PD-L1 IHC Research Articles

Genomics and transcriptomics of pancreatic adenosquamous carcinoma.

691 Background: Pancreatic cancer (PC) is anticipated to become the 2nd leading cause of cancer-related deaths by 2030. Pancreatic adenosquamous carcinoma (PASC) is a rare subtype of PC with histology including squamous adenomatous features (≥30% squamous histology), while pancreatic squamous cell cancer (PSCC) consists purely of squamous cells. We evaluated genomic, transcriptomic and prognostic differences between PASC, PSCC and pancreatic ductal adenocarcinoma (PDAC). Methods: 8951 PC tumors (83 PASC, 22 PSCC & 8846 PDAC) tested at Caris Life Sciences with WTS (Illumina, Novaseq) and NextGen DNA sequencing (NextSeq, 592 genes and NovaSeq, WES) and IHC were analyzed. Immune cell fraction was calculated by QuantiSeq. Overall survival (OS) analysis (time of tissue collection to last contact) was obtained from insurance claims and calculated with KM method. Statistical significance was determined using Chi-square/Fisher-Exact and adjusted for multiple comparisons (q<0.05). Results: No age or sex differences were seen between PDAC (median 68yo) and PASC (median 68.5yo), however PSCC was more prevalent in younger pts (median 62.5yo, q<0.05). Mutation rates of CDKN1B (3% vs 0.2%), SF3B1 (8 vs 2%), PTEN (5 vs 0.8%), BCL9 (4 vs 0.4%), AXIN1 (4 vs 0.2%) and CASP8 (3 vs 0.2%) and amplifications of AKT2 (5 vs 2%), ZNF384 (1 VS 0.1%), HMGA2 (7 vs 0.7%) and ROS1 fusions (1 vs 0.04%) all trended higher in PASC than PDAC (p<0.05). Several YAP signaling and squamous differentiation genes were significantly differentially expressed among histologies (table). PD-L1 IHC was higher in PASC vs PDAC (37 vs 14%, q<0.05); with a trend towards higher MSI-H (6 vs 1%) & TMB-H (6 vs 2%) in PASC vs PDAC (p<0.05). Higher infiltration of CD4+ T cells into the TME was seen in PASC vs PDAC (non-zero %, 73% vs 23%, q<0.05). Several immune genes ( CD274, IDO1, LAG3, & CTLA4) had higher expression in PASC vs PDAC (fold change 1.3-3.1, q<0.05). The MAP kinase pathway activation score (MPAS) was lower in PASC vs PDAC (median score -0.48 vs 0.25), while the IFN gamma signature was higher in PASC vs PDAC (median -0.24 vs -0.33) [both q<0.05]. There was increased OS for PASC compared to PSCC (12.6 vs 4.7 mo, HR=0.43, CI=0.25-0.75, p=0.002); no significant difference in OS was observed between PASC and PDAC. Conclusions: This is the largest molecular profiling analysis of PASC, which is characterized by unique genomic alterations, and is associated with higher PD-L1 expression, immune related gene expression, CD4+ T cell infiltration and IFN gamma signature, and lower MAPK activation. PASC is associated with better OS compared to PSCC. These findings may provide subtype-specific therapeutic opportunities for PASC and PSCC pts.[Table: see text]

Characterization of the cachexia pathway in pancreatic ductal adenocarcinoma.

699 Background: Cancer cachexia is characterized by progressive weight loss and skeletal muscle degradation, contributing to 33% of pancreatic ductal adenocarcinoma (PDAC) deaths. Targeting myostatin in the myostatin-activin pathway has been shown to reverse cachexia. Here, we present comprehensive clinical and molecular characterization of the myostatin-activin pathway in PDAC. Methods: 9,607 samples of PDAC tested at Caris Life Sciences (Phoenix, AZ) with WTS (Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES) were analyzed. Cachexia gene scores (GS) were calculated by averaging the positive z scores of activators and negative z scores of repressors in the myostatin-activin pathway. Activators were ACVR1B, ACVR1C, ACVR2A, ACVR2B, SMAD2, SMAD3, SMAD4, and TGFBR1, while repressors were SMAD1, SMAD5, SMAD6, SMAD7, SMURF1, and SMURF2. The top quartile (Q4) and bottom quartile (Q1) of GS were compared using chi-squared and Fisher-Exact tests. RNA deconvolution analysis with QuantiSEQ estimated cell infiltration in the tumor microenvironment (TME). Differences in overall survival (OS) were analyzed from insurance claims data and calculated from time of tissue collection using Kaplan-Meier estimates. Statistical significance was determined as a P-value adjusted for multiple comparisons ( q<0.05). Results: GS were higher in primary tumors compared to metastases (median: -0.71 vs -0.86, q<0.05). GS was associated with increased PD-L1 IHC expression (Q1 21.2% vs Q4 10.3%, q<0.001), but not TMB-high (1.9% vs 2.1%, q=1) or MSI-H status (1.1% vs 1.3%, q=1). GS correlated with increased expression of immune related genes ( CD274, CD80, IDO1, CD86, PDCD1, LAG3, CTLA4, HAVCR2, and IFNG, q<0.05). However, TME immune cell infiltration did not vary based on cachexia GS. Mutation rates of TP53 (Q1 79.4% vs Q4 73.7%), ARID1A (Q1 11.8% vs Q4 6.9%) and KRAS (Q1 92.7% vs Q4 86.4%) were associated with Q1-GS (all q<0.01), while STK11 mutations (1.1% vs 3.0%, q=0.001) were associated with Q4-GS. Spearman correlation linked cachexia GS with the lipid metabolizing genes UCP2 (rho=0.49) and UCP3 (rho=0.30), as well as the inflammatory markers CCL2/MCP-1(rho=0.34) and IL1B (rho=0.33). Decreased OS was seen with higher tumor expression of the myostatin-activin pathway activator, SMAD3 (6.6 vs 8.5 mo, HR=1.22, CI 1.12-1.33, P<0.001), and lower expression of the repressor SMAD7 (6.9 vs 8.9 mo, HR=0.91, CI 0.79-0.90, P=0.034) in metastatic PDAC. Conclusions: This is the largest molecular and clinical characterization of the myostatin-activin cachexia pathway in PDAC. Our data show that increased activation of the myostatin-activin pathway is associated with immune mediators, lipid metabolism, and inflammatory gene activation. Activators and repressors are significant predictors of survival in PDAC, suggesting possible novel therapeutic targets.

