Skip to main content
PMC home page
Annals of Oncology logoLink to Annals of Oncology
. 2019 Jan 8;30(3):424–430. doi: 10.1093/annonc/mdy550

Epigenomic promoter alterations predict for benefit from immune checkpoint inhibition in metastatic gastric cancer

R Sundar 1,2,3,#, K K Huang 2,#, A Qamra 2, K -M Kim 4, S T Kim 5, W K Kang 5, A L K Tan 2, J Lee 5,✉, P Tan 2,6,7,8,✉
PMCID: PMC6442650  PMID: 30624548

Abstract

Background

Utilization of alternative transcription start sites through alterations in epigenetic promoter regions causes reduced expression of immunogenic N-terminal peptides, which may facilitate immune evasion in early gastric cancer. We hypothesized that tumors with high alternate promoter utilization would be resistant to immune checkpoint inhibition in metastatic gastric cancer.

Patients and methods

Two cohorts of patients with metastatic gastric cancer treated with immunotherapy were analyzed. The first cohort (N = 24) included patients treated with either nivolumab or pembrolizumab. Alternate promoter utilization was measured using the NanoString® (NanoString Technologies, Seattle, WA, USA) platform on archival tissue samples. The second cohort was a phase II clinical trial of patients uniformly treated with pembrolizumab (N = 37). Fresh tumor biopsies were obtained, and transcriptomic analysis was carried out on RNAseq data. Alternate promoter utilization was correlated to T-cell cytolytic activity, objective response rate and survival.

Results

In the first cohort 8 of 24 (33%) tumors were identified to have high alternate promoter utilization (APhigh), and this was used to define the APhigh tertile of the second cohort (13 APhigh of 37). APhigh tumors exhibited decreased markers of T-cell cytolytic activity and lower response rates (8% versus 42%, P = 0.03). Median progression-free survival was lower in the APhigh group (55 versus 180 days, P = 0.0076). In multivariate analysis, alternative promoter utilization was an independent predictor of immunotherapy survival [hazard ratio 0.29, 95% confidence interval 0.099–0.85, P = 0.024). Analyzing tumoral evolution through paired pre-treatment and post-treatment biopsies, we observed consistent shifts in alternative promoter utilization rate associated with clinical response.

Conclusion

A substantial proportion of metastatic gastric cancers utilize alternate promoters as a mechanism of immune evasion, and these tumors may be resistant to anti-PD1 immune checkpoint inhibition. Alternate promoter utilization is thus a potential mechanism of resistance to immune checkpoint inhibition, and a novel predictive biomarker for immunotherapy.

Trial Registration

ClinicalTrials.gov Identifier: NCT#02589496

Keywords: epigenetic alternate promoter, immune checkpoint inhibition, pembrolizumab, immunotherapy, gastric cancer

Key Message

A substantial proportion of metastatic gastric cancers utilize alternate promoters as a mechanism of immune evasion, and these tumors may be resistant to anti-PD1 immune checkpoint inhibition. Alternate promoter utilization is thus a potential mechanism of resistance to immune checkpoint inhibition, and a novel predictive biomarker for immunotherapy.

Introduction

Gastric cancer is a leading cause of death and patients with metastatic disease have a survival that remains below 2 years despite current therapies. The Cancer Genome Atlas (TCGA) group recently described four subtypes for gastric cancer—high microsatellite instability (MSI), Epstein–Barr virus positive (EBV), genomically stable (GS) and chromosomally unstable (CIN) [1]. CIN and GS subtypes constitute a majority of tumors, with EBV and MSI being relatively rare. These classifications, however, have yet to be incorporated into clinical practice, and currently the only biomarkers of clinical value in metastatic gastric cancer are Her2, MSI and PD-L1 status [2–4]. Among treatment modalities, immune checkpoint inhibition (ICI) has made significant breakthroughs in several tumor types including gastric cancer. Pembrolizumab and nivolumab are monoclonal antibodies that bind to and block interactions between the immune checkpoint programmed cell death protein - 1 (PD-1) and its ligands (e.g. PD-L1, PD-L2). These ICI drugs are currently approved for patients who have failed at least two prior lines of systemic chemotherapy in metastatic gastric cancer [5, 6].

