Association Between Biomarkers and Clinical Outcomes of Pembrolizumab Monotherapy in Patients With Metastatic Triple-Negative Breast Cancer: KEYNOTE-086 Exploratory Analysis

PURPOSE

In the two-cohort phase II KEYNOTE-086 study (ClinicalTrials.gov identifier: NCT02447003), first-line and second-line or later pembrolizumab monotherapy demonstrated antitumor activity in metastatic triple-negative breast cancer (mTNBC; N = 254). This exploratory analysis evaluates the association between prespecified molecular biomarkers and clinical outcomes.

METHODS

Cohort A enrolled patients with disease progression after one or more systemic therapies for metastatic disease irrespective of PD-L1 status; Cohort B enrolled patients with previously untreated PD-L1-positive (combined positive score [CPS] ≥ 1) metastatic disease. The association between the following biomarkers as continuous variables and clinical outcomes (objective response rate [ORR], progression-free survival [PFS], and overall survival [OS]) was evaluated: PD-L1 CPS (immunohistochemistry), cluster of differentiation 8 (CD8; immunohistochemistry), stromal tumor-infiltrating lymphocyte (sTIL; hematoxylin and eosin staining), tumor mutational burden (TMB; whole-exome sequencing [WES]), homologous recombination deficiency-loss of heterozygosity, mutational signature 3 (WES), mutational signature 2 (apolipoprotein B mRNA editing catalytic polypeptide-like; WES), T-cell-inflamed gene expression profile (Tcell_infGEP; RNA sequencing), and 10 non-Tcell_infGEP signatures (RNA sequencing); Wald test P values were calculated, and significance was prespecified at α = 0.05.

RESULTS

In the combined cohorts (A and B), PD-L1 (P = .040), CD8 (P < .001), sTILs (P = .012), TMB (P = .007), and Tcell_infGEP (P = .011) were significantly associated with ORR; CD8 (P < .001), TMB (P = .034), Signature 3 (P = .009), and Tcell_infGEP (P = .002) with PFS; and CD8 (P < .001), sTILs (P = .004), TMB (P = .025), and Tcell_infGEP (P = .001) with OS. None of the non-Tcell_infGEP signatures were associated with outcomes of pembrolizumab after adjusting for the Tcell_infGEP.

CONCLUSION

In this exploratory biomarker analysis from KEYNOTE-086, baseline tumor PD-L1, CD8, sTILs, TMB, and Tcell_infGEP were associated with improved clinical outcomes of pembrolizumab and may help identify patients with mTNBC who are most likely to respond to pembrolizumab monotherapy.

338 Effects of pembrolizumab on the tumor microenvironment (TME) after one presurgery treatment cycle in patients with triple-negative breast cancer (TNBC): phase 1b KEYNOTE-173 study

Background

In the phase 3 KEYNOTE-522 trial, neoadjuvant pembrolizumab+chemotherapy followed by adjuvant pembrolizumab monotherapy resulted in a statistically significant improvement in pathologic complete response (pCR) and event-free survival, compared with neoadjuvant chemotherapy alone, in patients with early-stage TNBC. In the phase 1b KEYNOTE-173 (NCT02622074) trial—another neoadjuvant pembrolizumab+chemotherapy trial—we evaluated TNBC biopsy samples at baseline and collected after one cycle of neoadjuvant pembrolizumab, before the initiation of chemotherapy, to explore the features within the TME at both timepoints that might be potentially predictive of clinical response and the effects of a single cycle of pembrolizumab on cell populations within the TME.

Methods

Twenty paired samples (baseline and obtained following one cycle of pembrolizumab before the initiation of chemotherapy) were included. Multiplex immunohistochemistry analyzed deconvoluted cell fractions by spatial localization (tumor compartment, stromal compartment, or total tumor) using three 6-plex panels: T cell (CD3/CD8/FoxP3/Ki67/granzyme B/PD-1), myeloid cell (CD68/CD163/MHCII/arginase/CD33/CD11c), and natural killer cell (CD16/CD56/CD11b/CD20/CD3/CD45). Area under the receiver operating characteristic (AUROC) was used to assess associations between immune subsets and pCR. Analyses were descriptive, with top-ranked findings reported, and were deemed hypothesis generating; no claims of statistical significance are made.

Results

At baseline, 6 of 75 evaluated immune subsets (counting different compartments) showed 95% CIs of AUROC not crossing 0.5 with pCR. These include some myeloid cell populations within the tumor compartment (AUROC, 95% CI), specifically CD11c+ (macrophage and dendritic cell [DC]: 0.85, 0.63–1.00), CD11c+/MHCII+/CD163–/CD68– (DC: 0.76, 0.53–0.99), CD11c+/MHCII–/CD163–/CD68– (nonactivated/immature DC: 0.8, 0.54-1.00), and CD11c+/CD163+ (M2 macrophage: 0.77, 0.55–0.99). Other associations with pCR included baseline CD11c+/MHCII–/CD163–/CD68– (nonactivated/immature DC) within the total tumor (AUROC, 0.76; 95% CI, 0.51–1.00) and the baseline ratio of CD11c/CD3 within the tumor compartment (AUROC, 0.75; 95% CI, 0.52–0.98). Although T-cell associations were relatively weak, specific CD8 subsets, especially CD8+/granzyme B+/Ki67+, showed a trend toward association. Negative correlations between change from baseline and baseline values were observed; therefore, baseline detrending was applied to change from baseline values. One immune subset showed a negative association trend between change from baseline and pCR after baseline detrending: CD163+/MHCII+ (DC3) within the stroma (AUROC, 0.2; 95% CI, 0.0–0.42).

