Cancer-associated fibroblasts are the main contributors to epithelial-to-mesenchymal signatures in the tumor microenvironment

Epithelial-to-mesenchymal transition (EMT) is associated with tumor initiation, metastasis, and drug resistance. However, the mechanisms underlying these associations are largely unknown. We studied several tumor types to identify the source of EMT gene expression signals and a potential mechanism of resistance to immuno-oncology treatment. Across tumor types, EMT-related gene expression was strongly associated with expression of stroma-related genes. Based on RNA sequencing of multiple patient-derived xenograft models, EMT-related gene expression was enriched in the stroma versus parenchyma. EMT-related markers were predominantly expressed by cancer-associated fibroblasts (CAFs), cells of mesenchymal origin which produce a variety of matrix proteins and growth factors. Scores derived from a 3-gene CAF transcriptional signature (COL1A1, COL1A2, COL3A1) were sufficient to reproduce association between EMT-related markers and disease prognosis. Our results suggest that CAFs are the primary source of EMT signaling and have potential roles as biomarkers and targets for immuno-oncology therapies.

Myeloid Cell-associated Resistance to PD-1/PD-L1 Blockade in Urothelial Cancer Revealed Through Bulk and Single-cell RNA Sequencing

PURPOSE

To define dominant molecular and cellular features associated with PD-1/PD-L1 blockade resistance in metastatic urothelial cancer.

EXPERIMENTAL DESIGN

We pursued an unbiased approach using bulk RNA sequencing data from two clinical trials to discover (IMvigor 210) and validate (CheckMate 275) pretreatment molecular features associated with resistance to PD-1/PD-L1 blockade in metastatic urothelial cancer. We then generated single-cell RNA sequencing (scRNA-seq) data from muscle-invasive bladder cancer specimens to dissect the cellular composition underlying the identified gene signatures.

RESULTS

We identified an adaptive immune response gene signature associated with response and a protumorigenic inflammation gene signature associated with resistance to PD-1/PD-L1 blockade. The adaptive immune response:protumorigenic inflammation signature expression ratio, coined the 2IR score, best correlated with clinical outcomes, and was externally validated. Mapping these bulk gene signatures onto scRNA-seq data uncovered their underlying cellular diversity, with prominent expression of the protumorigenic inflammation signature by myeloid phagocytic cells. However, heterogeneity in expression of adaptive immune and protumorigenic inflammation genes was observed among single myeloid phagocytic cells, quantified as the myeloid single cell immune:protumorigenic inflammation ratio (Msc2IR) score. Single myeloid phagocytic cells with low Msc2IR scores demonstrated upregulation of proinflammatory cytokines/chemokines and downregulation of antigen presentation genes, were unrelated to M1 versus M2 polarization, and were enriched in pretreatment blood samples from patients with PD-L1 blockade-resistant metastatic urothelial cancer.

CONCLUSIONS

The balance of adaptive immunity and protumorigenic inflammation in individual tumor microenvironments is associated with PD-1/PD-L1 resistance in urothelial cancer with the latter linked to a proinflammatory cellular state of myeloid phagocytic cells detectable in tumor and blood.See related commentary by Drake, p. 4139.

Analyses of PD-L1 and Inflammatory Gene Expression Association with Efficacy of Nivolumab ± Ipilimumab in Gastric Cancer/Gastroesophageal Junction Cancer

PURPOSE

In advanced gastric cancer/gastroesophageal junction cancer (GC/GEJC), there is a need to identify biomarkers of response to therapies, such as immune checkpoint inhibitors.

PATIENTS AND METHODS

In post hoc exploratory analyses from CheckMate 032 (GC/GEJC cohort), we evaluated associations between nivolumab ± ipilimumab (NIVO ± IPI) efficacy and programmed death ligand 1 (PD-L1) expression, defined by tumor cells (% TC) or combined positive score (CPS; sum of PD-L1-staining TCs + immune cells, divided by total viable TCs, × 100) using the Dako PD-L1 IHC 28-8 pharmDx assay, or inflammatory gene expression.

RESULTS

There was a trend toward increased efficacy (objective response and overall survival) when PD-L1 expression was determined by CPS compared with % TC at higher cutoffs of ≥5 and ≥10 in the pooled analysis of all treatment regimens. In this analysis, 19% and 26% of patients with PD-L1-positive tumors at a CPS cutoff of ≥5 and ≥10, respectively, had an objective response compared with 8% and 9% of patients at the equivalent % TC cutoffs. Longer survival was demonstrated in patients with PD-L1-positive (defined by CPS cutoffs of ≥5 and ≥10) versus PD-L1-negative status. Similar results were observed in the NIVO 1 mg/kg + IPI 3 mg/kg subgroup. Multiple inflammatory gene signatures/transcripts, including a signature consisting of four genes (CD274, CD8A, LAG3, and STAT1), showed associations with response to NIVO ± IPI.

CONCLUSIONS

This study suggests a greater association of PD-L1 expression by CPS with NIVO ± IPI efficacy compared with % TC PD-L1 expression in patients with GC/GEJC. Inflammatory signatures were also associated with NIVO ± IPI response, warranting further investigation.See related commentary by Moutafi and Rimm, p. 3812.

