Next-generation DNA Sequencing Research Articles

Survival associations and driver oncogene overlap for copy-number amplifications of ERBB2, KRAS and MET in non-small cell lung cancer.

9084 Background: Next-generation sequencing (NGS) to detect targetable oncogenic driver mutations is standard of care in non-small cell lung cancer (NSCLC). However, the interpretation and clinical impact of copy number amplifications (CNA) of driver genes is less clear. ERBB2, KRAS and MET are oncogenic drivers with intersecting resistance mechanisms and evolving therapeutic landscapes. We examined the overlap between other oncogenic drivers and CNA of these genes, and the association of CNA with overall survival (OS), in a cohort of patients with NSCLC. Methods: NGS of DNA (592 genes or WES)/RNA (WTS) was performed for 5870 consecutive lung adenocarcinoma tumors submitted to Caris Life Sciences (Phoenix, AZ). The copy number of each exon was determined by calculating the average depth of the sample along with the sequencing depth of each exon. Driver oncogenes (Driver+/-) were defined as pathogenic fusions in ALK, NTRK (1,2 and 3), RET and ROS1 or pathogenic SNVs/indels in ALK, BRAF, EGFR, ERBB2, KRAS, MET. Tumor mutational burden high (TMB-H) was defined as 10 mutations per MB. The χ² test was applied as appropriate ( p < .05). Results: The cohort was 55% female; median age was 69 years (range 24-89) and similar across all cohorts. Incidence of CNA at different amplification thresholds was correlated with oncogene drivers. At CNA ≥6 for ERBB2 and KRAS, but CNA ≥4 for MET, a significant tendency for mutual exclusivity with driver mutations was observed (log2 odds ratio ERBB2: -3.1, KRAS: -1.1 , MET: -2.2 , all p < .001). Tumors meeting these thresholds were designated CNA-High ( Gene-H), versus CNA-Low ( Gene-L). Only one tumor harboured two CNA-H genes ( KRAS and MET). Independent of CNA status, the TMB-H frequency was decreased in Driver+ tumors (27% vs 52% in Driver-, p < .001). CNA-H tumors were associated with decreased OS (compared to CNA-L); for ERBB2-H, hazard ratio (HR) 1.74 (95% CI 1.44-2.10, p < .001), for KRAS-H HR 1.29 (1.16-1.65, p < .001), and for MET-H HR 1.46 (1.28-1.68, p < .001). Similar results were observed when stratified by driver mutation status (Table), with the exception of KRAS/Driver- tumors. Conclusions: These data suggest tumors with high CNA of ERBB2, KRAS and MET represent clinically distinct entities and are significantly less likely to harbour concurrent driver oncogenes. The association between increased copy number and worse OS suggests amplified tumors are an important area for therapeutic development. Overall survival (collection to last contact) for ERBB2, KRAS and MET CNA-High versus CNA-Low tumors, by Driver mutation status. [Table: see text]

The genomic and transcriptomic landscapes of chemotherapy naïve vs post-chemotherapy germ cell tumors.

5007 Background: Cisplatin resistance occurs in up to 30% of patients with advanced germ cell tumors (GCTs). Despite effective salvage treatment regimen, there is a 50% risk of cancer-specific death. Previously, we identified molecular features, including TP53 mutations and MDM2 amplification, associated with cisplatin resistance (Bagrodia, 2016). Herein, we evaluate the genomic and transcriptomic landscapes of chemotherapy naïve vs post-chemotherapy GCT’s to uncover genetic factors that might drive cisplatin resistance. Methods: GCTs (N = 138) were tested at Caris Life Sciences (Phoenix, AZ) with next-generation sequencing of DNA (592-gene or whole exome) and RNA (whole transcriptome). Prevalence was calculated for pathogenic SNVs/indels and copy number amplifications (CNA≥6 copies). Primary (N = 65) and metastatic (N = 73) sites were defined based on the biopsy site. A specialized genitourinary pathologist blindly reviewed selected H&E-stained slides and designated tumors as chemotherapy naïve (N = 66) or post-chemotherapy (N = 17) based on treatment related changes. Differentially regulated pathways were assessed by gene set enrichment analysis (GSEA). Transcriptomic signatures predictive of response to immunotherapy (T cell-inflamed) and platinum sensitivity (PSS, high score suggests increased platinum sensitivity) were applied. Mann-Whitney U and χ² tests were applied, with P-values adjusted for multiple comparisons. Results: Primary and metastatic GCT had slight differences between putative drivers of platinum therapy resistance: KIT (15% vs 1), TP53 (10% vs 17), KRAS (13% vs 9) mutations and MDM2 CNA (4% vs 3), all p > 0.05. When comparing post-chemo vs chemo naïve GCTs, no significant differences in the genomic landscape were observed (p > 0.05). Chemo naïve tumors were shown to be enriched for gene sets associated with KRAS signaling and epithelial-to-mesenchymal transition. Additionally, they trended towards a higher proportion of T cell-inflamed tumors as compared to post-chemo samples (32% vs 18, p = 0.36). The average PSS for chemo naïve was slightly higher than post-chemo (CN: 0.07 Arbitrary Units (AU), N = 62 vs PC: -1.82, N =16, p = .52). Chemo naïve tumors with TP53, KRAS, or KIT mutation or MDM2 CNA had a lower PSS (-4.16 AU, N= 18) than tumors with none of those alterations (2.17, N = 38), p = 0.05. Finally, chemo naïve tumors with MDM2 CNA (N=2, -16.26 AU) had the lowest PSS followed by tumors with mutations in TP53 (N=6, -2.04), KIT (N = 2, -1.77) and KRAS (N=9, -1.05) (mutations were mutually exclusive except for a concurrent KRAS and TP53 mutation). Conclusions: This study provides evidence that chemo naïve GCTs with mutations in TP53, KRAS, KIT or MDM2 CNA have higher expression of genes associated with resistance to platinum-based chemotherapy. These findings contribute to a better understanding of the molecular characteristics of GCTs and may inform prognosis and treatment.