Conclusions

Although the sample size in this exploratory analysis was small (n=20), myeloid cell populations within the tumor compartment at baseline show a promising association trend, as evaluated by AUROC, with pCR after neoadjuvant pembrolizumab+chemotherapy in early-stage TNBC. Changes in immune subsets following one cycle of pembrolizumab were not strongly associated with pCR.

Trial Registration

ClinicalTrials

gov, NCT02622074

Ethics Approval

The study protocol and all amendments were approved by the relevant institutional review board or ethics committee at each study site.

Consent

All patients provided written informed consent to participate in the clinical trial.

Abstract PD14-07: Association between biomarkers and response to pembrolizumab in patients with metastatic triple-negative breast cancer (mTNBC): Exploratory analysis from KEYNOTE-086

Background: In the phase 2 KEYNOTE-086 study (NCT02447003), pembrolizumab monotherapy had durable antitumor activity in a subset of patients with previously treated mTNBC (cohort A; n = 170) and in patients with previously untreated PD-L1-positive mTNBC (cohort B; n = 84). In this exploratory analysis of KEYNOTE-086, we evaluated the association between several biomarkers and response to pembrolizumab. Methods: Cohort A enrolled patients regardless of PD-L1 expression who had documented disease progression following ≥1 systemic therapy for metastatic disease. Cohort B enrolled patients with PD-L1-positive (combined positive score [CPS] ≥1) tumors who had not received prior systemic therapy for metastatic disease. Immunohistochemistry was used to measure PD-L1 CPS and CD8 density; H&E staining for percentage of stromal tumor infiltrating lymphocytes (sTILs); RNA-sequencing for 18-gene T-cell-inflamed gene expression profile (GEP), angiogenesis, and glycolysis signatures; and whole exome sequencing (paired tumor and germline) for TMB (TMB-H defined as ≥175 mut/exome), HRD-LOH (DNA damage), Signature 3, and APOBEC. Biomarkers were analyzed as continuous variables in cohorts A and B combined and individually. Area under the receiver operating characteristic curve was estimated between each biomarker and overall response rate (ORR). Wald test P-values were calculated using logistic regression adjusted for cohort and Eastern Cooperative Oncology Group performance status. Germline and somatic BRCA1/2 mutations were pooled; one-sided P-value was calculated using Fisher’s exact test. Spearman’s correlation was used for correlations. Results: Biomarker data were available in the following number of patients: 253 (99.6%; PD-L1), 204 (80.3%; CD8), 187 (73.6%; GEP), 171 (67.3%; TMB/HRD), 228 (89.8%; sTILs), 163 (64.2%; Signature 3/APOBEC), and 132 (52.0%; angiogenesis/glycolysis). When data from cohorts A and B were combined, PD-L1 CPS (median 2; IQR 0-10), CD8 (median 159; IQR 62-319), GEP (median -0.34; IQR -0.57 to -0.11), TMB (median 82 mut/exome; IQR 50-139), and sTILs (median 5; IQR 2-20) were significantly associated with ORR (Table). There were moderate correlations between PD-L1 and GEP (r = 0.532), PD-L1 and sTILs (r = 0.451), and GEP and sTILs (r = 0.490). No correlation was observed between TMB and PD-L1 (r = 0.038), GEP (r = -0.035), and sTILs (r = -0.031). When cohorts were combined, TMB was significantly associated with ORR, PFS, and OS after adjustment for PD-L1, GEP, CD8, or sTILs. HRD-LOH score, Signature 3, and APOBEC were not significantly associated with ORR (Table); P for BRCA1/2 was 0.2385. The angiogenesis signature was associated with lack of response while the glycolysis signature was associated with response to pembrolizumab (Table). Conclusions: In this exploratory biomarker analysis from KEYNOTE-086, higher levels of PD-L1, GEP, TMB, CD8 IHC, sTILs, and the glycolysis signature were associated with increased response to pembrolizumab monotherapy. These findings may help identify patients with mTNBC who are most likely to respond to pembrolizumab.

Table. Association of Biomarkers as Continuous Variables With Pembrolizumab Objective ResponseBiomarkerCombined Cohorts AUCCombined Cohorts P*Combined Cohorts Multitest corrected PCohort A AUCCohort B AUCPD-L10.6740.040-0.5440.654GEP0.7480.003-0.8370.561TMB0.6270.007-0.5480.710CD8 IHC0.760.000020.000120.850.68sTILs0.6710.012-0.6320.641HRD0.3940.874-0.5220.316Signature 30.6830.072-0.6970.736APOBEC0.5280.537-0.6740.623Angiogenesis0.6770.0090.0450.6610.731Glycolysis0.6120.0090.0360.8590.459AUC, area under the curve.*One-sided P-values are shown for all biomarkers except for Signature 3 and APOBEC, for which 2-sided P-values are shown.

Citation Format: Sherene Loi, Peter Schmid, Javier Cortes, David W. Cescon, Eric P. Winer, Deborah L. Toppmeyer, Hope S. Rugo, Michelino De Laurentiis, Rita Nanda, Hiroji Iwata, Ahmad Awada, Antoinette R. Tan, Roberto Salgado, Vassiliki Karantza, Petar Jelinic, Anran Wang, Lingkang Huang, Razvan Cristescu, Lakshman Annamalai, Jennifer Yearley, Jennifer Yearley, Sylvia Adams. Association between biomarkers and response to pembrolizumab in patients with metastatic triple-negative breast cancer (mTNBC): Exploratory analysis from KEYNOTE-086 [abstract]. In: Proceedings of the 2020 San Antonio Breast Cancer Virtual Symposium; 2020 Dec 8-11; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2021;81(4 Suppl):Abstract nr PD14-07.