A recent phase II study of single-agent pembrolizumab in metastatic gastric cancer described remarkable and durable response in the MSI and EBV subtypes [7]. In MSI tumors, ICI activity is associated with high tumor mutation load and neoantigen formation [8], with response rates for MSI gastric cancer ranging from 57% to 86% [3, 4, 7]. EBV subtype tumors are associated with tumoral immune-cell infiltration and high expression of PD-L1 in a majority of tumors [9], and in the phase II study described above, 100% of EBV-positive gastric cancers responded to pembrolizumab [7]. Although the sample size was small (n = 6), and further validation studies are required, it is thus possible that testing for EBV status may be incorporated into clinical workflows in the future to select patients for immunotherapy. Notably, while not as sensitive to ICI as EBV and MSI tumors, responses have also been reported in the CIN/GS subtypes, suggesting other mechanisms of response and primary resistance to ICI which are currently unexplored. Biomarker selection of gastric cancer for ICI is thus an area of urgent unmet clinical need, in particular for the CIN/GS subtypes.

Epigenetic activation of promoter elements allows for regulation of genomic transcription and plays a role in normal biologic function. Usage of alternate promoters allows initiation of transcription at different transcription start sites (TSS), resulting in production of altered mRNA transcripts and protein isoforms. Recently, somatic recruitment of alternate promoters in gastric cancer has been described as a mechanism of tumor immune-editing [10]. Specifically, reduced production of high-affinity major histocompatibility complex class I binding gastric cancer peptides through loss of immunogenic N-terminal peptides has been proposed as a mechanism of immune evasion, allowing nascent tumor formation. However, because this study was limited to early-stage gastric cancer (surgical resection samples), the applicability of these findings to metastatic (late-stage) gastric cancer remains unclear. Further, we hypothesized that metastatic gastric cancers with high alternate promoter utilization would be resistant to anti-PD-1 ICI.

Here, in two independent metastatic gastric cancer patient cohorts treated with ICIs, we monitored alternative promoter utilization using two different technology platforms (Nanostring and RNA-sequencing). We confirmed that similar relationships between alternative promoter utilization and intra-tumor immunity exist in metastatic gastric cancers. Moreover, by analyzing gastric tumors corresponding to the aforementioned phase II trial [7], we show that tumors with high alternate promoter utilization, and thereby lower predicted immunogenicity, are more resistant to ICI in a manner independent of previous ICI biomarkers such as mutational burden and PD-L1 CPS scores. Increased alternative promoter utilization may thus represent a novel biomarker of ICI response in metastatic gastric cancer.

Methods

REMARK criteria for validation of tumor biomarkers was followed in this study [11].

Clinical cohorts

Discovery cohort

Consecutive patients with metastatic gastric cancer treated with nivolumab or pembrolizumab treatment at Samsung Medical Centre, Seoul, Korea, were included in this cohort. ICIs were administered as salvage treatment in patients who failed to at least one cytotoxic regimen. Nivolumab 3 mg/kg was administered as a 1-hour infusion every 2 weeks and pembrolizumab 200 mg was administered as a 30-minute i.v. infusion every 3 weeks until disease progression or unacceptable toxicity. Ethics approval was obtained, and all patients provided written informed consent before archival tumor tissue specimens from primary tumors were collected and prospectively followed up for survival data.

Pembrolizumab trial cohort

Patients with histologically proven metastatic and/or recurrent gastric adenocarcinoma that had failure of at least one line of chemotherapy that included platinum/fluoropyrimidine were enrolled in this study. The trial was conducted in accordance with the Declaration of Helsinki and the Guidelines for Good Clinical Practice (ClinicalTrials.gov identifier: NCT#02589496). The trial protocol was approved by the Institutional Review Board of Samsung Medical Center (Seoul, Korea) and all patients provided written informed consent before enrollment. Pembrolizumab 200 mg was administered as a 30-minute i.v. infusion every 3 weeks until documented disease progression, unacceptable toxicity, or up to 24 months. Tumor responses were evaluated every two cycles according to RECIST 1.1 criteria. Details have been reported previously [7].