Distinct molecular features and potential therapeutic strategies for spinal cord gliomas.

2056 Background: Spinal cord gliomas (SCG) are rare primary CNS tumors with varied prognosis. Standard treatment remains poorly defined, as no clear survival advantage exists with extensive resection, radiation, or chemotherapy. Mounting evidence suggests that tumors from different regions of the CNS harbor unique molecular signatures. In a retrospective study, we comprehensively characterized molecular alterations to identify potential therapeutic strategies in the largest cohort of SCG reported to date. Methods: We performed centralized pathology review and analyzed SCG with next-generation sequencing of DNA (592 gene NextSeq or WES, Novaseq), RNA (WTS, NovaSeq) and pyrosequencing (MGMT promoter methylation, MGMTme). We estimated tumor microenvironment cell infiltration by quanTIseq & Epithelial-Mesenchymal Transition (EMT) by RNAseq. We used X2/Fisher’s-exact/Mann-Whitney U tests for comparison & determined significance (p < 0.05), adjusting for multiple comparison by the Benjamini-Hochberg method (q < 0.05). Results: We analyzed 39 surgically accessible SCG (19 high grade, 17 low grade, 3 NOS); ependymomas were excluded. The most common alterations were mutations in H3F3A (31%), TP53 (25%), ATRX (21%), NF1 (18%) & BRAF fusion (19%). H3F3A mutations occurred exclusively in high grade SCG (63%, q < 0.05) while BRAF fusions occurred exclusively in low grade SCG (60%, p < 0.05). TP53 mutations were more prevalent in high grade SCG (50% vs. 8%, p < 0.05). Compared to intracranial gliomas (n = 6732), SCG harbor significantly more frequent H3F3A (31% vs. 1.6%), KRAS mutations (10% vs. 0.7%) & BRAF fusions (19% vs. 0.7%) but less frequent TERT (4.3% vs. 66%) & PTEN (0 vs. 27%) mutations or EGFR amplification (0 vs. 28%) (all q < 0.05). SCG rarely harbor the canonical intracranial alterations IDH1 (2.6% vs. 17%), EGFRvIII (0 vs. 17%) or MGMTme (19% vs. 47%) (p < 0.05). SCG have greater penetration of immune infiltrates of DCs (median cell fraction/MCF: 13% vs. 9%), T regs (positive percent: 47% vs. 26%) & B cells (MCF: 9% vs. 7%), with less penetration of monocytes (MCF: 0 vs. 2%) (q < 0.05). EMT score associates with high grade glioma in intracranial disease (q < 0.05) but does not recapitulate in SCG (p > 0.05). Conclusions: Clinical management of SCG is currently drawn from experience with intracranial gliomas. Our results identify unique molecular features of SCG suggesting an underlying biology distinct from intracranial gliomas. In SCG, H3F3A mutations are exclusive to high grade while BRAF fusions are exclusive to low grade; SCG rarely harbor canonical intracranial alterations such as IDH1, EGFR or MGMTme; and SCG have greater penetration of DCs with less penetration of monocytes. We provide a biological explanation for limited effectiveness of current therapies in SCG with potential implications for chemotherapy, targeted and immunotherapy. Our work underscores the need for investigations dedicated uniquely to SCG.

Association of somatic mutations and histologic subtype/grade on prognosis and PD-L1 expression in mesothelioma.