Abstract C045: Uncovering molecular differences in pancreatic ductal adenocarcinoma tissues from Black and White patients in the US

Abstract Background. Pancreatic cancer is the 3rd leading cause of cancer-related deaths in the US, with a 5-year survival rate of approximately 12%. The occurrence of new pancreatic cancer cases has continued to rise in the last two decades, with the highest incidence observed in the Black and African American (BAA) population. While socioeconomic and environmental factors may contribute to observed racial disparities in the occurrence of pancreatic cancer, it is unclear whether there are molecular differences in pancreatic tumors across racially and ethnically diverse populations. The goal of this study was to identify key molecular differences between BAA and White patients’ pancreatic tumors. Methods. We analyzed 4620 de-identified records with a primary diagnosis of pancreatic ductal adenocarcinoma (PDAC) who underwent somatic NGS testing via a commercially available 595-648 gene DNA panel as well as whole transcriptome RNA-seq. Race was extracted from clinical records (e.g., order forms, patient records, etc.). Somatic mutations, microsatellite instability (MSI), and tumor mutational burden (TMB) were compared between White (n=3797) and BAA (n=452) samples by Chi-squared or Fisher’s Exact tests. False discovery rate (FDR) corrections were applied to pairwise comparisons. Results. The combined cohort had a median age of 68 years and 88% of records had stage 3 or 4 disease at the time of diagnosis. We found that alterations in TP53 (70% vs 62%, q=0.014) and KMT2C (6.9% vs 2.3% q<0.001) were at significantly higher frequencies in BAA vs White samples. Weakly significant differences were observed in BAA vs White samples based on alterations in KRAS (79% vs 74%, p=0.02), RNF43 (6% vs 3.7% p=0.022), and GNAS (0.2% vs 1.6%, p=0.022), all of which had q-values of 0.14 after FDR correction. When stratifying by detailed alterations, KRAS G12R was significantly more frequent in BAA than White (14% vs 10%, q=0.037), and GNAS R201H was more frequent in White vs BAA (0.9% vs. 0%, q=0.056) samples. Finally, when assessing biomarkers associated with immunotherapy response in solid tumors, we found that neither the frequency of TMB-H (≥ 10 mut/MB) nor MSI-high status were significantly different between BAA and White samples. Interestingly, in the subset of 676 PDAC patients who had PD-L1 IHC performed previously, BAA samples had a higher frequency of PD-L1 positivity compared with White samples (20% vs 12%, p=0.039). Conclusions. We assessed molecular differences between BAA and White patients’ tumors in the largest cohort of PDAC records reported to date. Our findings suggest that there are key molecular tumor characteristics which occur more frequently in BAA PDAC tumors. In particular, alterations in genes such as KMT2C, which were more frequent in our cohort of BAA tumors, are known contributors to pancreatic tumorigenesis and represent potential therapeutic targets. These findings highlight the importance of studying racially and ethnically diverse populations to better understand key molecular alterations. Citation Format: Emily Teslow, Minxuan Huang, Melissa Stoppler, Albert Levin, Ling Huang. Uncovering molecular differences in pancreatic ductal adenocarcinoma tissues from Black and White patients in the US [abstract]. In: Proceedings of the 16th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved; 2023 Sep 29-Oct 2;Orlando, FL. Philadelphia (PA): AACR; Cancer Epidemiol Biomarkers Prev 2023;32(12 Suppl):Abstract nr C045.

PD-L1 IHC 22C3 pharmDx Demonstrates Precision and Reproducibility in Triple-negative Breast Cancer

In the United States, the PD-L1 IHC 22C3 pharmDx indications for use have expanded to include the assay’s use as an aid in identifying triple-negative breast cancer (TNBC) patients for treatment with pembrolizumab (KEYTRUDA®). Analytical validation for TNBC was conducted by Agilent Technologies in support of the device performance and FDA-approval. Analytical validation studies included device precision (inter-instrument, inter-operator, inter-day, inter-lot, and intra-run) and external reproducibility (inter- and intra-site, inter- and intra-observer) using the Combined Positive Score (CPS) ≥ 10 cutoff. All precision and external reproducibility studies achieved confidence interval lower bound values of greater than 85% for positive, negative, and overall percent agreement. These analytical validation studies demonstrate that PD-L1 IHC 22C3 pharmDx is precise and reproducible in evaluating PD-L1 protein expression at a CPS ≥ 10 cutoff when used for testing TNBC tissues.