Nanostring analysis

NanoString nCounter Reporter CodeSets (NanoString Technologies, Seattle, WA, USA) were designed for 80 recurrent somatic alternate promoter-related genes, as well as immune-related genes corresponding to intra-tumoral cytolytic activity (CYT) [12], cytokines and immune checkpoints. At least two probes were designed for each gene to measure the expression of canonical and alternate promoter-driven transcripts. A canonical probe at the 5′ transcript marked by unaltered H3K4me3 and an alternate probe at the 5′ transcript of the somatic promoter [10]. Data analysis was carried out using the vendor-provided nCounter software (nSolver, NanoString Technologies, Seattle, WA, USA). Raw counts were normalized using the geometric mean of the internal positive control probes included in each CodeSet.

RNA sequencing

Tumor tissues were obtained between day −42 and day 1 before initiation of study treatment. If tumor content was estimated as more than 40% after thorough pathological review, tumor DNA and RNA were extracted from freshly obtained tissues using a QIAamp Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. We measured concentrations and 260/280 and 260/230 nm ratios with an ND1000 spectrophotometer (Nanodrop Technologies, Thermo-Fisher Scientific, MA, USA) and then further quantified DNA/RNA using a Qubit fluorometer (Life Technologies, CA, USA).

RNA transcriptomic analysis

RNAseq data were aligned to GENCODE v19 transcript annotation using TopHat and FPKM abundance measure were generated using Cufflinks. Transcripts were then merged across all samples and normalized using Cuffnorm. To analyze alternative promoter-associated expression, RNAseq reads were mapped against the genomic location previously identified by epigenomic profiling [10]. RNAseq mapping to these epigenome-defined promoter regions were then quantified, normalized by promoter length and by library size. Finally, fold changes in expression at each promoter site were computed between each tumor and the median expression level across all tumor samples.

PDL1 immunohistochemistry analysis, MSI status, EBV status, TCGA subtyping and tumor mutational burden were based on classifications used for the phase II cohort reported previously[7] and briefly summarized in supplementary Methods, available at Annals of Oncology online.

Statistical analysis

Associations of clinicopathologic features to histologic sub-classification was carried out using Fisher’s exact test. Progression-free survival (PFS) was calculated from the time of first dose of pembrolizumab or nivolumab to the time of disease progression or death, and overall survival (OS) was calculated from time of first dose of pembrolizumab or nivolumab to time of death. Kaplan–Meier (KM) curves and log rank test were used for survival analysis. The hazard ratio (HR) and its 95% confidence interval (CI) were evaluated for each analysis using Cox proportional hazards regression model. All analyses were done using R (3.4.1). In the validation cohort, samples with alternate promoter usage score greater than the 66th percentile were defined as high alternate promoter (APhigh) and remaining low alternate promoter (APlow).

Results

Alternate promoter utilization in metastatic gastric cancer

Our first cohort consisted of 24 metastatic gastric cancer patients treated with nivolumab and pembrolizumab (29 subjects were initially included, with 24 tumor samples passing quality control for sufficient tissue for Nanostring analysis). We used a customized Nanostring panel to measure transcripts associated with either the canonical or alternate promoter, as previously described [10]. Differentially expressed alternative promoters were defined as a promoter site showing <0.25×-fold change (for lost somatic promoters) or >4×-fold change in expression level (for gained somatic promoters) over the median across all samples. Using this algorithm, we found a third of the tumors (8/24) displayed high alternate promoter utilization in more than 10% of the sites (>8/80). We defined this group as APhigh while the rest were defined as APlow (supplementary Figure S1A, available at Annals of Oncology online).

Measurement of cytolytic T-cell activity has previously been described by studying expression of CD8A (CD8+ tumor infiltrating lymphocytes), granzyme A (GZMA) and perforin 1 (PRF1) [13, 14]. The APlow group demonstrated significantly increased expression of GZMA (P = 0.025), PRF1 (P = 0.011) and CD8A (P = 0.059) when compared with the APhigh group suggesting increased cytotoxic T-cell activity in the APlow group (supplementary Figure S1B, available at Annals of Oncology online). These findings are concordant with earlier results described in early gastric cancer [10], demonstrating that alternate promoter utilization in metastatic gastric cancers is also inversely related to antitumor immunity. Notably, despite the small sample size, heterogenous treatment regimens, and non-trial-based nature of this cohort, there was a trend for patients with APhigh tumors to have worse PFS compared with patients with APlow tumors (129 versus 389 days, HR 1.96, 95% CI 0.55–6.93, P = 0.29, supplementary Figure S1C, available at Annals of Oncology online). Based on these findings, we thus proceeded to further test this hypothesis in a separate cohort of uniformly treated patients.