Development and Performance of a CD8 Gene Signature for Characterizing Inflammation in the Tumor Microenvironment across Multiple Tumor Types

Across multiple tumor types, immune checkpoint inhibitors (ICIs) have demonstrated clinical benefit to patients with cancer, yet there is a need to identify predictive biomarkers of response to these therapies. A multiparameter gene expression profiling-based tumor inflammation assay may offer robust characterization of the tumor microenvironment, thereby extending the utility of single-gene analysis or immunohistochemistry (IHC) in predicting response to ICIs. The authors interrogated 1778 commercially procured, formalin-fixed, paraffin-embedded samples using gene expression profiling and pathology-assisted digital CD8 IHC. A machine-learning approach was used to develop gene expression signatures that predicted CD8⁺ immune cell abundance as surrogates for tumor inflammation in melanoma and squamous cell carcinoma of the head and neck samples. An assay for a 16-gene CD8 signature was developed and analytically validated across 12 tumor types. CD8 signature scores correlated with CD8 IHC in a platform-independent manner, and inflammation prevalence was similar between assay methods for all tumor types except prostate cancer and small cell lung cancer. In retrospective analyses, CD8 signature scores were associated with progression-free survival and overall survival with nivolumab in patients with urothelial carcinoma from CheckMate 275. This study demonstrated that the CD8 signature assay can be used to accurately quantify CD8⁺ immune cell abundance in the tumor microenvironment and has potential clinical utility for determining patients with cancer likely to respond to ICIs.

Molecular correlates of response to nivolumab at baseline and on treatment in patients with RCC

BACKGROUND

Nivolumab is an immune checkpoint inhibitor targeting the programmed death-1 receptor that improves survival in a subset of patients with clear cell renal cell carcinoma (ccRCC). In contrast to other tumor types that respond to immunotherapy, factors such as programmed death ligand-1 (PD-L1) status and tumor mutational burden show limited predictive utility in ccRCC. To address this gap, we report here the first molecular characterization of nivolumab response using paired index lesions, before and during treatment of metastatic ccRCC.

METHODS

We analyzed gene expression and T-cell receptor (TCR) clonality using lesion-paired biopsies provided in the CheckMate 009 trial and integrated the results with their PD-L1/CD4/CD8 status, genomic mutation status and serum cytokine assays. Statistical tests included linear mixed models, logistic regression models, Fisher's exact test, and Kruskal-Wallis rank-sum test.

RESULTS

We identified transcripts related to response, both at baseline and on therapy, including several that are amenable to peripheral bioassays or to therapeutic intervention. At both timepoints, response was positively associated with T-cell infiltration but not associated with TCR clonality, and some non-Responders were highly infiltrated. Lower baseline T-cell infiltration correlated with elevated transcription of Wnt/β-catenin signaling components and hypoxia-regulated genes, including the Treg chemoattractant CCL28. On treatment, analysis of the non-responding patients whose tumors were highly T-cell infiltrated suggests association of the RIG-I-MDA5 pathway in their nivolumab resistance. We also analyzed our data using previous transcriptional classifications of ccRCC and found they concordantly identified a molecular subtype that has enhanced nivolumab response but is sunitinib-resistant.

CONCLUSION

Our study describes molecular characteristics of response and resistance to nivolumab in patients with metastatic ccRCC, potentially impacting patient selection and first-line treatment decisions.

TRIAL REGISTRATION NUMBER

NCT01358721.

Combinatorial biomarkers to predict responses to immune checkpoint therapy in metastatic urothelial cancer

488

Background: Immune checkpoint therapy can produce durable anti-tumor responses in metastatic urothelial carcinoma (mUCC); however, the responses are not universal. Despite multiple approvals of immune checkpoint therapy in mUCC, we lack predictive biomarkers to guide patient selection. Therefore, there is a critical need to develop clinically useful biomarkers to refine patient selection. The single biomarker studies either focused on tumor mutations or immune response biomarkers, which may limit predictive power due to lack of integration between cancer cell biology and immune cell responses. The identification of biomarkers may require interrogation of both the tumor mutational status and the immune microenvironment. Methods: We performed retrospective multi-platform immuno-genomic analyses of pre-treatment tumor tissues in a discovery cohort (n = 31). Next, we tested the clinical relevance of ARID1A mutation and pre-treatment CXCL13 expression in two independent confirmatory cohorts (CheckMate275 and IMvigor210). Additionally, we performed reverse translational studies using murine model of bladder cancer to demonstrate direct association of the biomarkers in anti-PD-(L)-1 mediated anti-tumor immunity. Results: We identified genomic mutation of AT-rich interactive domain-containing protein 1A ( ARID1A) in tumor cells and expression of immune cytokine CXCL13 in the pre-treatment tumor tissues as two predictors of clinical responses. We found that ARID1A mutation and expression of CXCL13 in the baseline tumor tissues correlated with improved overall survival (OS) in both confirmatory cohorts (CheckMate275, CXCL13 data, n = 217; ARID1A data, n = 139, and IMvigor210, CXCL13 data, n = 348; ARID1A data, n = 275). Further, reverse translational studies revealed that CXCL13−/− tumor-bearing mice were resistant to immune checkpoint therapy whereas ARID1A knockdown enhanced sensitivity to immune checkpoint therapy in a murine model of bladder cancer. We then interrogated CXCL13 expression plus ARID1A mutation as a combination biomarker in predicting response to immune checkpoint therapy in CheckMate275 and IMvigor210. Combination of the 2 biomarkers in baseline tumor tissues showed improved OS compared to either single biomarker. Conclusions: Cumulatively, this study revealed that the combination of CXCL13 plus ARID1A mutation may improve patient selection in mUCC for immune checkpoint therapy.