8507 Background: Mesothelioma includes epithelioid, sarcomatoid, and biphasic histologic subtypes, and the epithelioid subtype is histologically graded based on mitotic count and nuclear atypia. Correlation between the most common somatic mutations ( BAP1, NF2, TP53, CDKN2A), and overall survival is poorly understood. High PD-L1 expression is a poor prognostic factor for survival and may be a predictor of response to immunotherapy. Few studies have evaluated the relationships between these factors. Methods: We performed a retrospective study in 217 patients with pleural, peritoneal and bicavitary mesothelioma as part of an IRB-approved biobanking protocol. Histologic subtype, grade of epithelioid mesothelioma, tumor stage, and overall survival time were determined. Next-generation DNA sequencing (NGS) was used to identify the prevalence of somatic mutations. PD-L1 immunohistochemistry was obtained on 184 samples and quantified as percentage of tumor cells. Histologic subtype, nuclear grade, clinical stage, PD-L1 expression, survival time, and the prevalence of each mutation were compared using chi-squared, Kruskal-Wallis, or Cox proportional hazard tests. All p-values were determined in R version 4.2.0 using a two-sided alpha error of 0.05 to mark statistical significance. Results: The most frequent somatic mutations were found in BAP1 (99/217, 45.6%), CDKN2A (47/217, 21.7%), TP53 (37/217, 17.1%), and NF2 (31/217, 14.3%). Patients with CDKN2A-mutated mesothelioma experienced shorter survival (p=0.032). TP53 mutations were associated with higher nuclear grade (p=0.015), higher mitotic count (p=0.013), and greater nuclear atypia (p=0.015). TP53 mutations were more prevalent in pleural than peritoneal mesothelioma (p=0.016) and in tumors without lymph node metastasis (p=0.05). As expected, patients with epithelioid mesothelioma had superior overall survival (p=0.02) and their tumors had lower PD-L1 expression (p=0.007) than either sarcomatoid or biphasic mesotheliomas. Neither PD-L1 expression nor mutations in BAP1, NF2, CDKN2A, or TERT were associated with nuclear grade, atypia, or mitotic count. No association was found between BAP1 or NF2 mutations and tumor stage, lymph node metastasis, or PD-L1 expression. Conclusions: To our knowledge, this is the first study of this scale to correlate PD-L1 expression, somatic mutation data, histologic features, and survival in mesothelioma patients. The most significant findings were that patients with CDKN2A mutations had shorter survival and TP53 mutations, which were more prevalent in pleural mesotheliomas, were associated with higher nuclear grade, higher mitotic count, and greater nuclear atypia. An ongoing analysis evaluates the contribution of germline mutations, which occurred in a significant proportion (15.7%) of this mesothelioma patient cohort.

FOXC1 expression in meningiomas.

2074 Background: Meningiomas account for nearly 1/3 of brain tumors and are characterized by a low mutational burden and dysregulation of DNA methylation. The promoter of FOXC1 is hypomethylated relative to dura in meningiomas of all grades and knockdown of FOXC1 in murine models causes craniofacial hypoplasia and absent frontal dura, suggesting a role as a meningeal identity gene. Here we perform a multi-omic analysis of FOXC1 expression in a large cohort of meningiomas. Methods: A total of 555 meningiomas (148 grade I, 255 grade II and 87 grade III) were analyzed with next-generation DNA sequencing (592 gene NextSeq panel; n=150 or Agilent WES whole exome sequencing, n=405) and whole mRNA (WTS, NovaSeq, 555) at Caris Life Sciences (Phoenix, AZ). Top quartile transcripts per million (TPM) for FOXC1 expression were considered high (FOXC1-H) and bottom quartile low (FOXC1-L). Immune cell infiltration was estimated using RNA deconvolution by QuantiSeq to calculate median cell fractions (MCF) or positive percentages (PP). Data were analyzed using X2/Fisher’s-exact/Mann-Whitney U tests. A p-value<0.05 after BH correction was considered significant. Results: Median FOXC1 expression was 33 TPM. Expression of FOXC1 transcript was higher in lower grade tumors (I: 34, II: 29, III: 27, q<0.05). NF2 mutations were significantly more common in FOXC1-L than FOXC1-H tumors (45% vs. 68%, q<0.05). Although a higher percentage of FOXC1-H meningiomas harbored TERT promoter mutations (2% vs. 9%, p<0.05) and a higher percentage of FOXC1-L meningiomas harbored a TRAF7 mutation (19%. Vs. 5.5%, p<0.05), these differences were not significant after correction for multiple comparisons. Programmed cell death ligand1 (PD-L1/CD274, 2.27 vs. 492) and MHC-I (HLA-A: 58 vs. 109, HLA-B: 59 vs. 74, and HLA-C: 61 vs. 101) transcripts were more highly expressed in FOXC1-H than FOXC1-L tumors. FOXC1-L tumors exhibited significantly more B cells (MCF: 5% vs. 4%), M1 Macrophages (MCF: 2% vs. 1%), CD8+ T cells (PP: 27% vs. 7%), and a higher interferon gamma transcript signature than FOXC1-H. Infiltrates of NK cells (MCF: 3.5% vs. 4.3%) and dendritic cells (MCF: 6% vs. 8%) were higher in FOXC1-H (all q<0.05). Conclusions: The craniofacial patterning transcription factor FOXC1 was found to be widely expressed in a large cohort of meningiomas and its expression is inversely correlated with WHO tumor grade. Multi-omic analysis reveals a significant relationship between FOXC1 transcript levels and key markers of the immune environment including PD-L1 transcript expression. These data suggest that FOXC1 is a marker of meningiomas in addition to its role as a meningeal identity marker during development. Functional studies are warranted to determine whether FOXC1 is necessary for meningioma growth, whether FOXC1 plays a direct immunomodulatory role in meningiomas, and whether the lower FOXC1 expression levels in high grade tumors represent a more embryonic phenotype.

Effect of germline mutations on somatic alteration landscapes in BRCA-associated cancers.