Immune phenotype of patients with stage IV metastatic inflammatory breast cancer

Background

Inflammatory breast cancer (IBC) is a rare but aggressive carcinoma characterized by severe erythema and edema of the breast, with many patients presenting in advanced metastatic disease. The “inflammatory” nature is not due to classic immune-mediated inflammation, but instead results from tumor-mediated blockage of dermal lymphatic ducts. Previous work has shown that expression of PD-L1 on tumor cells can suppress T cell activation in triple-negative (TN) non-IBC breast cancer. In the present work, we investigated immune parameters in peripheral blood of metastatic IBC patients to determine whether cellular components of the immune system are altered, thereby contributing to pathogenesis of the disease. These immune parameters were also compared to PD-1 and PD-L1 expression in IBC tumor biopsies.

Methods

Flow cytometry-based immune phenotyping was performed using fresh peripheral blood from 14 stage IV IBC patients and compared to 11 healthy age-similar control women. Immunohistochemistry for CD20, CD3, PD-1, and PD-L1 was performed on tumor biopsies of these metastatic IBC patients.

Results

IBC patients with Stage IV disease had lymphopenia with significant reductions in circulating T, B, and NK cells. Reductions were observed in all subsets of CD4⁺ T cells, whereas reductions in CD8⁺ T cells were more concentrated in memory subsets. Immature cytokine-producing CD56^bright NK cells expressed higher levels of FcγRIIIa and cytolytic granule components, suggesting accelerated maturation to cytolytic CD56^dim cells. Immunohistochemical analysis of tumor biopsies demonstrated moderate to high expression of PD-1 in 18.2% of patients and of PD-L1 in 36.4% of patients. Interestingly, a positive correlation was observed between co-expression levels of PD-L1 and PD-1 in tumor biopsies, and higher expression of PD-L1 in tumor biopsies correlated with higher expression of cytolytic granule components in blood CD4⁺ T cells and CD56^dim NK cells, and higher numbers of CD8⁺ effector memory T cells in peripheral blood. PD-1 expression in tumor also correlated with increased infiltration of CD20⁺ B cells in the tumor.

Conclusions

Our results suggest that while lymphocyte populations are severely compromised in stage IV IBC patients, an immune response toward the tumor had occurred in some patients, providing biological rationale to evaluate PD-1/PD-L1 immunotherapies for IBC.

Supplementary information

The online version contains supplementary material available at 10.1186/s13058-020-01371-x.

NCOG-44. NEUROLOGIC ASSESSMENT IN NEURO-ONCOLOGY (NANO) SCALE IN A PHASE II STUDY OF PEMBROLIZUMAB OR PEMBROLIZUMAB PLUS BEVACIZUMAB IN PATIENTS WITH RECURRENT GLIOBLASTOMA

PURPOSE

The neurologic assessment in neuro-oncology (NANO) scale was developed as a standardized metric to objectively measure neurologic function in brain tumor patients to complement radiographic assessment in defining overall outcome. A multicenter, phase 2 study of pembrolizumab with or without bevacizumab in patients with recurrent glioblastoma incorporated the NANO scale as an exploratory endpoint.

METHODS

Neurologic examination was evaluated at baseline and MRI assessments using the NANO scale until patients came off study. Statistical descriptive data analysis was performed using R (version 3.4.3). Correlation analysis utilized Fisher’s exact test.

RESULTS

NANO compliance rate was 94% in 80 patients accrued on the study. Of the 80 patients, 7 were missing NANO at baseline visit and were excluded from analysis for NANO response criteria. Fifteen patients did not have end of treatment NANO evaluation. Of 73 patients, 35 (48%) had a normal neurologic examination at baseline by NANO. Strength and language accounted for the majority of changes in neurologic function over the course of study treatment. Eighteen patients (25%) had neurologic progression by NANO, of whom 2 did not have concurrent radiographic progression. Three patients (pembrolizumab plus bevacizumab cohort) had a neurologic response associated with stable disease on MRI. NANO assessment prior to initiation of cycle 3 correlated with RANO response (p=0.011), change in KPS (p=0.002) and dexamethasone requirement (p=0.007) while those with NANO progression at this assessment had worse overall survival (291 vs 324 days), but this trend did not achieve statistical significance (p=0.2).

CONCLUSIONS

Evaluation of neurologic function by NANO scale was feasible in a multicenter prospective study in patients with GBM with a high compliance rate. The NANO scale objectively tracked stable neurologic function in most patients throughout the trial period and was associated with a trend for survival.

CTIM-32. PHASE II AND BIOMARKER STUDY OF PEMBROLIZUMAB OR PEMBROLIZUMAB PLUS BEVACIZUMAB FOR RECURRENT GLIOBLASTOMA PATIENTS

PURPOSE

Vascular endothelial growth factor (VEGF) is upregulated in glioblastoma and may contribute to immunosuppression. We performed a phase 2 study of pembrolizumab, a programmed death-1 (PD-1) blocking antibody alone or with the anti-VEGF antibody bevacizumab in recurrent glioblastoma with detailed analyses of biomarkers and patient neurologic function.

METHODS

Eighty bevacizumab-naive, recurrent glioblastoma patients were randomized to receive pembrolizumab with bevacizumab (cohort A, n=50) or pembrolizumab monotherapy (cohort B, n=30). The primary endpoint was six-month progression-free survival (PFS-6). Exploratory endpoints included evaluation of tumor PD-L1 expression, TIL density, immune activation gene expression signature and plasma cytokines with outcome. Changes in neurologic function were prospectively assessed using the Neurologic Assessment in Neuro-Oncology (NANO) scale.