Multiple SARS-CoV-2 Introductions Shaped the Early Outbreak in Central Eastern Europe: Comparing Hungarian Data to a Worldwide Sequence Data-Matrix

Severe Acute Respiratory Syndrome Coronavirus 2 is the third highly pathogenic human coronavirus in history. Since the emergence in Hubei province, China, during late 2019, the situation evolved to pandemic level. Following China, Europe was the second epicenter of the pandemic. To better comprehend the detailed founder mechanisms of the epidemic evolution in Central-Eastern Europe, particularly in Hungary, we determined the full-length SARS-CoV-2 genomes from 32 clinical samples collected from laboratory confirmed COVID-19 patients over the first month of disease in Hungary. We applied a haplotype network analysis on all available complete genomic sequences of SARS-CoV-2 from GISAID database as of 21 April 2020. We performed additional phylogenetic and phylogeographic analyses to achieve the recognition of multiple and parallel introductory events into our region. Here, we present a publicly available network imaging of the worldwide haplotype relations of SARS-CoV-2 sequences and conclude the founder mechanisms of the outbreak in Central-Eastern Europe.

Abstract 2117: Immune contexture of breast cancer subtypes in real-world molecular data

Background: Breast cancer has been extensively characterized molecularly with cohorts such as The Cancer Genome Atlas (TCGA). However, as the standard of care evolves, there is a need for contemporaneous datasets with more detailed treatment and outcomes data. The ORIEN Avatar database (M2Gen, Tampa, FL) consists of clinical and tumor sequencing data from treatment-naive or on-treatment biopsies. This study evaluates the quality of ORIEN Avatar's RNA-sequencing (RNASeq) data, interrogates the impact of prior treatment on PAM50 subtyping, and reports associations between breast cancer subtypes and estimated immune cell populations.

Methods: In the ORIEN Avatar dataset, RNASeq data were prepared using a multistep normalization process. We assessed the quality of normalized RNASeq data using 3 modalities. Using a principal-component (PC) analysis, we interrogated the association of gene expression with potential technical and biological factors, including processing batch effects, preservation methods, and treatment status. We evaluated concordance between immunohistochemistry (IHC)-based subtyping and intrinsic subtyping using the PAM50 gene expression signature. Prevalence of in silico-derived tumor-infiltrating immune cell compositions in ORIEN Avatar compared with TCGA breast cancer samples was investigated.

Results: The ORIEN Avatar breast cancer cohort comprised 560 patients with RNASeq data. The first 3 PCs (accounting for 11%, 9%, and 7% of total explained variance, respectively) were associated with biological factors, including PAM50 subtyping and HER2 receptor and hormone receptor (HR) status. PAM50 subtype was largely concordant with IHC subtype, per clinical record of biomarker testing, and prior treatment status did not impact PAM50 subtyping. We observed that 68% of basal samples were triple negative by IHC, 92% of luminal samples were HER2-negative and HR-positive by IHC, and 62% of HER2-enriched samples per RNASeq were HER2 receptor positive by IHC. Several in silico-derived tumor-infiltrating immune cell compositions, including activated CD4 memory T cells, M0 macrophages, and activated dendritic cells, were enriched in triple-negative breast cancer samples, consistent with TCGA. Correlative analysis of RNASeq data with other biomarker data and clinical outcome is planned.

Conclusions: Our results using real-world (RW) ORIEN Avatar RNASeq data were consistent with molecular profiling of breast cancer assessed using IHC and TCGA. This proof-of-concept study found no association between prior treatment status and PAM50 intrinsic subtyping. Our analysis provides a framework to assess the quality of RW molecular profiling and highlights the feasibility of leveraging harmonized molecular data to replicate and discover novel biological evidence.

Citation Format: Xuya Wang, Bin Li, Peter M. Szabo, Han Chang, Mustimbo Roberts, Alice M. Walsh. Immune contexture of breast cancer subtypes in real-world molecular data [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 2117.

Abstract 3200: Association between inflammatory gene expression signatures in blood or tumor samples and response to nivolumab in patients with SCCHN

Background: Tumor inflammation has been associated with response to nivolumab (NIVO) in patients with cancer. In patients with squamous cell carcinoma of the head and neck (SCCHN), inflammation in tumor tissue samples can be assessed with inflammatory gene expression signature scores determined by gene expression profiling (GEP); however, tissue availability may be limited. We evaluated the association between inflammatory signature scores, assessed by GEP using blood or tumor samples, and response to NIVO.

Methods: In post-hoc analyses, GEP was performed by RNA-seq on pre- (D1) and on-treatment (D43) whole blood samples and pretreatment tumor samples from patients with SCCHN treated with NIVO or investigator's choice (IC) therapy (CheckMate 141; NCT02105636). Scores for blood and tumor samples were calculated for multiple published tissue-derived inflammation signatures, and association with progression-free survival (PFS), overall survival (OS), and objective response (OR) was evaluated.

Results: GEP was evaluable in 278 D1 (188 NIVO, 90 IC) and 186 D43 (129 NIVO, 57 IC) blood samples. For a 13-gene inflammation signature, signature scores from pretreatment tumor and matched blood samples were correlated (r = 0.33). For D1 and D43 blood samples, both OS and PFS were positively associated with signature scores in at least 1 of the treatment groups, and treatment benefit with NIVO vs IC increased with increasing score. OR was positively associated with signature scores for D1 and D43 samples, especially for NIVO-treated patients (Table). Results were highly correlated between multiple inflammation signatures.