Comparison of SP142 and 22C3 PD-L1 assays in a population-based cohort of triple-negative breast cancer patients in the context of their clinically established scoring algorithms

BackgroundImmunohistochemical (IHC) PD-L1 expression is commonly employed as predictive biomarker for checkpoint inhibitors in triple-negative breast cancer (TNBC). However, IHC evaluation methods are non-uniform and further studies are needed to optimize clinical utility.MethodsWe compared the concordance, prognostic value and gene expression between PD-L1 IHC expression by SP142 immune cell (IC) score and 22C3 combined positive score (CPS; companion IHC diagnostic assays for atezolizumab and pembrolizumab, respectively) in a population-based cohort of 232 early-stage TNBC patients.ResultsThe expression rates of PD-L1 for SP142 IC ≥ 1%, 22C3 CPS ≥ 10, 22C3 CPS ≥ 1 and 22C3 IC ≥ 1% were 50.9%, 27.2%, 53.9% and 41.8%, respectively. The analytical concordance (kappa values) between SP142 IC+ and these three different 22C3 scorings were 73.7% (0.48, weak agreement), 81.5% (0.63) and 86.6% (0.73), respectively. The SP142 assay was better at identifying 22C3 positive tumors than the 22C3 assay was at detecting SP142 positive tumors. PD-L1 (CD274) gene expression (mRNA) showed a strong positive association with all two-categorical IHC scorings of the PD-L1 expression, irrespective of antibody and cut-off (Spearman Rho ranged from 0.59 to 0.62; all p-values < 0.001). PD-L1 IHC positivity and abundance of tumor infiltrating lymphocytes were of positive prognostic value in univariable regression analyses in patients treated with (neo)adjuvant chemotherapy, where it was strongest for 22C3 CPS ≥ 10 and distant relapse-free interval (HR = 0.18, p = 0.019). However, PD-L1 status was not independently prognostic when adjusting for abundance of tumor infiltrating lymphocytes in multivariable analyses.ConclusionOur findings support that the SP142 and 22C3 IHC assays, with their respective clinically applied scoring algorithms, are not analytically equivalent where they identify partially non-overlapping subpopulations of TNBC patients and cannot be substituted with one another regarding PD-L1 detection.Trial registration The Swedish Cancerome Analysis Network - Breast (SCAN-B) study, retrospectively registered 2nd Dec 2014 at ClinicalTrials.gov; ID NCT02306096.

1548P PD-L1 assessment in oesophageal and gastro-oesophageal adenocarcinoma with PD-L1 IHC 22C3 pharmDx and PD-L1 IHC 28-8 pharmDx assays on Autostainer Link 48: The University Hospitals Birmingham experience over a period of 18 months

1548P PD-L1 assessment in oesophageal and gastro-oesophageal adenocarcinoma with PD-L1 IHC 22C3 pharmDx and PD-L1 IHC 28-8 pharmDx assays on Autostainer Link 48: The University Hospitals Birmingham experience over a period of 18 months

1244P AI-based accurate PD-L1 IHC assessment in non-small cell lung cancer across multiple sites and scanners

1244P AI-based accurate PD-L1 IHC assessment in non-small cell lung cancer across multiple sites and scanners

CD3ζ as a novel predictive biomarker of PD-1 inhibitor resistance in melanoma

Malignant melanoma is the most lethal form of skin cancer, and its incidence rates are increasing in Europe, America, and Oceania countries. Despite immune checkpoint inhibitors, such as PD-1 inhibitors, have been shown to have significant therapeutic effects on malignant melanoma, many patients are unresponsive to these treatments, even emerged resistance. There is an urgent need to discover novel biomarkers that might distinguish resistant patients from responders. In this study, we used a series of bioinformatics analyses and experimental validation. The GSE65041 was used for differential expression analysis. Kaplan-Meier was used to assess the prognostic value. ESTIMATE, ssGSEA, EPIC, TIMER, quanTiseq and MCPcounter for estimation of immune infiltration in the tumor microenvironment. We eventually identified that CD3ζ was significantly down-regulated in IHC PD-L1(−) melanoma patients. Low level of CD3ζ expression possessed a poor prognosis. CD3ζ low expression population is significantly associated with lower immune infiltration. In vivo experiment, CD3ζ expression was significantly down-regulated in mice melanoma after intradermally injected with B16–F10R cells. Compared to their wildtype counterparts, melanoma resistant mice treated with nivolumab showed significant reductions in tumor volume and weight when adding CD3ζ. In vitro experiment, the addition of CD3ζ increased nivolumab effection on inhibiting B16–F10R cell viability. Our findings indicated that CD3ζ could be a novel predictive biomarker of PD-1 inhibitor resistance in melanoma.

Adequacy of cytology and small biopsy samples obtained with rapid onsite evaluation (ROSE) for predictive biomarker testing in non-small cell lung cancer