RESULTS

Pembrolizumab alone or with bevacizumab was well tolerated but of limited benefit. For cohort A, PFS-6 was 26.0% (95% CI: 16.3, 41.5), median OS was 8.8 months (95% CI: 7.7, 14.2), ORR was 20% and median duration of response was 48 weeks. For cohort B, PFS-6 was 6.7% (95% CI: 1.7, 25.4), median OS was 10.3 months (95% CI: 8.5, 12.5) and ORR was 0%. Factors associated with worsened OS included baseline dexamethasone use and increased post-therapy plasma VEGF (cohort A) and wild-type IDH1, unmethylated MGMT and increased baseline PlGF and sVEGFR1 levels (cohort B), but tumor immune markers were not informative. The NANO scale effectively predicted neurologic function.

CONCLUSIONS

Although well tolerated, pembrolizumab was ineffective both as monotherapy and with bevacizumab for recurrent glioblastoma. Nonetheless, radiographic responses to combinatorial therapy were durable. Baseline dexamethasone use and plasma cytokines but not tumor immunologic biomarkers were associated with outcome. Neurologic function evaluated by the NANO scale contributed to outcome assessment.

Development of a Humanized 3D Tumor Spheroid Models and Discovery of Effective Immunotherapy Strategies Against Resistant Microsatellite Stable Colorectal Cancers

Despite the current success of immunotherapies, only microsatellite instable (MSI) Colorectal Cancer (CRC) typically respond to T cell-driven immunotherapies while microsatellite stable (MSS) CRC displays poor response. Our goal is to identify resistance mechanisms in MSS-CRC to uncover novel therapeutic targets. Current syngeneic CRC models are largely immunogenic and lack complex structure. Therefore, we developed novel 3D humanized CRC mouse models which form complex human-like tumors which are typically absent in syngeneic systems. We investigated molecular profiles of human MSI-versus MSS-CRC tumor xenografts following Keytruda immunotherapy and identified an immunosuppressive profile in MSS models that was not observed in MSI-CRC. We evaluated the therapeutic potential of blocking inflammation in combination with other immuno-modulatory therapies in both cold and hot CRC models and identified a significant regression of tumor burden in MSS-CRC mice. Our studies provide a novel CRC in vivo model and a unique opportunity to better define how immune cells interact in human MSS- and MSI-CRC microenvironments, to enable clinical therapeutic implementation.

Pan-tumor genomic biomarkers for PD-1 checkpoint blockade-based immunotherapy

Programmed cell death protein-1 (PD-1) and programmed cell death ligand-1 (PD-L1) checkpoint blockade immunotherapy elicits durable antitumor effects in multiple cancers, yet not all patients respond. We report the evaluation of >300 patient samples across 22 tumor types from four KEYNOTE clinical trials. Tumor mutational burden (TMB) and a T cell-inflamed gene expression profile (GEP) exhibited joint predictive utility in identifying responders and nonresponders to the PD-1 antibody pembrolizumab. TMB and GEP were independently predictive of response and demonstrated low correlation, suggesting that they capture distinct features of neoantigenicity and T cell activation. Analysis of The Cancer Genome Atlas database showed TMB and GEP to have a low correlation, and analysis by joint stratification revealed biomarker-defined patterns of targetable-resistance biology. These biomarkers may have utility in clinical trial design by guiding rational selection of anti-PD-1 monotherapy and combination immunotherapy regimens.

Analytical validation of quantitative immunohistochemical assays of tumor infiltrating lymphocyte biomarkers

Tumor infiltrating lymphocytes (TIL), especially T-cells, have both prognostic and therapeutic applications. The presence of CD8+ effector T-cells and the ratio of CD8+ cells to FOXP3+ regulatory T-cells have been used as biomarkers of disease prognosis to predict response to various immunotherapies. Blocking the interaction between inhibitory receptors on T-cells and their ligands with therapeutic antibodies including atezolizumab, nivolumab, pembrolizumab and tremelimumab increases the immune response against cancer cells and has shown significant improvement in clinical benefits and survival in several different tumor types. The improved clinical outcome is presumed to be associated with a higher tumor infiltration; therefore, it is thought that more accurate methods for measuring the amount of TIL could assist prognosis and predict treatment response. We have developed and validated quantitative immunohistochemistry (IHC) assays for CD3, CD8 and FOXP3 for immunophenotyping T-lymphocytes in tumor tissue. Various types of formalin fixed, paraffin embedded (FFPE) tumor tissues were immunolabeled with anti-CD3, anti-CD8 and anti-FOXP3 antibodies using an IHC autostainer. The tumor area of stained tissues, including the invasive margin of the tumor, was scored by a pathologist (visual scoring) and by computer-based quantitative image analysis. Two image analysis scores were obtained for the staining of each biomarker: the percent positive cells in the tumor area and positive cells/mm² tumor area. Comparison of visual vs. image analysis scoring methods using regression analysis showed high correlation and indicated that quantitative image analysis can be used to score the number of positive cells in IHC stained slides. To demonstrate that the IHC assays produce consistent results in normal daily testing, we evaluated the specificity, sensitivity and reproducibility of the IHC assays using both visual and image analysis scoring methods. We found that CD3, CD8 and FOXP3 IHC assays met the fit-for-purpose analytical acceptance validation criteria and that they can be used to support clinical studies.

Patterns of PD-1, PD-L1, and PD-L2 expression in pediatric solid tumors

BACKGROUND

Significant antitumor effects have been observed in a variety of malignancies via blockade of immune checkpoints. Interaction of programmed death 1 (PD-1) with its ligands PD-L1 and PD-L2 suppresses T-cell function and restricts immune-mediated tumor killing. We examined expression of these proteins in children with solid tumors, as expression may serve as biomarkers of response to this class of drugs.