TableHazard ratio (95% confidence interval)aPFSOSBlood samplesD1D43D1D43NIVO vs ICLow inflammation scoreb1.1 (0.8–1.5)1.2 (0.8–1.8)0.9 (0.7–1.3)0.9 (0.6–1.3)High inflammation scorec0.8 (0.6–1.2)0.7 (0.4–1.1)0.7 (0.5–1.0)0.5 (0.3–0.8)High vs low inflammation scoredNIVO0.6 (0.5–0.8)0.6 (0.4–0.8)0.5 (0.4–0.6)0.5 (0.3–0.7)IC0.9 (0.6–1.2)1.1 (0.7–1.6)0.6 (0.5–0.9)0.8 (0.6–1.2)NIVO vs IC0.7 (0.5–1.1)0.6 (0.3–0.9)0.7 (0.5–1.1)0.6 (0.4–0.9)Area under the receiver operating characteristic curve for OR, %Blood samplesD1D43NIVO60.765.3IC52.655.7aCox proportional-hazards model.b25th inflammation score percentile.c75th inflammation score percentile.d75th vs 25th inflammation score percentile.D1, pretreatment samples; D43, on-treatment samples; IC, investigator's choice of therapy; NIVO, nivolumab; OS, overall survival; PFS, progression-free survival.

Conclusion: Inflammation signature scores correlated between SCCHN tumor and matched blood samples and were associated with outcomes, especially for NIVO treatment. Further study of inflammation signatures using blood-based GEP is warranted to further assess tumor inflammation and its association with response to NIVO in SCCHN.

Citation Format: Jun Li, Peter M. Szabo, Scott D. Chasalow, Han Chang, Abraham Apfel, Mustimbo Roberts, Jaclyn Neely. Association between inflammatory gene expression signatures in blood or tumor samples and response to nivolumab in patients with SCCHN [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 3200.

Abstract 1997: Comparison of gene expression profiling platforms: Translatability of tumor inflammation gene signatures

Background: A CD8 gene expression signature, developed on a tumor immunology panel (TIP) by BMS as a surrogate marker for inflammation in the tumor microenvironment, was associated with response to immune checkpoint inhibitors in a post-hoc analysis of patients with urothelial carcinoma (CheckMate 275). Similar associations in other tumor types have been assessed using other inflammation signatures (including a tumor inflammation signature [TIS] and a CD8+ T-cell abundance signature [CD8+ TCA]) and gene expression profiling (GEP) platforms (eg, RNA-sequencing [RNA-seq] or GEP panels). This study compares the performance of 4 different GEP panels for assessing tumor inflammation to determine if GEP can be used in a platform-independent manner.

Methods: RNA extracted from unstained commercial formalin-fixed, paraffin-embedded tissue sections (97 melanoma; 101 squamous cell carcinoma of the head and neck) was analyzed using RNA-seq (TruSeq RNA Exome, Illumina) and 2 GEP panels (PanCancer IO 360™ Panel [NanoString] and Oncomine™ Immune Response Research Assay [Thermo Fisher Scientific]). Extraction-free methods were used for the TIP and IOv2 GEP panels. Scores were derived for the CD8, TIS, and CD8+ TCA signatures. CD8+ T cells were quantified using pathologist-supervised, digital scoring–based analysis of CD8 immunohistochemistry (IHC) (Mosaic Laboratories). Programmed death ligand 1 (PD-L1) expression on tumor cells was assessed using the Dako PD-L1 IHC 28-8 pharmDx assay (Agilent).

Results: Correlations were comparable between RNA-seq and GEP panels for 1) expression of each gene in the CD8 signature and 2) CD8 signature scores. The inflammation signatures correlated with CD8 expression by IHC, but not with PD-L1 expression by IHC (Table).

RNA-seq vs panelGEP panelCorrelation of expression of genes in the CD8 signature (median r, IQR)aCorrelation of CD8 signature scores (r)IO 3600.87, 0.78–0.940.95Oncomine IRRA0.82, 0.66–0.860.93TIPb0.71, 0.44–0.830.89IOv2b0.79, 0.40–0.840.86Gene signature vs CD8 IHCGEP panelSignatureNumber of samples (n)Correlation of signature score vs CD8 IHC score (r)IO 360TISc1840.76IO 360CD8+ TCAd1840.69TIPbCD8c1700.80Gene signature vs PD-L1 IHCGEP panelSignatureNumber of samples (n)Correlation of signature score vs PD-L1 IHC score (r)IO 360TISc1870.23TIPbCD8c1720.20aMedian Pearson's r shown for gene-by-gene comparisons in the CD8 signature;bBMS custom panels;cSignature carries investigational use only status when derived using the panel specified;dResearch use only. GEP, gene expression profiling; IHC, immunohistochemistry; IO 360, PanCancer IO 360™ Panel (NanoString); IQR, interquartile range; Oncomine IRRA, Oncomine™ Immune Response Research Assay (Thermo Fisher Scientific); PD-L1, programmed death ligand 1; RNA-seq, RNA-sequencing; TCA, T-cell abundance; TIP, tumor immunology panel; TIS, tumor inflammation signature.

Conclusion: This study shows platform-independent consistency when assessing GEP-derived tumor inflammation signatures and demonstrates the feasibility of utilizing gene expression signatures across GEP platforms.

