Scoring PD-L1 Expression in Urothelial Carcinoma: An International Multi-Institutional Study on Comparison of Manual and Artificial Intelligence Measurement Model (AIM-PD-L1) Pathology Assessments

Assessing programmed death ligand 1 (PD-L1) expression on tumor cells (TCs) using Food and Drug Administration-approved, validated immunoassays can guide the use of immune checkpoint inhibitor (ICI) therapy in cancer treatment. However, substantial interobserver variability has been reported using these immunoassays. Artificial intelligence (AI) has the potential to accurately measure biomarker expression in tissue samples, but its reliability and comparability to standard manual scoring remain to be evaluated. This multinational study sought to compare the %TC scoring of PD-L1 expression in advanced urothelial carcinoma, assessed by either an AI Measurement Model (AIM-PD-L1) or expert pathologists. The concordance among pathologists and between pathologists and AIM-PD-L1 was determined. The positivity rate of ≥ 1%TC PD-L1 was between 20-30% for 8/10 pathologists, and the degree of agreement and scoring distribution for among pathologists and between pathologists and AIM-PD-L1 was similar both scored as a continuous variable or using the pre-defined cutoff. Numerically higher score variation was observed with the 22C3 assay than with the 28-8 assay. A 2-h training module on the 28-8 assay did not significantly impact manual assessment. Cases exhibiting significantly higher variability in the assessment of PD-L1 expression (mean absolute deviation > 10) were found to have patterns of PD-L1 staining that were more challenging to interpret. An improved understanding of sources of manual scoring variability can be applied to PD-L1 expression analysis in the clinical setting. In the future, the application of AI algorithms could serve as a valuable reference guide for pathologists while scoring PD-L1.

Deep computational image analysis of immune cell niches reveals treatment-specific outcome associations in lung cancer

The tumor immune composition influences prognosis and treatment sensitivity in lung cancer. The presence of effective adaptive immune responses is associated with increased clinical benefit after immune checkpoint blockers. Conversely, immunotherapy resistance can occur as a consequence of local T-cell exhaustion/dysfunction and upregulation of immunosuppressive signals and regulatory cells. Consequently, merely measuring the amount of tumor-infiltrating lymphocytes (TILs) may not accurately reflect the complexity of tumor-immune interactions and T-cell functional states and may not be valuable as a treatment-specific biomarker. In this work, we investigate an immune-related biomarker (PhenoTIL) and its value in associating with treatment-specific outcomes in non-small cell lung cancer (NSCLC). PhenoTIL is a novel computational pathology approach that uses machine learning to capture spatial interplay and infer functional features of immune cell niches associated with tumor rejection and patient outcomes. PhenoTIL's advantage is the computational characterization of the tumor immune microenvironment extracted from H&E-stained preparations. Association with clinical outcome and major non-small cell lung cancer (NSCLC) histology variants was studied in baseline tumor specimens from 1,774 lung cancer patients treated with immunotherapy and/or chemotherapy, including the clinical trial Checkmate 057 (NCT01673867).

Pharmacodynamic activity of BMS-986156, a glucocorticoid-induced TNF receptor-related protein agonist, alone or in combination with nivolumab in patients with advanced solid tumors

BACKGROUND

The success of immune checkpoint inhibitors has revolutionized cancer treatment options and triggered development of new complementary immunotherapeutic strategies, including T-cell co-stimulatory molecules, such as glucocorticoid-induced tumor necrosis factor receptor-related protein (GITR). BMS-986156 is a fully agonistic human immunoglobulin G subclass 1 monoclonal antibody targeting GITR. We recently presented the clinical data for BMS-986156 with or without nivolumab, which demonstrated no compelling evidence of clinical activity in patients with advanced solid tumors. Here, we further report the pharmacodynamic (PD) biomarker data from this open-label, first-in-human, phase I/IIa study of BMS-986156 ± nivolumab in patients with advanced solid tumors (NCT02598960).

MATERIALS AND METHODS

We analyzed PD changes of circulating immune cell subsets and cytokines in peripheral blood or serum samples collected from a dataset of 292 patients with solid tumors before and during treatment with BMS-986156 ± nivolumab. PD changes in the tumor immune microenvironment were measured by immunohistochemistry and a targeted gene expression panel.

