an encouraging antitumor activity in patients with nsqNSCLC and sqNSCLC. Electronic supplementary material The online version of this article (10.1007/s00262-020-02738-x) contains supplementary material, which is available to authorized users.


Introduction
Lung cancer is the leading cause of cancer-related mortality worldwide [1]. Nonsmall-cell lung cancer (NSCLC) represents 85% of the diagnosed lung cancer cases [2], and approximately 70% of the patients with NSCLC are nonsquamous at a diagnosis stage [3]. Platinum-based chemotherapy is the standard treatment for 1L advanced patients with NSCLC lacking targetable mutations [4,5]. However, the overall clinical outcomes are undesirable. The immunogenic properties of chemotherapies make it possible to combine chemotherapy with immunotherapy [6]. Pembrolizumab combining with platinum-based doublet chemotherapy (PT-DC) has shown an encouraging antitumor activity and manageable toxicity for 1L advanced NSCLC [5,7].
The biomarkers generated by the treatment response can be used to select patient. PD-L1 is a potential biomarker to predict the response to pembrolizumab monotherapy in 1L metastatic patients with NSCLC harboring PD-L1 tumor proportion score (TPS) ≥ 1% [8]. However, nivolumab showed an inconsistent result [9]. Regarding the combination therapies of PD-1 inhibitors (pembrolizumab or nivolumab) with chemotherapy in 1L setting for patients with NSCLC, there was not a clear association between treatment efficacy with PD-L1 expression [5][6][7]10].
Tumor mutation burden (TMB) has emerged as a novel biomarker to identify patients for immunotherapy [11]. In pembrolizumab or nivolumab monotherapy settings, patients with NSCLC and having a high TMB level showed an improved response and survival benefits [12,13]. Nevertheless, results are discrepant in the combination settings of these PD-1 inhibitors plus chemotherapy as 1L treatment [14,15].
T cell receptor (TCR) plays significant roles in antigen recognition with the main variable region of complementarity determining region 3 (CDR3) [16]. TCR diversity and clonality are suggested to indicate the clinical outcomes among immune checkpoint inhibitor (ICPi)-treated patients, but controversial results exist among different solid tumors [17,18]. In addition, studies are scarce in patients received the combination of a PD-1 inhibitor with chemotherapy.
Sintilimab is a highly selective anti-PD-1 antibody with a higher binding affinity to human PD-1 than pembrolizumab and nivolumab [19]. Our Phase 1a study indicated an acceptable safety profile without dose-limiting toxicities in solid tumors. The present phase 1b study described results from two cohorts on NSCLC, aiming to: (1) evaluate the safety and efficacy of the sintilimab plus chemotherapy (sintilimab-chemo) in previously untreated patients with nonsquamous or squamous NSCLC (nsqNSCLC/sqNSCLC) and (2) identify potential TMB or TCR biomarkers for this regimen. Previous results by the data cutoff date on Jan 15, 2019 was reported in 2019 ASCO [20]. Here, we updated the data at cutoff date on Apr 17, 2019 and performed biomarker analyses to provide more evidence for the combination therapy in 1L patients with nsqNSCLC/sqNSCLC.

Inclusion and exclusion criteria
Eligible patients were aged between 18 and 70 years; had histologically or cytologically confirmed nsqNSCLC (cohort D)/sqNSCLC (cohort E) that was unresectable, locally advanced, relapsed, or metastatic; at stage III or IV; and were previously untreated. The staging was based on the 7th edition of the TNM Classification of the International Association for Lung Cancer. Other major inclusion criteria were life expectancy of ≥ 12 weeks and at least one measurable lesion at baseline per Response Evaluation Criteria in Solid Tumors, version 1.1 (RECIST v1.1), Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1 and adequate organ function.
Patients with epidermal growth factor receptor (EGFR) mutation or anaplastic lymphoma kinase (ALK) rearrangement were excluded from the study. Detailed inclusion and exclusion criteria are available in the Supplementary Materials.
The study was approved by the independent ethics committee at each site and conducted following the Declaration of Helsinki and Good Clinical Practice guidelines. All patients provided written informed consent.

Study design
The cohort D/E were selected from a multi-center, openlabel, phase 1b study including six cohorts (NCT02937116). The D/E cohort was designed to preliminarily assess the efficacy and safety of sintilimab-chemo in patients with nsq/ sqNSCLC.
In cohort D, patients received up to 4 cycles of sintilimab (200 mg, IV) in combination with pemetrexed (500 mg/m 2 , IV, D1) and cisplatin (75 mg/m 2 , IV, D1) every 3 weeks. In cohort E, patients received up to 6 cycles of sintilimab in combination with gemcitabine (1250 mg/m 2 , IV, D1) and cisplatin (75 mg/m 2 , IV, D1) every 3 weeks. After combination therapy, patients received sintilimab maintenance therapy (sintilimab plus pemetrexed in cohort D, and sintilimab monotherapy in cohort E) as prior doses every 3 weeks for up to 24 months.