Citation Format: Esperanza Anguiano, Keyur Desai, Sheida Hayati, Peter M. Szabo, Saumya Pant, Lisu Wang, Nancy Zhang. Comparison of gene expression profiling platforms: Translatability of tumor inflammation gene signatures [abstract]. In: Proceedings of the Annual Meeting of the American Association for Cancer Research 2020; 2020 Apr 27-28 and Jun 22-24. Philadelphia (PA): AACR; Cancer Res 2020;80(16 Suppl):Abstract nr 1997.

ARID1A mutation plus CXCL13 expression act as combinatorial biomarkers to predict responses to immune checkpoint therapy in mUCC

Immune checkpoint therapy (ICT) can produce durable antitumor responses in metastatic urothelial carcinoma (mUCC); however, the responses are not universal. Despite multiple approvals of ICT in mUCC, we lack predictive biomarkers to guide patient selection. The identification of biomarkers may require interrogation of both the tumor mutational status and the immune microenvironment. Through multi-platform immuno-genomic analyses of baseline tumor tissues, we identified the mutation of AT-rich interactive domain-containing protein 1A (ARID1A) in tumor cells and expression of immune cytokine CXCL13 in the baseline tumor tissues as two predictors of clinical responses in a discovery cohort (n = 31). Further, reverse translational studies revealed that CXCL13^-/- tumor-bearing mice were resistant to ICT, whereas ARID1A knockdown enhanced sensitivity to ICT in a murine model of bladder cancer. Next, we tested the clinical relevance of ARID1A mutation and baseline CXCL13 expression in two independent confirmatory cohorts (CheckMate275 and IMvigor210). We found that ARID1A mutation and expression of CXCL13 in the baseline tumor tissues correlated with improved overall survival (OS) in both confirmatory cohorts (CheckMate275, CXCL13 data, n = 217; ARID1A data, n = 139, and IMvigor210, CXCL13 data, n = 348; ARID1A data, n = 275). We then interrogated CXCL13 expression plus ARID1A mutation as a combination biomarker in predicting response to ICT in CheckMate275 and IMvigor210. Combination of the two biomarkers in baseline tumor tissues suggested improved OS compared to either single biomarker. Cumulatively, this study revealed that the combination of CXCL13 plus ARID1A may improve prediction capability for patients receiving ICT.

Nivolumab in Patients with Advanced Platinum-resistant Urothelial Carcinoma: Efficacy, Safety, and Biomarker Analyses with Extended Follow-up from CheckMate 275

PURPOSE

We report efficacy and safety with extended follow-up, and exploratory biomarker analyses from the phase II CheckMate 275 trial to identify biomarkers of response to nivolumab in platinum-resistant metastatic or unresectable urothelial carcinoma (mUC).

PATIENTS AND METHODS

Patients received nivolumab 3 mg/kg once every 2 weeks until disease progression, unacceptable toxicity, or other protocol-defined reasons. The primary endpoint was objective response rate (ORR) per blinded independent review committee (BIRC; using RECIST v1.1) in all treated patients and by tumor PD-L1 expression. Key secondary endpoints were progression-free survival (PFS) per BIRC using RECIST v1.1 and overall survival (OS) in all patients and by PD-L1 expression. Exploratory endpoints included safety and biomarker analyses of tumor mutational burden (TMB), PD-L1, and previously identified mutational signatures.

RESULTS

Of 270 treated patients, 139 had evaluable TMB. With 33.7 months' minimum follow-up, ORR per BIRC, median PFS, and median OS [95% confidence interval (CI)] in all treated patients were 20.7% (16.1-26.1), 1.9 months (1.9-2.3), and 8.6 months (6.1-11.3), respectively. No new safety signals were identified. Higher TMB was associated (P < 0.05) with improved ORR [OR (95% CI): 2.13 (1.26-3.60)], PFS [HR: 0.75 (0.61-0.92)], and OS [HR: 0.73 (0.58-0.91)]. TMB combined with PD-L1 better predicted ORR, PFS, and OS than PD-L1 alone. Higher mutational signature 2 score was associated with better OS but did not improve the predictive value of TMB.

CONCLUSIONS

These results support the durable antitumor activity of nivolumab and suggest that TMB may enrich for better response in mUC. Future studies of TMB/PD-L1 as biomarkers for response to nivolumab in randomized trials are warranted.See related commentary by Swami et al., p. 5059.

Fibroblast Growth Factor Receptor 3 Alterations and Response to PD-1/PD-L1 Blockade in Patients with Metastatic Urothelial Cancer

Prior studies have demonstrated that fibroblast receptor 3 (FGFR3)-mutant urothelial cancers (UCs) are associated with decreased T-cell infiltration. As FGFR3 mutations are enriched in luminal-like UC and luminal-like UC has been shown to be relatively less responsive to PD-1/PD-L1 inhibition (checkpoint inhibition [CPI]), these data have led to the speculation that FGFR3 mutations may be causally related to poor T-cell infiltration and that UC patients harboring FGFR3 mutations may be suboptimal candidates for CPI. Using data derived from two clinical trials exploring CPI in metastatic UC, we demonstrate no statistically significant difference in response rates in patients with FGFR3-mutant versus wild-type UC. We present hypothesis-generating data, suggesting that similar response rates may be explained by a "balancing out" of previously identified independent positive and negative predictors of CPI sensitivity; that is, compared with FGFR3 wild-type UC, FGFR3-mutant UC is associated with a similar tumor mutational burden, lower T-cell infiltration, but also lower stromal/transforming growth factor beta (TGF-β) signals. Based on our findings, FGFR3 mutation status is not a biomarker of resistance to CPI. Indeed, the single-agent activity of both FGFR3 inhibitors and CPI in FGFR3-mutant UC, and potential non-cross resistance provide a strong pragmatic rationale for combination approaches. PATIENT SUMMARY: In this report, we examined the impact of a mutated gene found in a subset of urothelial cancers on response to treatment with immunotherapy. We found that patients with tumors harboring mutations in the gene FGFR3 respond to immunotherapy similarly to patients without such mutations.