METHODS

Sections cut from formalin-fixed paraffin-embedded (FFPE) tissue blocks were processed and evaluated for PD-1, PD-L1, and PD-L2 by immunohistochemistry (IHC) as well as by mRNA expression. A semiquantitative 0-5 IHC scoring system (0 = negative to 5 = very high) was applied, with scores incorporating combined prevalence of tumor cell and nontumor cell labeling. Expression profiling was performed using the NanoString nCounter™ system. Data analysis was performed using quantile normalization. All quantile-normalized data underwent subsequent log10 transformation.

RESULTS

One hundred twenty-four FFPE blocks were included in the analysis. PD-1, PD-L1, and PD-L2 IHC were not evaluable in 8, 0, and 12 blocks, respectively. PD-1, PDL-1, and PDL-2 expression was negative to moderate by both IHC (range 0-3) and mRNA expression (range 0-2.62). Correlation between IHC score and mRNA expression was poor for all three tested proteins (PD-1, r² = 0.06; PDL-1, r² = 0.007; and PDL-2, r² = 0.15).

CONCLUSIONS

Expression of PD-1, PD-L1, and PD-L2 is low in pediatric solid tumors. At low levels of expression, IHC score and mRNA expression correlate poorly. Current and planned clinical trials will determine whether this low level of expression predicts limited response to immune checkpoint inhibitors.

IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade

Programmed death-1-directed (PD-1-directed) immune checkpoint blockade results in durable antitumor activity in many advanced malignancies. Recent studies suggest that IFN-γ is a critical driver of programmed death ligand-1 (PD-L1) expression in cancer and host cells, and baseline intratumoral T cell infiltration may improve response likelihood to anti-PD-1 therapies, including pembrolizumab. However, whether quantifying T cell-inflamed microenvironment is a useful pan-tumor determinant of PD-1-directed therapy response has not been rigorously evaluated. Here, we analyzed gene expression profiles (GEPs) using RNA from baseline tumor samples of pembrolizumab-treated patients. We identified immune-related signatures correlating with clinical benefit using a learn-and-confirm paradigm based on data from different clinical studies of pembrolizumab, starting with a small pilot of 19 melanoma patients and eventually defining a pan-tumor T cell-inflamed GEP in 220 patients with 9 cancers. Predictive value was independently confirmed and compared with that of PD-L1 immunohistochemistry in 96 patients with head and neck squamous cell carcinoma. The T cell-inflamed GEP contained IFN-γ-responsive genes related to antigen presentation, chemokine expression, cytotoxic activity, and adaptive immune resistance, and these features were necessary, but not always sufficient, for clinical benefit. The T cell-inflamed GEP has been developed into a clinical-grade assay that is currently being evaluated in ongoing pembrolizumab trials.

IFN-γ-related mRNA profile predicts clinical response to PD-1 blockade

Programmed death-1-directed (PD-1-directed) immune checkpoint blockade results in durable antitumor activity in many advanced malignancies. Recent studies suggest that IFN-γ is a critical driver of programmed death ligand-1 (PD-L1) expression in cancer and host cells, and baseline intratumoral T cell infiltration may improve response likelihood to anti-PD-1 therapies, including pembrolizumab. However, whether quantifying T cell-inflamed microenvironment is a useful pan-tumor determinant of PD-1-directed therapy response has not been rigorously evaluated. Here, we analyzed gene expression profiles (GEPs) using RNA from baseline tumor samples of pembrolizumab-treated patients. We identified immune-related signatures correlating with clinical benefit using a learn-and-confirm paradigm based on data from different clinical studies of pembrolizumab, starting with a small pilot of 19 melanoma patients and eventually defining a pan-tumor T cell-inflamed GEP in 220 patients with 9 cancers. Predictive value was independently confirmed and compared with that of PD-L1 immunohistochemistry in 96 patients with head and neck squamous cell carcinoma. The T cell-inflamed GEP contained IFN-γ-responsive genes related to antigen presentation, chemokine expression, cytotoxic activity, and adaptive immune resistance, and these features were necessary, but not always sufficient, for clinical benefit. The T cell-inflamed GEP has been developed into a clinical-grade assay that is currently being evaluated in ongoing pembrolizumab trials.

Immune characterization of the programmed death receptor pathway in high-risk prostate cancer

40

Background: Programmed Death 1 (PD-1) is a T cell inhibitory receptor critical to a major immunomodulation pathway which has been implicated in tumor evasion of immune response. PD-1 and its ligand PD-L1 have been found to be expressed in many tumor types, and this expression has led to the development of drugs targeting the PD-1 pathway. The goal of this study was to understand the expression of PD-1, and PD-L1 in high grade prostate cancer tissue, and correlate the expression with disease and patients characteristics. Methods: Immunohistochemistry for PD-1 (CD279), PD-L1 (B7-H1), and CD3 was performed on prostatectomy/biopsy tissue samples taken from 25 men with high grade (Gleason 8-10) prostate cancer using anti-PD-L1 clone 22C3 (Merck) and anti-PD-1 clone NAT105 (Cell Marque) . CD3 was used as a specific marker for T cell infiltration. Charts were then retrospectively reviewed for patient and disease characteristics including age at diagnosis, race, Gleason score, prostatic specific antigen (PSA) level at diagnosis, number of positive cores at biopsy, volume of tissue on biopsy and/or prostatectomy involved by cancer, clinical TNM stage, pathologic TNM stage, biochemical recurrence, or metastasis. Statistical analyses were done to correlate these patient and disease characteristics with PD-1, PD-L1, and CD3 expression. Results: A score of 3-5 on the semi-quantitative 0-5 score was deemed “high” expression whereas a score of 0-2 was deemed “low” expression. Of the 25 samples, 2 (8%) scored high for PD-1 expression, 2 (8%) scored high for PD-L1 expression, and 18 (72%) scored high for CD3 expression. With independent t-tests there was found to be no statistically significant correlation between any of the variables we collected and expression of PD-1, PD-L1, or CD3. Conclusions: We found an overall low expression of PD-1 and PD-L1, and a concurrent high expression of CD3+ T cells in high risk prostate cancer tissue. No significant correlations were made between expression of PD-1, PD-L1, or CD3 and patient and disease characteristics. The elevated T cell content in high risk prostate cancer is particularly interesting with the continued emergence of new drugs focused on enhancing efficacy of the immune response in cancer.