Alternate promoter utilization as a predictor of response and survival with pembrolizumab treatment

For the second cohort, we used transcriptomic data from the phase II study described earlier [7]. Transcriptomic data from pretreatment biopsy samples and matched clinical data was available for 37 subjects and used for analysis. The median age was 57 years, 73% were male (N = 27), four (11%) were EBV-positive and four were MSI (11%) with the rest defined as CIN or GS TCGA subtype. Complete or partial responses to therapy was seen in 11 subjects (30%) (supplementary Table S1, available at Annals of Oncology online). Using 2732 somatic alternate promoter sites previously identified in gastric cancer [10], differentially expressed alternative promoters were defined similar to the first cohort (<0.25×-fold change for known somatically lost promoters or >4×-fold change for known somatically gained promoters). Notably, we have previously described good concordance between RNAseq and Nanostring platforms for assessment of alternate promoter utilization [10]. The sum of differentially expressed sites in each sample was calculated to define an alternate promoter usage score (Figure 1A). Scores ranged from 37 to 426 (median 136) (Figure 1B). Using data from the first cohort to guide cut-off points, the APhigh group was defined as samples >66% centile (n = 13), while APlow constituted the remaining samples.

Figure 1.

Figure 1.

Open in a new tab

Alternate promoter utilization in gastric cancer (pembrolizumab trial cohort, n = 37). (A) Heatmap of alternate promoter utilization. Transcript with higher than fourfold expression level compared with the median level in all tumor and mapping to the previously identified gain alternative promoter site were considered as gained alternative promoter (marked red in the heatmap). Transcript with lower than fourfold expression level compared with the median level in all tumor and mapping to the previously identified lost alternative promoter site were considered as lost alternative promoter (marked blue in the heatmap). (B) Alternative promoter utilization score is calculated as the sum of gained and lost alternative promoter in each sample. High alternate promoter utilization was defined as those >66th centile. (C) Association between high alternate promoter (APhigh) group and low alternate promoter (APlow) group with T-cell immune correlates. APhigh group are in red, whereas those in APlow group are in blue. Depicted are the expression of T-cell markers CD8A (P = 0.0037) and the T-cell cytolytic markers GZMA (P = 0.0055) and PRF1 (P = 0. 016). APhigh group shows lower expression of immune markers. (D) Waterfall plot of response to pembrolizumab according to APhigh (red) and APlow (blue) subgroups. Y axis represents percentage of maximum tumor reduction assessed according to RECIST 1.1 criteria.

The APhigh group had no statistically significant differences in clinicopathological characteristics compared with the APlow group for age, gender or histological subtype. No differences were also detected between the two groups between TCGA subtypes, mutational load and PDL1 CPS scores (supplementary Table S2, available at Annals of Oncology online). The APlow group demonstrated significantly increased expression of CD8A (P = 0.0037), GZMA (P = 0.0055) and PRF1 (P = 0.016) when compared with the APhigh group suggesting increased cytotoxic T-cell activity in the APlow group (Figure 1C). Objective response rate, defined as either partial response (PR) or complete response (CR) to therapy, was higher in the APlow group than in the APhigh group (10/24 versus 1/13, P = 0.03) (Figure 1D). Of note, in the APhigh group, the only response was in an MSI subtype tumor. Median PFS was 55 days in the APhigh group compared with 180 days in the APlow group (log rank P = 0.0076) (Figure 2A and B). The APlow group had 17% EBV (n = 4) and 12% MSI (n = 3) TCGA subtype samples, while the APhigh group had only 8% MSI (n = 1) and no EBV samples (Figure 2C). As previously shown, PFS between the various TCGA subtypes were different (P = 0.0026) (supplementary Figure S2, available at Annals of Oncology online), with the MSI and EBV subtypes having a significantly longer survival [491 days (MSI/EBV) versus 80 days (CIN/GS)]. Notably, among the CIN/GS subtype, PFS was also statistically significantly different between the APlow and APhigh groups [48 days (CIN/GS APhigh) versus 161 days (CIN/GS APlow), P = 0.0019] (Figure 2D). OS data were not mature at the time of reporting of this article, but with the existing data, a trend toward improved survival was seen in the APlow group (340 versus 292 days, P = 0.16). Multivariate analysis of clinicopathologic and alternate promoter utilization revealed high alternate promoter utilization as an independent predictive factor for PFS with pembrolizumab [HR 0.29, 95% CI 0.099–0.85, P = 0.024) (supplementary Table S3, available at Annals of Oncology online).