Abstract 2673: Association of PD-L1 combined positive score and immune gene signatures with efficacy of nivolumab (NIVO) ± ipilimumab (IPI) in patients with metastatic gastroesophageal cancer (mGEC)

Background: In the CheckMate 032 phase 1/2 study, NIVO ± IPI demonstrated clinically meaningful antitumor activity in patients with chemotherapy-refractory mGEC. Archival or fresh tumor biopsies were analyzed to determine whether expression of PD-L1 and selected immune gene signatures were predictive of response to NIVO ± IPI. Methods: Pooled analyses included all treatment regimens (NIVO 3 mg/kg, NIVO 1 mg/kg + IPI 3 mg/kg, NIVO 3 mg/kg + IPI 1 mg/kg, and patients treated with NIVO 1 mg/kg + IPI 1 mg/kg in the dose-escalation phase). PD-L1 immunohistochemistry (IHC; Dako PD-L1 IHC 28-8 pharmDx assay) was used to evaluate tumor PD-L1 expression, referred to as tumor proportion score (TPS). Combined positive score (CPS) was determined by evaluating PD-L1 expression on previously stained IHC slides using the CPS algorithm. Gene expression profiling (GEP) by RNA sequencing was used to evaluate immune cell activation and infiltration signatures, a Bristol-Myers Squibb (BMS) inflammatory signature, and PD-L1 gene expression. Results: The pooled CPS (N = 104), TPS (N = 130), and GEP (N = 40) cohorts were mostly representative of the overall CheckMate 032 mGEC cohort (N =163). At a median (range) follow-up of 23.4 (17.0-35.4) months, CPS at higher cutoffs correlated better with efficacy and had higher prevalence than TPS in all analyses (Table). For all immune gene signatures examined, responders had higher signature scores in aggregate. For the BMS inflammatory signature, the association between signature score and response was significant (P = 0.004; false discovery rate = 0.037). Conclusions: CPS demonstrated stronger association with efficacy of NIVO ± IPI than TPS in mGEC. Prevalence of CPS was higher than that of TPS. Among immune gene signatures examined, the BMS inflammatory signature achieved the best association with efficacy and warrants further investigation.

All treatments combinedMethodNCutoffResponse rate, %Prevalence, %All patients163NA9.8100CPS1041026.533519.250114.168TPS130109.1857.710117.531NIVO 1 mg/kg + IPI 3 mg/kgMethodNCutoffResponse rate, %Prevalence, %All patients49NA20.4100CPS331054.533541.252128.076TPS421002502140.024CPS = combined positive score (number of PD-L1-staining tumor cells, lymphocytes, and macrophages relative to all viable tumor cells x 100); IPI = ipilimumab; NA, not applicable; NIVO = nivolumab; TPS = tumor proportion score

Citation Format: Ming Lei, Nathan Siemers, Dimple Pandya, Han Chang, Teresa Sanchez, Cecile Dorange, Christopher Harbison, Peter M. Szabo, Yelena Janjigian, Patrick A. Ott, Padmanee Sharma, Johanna Bendell, Jeffry Evans, Filippo de Braud, Ian Chau, Zachary Boyd. Association of PD-L1 combined positive score and immune gene signatures with efficacy of nivolumab (NIVO) ± ipilimumab (IPI) in patients with metastatic gastroesophageal cancer (mGEC) [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2019; 2019 Mar 29-Apr 3; Atlanta, GA. Philadelphia (PA): AACR; Cancer Res 2019;79(13 Suppl):Abstract nr 2673.

CD8+ T cells in tumor parenchyma and stroma by image analysis (IA) and gene expression profiling (GEP): Potential biomarkers for immuno-oncology (I-O) therapy

2594

Background: Distribution patterns of CD8+ T cells within the tumor microenvironment (TME) can be assessed by IA, which may reflect underlying tumor biology and serve as a potential biomarker to assess the utility of I-O therapy. These patterns are variable and may be classified as immune desert (minimal infiltrate), excluded (T cells confined to tumor stroma or to the invasive margin), or inflamed (T cells diffusely infiltrating tumor parenchyma and stroma). We hypothesized that association of a GEP signature with abundance of parenchymal and stromal T-cell infiltrates may identify biomarkers of response or resistance to I-O therapy. To test this, we applied an AI-powered IA platform to quantify CD8+ T cells by geographical location and used GEP to define both CD8 abundance and associated geographic localization to tumor parenchyma and stroma. Methods: We performed an analysis using a tumor inflammatory GEP assay and CD8 immunohistochemistry on procured specimens (335 melanoma, 391 SCCHN). Digitized slides were used to train a convolutional neural network to quantify the number of CD8+ T cells in stroma, tumor parenchyma, parenchyma-stromal interface, and invasive margin. Generalized constrained regression models were used to predict GEP signatures specifically for stromal and parenchymal CD8+ T cells. Results: Parenchymal and stromal GEP scores were highly concordant with CD8+ infiltrate geography (adj- r2: 0.67, 0.65, respectively; P ≤ 0.01). Little overlap existed between gene sets associated with parenchymal and stromal CD8 T-cell geographies. CSF1R and NECTIN2 gene expression was observed to correlate inversely with parenchymal localization and directly with stromal CD8+ T-cell abundance. Conclusions: GEP signatures can be identified that are concordant with various CD8+ T-cell localization patterns in melanoma and SCCHN, demonstrating that GEP-IA can be developed to identify the immune status of interest in the TME. The specific genes identified have potential to elucidate mechanisms of resistance and/or inform I-O targets that can be further evaluated in relation to clinical significance in future studies.