10509 Background: The interaction between germline alterations in homologous recombination repair (gHRR) genes and tumoral genomic landscapes is poorly understood. We hypothesized that gHRR alterations may influence somatic mutational landscapes in BRCA-associated cancers. Methods: Using a large de-identified real-world dataset of patients undergoing matched tumor/normal next-generation DNA sequencing (Tempus xT), we compared somatic genomic landscapes of tumors arising in individuals with germline HRR mutations (g BRCA1, g BRCA2, g PALB2, g ATM, g CHEK2) versus their sporadic counterparts, across four BRCA-associated cancers (breast, ovary, pancreas, prostate). Results: In breast cancer (N = 6955; gHRR-altered 5.9%, sporadic 94.1%), somatic TP53 muts were enriched in g BRCA1 pts, and depleted in g ATM and g CHEK2 pts (all P < 0.001). ESR1 muts were depleted in g BRCA1 pts, and enriched in g ATM and g CHEK2 pts (all P < 0.05). PIK3CA muts were depleted in g BRCA1/ 2 pts, and enriched in g ATM and g CHEK2 pts (all P < 0.05). FGFR1 muts were depleted in g BRCA1/2 and g PALB2 pts, and enriched in g ATM pts (all P < 0.01). HER2 and CDH1 alterations were depleted in g BRCA1 pts (P < 0.05). In ovarian cancer (N = 4244; gHRR-altered 6.2%, sporadic 93.8%), somatic TP53 muts were enriched in g BRCA1/ 2 pts, and depleted in g ATM and g CHEK2 pts (all P < 0.05). ESR1 muts were enriched in g ATM pts (P < 0.001). PIK3CA muts were depleted in g BRCA1/ 2 pts, and enriched in g ATM pts (all P < 0.05). KRAS muts were depleted in g BRCA1/ 2 pts (P < 0.05). In pancreatic cancer (N = 5386; gHRR-altered 3.9%, sporadic 96.1%), somatic TP53 muts were enriched in g BRCA1 pts, and depleted in g ATM and g PALB2 pts (all P < 0.05). KRAS muts were enriched in g BRCA1/2 and g ATM pts (P < 0.05). Unexpectedly, ESR1 muts were enriched in g PALB2 and g CHEK2 pts (P < 0.05). In prostate cancer (N = 4378; gHRR-altered 4.5%, sporadic 95.5%), somatic TP53 muts trended higher in g BRCA1 pts, and lower in g ATM and g PALB2 pts (P = NS). TMPRSS2-ERG fusions were depleted in g BRCA1 and g BRCA2 pts (P < 0.05). FOXA1 muts were enriched in g BRCA2 pts (P < 0.05), while BRAF and CDK12 muts were enriched in g CHEK2 pts (P < 0.05 and P = 0.07). Median TMB was higher in BRCA-associated cancers with g BRCA1/2 and gPALB2 muts relative to sporadic cancers (all P < 0.05), but not in pts with g ATM or g CHEK2 muts. There were no differences in MMR mutations or MSI status. Conclusions: Across four BRCA-associated cancers, TP53 muts are enriched in BRCA1 pts and depleted in g ATM pts. In breast/ovarian cancers, PIK3CA muts are depleted in g BRCA1/ 2 pts, while ESR1 muts are enriched in g ATM pts. KRAS muts are enriched in g BRCA1/2-altered pancreas cancers, but are depleted in g BRCA1/2-altered ovarian cancers. g BRCA2-altered prostate cancers are enriched in FOXA1 muts, and depleted in ERG fusions. These data suggest that gHRR-mutated cancers have distinct genomic landscapes compared to their sporadic counterparts; this may influence therapeutic considerations.

KRAS G12C-mutated pancreatic cancer: Clinical outcomes based on chemotherapeutic regimen.

4150 Background: Frontline treatment for advanced pancreatic ductal adenocarcinoma (PDAC) has been either 5-fluorouracil, oxaliplatin and irinotecan (FOLFIRINOX) or gemcitabine and nab-paclitaxel (GP) for the past decade. While the NAPOLI-3 trial, utilizing liposomal irinotecan, highlighted the superiority of a triplet regimen over GP, the question remains whether certain subgroups may derive particular benefit from GP. Pre-clinical data in lung cancer suggest that KRAS G12C may facilitate enhanced DNA adduct removal after platinum chemotherapy and confer resistance to this drug class. While multiple KRAS G12C inhibitors have shown early promise in PDAC, multi-agent chemotherapy remains the frontline standard and will likely remain an important therapeutic tool. This study aimed to investigate clinical outcomes after platinum and non-platinum-based chemotherapy in patients with advanced KRAS G12C-mutated PDAC. Methods: PDAC samples were tested using whole transcriptome sequencing (WTS; Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES) at Caris Life Sciences (Phoenix, AZ). Significance was determined by X2 and Fisher-Exact and p adjusted for multiple comparisons (q). Real-world overall survival (rwOS) was obtained from insurance claims data and calculated from first of treatment to last contact with comparison done by Kaplan-Meier test. Results: A total of 5,555 PDAC patients harboring KRAS pathogenic variants were identified, including 109 with a G12C mutation. For KRAS G12C mutants, median overall survival was 470 days for those who received GP (n=22) compared to 240 days for patients who received FOLFIRINOX (n=9) with a median difference of 230 days (HR 0.32, CI 0.12-0.82, p=0.013). In contrast, median overall survival was higher for patients harboring G12D (392 vs 297 days, p=0.003), G12V (500 vs 356 days, p=0.012) or G12R (469 vs 391 days, p=0.095) mutations who received FOLFIRINOX compared to GP. KRAS G12C mutated PDAC showed the highest rate of PDL1+ staining (27%) compared to other variants (G12R 13% and G12D 19%, q<0.01). TMB-H and MSI-H prevalence were highest in G12C compared to other variants (no statistical significance observed). Conclusions: In patients with advanced PDAC and a G12C mutation, median overall survival appears significantly longer in those treated with GP compared to FOLFIRINOX. The opposite trend was seen in patients with other KRAS variants including G12D, G12V, and G12R, consistent with the recently presented NAPOLI-3 trial. PDL1 staining was also highest in the KRAS G12C cohort. While this is the largest reported analysis of outcomes to frontline chemotherapy in KRAS G12C-mutated PDAC, the sample size is small and needs validation in additional datasets. Next steps include evaluating DNA mutational status and RNA expression of genes involved in DNA repair in G12C vs other KRAS variants to better understand the outcome data presented here.