Figure 2.

Figure 2.

Open in a new tab

Survival curves based on alternate promoter utilization. (A) Kaplan–Meier curve of progression-free survival comparing high alternate promoter (APhigh) group versus low alternate promoter (APlow) group. (B) Swimmer plot. X axis represents the duration of pembrolizumab therapy for each patient. APhigh (red) and APlow (blue) subgroups depicted. (C) Distribution of TCGA subtypes among APhigh group and APlow group. In the APlow group, chromosomally unstable (CIN) (n = 6, 25%), genomically stable (GS) (n = 11, 46%), Epstein–Barr virus positive (EBV) (n = 4, 17%), microsatellite instability (MSI) (n = 3, 12%). In the APhigh group, CIN (n = 7, 54%), GS (n = 5, 38%), MSI (n = 1, 8%) and EBV (n = 0). (D) Kaplan–Meier curve of progression-free survival comparing TCGA subtypes split by APhigh group versus APlow group.

Alternate promoter utilization evolution post-treatment with pembrolizumab

Paired biopsy samples were available for eight subjects from the second cohort, providing an opportunity to monitor tumor evolution as a consequence of ICI therapeutic pressure. Post-treatment biopsies were taken from the primary stomach tumor at the point of progression on pembrolizumab. Of these eight subjects, two had PR, with duration of response of 211 and 491 days (both APlow), one had stable disease (SD), with duration of response for 167 days (APlow), and five had progressive disease (PD) (APhighN = 3; APlowN = 2) as best response. Interestingly, we noted very consistent shifts in the directionality of alternative promoter utilization based on clinical responses. Specifically, tumors with PR and SD exhibited ×1.5 or higher increase in alternate promoter usage score in the post-treatment biopsy samples compared with pre-treatment biopsy samples, while all five tumors with PD exhibited reductions in alternate promoter usage scores in the post-treatment biopsy sample (Fisher’s exact test, P = 0.018, supplementary Figure S3, available at Annals of Oncology online). While the precise mechanism underlying these changes remains to be elucidated, these results further support a relationship between alternative promoter landscapes and ICI therapeutic pressure.

Discussion

Our data suggest that similar to early gastric cancer, alternative promoter utilization occurs in metastatic gastric cancer and may function as a novel mechanism of resistance to ICI through epigenetic immune-editing. In this model, somatic recruitment of alternate promoter elements by gastric cancers leads to production of proteins with significantly reduced immunogenicity, leading to immune evasion. Supporting this model, high alternate promoter utilization is associated with lower cytotoxic T-cell activity, response rates and survival in gastric cancers treated with anti-PD-1 therapy. In our previous analysis of alternate promoters in primary resection samples of gastric cancer reported previously [10], we found that most of the alternate promoters observed in the primary gastric cancer cohorts (including the TCGA [1] and ACRG [15]) were also identifiable in gastric cancer cell lines. Genes associated with the alternative promoters also do not show obvious immune or tumor microenvironment associations, consistent with AP usage likely being tumor-specific rather than occurring in the tumor microenvironment.

We found that nearly a third of gastric tumors express high levels of these alternate promoter isoforms. Interestingly, alternate promoter utilization does not associate with common clinicopathological prognostic factors such as age, gender or histological subtype, nor with other immune-related markers such as PD-L1 IHC scores or mutational load. The APhigh group had only one MSI tumor and no EBV tumors, suggesting that this mechanism of resistance appears to be largely restricted to the CIN/GS subgroups. Analysis of the CIN/GS APhigh group’s PFS survival curves suggest a clear lack of benefit from pembrolizumab, with most patients progressing within 3 months of treatment. These findings form the basis of considering alternate promoter utilization as a predictive biomarker for selecting patients for ICI.