Mutational load (ML) and T-cell-inflamed microenvironment as predictors of response to pembrolizumab

1

Background: ML is associated with response to anti CTLA-4 and PD-1/L1 immunotherapies in some cancers, likely due to increased tumor antigenicity via generation of neoepitopes not subject to central immune tolerance. An IFNγ centric gene expression profile (GEP), characteristic of a T cell inflamed microenvironment, is also related to response to PD-1/PD-L1 therapies. The association between ML and outcome, and independent predictive values of ML and GEP in pembrolizumab treated patients across 21 tumor types were assessed. Methods: Whole exome and RNA sequencing were performed on FFPE specimens pooled from 2 multicohort advanced solid tumor trials (KEYNOTE 012 [n = 39] and 028 [n = 80]). The previously defined GEP score is a weighted sum of normalized expression values for 18 genes. ML and neoantigen analysis tools included GATK, MuTect and NetMHC. Statistical testing was adjusted for multiple testing. ML and GEP association was also explored in Moffitt and TCGA datasets. Results: ML and neoantigen load were significantly associated with objective response (OR) (AUROC = 0.76 and 0.78; p = 0.0036 and 0.0083, respectively). Median numbers of mutations were 180 in responders and 61 in nonresponders. The overall response rate (ORR) in all patients was 15%. At a ML cutoff of 102 (based on ROC Curve YoudenIndex), the ORR was 32.3% (prevalence 31.0%; NPV 92.8%). GEP was also significantly associated with OR (AUROC = 0.76; p = 0.0071). ML and GEP were modestly correlated (r = 0.28), consistent with associations between ML and GEP in Moffitt (r = 0.11) and TCGA databases (r = 0.29). When jointly modeled, ML was significantly associated with response (p = 0.0078) after adjusting for GEP (also significant; p = 0.0251). Data from additional patients will also be presented. Conclusions: These data show that ML and GEP are independently predictive biomarkers of response to pembrolizumab in multiple tumor types, suggesting that tumor antigenicity and T cell infiltration may provide complementary information for expected pembrolizumab activity. As measures of distinct, yet common features of a PD-1 checkpoint blockade responsive tumor, ML and GEP may have utility in characterizing responses to anti PD-1 and other cancer immunotherapies. Clinical trial information: NCT01848834; NCT02054806.

Correlation of T-cell infiltration and clonality with PD-L1 expression in soft tissue sarcomas

23

Background: The success of immunotherapy has raised new issues regarding the selection of patients, design of combination strategies, and sequencing of various regimens. Sarcomas have poor outcomes in the metastatic setting but may be amenable to immune therapies. However, we currently have limited knowledge of the immunologic profiles of different soft tissue sarcoma (STS) subtypes. Methods: We identified patients with the relatively common STS subtypes: leiomyosarcoma (LMS), undifferentiated pleomorphic sarcoma (UPS), synovial sarcoma (SS) and liposarcoma. Formalin fixed paraffin embedded (FFPE) tumor samples from 81 patients underwent gene expression analysis, immunohistochemistry for PD-1 and PD-L1, and sequencing of the T cell receptor Vβ region. Differences in liposarcoma subsets were also evaluated. Results: UPS and LMS had high expression levels of genes related to antigen presentation and T cell infiltration. UPS had higher levels of PD-L1 (p ≤ 0.001) and PD-1 (p ≤ 0.05) on IHC. UPS also had the highest T cell infiltration based on TCR sequencing, significantly more than SS, which had the lowest (p ≤ 0.05). UPS and LMS both had higher clonality compared with SS and liposarcoma (p ≤ 0.05). A model adjusted for STS histologic subtype found that for all sarcoma T cell infiltration and clonality were highly correlated with PD-1 and PD-L1 staining levels (p ≤ 0.01). Conclusions: In a model adjusted for sarcoma histologic subtypes, T cell infiltration and clonality were highly correlated with PD-1 and PD-L1 expression, consistent with the emerging view of tumor immunity that highly inflamed tumors acquire inhibitory ligands to evade tumor-specific T cells. UPS, which is a more highly mutated STS subtype, provokes a strong immune response evidenced by multiple inflammatory features suggesting that it may be well-suited to checkpoint inhibitor based approaches. SS and liposarcoma subsets are less highly mutated but do express immunogenic self-antigens therefore strategies to improve antigen presentation and T cell infiltration may be valuable for allowing immunotherapeutic success in these tumor types.

Abstract CT096: Clinical activity, immune and viral correlates of PD-1 blockade with pembrolizumab as first systemic therapy in patients with advanced Merkel cell carcinoma

Background: Merkel cell carcinoma (MCC) is an aggressive skin cancer linked to UV exposure and the Merkel cell polyomavirus (MCPyV). Cytotoxic chemotherapy is associated with high response rates in patients (pts) with metastatic MCC, although responses are seldom durable and median progression-free survival is only 13 weeks. MCC may be responsive to PD-1 pathway blockade as tumors often express PD-L1 and MCPyV-specific T cells express the inhibitory co-receptor PD-1.