Association of an inflammatory gene signature with CD8 expression by immunohistochemistry (IHC) in multiple tumor types

2593

Background: A multiparameter tumor inflammation assay based on gene expression profiling (TIA-GEP) can extend the utility of IHC to interrogate the tumor microenvironment (TME). Using CD8 expression assessed by IHC (CD8-IHC) as a surrogate for inflammation, statistical modelling was used to develop a specific gene signature on the TIA-GEP panel to predict CD8-IHC. The correlation between TIA-GEP and CD8-IHC and the prevalence of inflammation were explored across multiple tumor types. Methods: Levels of inflammation were measured by CD8-IHC and TIA-GEP on 1778 procured samples across 12 tumor types. Quality control metrics involved sample input quality, technical errors, and inter-run variability. Generalized linear models were used to identify an inflammation score that predicts the CD8-IHC score in melanoma and SCCHN tissue. The predictive accuracy of this signature was also examined in 10 additional tumor types. Results: Assessment of TME inflammation by CD8-IHC was consistent with that observed by TIA-GEP in multiple tumor types. The range of inflammation varied across different tumor types, with relatively lower inflammation range and scores in SCLC, ovarian, and prostate cancers, and higher values in NSCLC, melanoma, SCCHN, and gastric cancers. R2 x 100 values reflecting percent variation in CD8-IHC associated with TIA-GEP ranged from 62.4% to 79.2% ( P < 0.0001) for all tumor types except prostate cancer (32.5%). Low correlation in prostate cancer may be a result of low prevalence of inflammation by CD8-IHC. Estimated linear regression slopes between CD8-IHC and TIA-GEP ranged from 0.74 in SCLC to 1.27 in gastric cancer. Conclusions: The results suggest that the inflammation signature is a robust potential diagnostic tool predicting inflammation in the TME. The inflammation signature not only correlates with CD8-IHC for multiple tumor types, but also leverages the alternative benefits associated with TIA-GEP, which include information related to tumor inflammation-associated biomarkers and flexibility in exploring the value of other genomic signatures.

EMT- and stroma-related gene expression and resistance to PD-1 blockade in urothelial cancer

Cancers infiltrated with T-cells are associated with a higher likelihood of response to PD-1/PD-L1 blockade. Counterintuitively, a correlation between epithelial-mesenchymal transition (EMT)-related gene expression and T-cell infiltration has been observed across tumor types. Here we demonstrate, using The Cancer Genome Atlas (TCGA) urothelial cancer dataset, that although a gene expression-based measure of infiltrating T-cell abundance and EMT-related gene expression are positively correlated, these signatures convey disparate prognostic information. We further demonstrate that non-hematopoietic stromal cells are a major source of EMT-related gene expression in bulk urothelial cancer transcriptomes. Finally, using a cohort of patients with metastatic urothelial cancer treated with a PD-1 inhibitor, nivolumab, we demonstrate that in patients with T-cell infiltrated tumors, higher EMT/stroma-related gene expression is associated with lower response rates and shorter progression-free and overall survival. Together, our findings suggest a stroma-mediated source of immune resistance in urothelial cancer and provide rationale for co-targeting PD-1 and stromal elements.

STK11/LKB1 Mutations and PD-1 Inhibitor Resistance in KRAS-Mutant Lung Adenocarcinoma

KRAS is the most common oncogenic driver in lung adenocarcinoma (LUAC). We previously reported that STK11/LKB1 (KL) or TP53 (KP) comutations define distinct subgroups of KRAS-mutant LUAC. Here, we examine the efficacy of PD-1 inhibitors in these subgroups. Objective response rates to PD-1 blockade differed significantly among KL (7.4%), KP (35.7%), and K-only (28.6%) subgroups (P < 0.001) in the Stand Up To Cancer (SU2C) cohort (174 patients) with KRAS-mutant LUAC and in patients treated with nivolumab in the CheckMate-057 phase III trial (0% vs. 57.1% vs. 18.2%; P = 0.047). In the SU2C cohort, KL LUAC exhibited shorter progression-free (P < 0.001) and overall (P = 0.0015) survival compared with KRAS^MUT;STK11/LKB1^WT LUAC. Among 924 LUACs, STK11/LKB1 alterations were the only marker significantly associated with PD-L1 negativity in TMB^{Intermediate/High} LUAC. The impact of STK11/LKB1 alterations on clinical outcomes with PD-1/PD-L1 inhibitors extended to PD-L1-positive non-small cell lung cancer. In Kras-mutant murine LUAC models, Stk11/Lkb1 loss promoted PD-1/PD-L1 inhibitor resistance, suggesting a causal role. Our results identify STK11/LKB1 alterations as a major driver of primary resistance to PD-1 blockade in KRAS-mutant LUAC.Significance: This work identifies STK11/LKB1 alterations as the most prevalent genomic driver of primary resistance to PD-1 axis inhibitors in KRAS-mutant lung adenocarcinoma. Genomic profiling may enhance the predictive utility of PD-L1 expression and tumor mutation burden and facilitate establishment of personalized combination immunotherapy approaches for genomically defined LUAC subsets. Cancer Discov; 8(7); 822-35. ©2018 AACR.See related commentary by Etxeberria et al., p. 794This article is highlighted in the In This Issue feature, p. 781.