RESULTS

BMS-986156 + nivolumab induced a significant increase in peripheral T-cell and natural killer (NK) cell proliferation and activation, accompanied by production of proinflammatory cytokines. However, no significant changes in expression of CD8A, programmed death-ligand 1, tumor necrosis factor receptor superfamily members, or key genes linked with functional parameters of T and NK cells were observed in tumor tissue upon treatment with BMS-986156.

CONCLUSIONS

Despite the robust evidence of peripheral PD activity of BMS-986156, with or without nivolumab, limited evidence of T- or NK cell activation in the tumor microenvironment was observed. The data therefore explain, at least in part, the lack of clinical activity of BMS-986156 with or without nivolumab in unselected populations of cancer patients.

Image analysis reveals molecularly distinct patterns of TILs in NSCLC associated with treatment outcome

Despite known histological, biological, and clinical differences between lung adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC), relatively little is known about the spatial differences in their corresponding immune contextures. Our study of over 1000 LUAD and LUSC tumors revealed that computationally derived patterns of tumor-infiltrating lymphocytes (TILs) on H&E images were different between LUAD (N = 421) and LUSC (N = 438), with TIL density being prognostic of overall survival in LUAD and spatial arrangement being more prognostically relevant in LUSC. In addition, the LUAD-specific TIL signature was associated with OS in an external validation set of 100 NSCLC treated with more than six different neoadjuvant chemotherapy regimens, and predictive of response to therapy in the clinical trial CA209-057 (n = 303). In LUAD, the prognostic TIL signature was primarily comprised of CD4⁺ T and CD8⁺ T cells, whereas in LUSC, the immune patterns were comprised of CD4⁺ T, CD8⁺ T, and CD20⁺ B cells. In both subtypes, prognostic TIL features were associated with transcriptomics-derived immune scores and biological pathways implicated in immune recognition, response, and evasion. Our results suggest the need for histologic subtype-specific TIL-based models for stratifying survival risk and predicting response to therapy. Our findings suggest that predictive models for response to therapy will need to account for the unique morphologic and molecular immune patterns as a function of histologic subtype of NSCLC.

Density patterns of tumor-infiltrating lymphocytes and association with objective response to nivolumab in patients with lung adenocarcinoma from CheckMate 057

2662

Background: Immune checkpoint inhibitors (ICIs) are currently approved for use as therapy in advanced stage lung adenocarcinoma (LUAD). ICIs can decrease risk of progression by up to 60% when compared to chemotherapy, but only about 20% of patients (pts) show response. Given that high levels of tumor-infiltrating lymphocytes (TILs) have been shown to be associated with better prognosis, here, we assess whether computationally derived TIL density measures on digitized H&E images can predict RECIST derived response to nivolumab in LUAD in Checkmate 057 (CM057). CM057 is a clinical trial designed to compare the overall survival of metastatic non-squamous non-small cell lung cancer subjects treated with either nivolumab or docetaxel after failing platinum-based chemotherapy. Methods: H&E-stained samples of 683 LUAD pts were collected from TCGA (n = 421), University of Bern (UBern) (n = 100), and CM057 (n = 162). Tumor response was assessed using RECIST v1.1. Samples were digitized as whole slide images. 294 pts randomly selected from TCGA formed the training set. The remaining 389 pts were used for validation in response to nivolumab in CM057 and prognosticating overall survival (OS) in UBern and TCGA. Computerized algorithms automatically identified TILs and extracted features related to quantity and compactness of TILs with respect to other surrounding nuclei. The top 6 features, determined by least absolute shrinkage and selection operator, were used to train Cox regression models that assign a death risk score to each patient. Pts with risk scores higher than the training median score were considered “high risk” or “non-responders” while pts with lower scores were considered “low risk” or “responders”. Results: The classifier predicted objective response in CM057 with an AUC = 0.61. Additionally, survival analysis showed that the model was prognostic for OS with hazard ratios of 2.38 (confidence interval (CI): 1.32-4.29, p = 0.01, n = 127) in TCGA and 2.37 (CI: 1.32-4.25, p < 0.01, n = 100) in UBern. Conclusions: A computerized image analysis model based on measurements of TIL density showed association with response to treatment in LUAD pts who received nivolumab and was prognostic of OS. Although the AUC was not high, the results suggest analysis of TILs has potential to identify pts who will respond to treatment. Future work will include training a classifier using response to treatment as endpoint and combining the TIL measures with other biomarkers like TMB or PD-L1.