Complete biomarker workup of non-small cell lung cancer (NSCLC) specimens is essential for appropriate and timely clinical management decisions. This can be challenging to achieve from small cytology and histology specimens, with increasing numbers of molecular and immunohistochemical biomarkers required. We conducted a 5 year retrospective audit of cases at our institution to assess the diagnostic and biomarker testing adequacy rates, particularly those specimens obtained with rapid onsite evaluation (ROSE), performed by a cytopathologist and a cytology scientist or pathology trainee, including all endobronchial ultrasound guided transbronchial needle aspirations (EBUS-TBNA), CT guided lung fine needle aspirations (FNA) and CT guided lung core biopsies. A total of 5,354 cases were identified, of which 92.2% had sufficient material for diagnosis. Of the 1506 cases identified with a recorded diagnosis of lung adenocarcinoma or NSCLC, not otherwise specified, 1001 (66.5%) had biomarker testing requested. Sufficient material was available in 89.5% of cases for a complete biomarker workup which included EGFR and KRAS mutational testing (all cases), ALK, ROS1 and PD-L1 immunohistochemistry (all cases), and ALK and ROS1 FISH (as required). For EGFR and KRAS mutational testing across both cytology and histology specimens, 99% of cases were sufficient. Of the samples in which a complete biomarker workup was unable to be performed, approximately half were only insufficient due to inadequate numbers of tumour cells for PD-L1 immunohistochemistry. Excluding PD-L1 IHC, 952 (95.1%) of samples obtained with ROSE were sufficient for the remainder of the testing requirements. Next generation sequencing using a 33 gene custom AmpliSeq panel was achieved in up to 72% of cases. In conclusion, small cytology and histology specimens obtained with ROSE are suitable for predictive biomarker testing in NSCLC, although attention needs to be paid to obtaining sufficient cells (>100) for PD-L1 immunohistochemistry.

Validation of Immunotherapy Response Score as Predictive of Pan-solid Tumor Anti-PD-1/PD-L1 Benefit.

This study confirms the utility of the integrative IRS biomarker for predicting anti-PD-L1/PD-1 benefit. IRS significantly improved upon currently available biomarkers, including PD-L1 IHC, TMB, and MSI status. Additional utility for informing on chemotherapy, anti-PD-L1/PD-1, and anti-PD-L1/PD-1 plus chemotherapy treatments decisions is shown.

Performance of endobronchial ultrasound transbronchial needle aspiration as the first nodal staging procedure for the determination of programmed death ligand-1 expression in non-small cell lung cancer patients.

The determination of the programmed death ligand-1 (PD-L1) expression is part of the diagnostic algorithm for advanced non-small cell lung cancer (NSCLC) patients. We aimed to analyze the diagnostic performance of EBUS-TBNA performed as first-choice nodal staging procedure for the determination of PD-L1 expression in NSCLC patients. Longitudinal-prospective study including NSCLC patients diagnosed between January 2018 and October 2019, for whom a primary tumor biopsy sample and an EBUS-TBNA cytological malignant sample were available. Samples with fewer than 100 malignant cells were considered inadequate. PDL-1 IHC 22C3 pharmDx antibody was used. The percentage of tumor cells expressing PD-L1, setting 1% and 50% as cutoff points, was collected. The weighted kappa coefficient was used to assess the concordance of PD-L1 expression. The PD-L1 expression was compared inprecision terms. From a total of 43 patients, 53 pairs of samples were obtained, of which 23 (43.4%) were adequate and included for analysis. The weighted kappa coefficient for PD-L1 expression was 0.41 (95% CI 0.15-0.68) and 0.56 (95% CI 0.23-0.9) for cutoff values ≥ 1% and ≥ 50%, respectively. In advanced stages, the weighted kappa coefficient was 0.6 (95% CI 0.3-0.9) and 1 (95% CI 1-1) for PD-L1 expression cutoff values ≥ 1% and ≥ 50%, respectively. EBUS-TBNA showed a sensitivity, specificity, positive predictive value, and negative predictive value of 1 to detect PDL-1 expression ≥ 50% in advanced stages. EBUS-TBNA performed as first nodal staging procedure in advanced NSCLC patients provides reliable specimens for the detection of PD-L1 expression ≥ 50% and could guide immunotherapy.

Comparison of PD-L1 IHC 22C3 PharmDx Combined Positive Score (CPS) in Primary Versus Metastatic Nodal Squamous Cell Carcinomas of the Head and Neck: Is There a Significant Difference?

PD-L1 IHC 22C3 pharmDx is an FDA-approved companion test to select patients for anti-PD-L1 immunotherapy. In head and neck squamous cell carcinoma PD-L1 expression is determined using a Combined Positive Score (CPS), which evaluates expression in tumor cells and tumor-associated leukocytes. We hypothesized that in nodal metastasis, the CPS should be higher given their inherent higher proportion of leukocytes. A significant difference in CPS between sites would mean that the tissue chosen for PD-L1 testing would impact patient eligibility for therapy. Currently, guidelines about which tissue should be tested do not exist. PD-L1 22C3 IHC was performed in the primary and nodal metastases of 35 head and neck squamous cell carcinoma, and a CPS was generated by 3 pathologists. Mean CPS was higher at the primary than the nodal metastasis: 47.2 versus 42.2; however, the difference was not statistically significant: P=0.259 . By therapeutic groups: negative (CPS <1), low (CPS 1-19) and high (CPS≥20), low-expression was more common in the primary: 40 vs. 26%, and high-expression in the nodal metastasis: 74 vs. 60% but this difference was not statistically significant: P=0.180. Stratified by positive versus negative (CPS <1 vs. ≥1), there were no differences between sites. Interobserver agreement for CPS among the 3 raters was slight for both sites: ƙ = 0.117 and 0.025, fair if stratified by therapeutic group: ƙ = 0.371 and 0.318, and near perfect if stratified as negative versus positive: ƙ = 0.652 and 1. There were no statistically significant differences in CPS between primary and nodal metastases independent of how the CPS was stratified.

Equivalence of laboratory-developed test and PD-L1 IHC 22C3 pharmDx across all combined positive score indications.