Despite initial enthusiasm for ICIs in gastric cancer following the ATTRACTION-2 [5] and KEYNOTE-059 [6] studies, newer randomized phase III studies have yielded disappointing results. Pembrolizumab failed to demonstrate benefit compared with the second-line paclitaxel chemotherapy [16], while avelumab (an anti-PDL1 antibody) failed when compared with chemotherapy in the third line [17]. However, from the large number of patients treated on these studies—one common pattern appears to emerge: there are a small group of patients who derive great benefit from exposure to immunotherapy and appear to have prolonged, durable responses, of several months to years. This phenomenon, unique to immunotherapy, has led to rapidly increasing demands by patients and utilization by clinicians, however low the chances of success. Cost of immunotherapy is a major prohibitive factor, leading to socio-economic imbalances. In this treatment landscape, negative predictive biomarkers (that identify subjects with clear lack of benefit from specific therapy) are of particular interest. An example of commonly used negative predictive biomarkers in clinical practice in gastrointestinal cancers are RAS mutations to identify lack of benefit from anti-EGFR therapy in metastatic colorectal cancer [18]. MSI is one of the few established biomarkers to identify gastric cancer patients who benefit from ICI, and this remains a primary-site agnostic indication for this therapy [19]. Biomarkers such as PD-L1 expression using immunohistochemistry and tumor mutational burden have been explored, but conflicting data exist and there is currently a lack of consensus on the utility of these biomarkers with varying results according to different ICIs (i.e. pembrolizumab vs nivolumab) [20]. Further validation of the findings in our study is required in larger clinical data sets. However, the availability of fresh tissue biopsies to carry out RNAseq is limited and platforms such as Nanostring may need to be optimized to analyze transcriptomic data from archival tissue. Optimization of the bioinformatic analysis pipeline would also be required to allow this to be brought into routine clinical practice. Finally, biomarker-driven randomized studies would need to be conducted to fully establish the utility of this biomarker. In conclusion, alternate promoter utilization may potentially be a powerful negative predictive biomarker for selecting patients for clinical trials involving immune checkpoint inhibitors in addition to existing positive biomarkers such as MSI and EBV in gastric cancer.

Supplementary Material

Supplementary Data
Click here for additional data file. (191.3KB, zip)

Acknowledgements

We would like to thank the patients and their families for participating in our trials and magnanimously providing consent for the utilization of tissue to further the science.

Funding

RS is supported by a National Medical Research Council (NMRC) Fellowship (NMRC/Fellowship/0059/2018), Singapore. PT is supported by core funding from Duke-NUS Medical School and the Biomedical Research Council, Agency for Science, Technology and Research (no grant numbers apply). JL is supported by a grant from the Korean Health Technology R&D Project, Ministry of Health & Welfare, Republic of Korea (HI16C1990 and HI14C3418). This work was also supported by National Medical Research Council grants (TCR/009-NUHS/2013, NR13NMR111OM and NMRC/STaR/0026/2015).

Disclosure

The subject matter in this article has been submitted as a Technology Disclosure to the institutional Technology Transfer Office, for potential intellectual property protection. The authors have declared no conflicts of interest.