Characterization of MCL-1 in patients with colorectal cancer (CRC): Expression, molecular profiles, and outcomes.

3085 Background: Myeloid cell leukemia 1 (MCL-1) is a member of the BCL-2 protein family and is anti-apoptotic/pro-survival in function. Dysregulation of MCL-1 expression has been reported in several solid tumors, including lung and breast cancer. In CRC, MCL-1 has been associated with resistance to chemotherapeutic drugs and multi-kinase inhibitor regorafenib. Our study aimed to characterize the molecular features associated with MCL-1 gene expression in CRC. Methods: 28,576 CRC samples were analyzed by Caris Life Sciences (Phoenix, AZ) with WTS (Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES). MCL-1 expression was stratified by quartiles where top quartile transcripts per million (TPM) were considered high (Q4) and bottom quartile low (Q1). Cell infiltration (CI) in the tumor microenvironment (TME) was estimated by RNA deconvolution analysis using QuantiSEQ. Interferon-gamma and T-cell inflamed signatures were also calculated from RNA data. X2 and Fisher-Exact tests were used, and statistical significance was determined as a P-value adjusted for multiple comparisons ( q < 0.05). Real world survival was obtained from insurance claims data and Kaplan-Meier estimates were calculated for molecularly defined patients. Results: MCL-1 expression was significantly increased in rectal tumors compared to other primary tumor sites (median TPM [mTPM] 20.8 vs 18.9 vs 18.7, in rectal vs left-sided vs right-sided, respectively, q < 0.0001). MCL-1 high tumors were associated with increased rate of PD-L1 IHC levels (Q1 vs Q4: 4.2% vs 9.7% [all CRC]; 3.7% vs 8.9% [pMMR/MSS], q < 0.0001), while MCL-1-low showed marginally increased rates of TMB-high (8.0% vs 7.0% [all CRC]; 3.3% vs 2.6% [pMMR/MSS], P < 0.05). MCL-1-high expression was associated with a higher T-cell inflamed signature and IFN score ( q < 0.0001). MCL-1 high was associated with higher mutation rates of CDKN2A, BRCA2, KRAS and GNAS, while lower mutation rates of TP53, PIK3CA, PTEN, BRAF, APC, FBXW7, AMER1, SOX9, and SMAD2 and copy number amplifications in several genes ( q < 0.05). MCL-1 high TME had higher CI of M1 and M2 macrophages, monocytes, B cells, NK cells, and T-reg infiltration, while dendritic cells and CD4+ T-cell were lower ( q < 0.05). Patients with high MCL-1 expressing CRC had longer survival (Q4 vs Q1: 33.5 vs 24.8 months, HR 0.72, 95% CI 0.67-0.77, P < 0.0001) and longer time-on-treatment with regorafenib although with limited clinical benefit (Q4 vs Q1: 2.03 vs 1.06 months, HR: 0.67, 95% CI: 0.50-0.89, P = 0.005). Conclusions: Our data show a strong correlation between distinct immune biomarkers, TME cell infiltration and MCL-1 expression in CRC. Furthermore, increased tumor MCL-1 expression improved patient prognosis and treatment outcomes. These findings suggest a key clinical role for MCL-1 as an important modulator of anti-tumor immunity and TME and a potential biomarker in CRC.

Molecular characterization of nectin-4 in non-small cell lung cancer subtypes.