387 The utility of AI-powered spatial classification of intratumoral CD8+ immune-cell distribution in predicting overall survival in patients with melanoma as part of the checkMate 067 clinical trial

Background

Spatial patterns of CD8+ T cells in the tumor microenvironment are associated with clinical outcomes in patients with advanced solid tumors. However, attempts to quantify spatial topology are hindered by challenges in manual scoring, heterogeneous immune-cell infiltrates, and interpathologist variability. Artificial intelligence (AI)–powered analysis can quantify CD8 topology in a biologically meaningful, reproducible, and scalable way. Using an AI-driven algorithm, we retrospectively assessed CD8 topology as a biomarker of response to immunotherapy in patients with advanced melanoma.

Methods

We trained a random forest classifier to predict CD8 topology using parenchymal and stromal CD8+ immune-cell measurements derived from a deep-learning platform (PathAI, Boston, MA). For model validation, pathologists manually classified CD8 immunohistochemistry (C8/144B, Agilent, Santa Clara, CA) in melanoma samples into inflamed (CD8+ cells in tumor parenchyma), excluded (CD8+ cells restricted to stroma), and desert (deficient in CD8+ cells) patterns. We explored the association with overall survival (OS) in a subset of patients with previously untreated metastatic melanoma who received nivolumab + ipilimumab (NIVO+IPI, n=102) or NIVO alone (n=107) in the CheckMate 067 phase 3 trial. Retrospective analysis of baseline AI-defined CD8 topology was performed alone and combined with manually scored programmed death ligand 1 (PD-L1) expression on tumor cells.

Results

Classifier model predictions were concordant with manual scoring (determined by a consensus of pathologists) and non-inferior to the agreement between 2 pathologists, via Cohen’s kappa coefficient k=0.79 and k=0.65, respectively. No statistically meaningful differences in outcomes were observed between CD8-excluded and CD8-inflamed phenotypes within the PD-L1 ≥1% population. However, patients with PD-L1 <1%/CD8-excluded tumors exhibited longer median OS compared with those with PD-L1 <1%/CD8-inflamed (table 1). 38% (40/104) of PD-L1 <1% tumors were CD8-excluded. Within PD-L1 <1%, patients with an excluded phenotype also exhibited lower frequency of severe adverse events (grade ≥3) than patients with inflamed phenotype following treatment: NIVO+IPI, 75% (n=20) vs 91% (n=11); NIVO, 61% (n=18) vs 80% (n=15). Compared with PD-L1 status, the composite biomarker (AI-classified CD8-excluded plus PD-L1 ≥1%) identified a larger group of patients who had greater survival benefit with NIVO+IPI or NIVO alone (table 2).

Abstract 387 Table 1

Immunotherapy outcomes by CD8+ topology in PD-L1<1% melanoma

In a subset of patients with melanoma and tumor cell PD-L1 expression <1% in the CheckMate 067 clinical trial, those with a CD8-excluded phenotype demonstrated longer overall survival compared with those with a CD8-inflamed phenotype when treated with NIVO±IPI.

Abstract 387 Table 2

Composite biomarker outcomes in Checkmate 067

In patients with melanoma in the CheckMate 067 clinical trial, the composite biomarker (AI-classified CD8-excluded phenotype plus PD-L1 expression ≥1%) identified more biomarker-positive patients and demonstrated increased overall survival benefit vs PD-L1 status alone for patients treated with NIVO±IPI. Hazard ratios represent patients with a PD-L1 expression of ≥1% compared with PD-L1 <1% or patients with a PD-L1 expression of ≥1% and CD8-excluded phenotype compared with PD-L1 expression <1% and not CD8-excluded.

Conclusions

This study explores the utility of combining AI-powered CD8 topology classifications with PD-L1 expression as a composite biomarker associated with immunotherapy response. In patients with PD-L1 <1% melanoma, median OS with NIVO+IPI was significantly longer in patients with CD8-excluded tumors than with an inflamed phenotype. Further studies are underway to identify mechanisms underlying responses to NIVO+IPI.

Acknowledgements

We would like to thank the team at PathAI for development of the AI classifier, and Dako, an Agilent Technologies, Inc. company, for collaborative development of the PD-L1 IHC 28-8 pharmDx assay. Editorial support was provided by Emily Motola, PharmD, and Matthew Weddig of Spark Medica Inc.