Methods: In this open-label, phase II trial by the Cancer Immunotherapy Trials Network (NCT02267603), adults with advanced/metastatic MCC and no prior systemic therapy were enrolled. Pembrolizumab, a humanized IgG4 anti-PD-1 monoclonal antibody, was given IV every 3 weeks (2 mg/kg). Response was assessed by RECIST 1.1 every 9-12 weeks. Presence of MCPyV was assessed by measuring serum antibody titers to the MCPyV oncoprotein and/or quantifying tumor oncoprotein expression by immunohistochemistry (IHC).

Results: Twenty-six pts were treated (median duration 22 weeks, range 4 to 49 weeks as of 1/8/16 dataset). Twenty-three pts had at least one documented radiologic or clinical assessment of response. Fifteen of 23 evaluable pts (65%) responded (4 complete responses and 11 partial responses); 2 additional pts had stable disease and 6 had progressive disease as best response. Fourteen of 15 responses (93%) were ongoing at last follow-up. Ten responders had at least one follow-up evaluation with response duration ranging from 8+ to 27+ weeks beyond initial documented response. Treatment was generally well tolerated with mostly grade 1 or 2 adverse events. Two pts developed severe drug-related toxicities including grade 4 myocarditis (n = 1) and grade 4 transaminase elevation (n = 1), both of which improved with corticosteroid treatment. Both pts have ongoing tumor responses despite discontinuation of pembrolizumab. Sixteen of 26 pts (62%) were classified as having virus-positive MCC. Among the 23 evaluable pts, 9 of 14 (64%) of those with virus positive tumors responded, while 6 of 9 (67%) of pts with virus negative tumors responded. Analyses of T cells specific to MCPyV and titers of oncoprotein antibodies during therapy, as well as IHC characterization of tumor biopsies including PD-L1 expression are ongoing.

Conclusions: First-line therapy with pembrolizumab in pts with advanced MCC is associated with frequent (65%) and durable responses. Notably, the response rate was similar among pts with virus-positive and virus-negative tumors. While virus-positive tumors are known to express highly antigenic MCPyV oncoproteins, virus-negative tumors are typically UV-induced and display strikingly elevated mutational burdens generating putative neo-antigens. Thus, potentially through distinct mechanisms, both virus-positive and virus-negative MCC tumors appear to be immunogenic and frequently responsive to anti-PD-1 therapy.

Citation Format: Paul Nghiem, Shailender Bhatia, Evan Lipson, Ragini Kudchadkar, Natalie Miller, Adil Daud, Steven Fling, Philip Friedlander, Harriet Kluger, Holbrook Kohrt, Lisa Lundgren, Kim Margolin, Thomas Olencki, Sunil Reddy, Erica Shantha, William Sharfman, Elad Sharon, Michi Shinohara, John Thompson, Jennifer Yearley, Suzanne Topalian, Martin Cheever. Clinical activity, immune and viral correlates of PD-1 blockade with pembrolizumab as first systemic therapy in patients with advanced Merkel cell carcinoma. [abstract]. In: Proceedings of the 107th Annual Meeting of the American Association for Cancer Research; 2016 Apr 16-20; New Orleans, LA. Philadelphia (PA): AACR; Cancer Res 2016;76(14 Suppl):Abstract nr CT096.

Abstract P4-04-01: PDL-1 expression in primary breast cancers with germline mutations in BRCA 1 and 2

Background: The PD-1/PDL-1 pathway is a major inhibitory regulator of the immune response to tumors. Tumors responding to anti-PD-1 therapy appear to be characterized by a high "load" of somatic mutations, e.g.carcinogen-induced cancers (melanoma, lung) and tumors with a high intrinsic mutation rate (colorectal cancers with defects in DNA mismatch repair).High mutational load in these tumors may cause the expression of neoantigens which may induce an immune response unless the inhibitory PD-1/PDL-1 pathway is upregulated.Tumors in carriers of germline mutations in BRCA1/2 lack effective DNA repair and are genomically unstable, with a high mutational load and the possible expression of neoantigens. Breast cancers arising in carriers of germline mutations in BRCA1 and BRCA2 may rely on the PD-1/PDL-1 pathway to avoid immune destruction. Methods: Samples from thirty (30) treatment-naïve, primary breast cancers from 30 women with known germline mutations in BRCA 1 or BRCA2 were identified by records review in the Cedars-Sinai Department of Pathology, and selected for analysis of PD-1 and PDL-1 expression. Samples were analyzed for PD1 and PDL-1 expression utilizing immunohistochemical staining. Sections from FFPE tissue blocks were analysed with anti-PD-L1 clone 22C3(Merck); anti-PD-L1 clone SP142; or anti-PD-1 clone NAT105. IHC Scoring analyzed both tumor cell and non-tumor cell (inflammatory) infiltrate. A 0-5 scoring system (0= neg, 1 = rare, 2 = low, 3 = mod, 4 = high, 5 = very high) was applied. Tumors with IHC scores of 2 or greater were considered "positive" for expression of PD1 or PDL-1. Clinical Data: Age range 39-90 years, median age 61. Twenty(20) tumors were from BRCA 1 mutations carriers, and 10 were BRCA2. Sixteen (16) were basal type breast cancer (13 BRCA1, 3 BRCA2), 14 were estrogen receptor positive(ER+)(7 BRCA1, 7BRCA2), with 7 Luminal A and 7 Luminal B. There were no HER2 -amplified tumors in the cohort.Results: PD1 expression was observed in 11/30 (37%) of the cohort, and PDL-1 expression was detected in 21/30 (70%) of the cohort. PDL-1 expression was primarily seen in non-tumor, infiltrating immune cells. PDL-1 expression was seen in 15/20 (75%) of tumors from BRCA1 mutation carriers, and 6/10 (60%) of tumors from BRCA2 carriers. PDL-1 expression was present in 13/16 (81%) basal tumors and 8/14 (57%) ER+ tumors. Within the cohort of PDL-1 expressing cancers, 7/21 (33%) were scored as "high" or "very high", 5 basal breast cancers and 2 ER+ cancers. Conclusion: The current study has identified a high rate of PDL-1 expression in untreated primary breast cancers with germline BRCA1 and BRCA2 mutations, regardless of intrinsic subtype. Although the highest rate of expression was seen in basal breast cancers (81%), the majority of ER+ breast cancers, both Luminal A and Luminal B, also expressed PDL-1. Anti-PD-1 therapy has yielded a response rate of 19% in metastatic triple negative breast cancer, unselected for BRCA mutation.The current study suggests that tumors in carriers of germline mutations in BRCA1 or BRCA2 , regardless of intrinsic subtype, may rely on immune checkpoint inhibition for their growth and survival, and that therapy directed against the PD-1/PDL-1 pathway may be of benefit in this cohort of patients.