We conducted an analysis across multiple PD-L1 combined positive score (CPS) indications to establish concordance of a 22C3 antibody-based laboratory-developed test (LDT) on the Ventana BenchMark XT or BenchMark ULTRA platform and the regulatory-approved PD-L1 IHC 22C3 pharmDx in cervical cancer (CC), esophageal squamous cell carcinoma (ESCC), head and neck squamous cell carcinoma (HNSCC), triple-negative breast cancer (TNBC), and urothelial carcinoma (UC). Tumor specimens from each tumor type were stained with 22C3 antibody and scored using the 22C3 antibody-based LDT, and scores were compared with those using PD-L1 IHC 22C3 pharmDx. PD-L1 status was measured by the pathologist using CPS as a continuous score and using clinically relevant cutoffs (CC, ≥1 and ≥10; HNSCC, ≥1 and ≥20; ESCC, TNBC, and UC, ≥10). The agreement between the BenchMark platforms and PD-L1 IHC 22C3 pharmDx was assessed by intraclass correlation coefficient (ICC) and a contingency table for clinical interpretation. A total of 522 samples were evaluated for the pan-tumor analysis (CC, n = 77; ESCC, n = 80; HNSCC, n = 126; TNBC, n = 118, UC, n = 121). Most clinical interpretations of PD-L1 status were concordant between the BenchMark XT and PD-L1 IHC 22C3 pharmDx for all five tumor types with regard to negative percentage agreement (NPA; 83-97%), positive percentage agreement (PPA; 86-100%), and overall percentage agreement (OPA; 90-97%); the ICC by tumor type was high (≥0.88). Importantly, the pan-tumor ICC was 0.95 (95% CI 0.94-0.96). Thirty additional TNBC samples were evaluated using the BenchMark ULTRA and PD-L1 IHC 22C3 pharmDx; the NPA, PPA, and OPA were 100%. The 22C3 antibody-based LDT on Ventana BenchMark XT and BenchMark ULTRA platforms demonstrated high concordance with the regulatory-approved PD-L1 IHC 22C3 pharmDx across multiple tumor types. These findings suggest the comparability of PD-L1 IHC 22C3 pharmDx with an LDT based on the 22C3 antibody.

Implications of GITR expression profile for precision immunotherapeutics.

e14561 Background: Glucocorticoid-induced TNFR-related protein (GITR) is a type I transmembrane protein expressed mostly on CD25+CD4+ T-regs and upregulated on all T-cells upon activation. It serves as a T-cell co-stimulatory receptor and its role in cancer treatment is being explored. While pre-clinical data using combination therapies involving GITR agonism showed promise, early clinical trials have not shown improved outcomes to date. Methods: RNA expression levels of 397 genes in 514 patients with a variety of cancer types were assessed and analyzed at OmniSeq. Transcript levels were normalized to internal housekeeping gene profiles and ranked as percentiles using a reference population. High expression was defined as ≥ 75th percentile, and low-moderate as &lt;75th percentile. Univariate p-values were calculated using Welch t-test or Fischer’s exact test. Multivariate p-values were calculated using logistical regression. Results: A total of 99 patients (19.3%) had GITR high expression. Cancers with the highest proportion of GITR high expression were lung (7/20, 35%) and breast (19/49, 39%), and both these cancers showed a statistically significant association with GITR high expression on multivariate analysis (lung: OR 3.5, p = 0.04, 95% confidence interval (CI) 1.06 – 10.06 CI and breast: OR 4.59, p &lt;0.001, 95% CI: 2.11 – 9.85). 30% of lung and 30.6% of breast cancer patients had a GITR high with ligand low-moderate expression profile; these cancers had the highest proportion of patients with this profile. GITR high expression showed significant association with high RNA expression of multiple known immune checkpoint proteins (PD-1, PD-L1, CTLA4, LAG3, OX40, ICOS and CD137, but not TIM3) as well as PD-L1 IHC ≥1 on univariate analysis, but statistically significant association was only seen with PD-L1 IHC ≥1 (p = 0.007) and OX40 high expression (p = 0.001) on multivariate analysis. Conclusions: While clinical trials with GITR agonism are ongoing, our data suggests selection of certain types of malignancies depending on GITR status may be required. Moreover, there is a correlation between high GITR and high OX40 RNA expression, which could imply increased efficacy using this targeted combination. Further clinical development based on this immune-profiling is warranted. [Table: see text]

The impact of single gene testing (SGT) on subsequent comprehensive genomic profiling (CGP) success in community oncology practice for advanced non-small cell lung cancer (NSCLC): Results from a prospective observational reference laboratory testing program.