Trial Registration

Clinicaltrials

gov number NCT01844505

Ethics Approval

The study protocol and all amendments were approved by local institutional review boards, and the protocol was conducted in accordance with the Declaration of Helsinki and Good Clinical Practice Guidelines, as defined by the International Conference on Harmonisation of Technical Requirements for Pharmaceuticals for Human Use. All patients provided written informed consent before enrollment.

821 Machine learning models can quantify CD8 positivity in lymphocytes in melanoma clinical trial samples

Background

An accurate histological characterization of immune cells in the tumor microenvironment is essential for developing novel immune oncology targeted therapies and can assist in guiding patient treatment decisions. However, immune phenotyping is subject to challenges of manual scoring and inter-pathologist scoring variability. To support pathologist-scored immune phenotyping across tumor types, we are developing machine learning (ML)-based models that can identify and quantify CD8+ lymphocytes within the stromal and parenchyma regions of tumors from non-small cell lung cancer, renal cell carcinoma, breast cancer, gastric cancer, head and neck squamous cell carcinoma, urothelial carcinoma, and melanoma. Here, we focus on the ML model for melanoma showing recent results for ML-based identification and quantification of CD8+ lymphocytes and concordance with manual pathologic assessment in data derived from clinical trials.

Methods

ML algorithms were developed to quantify CD8+ lymphocytes in melanoma using 200 samples from a commercial dataset containing both primary and metastatic melanoma cases. Models were trained using the PathAI research platform on digitized whole slide images (WSI) stained for CD8 using clone C8/144b (Dako), and annotations were provided by the PathAI network of expert pathologists. Training included identification of slide artifacts, parenchyma, cancer stroma, and necrosis, as well as CD8+ lymphocytes and other CD8– cell types. Examples of melanin, such as pigmented macrophages, were added to non-CD8+ cell types. To evaluate the performance of the ML model, model-predicted CD8+ counts were compared to a consensus count from five independent pathologists for representative regions (“frames”) using the Pearson correlation. This was done in 112 held-out test frames from 90 WSI baseline samples from three clinical trials of immunotherapy treatment in individuals with metastatic melanoma. Inter-pathologist agreement among the five pathologists was also calculated.

Results

ML-based quantitation of CD8 positivity in lymphocytes showed high concordance with manual pathologist consensus counts. In frames validation of CD8+ counts on the test set of WSI, there was high correlation between the ML model and pathologist consensus counts (r=0.92 [95% CI 0.88–0.94]). This correlation was comparable to the agreement among the five expert pathologists (r=0.88 [95% CI 0.85–0.91]).

Conclusions

ML model-predicted CD8+ cell counts are highly concordant with pathologist scores on WSI samples from melanoma-focused clinical trials. These data demonstrate the capability of AI-powered digital pathology for accurate and reproducible quantitation of CD8+ lymphocytes in clinical trial samples, contributing to improved evaluation of the tumor microenvironment and targeted development of therapeutics.

Abstract 2017: Association of digital and manual quantification of tumor PD-L1 expression with outcomes in nivolumab-treated patients

Background: Programmed death ligand 1 (PD-L1) expression on tumor cells (TC), detected by immunohistochemistry (IHC), is associated with response to programmed death-1 (PD-1)/PD-L1 inhibitors in some tumor types. Manual review of PD-L1–positive (PD-L1+) tumors can be subjective, with the potential for misclassification of PD-L1–low tumors as PD-L1–negative due to weak positivity. We compared artificial-intelligence (digital) and manual scoring methods and assessed the association of PD-L1 expression with clinical outcomes in nivolumab (NIVO)-treated patients with urothelial carcinoma (UC) and melanoma (MEL).

Methods: PD-L1 expression was determined in baseline samples from NIVO monotherapy-treated patients with UC (CM275, NCT02387996) and MEL (CM067, NCT01844505; CM238, NCT02388906) using the Dako PD-L1 IHC 28-8 pharmDx assay. PD-L1+ TC were scored using digital (PathAI research platform) and manual (LabCorp) methods. Prevalence of tumors with PD-L1+ TC ≥ 1% and ≥ 5% and associations between PD-L1 expression and outcomes with NIVO were evaluated.