Endpoints and assessments
The responses to therapy were evaluated at baseline and every 9 weeks after initial dose, using computed tomography (CT) or magnetic resonance imaging (MRI) by the investigator per RECIST v1.1. For patients achieved an initial complete remission (CR) or partial remission (PR), the response was confirmed by radiographic evaluation after 4 weeks, and thereafter was assessed every 9 weeks until progression. Safety was assessed mainly by monitoring adverse events (AEs) throughout the trial. AEs were graded in severity based on the National Cancer Institute Common Terminology Criteria (NCICTC version 4.03) guidelines.
The primary endpoints were objective response rate (ORR), defined as the proportion of patients who achieved a CR or PR; and safety profile. The secondary endpoints included PFS, OS, disease control rate (DCR), duration of response (DOR) and time to response (TTR). The exploratory endpoint was the potential correlation between the biomarkers, such as PD-L1, TMB or TCR, and the clinical efficacy of the combination therapy in patients with nsq/ sqNSCLC.

PD-L1 immunohistochemical assay and scoring
PD-L1 immunohistochemical was detected with the Dako PD-L1 IHC 22C3 pharmDx, on the Autostainer Link 48 (clone 22C3, Dako, Carpinteria, CA). Details are provided in the supplementary materials. PD-L1 protein staining was determined by the TPS, which was calculated as the percentage of tumor cells staining with PD-L1 (0-100%). TPS ≥ 1% was defined as PD-L1 positive.

Tumor mutation burden analysis
The formalin-fixed paraffin-embedded tumor tissues and the matched peripheral blood samples were collected from patients at baseline. After DNA extraction and shearing into fragments, the DNA libraries were constructed with a designed gene panel. Then, the samples were performed with paired-end sequencing on the Illumina HiSeq X-Ten platform (Illumina, San Diego, USA). TMB was measured in mutations per one million coding bases (Mb), including somatic single-nucleotide variants and indel mutations in the whole exome. The detailed sequencing methods are provided in supplementary materials. Patients were divided into TMB-high (TMB-H) and TMB-low (TMB-L) groups based on the median TMB cutoff.

Analysis of TCR repertoire
Total RNA of total CD8 + T cells were extracted using the RNeasy Plus Mini Kit (Qiagen, Valencia, CA, USA) according to the manufacture's instruction. Then, RNA samples were performed with high-throughput sequencing of TCR using the ImmuHub ® TCR profiling system at a deep level (ImmuQuad Biotech, Hangzhou China), as described in the supplementary methods and our previous reports [21,22]. Shannon's index of diversity and clonality index were calculated using formulas described in the supplementary methods.

Statistical analyses
Twenty patients were planned to enroll in each cohort (D or E). Patients who received 1 or more drug dose were enrolled in the efficacy and safety analysis sets. ORR and DCR were estimated using the binomial distribution, and the two-sided 95% exact confidence intervals (CIs) were calculated by the Clopper-Pearson method. The time-to-event endpoints (median PFS, OS, DOR, TTR, and PFS and OS rates at 6 and 12 months) were assessed by Kaplan-Meier product-limit method. The survival curves were estimated by the log-rank test.
Student's t test (two-sided) was used to evaluate the TCR diversity and clonality index pre-and post-treatment, by GraphPad Prism version 6.0 (La Jolla, CA, USA). Other statistical analyses were conducted with the SAS software (version 9.2 or higher). P < 0.05 was the cutoff for significance.

TCR repertoire and efficacy
Owing to the small patient sample size, we merged patients in the two cohorts to perform the TCR correlation analysis. At the time of analysis, the patients who had progressed and out of the study were classified into PD group, and the remaining were into disease control (DC) group. Overall, 11 patients had evaluable TCR, 5 in DC group and 6 in PD group.
Patients in DC group showed an increased TCR Change (TCRC) clonality index (TCRC clonality ), namely the ratio of TCR clonality index post-and pre-treatment, when compared with those in PD group. However, the difference was not statistically significant (P = 0.089, Fig. 4a). The TCRC diversity index (TCRC diversity , the ratio of TCR diversity index post-and pre-treatment) in the DC group was significantly decreased when compared with the PD group (P = 0.014, Fig. 4b). The TCR clonality in the DC group was increased post-treatment when compared with pre-treatment, while that in PD group had an approximately stable change trend (Fig. 4c).
To better reflect the association between the treatment efficacy with the changing trend of TCR, we used the TCRC clonality or TCRC diversity of 1 as cutoff, and found a high TCRC clonality (> 1) was significantly associated with an improved OS (P = 0.007, Fig. 4e); in contrast, a low TCRC diversity (< 1) Fig. 4 Association of CD8 + T cell receptor (TCR) and treatment efficacy. a TCRC clonality (the ratio of TCR clonality index post-and pre-treatment) between disease control (DC) group and progressive disease (PD) group; b TCRC diversity (the ratio of TCR diversity index post-and pre-treatment) between DC and PD groups; c TCR clonality tracking within the treatment of every patient; d, e progression-free survival (PFS) and overall survival (OS) stratified by TCRC clonality (> 1 vs. < 1); F and G, PFS and OS stratified by TCRC diversity (> 1 vs. < 1) had an evident association with a prolonged PFS (P = 0.018, Fig. 4f).