e21119 Background: Nectin-4 (Nectin cell adhesion molecule-4) is a tumor associated antigen overexpressed in multiple solid tumor types. Studies show that over 60% of NSCLC are nectin-4 overexpressed. As enfortumab vedotin, an antibody-drug conjugate directed against nectin-4, has been FDA approved for advanced urothelial carcinoma, this holds treatment implications. Here, we investigated nectin-4 expression in NSCLC subtypes. Methods: 27,607 NSCLC tumors were tested for whole transcriptome sequencing (WTS, NovaSeq) and NextGen DNA sequencing (592 gene panel or whole exome sequencing, WES) at Caris Life Sciences (Phoenix, AZ). Tumors were analyzed overall in quartiles by nectin-4 expression (Q4-greatest, Q1-least) and further analyzed in the adenocarcinoma (AD) and squamous carcinoma (SQ) cohorts. PD-L1 expression was tested by IHC (Dako 22C3). Immune cell populations were estimated by QuanTISeq. Statistical significance was determined using Chi-square/Fisher-Exact/Mann-Whitney U tests and adjusted for multiple comparisons (q < 0.05). Results: We saw significantly greater expression of nectin-4 in SQ (median TPM- 15.68) versus AD tumors (median TPM - 9.88) (p < 0.05). We saw a greater prevalence of NFE2L2, STK11, ERBB2 mutations and genomic loss of heterozygosity (LOH) but a lower percentage of PD-L1+ tumors in Q4 vs Q1 tumors (all q-value < 0.05, Table 1). mRNA expression of immune checkpoints PDL1, CTLA4, CD80, CD86, HAVCR2, and IDO1 was increased in Q4 tumors (1.1-1.3-fold, all q < 0.05). Q4 tumors had reduced estimates of M1 macrophages, regulatory T cells (Tregs), and CD8+ T cells and increased estimates of neutrophils and NK cells. The association of nectin-4 expression with the immune landscapes of AD and SQ was similar. Q4 tumors in AD and SQ displayed increased expression of the above-mentioned immune checkpoints (q < 0.05). In AD and SQ, estimates of M1 macrophages, Tregs, and CD8+ T cells were reduced, but neutrophils and NK cells were increased in Q4 tumors (q < 0.05). The pattern of molecular alterations was different in AD and SQ. While KEAP1, STK11 and ERBB2 mutations were significantly enriched in AD Q4 tumors, NFE2L2 mutations were significantly enriched in SQ Q4 tumors (Table 1). Prevalence of LOH was higher and PD-L1+ tumors was lower in Q4 tumors in both AD and SQ (Table 1). Conclusions: NFE2L2 alterations in SQ and STK11, KEAP1 mutations in AD, which are associated with poor immunotherapy (IO) response, were enriched in Q4 tumors. This suggests that the immune landscape associated with high nectin-4 expression in NSCLC may be driven by the dysregulated KEAP1-NFE2L2 pathway. Targeted therapy towards nectin-4 may benefit these NSCLC subtypes that do not respond well to IO. [Table: see text]

Molecular and clinical correlates of DSCR1 expression in pancreatic cancer (PDAC).

4151 Background: Individuals with Down syndrome (DS) have a lower risk for solid tumors and angiogenesis related diseases. DSCR1 is highly upregulated in DS patients and its product, calcipressin-1, was shown to suppress angiogenesis and reduce cancer risk. High DSCR1 expression has been reported to decrease PDAC growth and metastasis in animal models. Here, we analyzed the molecular features and clinical outcomes associated with DSCR1 gene expression in PDAC. Methods: 8352 tumor samples tested at Caris Life Sciences (Phoenix, AZ) with WTS (Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES) were analyzed. Top quartile transcripts per million for DSCR1 expression were considered high (Q4) while bottom quartile low (Q1). Cell infiltration in the tumor microenvironment (TME) was estimated using QuantiSEQ. Interferon-gamma (IFG) and T-cell inflamed (TIS) signatures were calculated from RNA data. Statistical significance was determined as P-value adjusted for multiple comparisons ( q < .05). Real world survival was obtained from insurance claims data ( N = 4223) and Kaplan-Meier estimates were calculated. Results: DSCR1 expression was higher in primary tumors than metastases ( q < .05). No significant differences were observed between high vs low DSCR1 PDAC in immune-related biomarkers (TMB, dMMR/MSI-H and PD-L1 protein), gene mutations and copy number alterations except for KRAS mutations which were more frequent in DSCR1 Q4 (93 vs 86%, Q4 vs Q1, q < .0001). Gene set enrichment analysis showed that DSCR1 high tumors were enriched in alterations of several pathways including NOTCH signaling, DNA repair, IFG response, myogenesis and adipogenesis ( P < .05, false discovery rate < .25). B cells, M1 and M2 macrophages, neutrophils, NK cells, and Tregs were more abundant in the TME of tumors with high DSCR1 while dendritic cells, CD4+ T cells and monocytes were lower ( q < .05). DSCR1 Q4 was associated with higher TIS score (50% inflamed vs 3.6%, q < .05) and positively associated with immune-related gene expression including CTLA4, IDO1, CD80, PD-L1, LAG3, CD86, TIM3, IFG, PD-1, and PD-L2 (fold change: 2-4, q < .0001). Overall, DSCR1 expression above median was associated with longer median OS (17 vs 11 months, HR 0.89 [0.83-0.95], P < .0001). When stratified by quartiles, DSCR1 Q4 was associated with longer time on treatment [ToT] with gemcitabine/nab-paclitaxel (HR 0.75 [0.63-0.89], P = .001), and marginal benefit on ToT (HR 0.81 [0.65-1.0]) but longer survival in FOLFIRINOX treated patients (HR 0.73 [0.58-0.92], P = .008). Conclusions: This is the first and most extensive profiling study to investigate DSCR1 expression in PDAC. Our data show a strong association between tumor DSCR1 gene expression, several pathway alterations, immune-related gene expression, TME cell infiltration and patient survival. These findings suggest DSCR1 as a candidate prognostic biomarker and as a potential treatment target in PDAC.

Comprehensive profiling of clock genes expression in hepatocellular carcinoma (HCC).