Results: Prevalence of UC and MEL tumors with ≥ 1% and ≥ 5% PD-L1+ TC was higher for digital vs manual scoring (Table). For all samples, digital and manual scoring was associated with response to NIVO for PD-L1 ≥ 1% and ≥ 5%, and associations were similar between digital and manual scoring (Table). Digital and manual PD-L1 scoring correlated across samples from all trials (Kendall's tau range: 0.57–0.62).

TablePrevalence PD-L1+ TC ≥ 1%, n (%)Evaluable samples, nDigitalManualP valueSamples ≥ 1% by digital onlyCM275241166 (69)113 (47)1.61 × 10−658 (24)CM067264173 (66)160 (61)0.27936 (14)CM238377307 (81)259 (69)7.61 × 10−566 (18)PD-L1+ TC ≥ 1% vs &lt; 1%DigitalManualORR, odds ratio (95% CI)CM275a2.15 (0.98–4.70)1.60 (0.82–3.14)CM067b1.99 (1.19–3.35)1.89 (1.12–3.18)Survival, hazard ratio (95% CI)CM275 (OS)a0.67 (0.48–0.92)0.66 (0.48–0.90)CM067 (OS)b0.57 (0.41–0.80)0.71 (0.50–1.00)CM238 (RFS)c0.53 (0.36–0.77)0.83 (0.57–1.21)Prevalence PD-L1+ TC ≥ 5%, n (%)Evaluable samples, nDigitalManualP valueSamples ≥ 5% by digital onlyCM27524190 (37)74 (31)0.14928 (12)CM067264103 (39)76 (29)0.01736 (14)CM238377234 (62)139 (37)7.54 × 10−12104 (28)PD-L1+ TC ≥ 5% vs &lt; 5%DigitalManualORR, odds ratio (95% CI)CM275a3.50 (1.76–6.98)2.37 (1.18–4.73)CM067b2.33 (1.40–3.86)1.77 (1.01–3.09)Survival, hazard ratio (95% CI)CM275 (OS)a0.50 (0.36–0.71)0.58 (0.41–0.83)CM067 (OS)b0.67 (0.47–0.96)0.74 (0.51–1.09)CM238 (RFS)c0.50 (0.35–0.70)0.52 (0.36–0.76)Database lock 2019: CM275, June 14; CM067, January 18; CM238, April 3.aAdjusted for ECOG performance status, liver metastatic status, and hemoglobin.bAdjusted for ECOG performance status, liver metastatic status, lactate dehydrogenase, and BRAF mutation.cAdjusted for ECOG performance status, AJCC stage, lactate dehydrogenase, and BRAF mutation.AJCC, American Joint Committee on Cancer; CI, confidence interval; ECOG, Eastern Cooperative Oncology Group; ORR, objective response rate; OS, overall survival; PD-L1, programmed death ligand 1; RFS, recurrence-free survival; TC, tumor cells.

Conclusion: In post-hoc exploratory analyses, digital scoring of PD-L1 expression identified higher prevalence of PD-L1+ tumors and shows good association with response to NIVO in UC and MEL samples compared with manual scoring. Digital quantification demonstrated higher sensitivity at low levels of PD-L1 expression and may identify patients who could benefit from NIVO. Further study of the association with clinical outcomes is warranted and exploratory studies are ongoing to assess the performance of digital scoring in additional tumor types.

Citation Format: Chunzhe Duan, Michael Montalto, George Lee, Dimple Pandya, Daniel Cohen, Han Chang, Hao Tang, Nishant Agrawal, Hunter Elliott, Benjamin Glass, Ilan Wapinski, Robin Edwards, Andrew H. Beck, Vipul Baxi. Association of digital and manual quantification of tumor PD-L1 expression with outcomes in nivolumab-treated patients [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 2017.

Elevated serum interleukin-8 is associated with enhanced intratumor neutrophils and reduced clinical benefit of immune-checkpoint inhibitors

Serum interleukin-8 (IL-8) levels and tumor neutrophil infiltration are associated with worse prognosis in advanced cancers. Here, using a large-scale retrospective analysis, we show that elevated baseline serum IL-8 levels are associated with poor outcome in patients (n = 1,344) with advanced cancers treated with nivolumab and/or ipilimumab, everolimus or docetaxel in phase 3 clinical trials, revealing the importance of assessing serum IL-8 levels in identifying unfavorable tumor immunobiology and as an independent biomarker in patients receiving immune-checkpoint inhibitors.

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.