6506 Background: CGP concurrently tests tumor tissue for guideline-recommended predictive biomarkers for optimal therapy selection and identification of clinical trials in NSCLC patients. A common practice in the community oncology setting is to first order SGT, then consider CGP when SGT results are negative or after treatment failure. Given the limited tissue available and the scope of alterations not tested by SGT, we sought to characterize the effects of prior SGT on subsequent CGP testing success, and therapeutic opportunities identified by CGP beyond SGT for NSCLC patients in the community setting. Methods: Reference laboratory information systems were used to prospectively contact clinicians across 80 community practices who ordered at least 1 SGT for guideline-recommended genomic variants in NSCLC. CGP was offered for their patients prior to SGT or upon receipt of negative SGT results. SGT included individual assays for BRAF, EGFR, KRAS, MET exon 14 skipping mutations; ALK, RET, and ROS1 rearrangements; and PD-L1 IHC . CGP included DNA-seq for mutations, copy number variants in 523 genes including guideline-recommended genomic variants, MSI, and tumor mutational burden; RNA-seq for rearrangements/fusions/splice variants; and PD-L1 IHC. Results: Among a total of 580 NSCLC patients with CGP ordered (May 2021-December 2022), 168 (29%) had ≥1 SGT ordered prior to CGP (median=5). No patients had all SGT performed, with untested cases ranging from 10% for ALK to 90% for MET exon 14. The same FFPE tissue block was used for CGP in 150/168 (89%) of cases with prior negative SGT. Compared to CGP-only cases, CGP for cases with prior negative SGT was canceled twice as often at tissue review (16% vs 7%; p=.001), had higher DNA extraction failures (13% vs 8%; p=.09), and lower DNA sequencing success (70% vs 83%, p&lt;.001). CGP detected guideline-recommended variants in 51% of all cases. Among prior negative SGT cases, frequencies were higher in genes where SGT was performed less often and is less sensitive than CGP RNA-seq ( RET fusions , MET exon 14 skipping). CGP also identified guideline-recommended variants in genes with no SGT offered during the study period ( ERBB2 mutations , NTRK2/3 fusions), as well as variants with emerging evidence for FDA expedited program-designated clinical trial therapies in 28% of patients, including NRG1 fusions and BRAF non-V600E , KEAP1, KRAS non-G12C, NFE2L2, STK11, and TP53 Y220C mutations. Conclusions: Prior negative SGT doubled subsequent CGP test cancellations for NSCLC due to tissue insufficiency and increased CGP DNA extraction failures. SGT practice in the community oncology setting does not meet practice guideline recommendations and negatively impacts the potential benefit of subsequent CGP for NSCLC patients.

Co-mutational status and PD-L1 expression in KRAS mutant non-small cell lung cancer (NSCLC): Role in treatment selection and association with clinical outcomes.

9038 Background: In NSCLC, different KRAS mutations (mt) define unique subgroups and certain co-mutations (co-mt) could have prognostic and therapeutic implications. Retrospective studies associate poor prognosis with co-mts in STK11 and KEAP1 with KRAS. It is unclear if PD-L1 expression and co-mt in KRAS mt NSCLC impacts outcomes. Methods: Molecular profiles of 27748 NSCLC tumors were tested with next-generation sequencing (Caris Life Sciences, Phoenix, AZ) and classified by KRAS mt. PD-L1 IHC (22C3) was reported as TPS. Real-world post-immunotherapy (IO) overall survival (OS) was obtained from insurance claims and calculated from start of an immune check-point inhibitor (with or without chemotherapy) to the last day of follow-up. Results: In all, 7634 (28%) samples had a KRAS mt; most common being G12C (11%), G12V (5.3%), G12D (3.9%) and G13X (2.0%). The most frequent co-mt was KRAS-TP53 (KP) (46%), similar in all KRAS subtypes. KRAS-STK11(KL) was in 23% of KRAS mt, enriched in G13X (33%) and lowest in G12D (16%). KRAS-KEAP1(KK) was co-mt in 10% of KRAS mt, highest in G13X (16%) and lowest in G12D (8%). KRASmt/ STK11wt/ KEAP1wt/ TP53wt (K only) was 27% of KRAS mt, highest in G12D (36%) and lowest in G13X (16%). A small subgroup, 5.6% was KRAS-STK11-KEAP1 co-mt (KKL). The majority of pts in the KL (61%), KK (52%) and KKL (62%) cohorts had TPS &lt;1%. In 1723 KRAS mt NSCLC pts treated with IO, KL, KK and KKL had significantly worse post-IO OS than KP and K only (table); KKL had the lowest OS. In the TPS≥1% KRAS mt group, KL, KK and KKL had worse post-IO OS than KP and K-only. In TPS&lt;1% group, KL and KK had poor post-IO OS and K-only had favorable outcomes. However, KP had worse post-IO OS compared to K-only in TPS&lt;1%. In G12C and G12V, similar trend with worse post-IO OS in KL, KK and KKL, compared to KP and K only was noted. Conclusions: We report a large real-world dataset evaluating outcomes with check-point inhibitors in NSCLCs with KRAS and specific co-mts. KL, KK and KKL subgroups demonstrate universally poor outcomes in all KRAS subtypes irrespective of PD-L1. Pts with KP co-mts have adverse post-IO outcomes in TPS&lt;1% but favorable in TPS≥1%. TPS score remains a predictive marker of IO outcomes in KP, but not in KL, KK and KKL. These observations emphasize that both PD-L1 and co-mts have a clear association with clinical outcomes in KRAS-mt NSCLC, and must be used in predictive models for individualized therapy. [Table: see text]

AI-based radiomic biomarkers to predict PD-(L)1 immune checkpoint inhibitor response within PD-L1 high/low/negative expression categories in stage IV NSCLC.