4108 Background: The circadian clock mechanism controls the physiological homeostasis of the liver and plays a key role in hepatocarcinogenesis. Recent evidence unveiled core clock proteins as novel therapeutic targets in cancer. Our group showed that clock regulators BMAL1 and CLOCK can promote proliferation of liver cancer cells by modulating the cell cycle checkpoint kinase Wee1. Here we further evaluated the molecular landscape of clock pathway alterations in HCC leveraging multi-platform profiling of patient tumor samples. Methods: 780 HCC tested at Caris Life Sciences (Phoenix, AZ) with WTS (Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES) were analyzed. Clock gene Score (CS) was determined using expression of core clock genes Z scores (positives of CLOCK, ARNTL, RORA/B/C and negatives of repressors CRY1/2, PER1/2/3, REVERBA/B) stratified by quartiles. xCell was used to quantify cell infiltration in the tumor microenvironment (TME). Significance was determined as P-values and adjusted for multiple testing ( q) of < .05. Gene expression profiles were analyzed for transcriptional signatures predictive of response to immunotherapy including the T cell inflamed score (TIS) and IFG score. Real world survival was obtained from insurance claims data and Kaplan-Meier estimates were calculated for comparison. Results: CS was higher in metastatic sites than primary tumors (median transcripts per million [TPM]: 0.81 vs 0.37, P < .05). No significant differences in patient age and sex were observed between CS Q1 (lowest) and Q4 (highest) cohorts, although a trend towards a higher frequency of males was observed in Q4 (76% vs 68%, Q4 vs Q1, P = .07). CS was positively associated with telomerase subunit TERT mutations (64% vs 52%, Q4 vs Q1, P = .04) and negatively correlated with FGF3 copy number amplification (2% vs 6%, P = .04) and WEE1 gene expression (median TPM: 15 vs 28, q < .05). No dMMR/MSI-H tumors were observed in our series and there were no significant associations with tumor mutational burden and PD-L1 protein expression. Expression of immune related genes was lower in tumors with high CS, including IDO1, CD80, PD-L1, LAG3, CD86, TIM3, PD-1 and PD-L2 (fold change: 0.57-0.67 q < .05). NK cell infiltration in the TME and the TIS score were also significantly lower in CS-high HCC ( q < .05). Individually, lower CLOCK and CRY1 tumor mRNA expression were associated with longer OS (Q1 vs Q4: CLOCK HR 0.71, 95%CI [0.51-0.98], P = .04 and CRY1 HR 0.70 [0.51-0.95], P = .02, respectively). Conclusions: This is the most extensive profiling study to investigate the expression of clock genes in HCC. Our data show that clock genes expression impacts patient survival and is associated with alterations in immune-related gene expression and TIS score which suggest a role in the modulation of anti-tumor immunity. These results support the clock pathway role as a oncogenic driver and its potential as a therapeutic target in HCC.

Clinical implications of molecular alterations in intraductal carcinoma of the prostate.

5024 Background: Intraductal carcinoma of the prostate (IDC-P) is a recent addition to the 2016 WHO classification system and little is known about the nature of its true incidence and biology. Clinically, it has been observed to be associated with higher grade group tumors and a more aggressive disease course with a high risk of local recurrence and distant metastasis. However, the molecular underpinnings which define IDC-P are not well elucidated. Initial case series with molecular profiling have suggested an enrichment of alterations in DNA damage/repair (DDR), PI3K, WNT and MAPK pathways. We now present comprehensive molecular profiling data from the largest cohort of IDC-P reported to date in order to better characterize this relatively new histologic entity. Methods: Eligible patients included those with radical prostatectomy tumor tissue labeled as IDC-P on testing requisition with available tissue-based DNA and RNA sequencing utilizing a commercially available CLIA-certified assay (Caris Life Sciences). Prostatic tumor specimens were analyzed using NextGen DNA sequencing (NextSeq, 592 genes and NovaSeq, WES). Digitized H&E slides underwent central pathological review by a board certified pathologist (SW) specializing in genitourinary malignancies to confirm presence of intraductal histology according to the 2022 WHO classification. Results: 31 of 44 submitted cases were confirmed to be IDC-P on central review. Median age of patients was 64 yrs (range 44-79). The majority of cases (26/31 - 84%) were Grade Group 5 and all were Grade Group 3 or higher. Overall the most frequently detected alterations were in TP53 (39%), TMPRSS2 fusion (39%), FOXA1 (16%), CDK12 (10%), PIK3CA (10%), SPOP (10%), ATM (7%), AKT1 (7%) and RB1 (5%). The most commonly altered pathways were in cell cycle regulators (50%), DDR ( BRCA1/2, ATM, CDK12) (23%), and PI3K/AKT/MTOR (23%) and WNT (10%). The FOXA1 mutation rate in this cohort was higher compared to the localized prostate cancer cohort from the TCGA (16% - 5/31 vs 6% - 82/1325). In the IDC-P cohort, there were no tumors which demonstrated mismatch-repair deficiency (dMMR) or microsatellite instability and none of the profiled tumors had a high tumor mutational burden (≥10 mutations/megabase). Conclusions: This data suggests that IDC-P harbors targetable molecular alterations in DDR genes ( BRCA1/2 and CDK12) with approved targeted agents (e.g. PARP inhibitors). These findings should underscore the importance of germline and somatic mutation testing when IDC-P is identified. We also found enrichment of oncogenic FOXA1 mutations in IDC-P which are associated with prostate cancer progression, castrate-resistance and an unfavorable prognosis. Further efforts are ongoing to expand this cohort of IDC-P cases and compare them with a matched cohort of pure adenocarcinoma cases and will include analysis of key RNA expression signatures.

Not all treated KRAS-mutant pancreatic adenocarcinomas are equal: KRAS G12D and survival outcome.