Discussion
Due to the immunogenic properties of chemotherapy, immunotherapy in combination with chemotherapy (immunotherapy-chemo) approach has been widely studied. Our phase 1b study preliminarily suggested that sintilimab-chemo had a tolerable safety profile and may improve tumor response in 1L advanced patients with nsq/sqNSCLC.
In our study, most frequent AEs for the combination therapy in two cohorts were nonhematological, such as nausea (42.9%) which was comparable to historical data of pembrolizumab-or nivolumab-chemo (29-56%) in advanced patients with NSCLC [5,6,23]. Hematological events also occurred in our study, such as anemia (42.9% in cohort D and 85.0% in cohort E), which was similar to previous data of gemcitabine-cisplatin chemotherapy and pembrolizumab-chemo (45-47%) [5,24]. The occurrences of neutrophil count decreased (80.0% vs. 47.6%) and anemia (85.0% vs. 42.9%) were higher in cohort E (sintilimab plus cisplatin + gemcitabine) than that in cohort D (sintilimab plus cisplatin + pemetrexed). The different occurrences were possibly caused by different chemotherapy regimens. This phenomenon was observed in 1L nivolumab (10 mg/kg) plus the same chemotherapy regimens (17% vs. 0% and 50% vs. 13%, respectively) [6]. Rash was one of the most frequent grade 3 or 4 AEs in cohort D (no more than 10%), consistent with historical data of pembrolizumab (1.7%) [5]. Lung infection was also the frequent AE of grade 3 or 4 (10%), but it was manageable and did not cause any death. In cohort E, the most common AEs were decreased WBC count and decreased neutrophil count (less than 50%), which were also comparable to nivolumab-and pembrolizumab-chemo (less than 20%) settings [5,6].
In this study, most sintilimab-related AEs were grade 1-2 in severity, such as ALT increased, fatigue, rash, and hypothyroidism, which were also previously reported relating to nivolumab [6]. Except for one patient showing grade 3 immune-related AEs (rash), most immune-related AEs were grade 1-2, mainly affecting the skin and pulmonary organs. No AEs of grade 5 were reported in our study, indicating a tolerable safety profile of this combination strategy.
Combination therapy of PD-1 inhibitors and chemotherapy could enhance the anti-tumor activity of chemotherapy and improve the treatment response. In 1L patients with nsqNSCLC receiving pembrolizumab-chemo (carboplatin and pemetrexed), the ORR was 75% (18/24) [23]. In 1L nivolumab-chemo (PT-DC) treated patients with NSCLC, the confirmed ORR ranged from 33 to 47% across different dose groups [6]. Sintilimab-chemo in our study showed an increased response than nivolumab-chemo, with an ORR of 68.4% (cohort D) or 64.7% (cohort E). Regarding the association between PD-L1 expression and response, the patients with NSCLC had similar ORRs to pembrolizumabchemo among different PD-L1 TPS groups [23,25]. In addition, in nivolumab-chemo treated advanced patients with NSCLC, the confirmed ORR was not affected by different expression levels of PD-L1 [6]. Notably, in a phase 3 study, pembrolizumab-chemo showed a superior response benefit over chemotherapy in metastatic patients with nsqNSCLC, regardless of the tumor PD-L1 expression [5]. Consistent with these preliminary results, our study indicated no significant correlations between PD-L1 expression and responses to sintilimab-chemo in both patients with nsqNSCLC and sqNSCLC.
The immunotherapy-chemo strategy also revealed an improved long-term survival benefit. The 1L nivolumabchemo (PT-DC) had a promising survival benefit for advanced patients with NSCLC, with a median PFS ranging from 4.8 to 7.1 months and a median OS from 11.6 to 19.2 months, both of which were longer than previous data of PT-DC alone [6]. Pembrolizumab-chemo (pemetrexed + cisplatin/carboplatin) significantly prolonged the median PFS when compared with chemotherapy alone for previously untreated metastatic patients with nsqNSCLC lacking EGFR or ALK mutations (8.8 m vs. 4.9 m) [5]. Our results showed sintilimab-chemo had an improved longterm survival outcome with an estimated median PFS of 12.6 months in patients with nsqNSCLC and 6.5 months in patients with sqNSCLC. Notably, the 12-month OS rate (70.0% or 62.0%) in this study was similar to the historical data of pembrolizumab-or nivolumab-chemo (69.2% or 50-87%, respectively) [5,6]. However, it could not make an assertive conclusion due to the immature data.