1517 Background: Recent efforts exploring the utility of quantitative imaging (radiomic) biomarkers to predict response to immune checkpoint inhibitors (ICI) have shown promise to provide a more accurate and scalable method than biopsy-based PD-L1 IHC and tumor mutation burden. One of the challenges of evaluating ICI biomarkers on retrospective real-world data is the inherent association between a physician’s treatment choice and PD-L1 expression status. This work investigates radiomics-based multi-modal biomarkers within patient cohorts receiving first-line therapy (ICI monotherapy, ICI plus chemotherapy, and chemotherapy), and in cohorts of all patients receiving ICI (all-lines) subdivided by PD-L1 expression. Methods: Using a large multi-institutional real-world dataset, we analyzed radiomic characteristics of 6,295 primary and metastatic lesions from 1,206 stage IV NSCLC patients treated with PD-(L)1 ICIs from nine institutions across the US and Europe. Patients with unavailable imaging follow-up or with EGFR/ALK oncogenic driver mutations were excluded from analysis, resulting in a total dataset of 791 subjects randomly assigned to training (N = 541) and validation sets (N = 250). Radiomic data was extracted from baseline CT scans capturing tumor heterogeneity, spicularity, and burden in the lung, lymph nodes, and liver. A multi-modal ensemble classifier combining demographic features, PD-L1 TPS, and radiomic data was developed to predict response to ICI therapy per RECIST 1.1 criteria. The model's performance was evaluated in terms of the area under the receiver operating characteristic curve (ROC-AUC) and compared to PD-L1 IHC using the two-tailed DeLong test. Results: In first-line cohorts, the model identified responders with an ROC-AUC of 0.83 (0.72-0.94, P = 0.005; N = 80, 81% PD-L1 &gt; = 50%) in ICI monotherapy, 0.65 (0.47-0.83, P = 0.28; N = 78, 71% PD-L1 &lt; 50%) in ICI plus chemotherapy, and 0.38 (0.002-0.76, P = 0.53; N = 37) in chemotherapy. In all-lines cohorts subdivided into PD-L1 high/low/negative expression categories, the model identified responders with an ROC-AUC of 0.72 (0.57-0.87, P = 0.005, N = 102), 0.74 (0.58-0.89, P = 0.003, N = 71), and 0.60 (0.46-0.74, P = 0.16, N = 77), respectively. Conclusions: Our study demonstrates that multi-modal models can predict ICI response with high performance. The stronger performance in patients receiving ICI therapy alone indicates the model’s predictive, rather than prognostic, power. Furthermore, the model demonstrated good performance in identifying ICI responders within patient sub-cohorts defined by PD-L1 expression status. These insights may be used to inform clinical decision-making, such as escalation or de-escalation of concurrent chemotherapy in stage IV NSCLC patients. In our future work, we will investigate these questions in larger cohorts and prospective studies.

Combining plasma and tissue next-generation sequencing provides optimal treatment selection for metastatic NSCLC: A comparative study.

e21193 Background: 64% of all NSCLC patients in the USA with targetable mutations do not receive therapy due to clinical practice gaps around biomarker identification, including guideline-recommended alterations. Similar diagnostic and access challenges are also seen in Asia, but the scope of these challenges may differ due to the different mutation profiles of NSCLC. To determine the biomarker gaps involved, we conducted a study evaluating 38 metastatic NSCLC patients who received 3 diagnostic molecular workflows, and compared all 3 results to determine potential practice gaps. This is the first study to our knowledge that evaluates 3 different diagnostic workflows for one cohort to determine relative performance. Methods: 386 newly diagnosed and pathologically confirmed patients with metastatic lung cancer were managed at our hospital over Aug 21 - Sep 22. 38 metastatic non-squamous NSCLC patients were recruited prospectively. The patients received 1 plasma NGS tests (LiquidHALLMARK, Lucence, Singapore), 1 tissue comprehensive genomic profiling (CGP) test, and a panel of single-gene molecular tests (Idylla EGFR, ALK-IHC, ROS1-FISH, PD-L1 IHC). Results: Of the 38 patients recruited for the prospective trial, liquid biopsy identified 24 actionable alterations (TAT 3-13 days), tissue NGS identified 23 actionable alterations (TAT 11-27 days) and single gene testing approach provided 17 actionable alterations (TAT variable). 6 patients did not have available tissue reports. In terms of patient numbers, liquid biopsy detected at least one G9 biomarker in 21/38 patients and tissue NGS detected at least one G9 biomarker in 22/32 patients. Combining tissue NGS and liquid biopsy provided 26/38 (68.4%) patients with targeted treatment options. Single gene testing identified at least one G9 biomarker in 16/38 (42.1%) patients. Combining tissue NGS, liquid biopsy and single gene test improved sensitivity to 73.7% (28/38 patients). Overall, an estimated 1 in 4 NSCLC patients (26.4%) would have been identified with a diagnostic strategy using combined tissue and liquid NGS, over current conventional single gene testing approaches. Conclusions: Combining tissue and liquid NGS profiling provides the best treatment selection outcomes for patients with metastatic NSCLC. Our study highlighted that a quarter of NSCLC patients in Singapore would be identified for eligible targeted treatments from adopting a nationwide NGS testing approach, ideally tissue and liquid combined, to improve cancer services.

Global prevalence of CLDN18.2 in patients with locally advanced (LA) unresectable or metastatic gastric or gastroesophageal junction (mG/GEJ) adenocarcinoma: Biomarker analysis of two zolbetuximab phase 3 studies (SPOTLIGHT and GLOW).