References

  • 1.Cancer Genome Atlas Research Network. Comprehensive molecular characterization of gastric adenocarcinoma. Nature 2014; 513: 202–209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 2. Bang YJ, Van Cutsem E, Feyereislova A. et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomised controlled trial. Lancet 2010; 376(9742): 687–697. [DOI] [PubMed] [Google Scholar]
  • 3. Fuchs CS, Doi T, Jang RW. et al. Safety and efficacy of pembrolizumab monotherapy in patients with previously treated advanced gastric and gastroesophageal junction cancer: phase 2 Clinical KEYNOTE-059 Trial. JAMA Oncol 2018; 4(5): e180013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 4. Le DT, Durham JN, Smith KN. et al. Mismatch repair deficiency predicts response of solid tumors to PD-1 blockade. Science 2017; 357(6349): 409–413. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5. Kang YK, Boku N, Satoh T. et al. Nivolumab in patients with advanced gastric or gastro-oesophageal junction cancer refractory to, or intolerant of, at least two previous chemotherapy regimens (ONO-4538-12, ATTRACTION-2): a randomised, double-blind, placebo-controlled, phase 3 trial. Lancet 2017; 390(10111): 2461–2471. [DOI] [PubMed] [Google Scholar]
  • 6. Muro K, Chung HC, Shankaran V. et al. Pembrolizumab for patients with PD-L1-positive advanced gastric cancer (KEYNOTE-012): a multicentre, open-label, phase 1b trial. Lancet Oncol 2016; 17(6): 717–726. [DOI] [PubMed] [Google Scholar]
  • 7. Kim ST, Cristescu R, Bass AJ. et al. Comprehensive molecular characterization of clinical responses to PD-1 inhibition in metastatic gastric cancer. Nat Med 2018; 24(9): 1449–1458. [DOI] [PubMed] [Google Scholar]
  • 8. Le DT, Uram JN, Wang H. et al. PD-1 blockade in tumors with mismatch-repair deficiency. N Engl J Med 2015; 372(26): 2509–2520. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9. Sundar R, Qamra A, Tan ALK. et al. Transcriptional analysis of immune genes in Epstein-Barr virus-associated gastric cancer and association with clinical outcomes. Gastric Cancer 2018; 21(6): 1064–1070. [DOI] [PubMed] [Google Scholar]
  • 10. Qamra A, Xing M, Padmanabhan N. et al. Epigenomic promoter alterations amplify gene isoform and immunogenic diversity in gastric adenocarcinoma. Cancer Discov 2017; 7(6): 630–651. [DOI] [PubMed] [Google Scholar]
  • 11. McShane LM, Altman DG, Sauerbrei W. et al. REporting recommendations for tumour MARKer prognostic studies (REMARK). Br J Cancer 2005; 93(4): 387–391. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 12. Rooney MS, Shukla SA, Wu CJ. et al. Molecular and genetic properties of tumors associated with local immune cytolytic activity. Cell 2015; 160(1-2): 48–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13. Johnson BJ, Costelloe EO, Fitzpatrick DR. et al. Single-cell perforin and granzyme expression reveals the anatomical localization of effector CD8+ T cells in influenza virus-infected mice. Proc Natl Acad Sci USA 2003; 100(5): 2657–2662. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14. 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(7528): 563–567. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15. Cristescu R, Lee J, Nebozhyn M. et al. Molecular analysis of gastric cancer identifies subtypes associated with distinct clinical outcomes. Nat Med 2015; 21(5): 449–456. [DOI] [PubMed] [Google Scholar]
  • 16. Shitara K, Özgüroğlu M, Bang Y-J. et al. Pembrolizumab versus paclitaxel for previously treated, advanced gastric or gastro-oesophageal junction cancer (KEYNOTE-061): a randomised, open-label, controlled, phase 3 trial. Lancet 2018; 392(10142): 123–133. [DOI] [PubMed] [Google Scholar]
  • 17. Bang YJ, Ruiz EY, Van Cutsem E. et al. Phase 3, randomised trial of avelumab versus physician's choice of chemotherapy as third-line treatment for patients with advanced gastric or gastro-oesophageal junction cancer: primary analysis of JAVELIN Gastric 300. Ann Oncol 2018; 29(10): 2052–2060. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 18. Douillard JY, Oliner KS, Siena S. et al. Panitumumab-FOLFOX4 treatment and RAS mutations in colorectal cancer. N Engl J Med 2013; 369(11): 1023–1034. [DOI] [PubMed] [Google Scholar]
  • 19. Boyiadzis MM, Kirkwood JM, Marshall JL. et al. Significance and implications of FDA approval of pembrolizumab for biomarker-defined disease. J Immunother Cancer 2018; 6(1): 35. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20. Lin EM, Gong J, Klempner SJ, Chao J.. Advances in immuno-oncology biomarkers for gastroesophageal cancer: programmed death ligand 1, microsatellite instability, and beyond. World J Gastroenterol 2018; 24(25): 2686–2697. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supplementary Data
Click here for additional data file. (191.3KB, zip)

Articles from Annals of Oncology are provided here courtesy of Oxford University Press

RESOURCES