A clonal expansion of the neoantigen-specific T cells is expected after the response to anti-PD-1/PD-L1 regimens [16]. In multiple cancers, increased TCR clonality after ICPs is associated with an improved treatment efficacy [26]. The TCR-Vβ underwent expansions after neoadjuvant immunotherapy with atezolizumab in NSCLC [27]. Increased CD4 + TCR repertoire clonality was correlated with a high density of tertiary lymphoid structure B cells, which was a biomarker of a higher OS in NSCLC [28]. Before treatment, high TCR diversity indicates a better immune status, with the mechanism that high diversity could preclude the magnitude of immune escape via increasing more potential tumor-specific T cells, which can control tumor cell growth and recognize corresponding antigens [17]. In patients with NSCLC, those with an increased peripheral PD-1 + CD8 + (double-positive PD-1 and CD8) clonality after ICPi treatment, and with a high PD-1 + CD8 + diversity pre-ICPi exhibited a better response and a longer PFS, as compared to those with low clonality and diversity [16]. Consistent with this finding, in our study, the higher TCRC clonality (which reflected an increased TCR clonality post-treatment) or lower TCRC diversity (indicating a high TCR diversity pretreatment) had an evident association with a prolonged OS or PFS benefit from the sintilimab-chemo treatment. Nevertheless, further evidence is required to support the potential use of these indexes as effective predictors for the combination strategy of PD-1 inhibitor and chemotherapy.
It was suggested that mutational landscape of NSCLCs might affect the response to anti-PD-1 therapy, and TMB might be a promising biomarker for selecting appropriate patients [12,29]. A high TMB is commonly considered to promote neoantigens formation, and the most mutated tumors are likely to be the most immunogenic ones [30]. Although FDA has accepted a priority review to a supplemental Biologics License Application for pembrolizumab for the 1L treatment of solid tumors among TMB-H (≥ 10 mut/Mb) populations based on Keynote 158 [31], TMB is not the standard biomarker for predicting the efficacy of PD-1/PD-L1 inhibitors. Moreover, the positive correlation between high TMB and the improved response of PD-1 inhibitors was generally acknowledged in the monotherapy [13,32]. Meanwhile, the prediction of TMB on response to ICPi is inconclusive since the mutation ranges are overlapped between responders and nonresponders [30]. In patients treated with immunotherapy-chemo, inconsistent results also existed on the relationship between TMB and the treatment efficacy. A high TMB (≥ 10 mut/Mb) was associated with a prolonged PFS in 1L nivolumab-chemo-treated patients with advanced NSCLC [14], whereas a high tissue TMB (≥ 175 mut/exome) did not present a significant correlation with the efficacy in 1L pembrolizumab-chemo-treated patients with metastatic nsq/sqNSCLC [15]. In our study, in both nsqNSCLC and sqNSCLC cohorts, patients with a high TMB (≥ 4.25) did not show a significantly better response to sintilimab-chemo.
Despite the promising findings, there were some limitations. The antitumor activity of sintilimab-chemo was reported in a single-arm phase 1 study with a small sample size, and should be proven in large populations. Besides, the biomarker results, such as PD-L1, TMB, and TCR also require further confirmation because of the small patient samples. Meanwhile, very few patients had evaluable TCR, so the interpretations about TCR need to be cautious. Nonetheless, our study provides preliminary evidence for sintilimab-chemo as 1L treatment in patients with advanced nsq/ sqNSCLC. Two phase 3 studies are currently ongoing to evaluate the combination therapy in 1L patients with nsqN-SCLC (NCT03607539) and sqNSCLC (NCT03629925), respectively.
In conclusion, sintilimab in combination with pemetrexed-cisplatin or with gemcitabine-cisplatin showed manageable toxicity and an encouraging antitumor activity in patients with nsqNSCLC and sqNSCLC, regardless of PD-L1 expression or TMB level. A phase 3 study investigating sintilimab-chemo as 1L treatment in patients with NSCLC, irrespective of PD-L1 expression, is currently ongoing and has achieved endpoint in the interim analysis.