4035 Background: There is an unmet need for additional therapies to treat patients (pts) with locally advanced (LA) unresectable or metastatic gastric or gastroesophageal junction (mG/GEJ) adenocarcinoma. Claudin-18.2 (CLDN18.2), a new validated target in these pts, is expressed in normal gastric mucosa cells and often retained in G/GEJ tumor cells. There are limited data on the global prevalence of CLDN18.2 in tumors of pts with LA unresectable or mG/GEJ adenocarcinoma. The SPOTLIGHT (NCT03504397) and GLOW (NCT03653507) phase 3 studies demonstrated clinically meaningful and statistically significant improvement in PFS and OS with the CLDN18.2-targeted antibody zolbetuximab + chemotherapy (mFOLFOX6 or CAPOX, respectively) vs placebo + chemotherapy as 1L therapy in pts with CLDN18.2+/HER2− disease. We report the biomarker analysis of these studies. Methods: Pts with LA unresectable or mG/GEJ adenocarcinoma were screened for CLDN18.2 status by IHC before enrollment. Tumors were defined as CLDN18.2+ if they had ≥75% of tumor cells with moderate-to-strong membranous CLDN18 staining per the investigational VENTANA CLDN18 (43-14A) RxDx Assay. HER2 status was evaluated per central or local assessment. Ad hoc PD-L1 IHC was performed via the Dako PD-L1 IHC 28-8 pharmDx assay for a subgroup of enrolled pts. Results exclude pts from mainland China and pts missing data for disease type, Lauren classification, or tumor collection site. Results: Across SPOTLIGHT and GLOW, 3576 pts had valid CLDN18 IHC results; 1399 (39.1%) had CLDN18.2+ tumors. CLDN18.2 prevalence was 43.7% (513/1175) in female pts and 36.9% (886/2401) in male pts. CLDN18.2 prevalence was 44.0% (671/1524) in pts in Europe/Middle East, 37.7% (183/485) in pts in N. America, and 36.5% (479/1314) in pts in Asia Pacific. CLDN18.2 prevalence was 41.0% (1056/2576) in pts with gastric adenocarcinoma and 37.3% (302/809) in pts with GEJ adenocarcinoma. CLDN18.2 prevalence trended higher in pts with diffuse (48.9%, 479/980) vs intestinal (38.9%, 265/682) disease. CLDN18.2 prevalence was 41.1% (175/426) in tumors collected from metastatic sites and 38.6% (1195/3094) in tumors collected from primary sites. CLDN18.2 prevalence was similar across other clinical and histopathological characteristics. Of 2908 HER2− pts with valid CLDN18 IHC results, 1264 (43.5%) had CLDN18.2+ tumors. Of 599 enrolled pts tested for PD-L1 expression, 104 (17.4%) had a PD-L1 CPS ≥5. Conclusions: The SPOTLIGHT and GLOW phase 3 studies are the largest data sources to date for determining global CLDN18.2 prevalence. CLDN18.2 is a high prevalence biomarker, with high tumor expression in ~35–40% of pts with LA unresectable or mG/GEJ adenocarcinoma, for whom zolbetuximab + chemotherapy represents a potential 1L therapy. Clinical trial information: NCT03504397 , NCT03653507 .

Clinical and molecular features of long-term response to anti-PD-(L)1 therapy in patients with advanced non-small cell lung cancer.

9107 Background: PD-(L)1 blockade can generate long-term responses in subsets of patients (pts) with advanced non-small cell lung cancer (NSCLC), but the clinical and molecular features of long-term responders (LTR) are not well established. Whereas features predictive of initial response have been rigorously explored, long-term follow up in large cohorts is required to identify features distinguishing LTR from short-term responders (STR). Methods: We analyzed pts with advanced NSCLC treated with anti-PD-(L)1 therapy at three institutions (MSK, DFCI, MDACC) between 2011 and 2022; responses were assessed by RECIST. LTR was defined as partial response (PR) or complete response (CR) ≥ 24 months; STR was defined as PR/CR &lt; 12 months followed by progressive disease (PD). PD-L1 IHC, tumor mutational burden (TMB), next generation sequencing (NGS), and whole exome sequencing (WES) results from subsets of pts were analyzed. High TMB was defined by harmonized TMB z-score ≥ 0 (≥ median TMB in each cohort: ≥ 7.9 for MSK, ≥ 10.5 for DFCI). Results: Among 3240 pts overall (MSK = 1536, DFCI = 1238, MDACC = 466), LTR was achieved in 267 (8.2%, 95% CI 7.3 to 9.2%) and STR was achieved in 213 (6.6%, 95% CI 5.7 to 7.5%), with similar rates from each site. 5-year overall survival (OS) was 83% among LTR and 20% among STR. In univariate analyses, non-squamous histology (non-SQ: odds ratio [OR] 2.13, p = 0.005), peripheral blood derived neutrophil to lymphocyte ratio (dNLR) &lt; 3.0 (OR 2.15, p &lt; 0.001), and high TMB (OR 2.25, p = 0.001) were associated with LTR compared to STR; high PD-L1 (PD-L1 ≥ 50%: OR 1.19, p = 0.494) was not. In multivariate analyses, non-SQ histology (OR 2.56, p = 0.047) and high TMB (OR 2.73, p = 0.001) remained independently associated with LTR over STR. Among LTR compared to patients with best response of PD, dNLR &lt; 3.0 (OR 1.72, p = 0.030), high PD-L1 (OR 5.23, p &lt; 0.001), and high TMB (OR 2.55, p = 0.001) associated with LTR. LTR showed deeper radiographic responses compared to STR (best response of -50% or greater: OR 3.88, p &lt; 0.01). Among 1798 pts with NGS, pathogenic ARID1A alterations were numerically higher among LTR compared to STR, but this was not significant when accounting for multiple hypothesis testing (14% vs 2%, p = 0.022, q = 0.196). Among 73 pts with available WES, both clonal (p = 0.037) and subclonal TMB (p = 0.010) were significantly higher among LTR compared to STR. Conclusions: We report outcomes from &gt; 10 years of advanced NSCLC pts treated with anti-PD(L)1 therapy in a large multicenter cohort, with ongoing response at 2 years associated with &gt; 80% 5-year OS. High TMB was a strong independent predictor of LTR over STR. Non-squamous histology, low dNLR, and depth of response also correlate with LTR, while no individual genomic alterations clearly distinguish LTR. These features may assist in identifying pts more likely to derive durable benefit from anti-PD(L)1 therapy without treatment intensification.