Citation Format: Audeh MW, Dadmanesh F, Yearley J. PDL-1 expression in primary breast cancers with germline mutations in BRCA 1 and 2. [abstract]. In: Proceedings of the Thirty-Eighth Annual CTRC-AACR San Antonio Breast Cancer Symposium: 2015 Dec 8-12; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2016;76(4 Suppl):Abstract nr P4-04-01.

Dissecting mechanisms underlying the pathogenesis of eosinophilic esophagitis (HYP6P.263)

Eosinophilic esophagitis (EoE) is an allergic disease characterized by esophageal eosinophilia, inflammation, and dysfunction. EoE has become increasingly common, but current management strategies are nonspecific. Thus, there is an urgent need to identify new pathways that could be targeted to treat EoE. Recently, EoE was associated with a gain-of-function polymorphism in the gene that encodes thymic stromal lymphopoietin (TSLP), a cytokine that promotes allergic inflammation and peripheral basophilia. However, how TSLP and basophils might contribute to the development of eosinophil responses during EoE remains unknown. Here, we employed a new murine model of EoE-like disease to investigate the role of TSLP and basophils in promoting esophageal eosinophil responses. Development of esophageal eosinophil responses was dependent on TSLP-elicited basophils, and antibody-mediated neutralization of TSLP or depletion of basophils ameliorated established esophageal eosinophilia. In addition, we examined how sort-purified human basophils influence eosinophil responses in vitro. Finally, elevated TSLP levels and exaggerated basophil responses observed in esophageal biopsies from EoE patients correlated with eosinophil responses. Together, these data indicate that TSLP-elicited basophil responses may play a key role in mediating eosinophil responses in EoE, suggesting that the TSLP-basophil axis could represent a new and promising therapeutic target to treat this disease.

A diet high in saturated fat and cholesterol accelerates simian immunodeficiency virus disease progression

Several lines of evidence suggest that dietary fat and cholesterol may play a role in the pathogenesis of human immunodeficiency virus (HIV) infection and disease progression. We examined the effect that an atherogenic diet (AD) high in saturated fatty acids and cholesterol has on disease progression and systemic inflammation in the simian immunodeficiency virus (SIV)-infected macaque model of acquired immunodeficiency syndrome. Macaques fed an AD had significantly more rapid disease progression, resulting in an increased risk of SIV-related death compared with that in control macaques (hazard ratio, 5.4 [95% confidence interval, 1.7-17.0]; P<.001). Peak viral load was higher in the AD group compared with control values, but further statistically significant differences were not detected at viral set point. The baseline plasma interleukin-18 level after 6 months of the AD was predictive of disease progression. Our findings may have important implications for HIV-infected individuals, because they suggest that dietary changes and manipulation of lipid metabolism could offer potential benefits by slowing disease progression.

Cytokine Modulation of the Innate Immune Response in Feline Immunodeficiency Virus–Infected Cats

BACKGROUND

In vitro data suggest that innate immune function in human immunodeficiency virus type 1-infected patients is compromised; however, in vivo studies are lacking. Feline immunodeficiency virus (FIV) infection in cats provides an excellent model to explore innate immune function in vivo. The innate response against Listeria monocytogenes is well understood, making it a useful immune probe.

METHODS

Recombinant L. monocytogenes carrying eukaryotic expression plasmids for feline tumor necrosis factor (TNF)- alpha , interleukin (IL)-10, interferon (IFN)- gamma , and IL-15 were created to determine whether specific cytokines would modulate innate immune function. L. monocytogenes was delivered subcutaneously, and local lymph nodes were evaluated for size, cell subpopulations, and L. monocytogenes burden. Two months later, memory responses were evaluated by IFN- gamma enzyme-linked immunospot assay.

RESULTS

FIV-positive cats had significantly less lymph-node enlargement and a greater L. monocytogenes burden than FIV-negative control cats. TNF- alpha improved listericidal activity in FIV-negative control cats but not in FIV-positive cats, whereas IL-10 modestly reduced function in FIV-negative control cats. IFN- gamma improved memory responses but not clearance of L. monocytogenes. IL-15 improved innate function in FIV-positive cats and increased the percentage of natural killer cells.

CONCLUSIONS

Lentivirus infection impairs innate immune function in vivo, and IL-15 can significantly restore function. We hypothesize that altered dendritic-cell function and increased regulatory T cell activity may underlie the innate immune defect in HIV infection.