Limitations of high throughput methods for miRNA expression profiles in non-functioning pituitary adenomas

Microarray, RT-qPCR based arrays and next-generation-sequencing (NGS) are available high-throughput methods for miRNA profiling (miRNome). Analytical and biological performance of these methods were tested in identification of biologically relevant miRNAs in non-functioning pituitary adenomas (NFPA). miRNome of 4 normal pituitary (NP) and 8 NFPA samples was determined by these platforms and expression of 21 individual miRNAs was measured on 30 (20 NFPA and 10 NP) independent samples. Complex bioinformatics was used. 132 and 137 miRNAs were detected by all three platforms in NP and NFPA, respectively, of which 25 were differentially expressed (fold change > 2). The strongest correlation was observed between microarray and TaqMan-array, while the data obtained by NGS were the most discordant despite of various bioinformatics settings. As a technical validation we measured the expression of 21 selected miRNAs by individual RT-qPCR and we were able to validate 35.1%, 76.2% and 71.4% of the miRNAs revealed by SOLiD, TLDA and microarray result, respectively. We performed biological validation using an extended number of samples (20 NFPAs and 8 NPs). Technical and biological validation showed high correlation (p < 0.001; R = 0.96). Pathway and network analysis revealed several common pathways but no pathway showed the same activation score. Using the 25 platform-independent miRNAs developmental pathways were the top functional categories relevant for NFPA genesis. The difference among high-throughput platforms is of great importance and selection of screening method can influence experimental results. Validation by another platform is essential in order to avoid or to minimalize the platform specific errors.

Overexpression of GRß in colonic mucosal cell line partly reflects altered gene expression in colonic mucosa of patients with inflammatory bowel disease

The glucocorticoid receptor (GR) plays a crucial role in inflammatory responses. GR has several isoforms, of which the most deeply studied are the GRα and GRß. Recently it has been suggested that in addition to its negative dominant effect on GRα, the GRß may have a GRα-independent transcriptional activity. The GRß isoform was found to be frequently overexpressed in various autoimmune diseases, including inflammatory bowel disease (IBD). In this study, we wished to test whether the gene expression profile found in a GRß overexpressing intestinal cell line (Caco-2GRß) might mimic the gene expression alterations found in patients with IBD. Whole genome microarray analysis was performed in both normal and GRß overexpressing Caco-2 cell lines with and without dexamethasone treatment. IBD-related genes were identified from a meta-analysis of 245 microarrays available in online microarray deposits performed on intestinal mucosa samples from patients with IBD and healthy individuals. The differentially expressed genes were further studied using in silico pathway analysis. Overexpression of GRß altered a large proportion of genes that were not regulated by dexamethasone suggesting that GRß may have a GRα-independent role in the regulation of gene expression. About 10% of genes differentially expressed in colonic mucosa samples from IBD patients compared to normal subjects were also detected in Caco-2 GRß intestinal cell line. Common genes are involved in cell adhesion and cell proliferation. Overexpression of GRß in intestinal cells may affect appropriate mucosal repair and intact barrier function. The proposed novel role of GRß in intestinal epithelium warrants further studies.

Randomized Trial of Oral Cyclophosphamide and Veliparib in High-Grade Serous Ovarian, Primary Peritoneal, or Fallopian Tube Cancers, or BRCA-Mutant Ovarian Cancer

PURPOSE

Veliparib, a PARP inhibitor, demonstrated clinical activity in combination with oral cyclophosphamide in patients with BRCA-mutant solid tumors in a phase I trial. To define the relative contribution of PARP inhibition to the observed clinical activity, we conducted a randomized phase II trial to determine the response rate of veliparib in combination with cyclophosphamide compared with cyclophosphamide alone in patients with pretreated BRCA-mutant ovarian cancer or in patients with pretreated primary peritoneal, fallopian tube, or high-grade serous ovarian cancers (HGSOC).

EXPERIMENTAL DESIGN

Adult patients were randomized to receive cyclophosphamide alone (50 mg orally once daily) or with veliparib (60 mg orally once daily) in 21-day cycles. Crossover to the combination was allowed at disease progression.

RESULTS

Seventy-five patients were enrolled and 72 were evaluable for response; 38 received cyclophosphamide alone and 37 the combination as their initial treatment regimen. Treatment was well tolerated. One complete response was observed in each arm, with three partial responses (PR) in the combination arm and six PRs in the cyclophosphamide alone arm. Genetic sequence and expression analyses were performed for 211 genes involved in DNA repair; none of the detected genetic alterations were significantly associated with treatment benefit.

CONCLUSION

This is the first trial that evaluated single-agent, low-dose cyclophosphamide in HGSOC, peritoneal, fallopian tube, and BRCA-mutant ovarian cancers. It was well tolerated and clinical activity was observed; the addition of veliparib at 60 mg daily did not improve either the response rate or the median progression-free survival.