4020 Background: KRAS is an oncogenic driver in pancreatic ductal adenocarcinoma (PDAC) with mutations identified in > 90% of cases. G12D is the most frequent variant, followed by G12V and G12R. We recently reported on the prognostic impact of distinct KRAS mutations. The current study utilized a large clinical and genomic database, to further explore and characterize the prognostic and molecular differences between KRAS variants, focusing on KRAS G12D and G12R. Methods: PDAC samples were tested using whole transcriptome sequencing (WTS; Illumina NovaSeq) and NextGen DNA sequencing (NextSeq, 592 Genes and NovaSEQ, WES) at Caris Life Sciences (Phoenix, AZ). Transcriptomic signatures including MPAS (MAPK activation score), T-cell inflamed score and tumor micro environment (TME) characterization were calculated on WTS data. Significance was determined by X2 and Fisher-Exact and p-value was adjusted for multiple comparisons (q). Real-world overall survival (rwOS) obtained from insurance claims data was calculated from tissue collection to last contact (comparison done by Kaplan-Meier test). Results: 5,555 PDAC patients harboring either KRAS G12D (n = 2,671), G12V (n=1,871) G12R (n = 904) or G12C (n = 109) variants were identified. Patients with KRAS G12R mutant tumors had significantly longer OS compared to G12D (396 vs 311 days, HR 0.81, CI 0.74-0.88, p<0.0001). There was no difference among KRAS variants in the rate of TP53, CDKN2A and SMAD4 mutations. ARID1A and KMT2D were more frequently mutated in KRAS G12D vs. G12R. The MPAS gene signature reflecting MAPK pathway activation trended lower in G12R vs. G12D. Expression of multiple genes in this pathway was statistically lower in the G12R cohort. Immune profiling suggested that: PDL1 expression is significantly lower in G12R vs G12D (13% vs 19%), TMB-H and dMMR were comparable in G12D vs G12R and several glucose and glutamine metabolism genes were significantly lower in expression when comparing G12R vs G12D - table. OS was improved within the G12R cohort for those patients on metformin (n=273 patients) (416 vs 388 days, HR 0.84, CI 0.72-0.99, p =0.037) while no difference was seen in those with KRAS G12D based on metformin use. Conclusions: Patients with G12D mutations have significantly lower survival compared to G12R. Significant molecular differences were seen in MAPK pathway gene expression, markers of immune activation, and genes involved in glucose and glutamine metabolism. Intriguingly, metformin use appeared to impact survival in the KRAS G12R subgroup. We aim to further explore distinct vulnerabilities based on MAPK pathway activation and dysregulated metabolism. Based on this data, future studies should address the KRAS mutation status and explore distinct therapeutic vulnerabilities. [Table: see text]

Characterizing the role of PIM kinases in the prostate tumor immune microenvironment.

e17080 Background: The Proviral Integration site for Moloney murine leukemia virus (PIM) kinases, a family of pro-survival kinases ( PIM1, PIM2, and PIM3), are frequently overexpressed in prostate cancer (PC) and associated with PC aggressiveness and immune evasion. Preclinical data indicate that PIM1 overexpression increases macrophage infiltration and survival using in vivo models of PC. Here, we characterized the association of PIM1, PIM2, and PIM3 with PC immune signatures. Methods: 77 primary prostate and 11 metastatic lymph node samples from treatment-naive metastatic hormone-sensitive PC pts were analyzed by DNA next-generation sequencing (592-gene, NextSeq; WES, NovaSeq) and Whole Transcriptome Sequencing (WTS; NovaSeq) (Caris Life Sciences, Phoenix, AZ). PC with PIM1/PIM2/PIM3-high(H) and -low(L) expression were classified by top and bottom quartile, respectively. Androgen receptor (AR) signature and Neuroendocrine Prostate Cancer (NEPC) score were calculated based on the expression level of previously defined gene signatures (Hieronymus et al. 2006, Beltran et al. 2016). Pathway enrichment was determined by GSEA (Broad Inst). Immune cell fractions were calculated by deconvolution of WTS using Quantiseq. Statistical significance was determined by chi-square and Mann-Whitney U (p < 0.05) and adjusted for multiple comparisons (q < 0.05). Results: PIM1/PIM2/PIM3-H PC had a higher median MAPK activation score (MPAS) compared to PIM1/PIM2/PIM3-L PC (0.9, 1.3, 0 vs -1.6, -1.4, -1.1 respectively, all p < 0.05). There was no difference in AR signature or NEPC score between PIM-H and PIM-L PC. PIM3-H PC exhibited higher PSA expression (3.15-fold, p < 0.05). PIM1/PIM2-H PC had higher AR expression (1.9 and 3.5-fold, respectively, all p < 0.05). PIM2/PIM3-H PC had enrichment of PI3K/AKT/mTOR, IL2/STAT5, IL6/JAK/STAT3, KRAS, TGFβ, NOTCH, hypoxia signaling, and ROS, P53, OXPHOS pathway and EMT (NES: 1.3-1.5, all FDR < 0.25). PIM1/PIM2-H PC had higher expression of immunostimulatory genes ( IL1β, TNF, and TNFSF13, FC: 1.1-3.1, p < 0.05). PIM1/PIM2/PIM3-H PC had higher expression of MHC class I ( HLA-A, HLA-B, HLA-C, B2M, FC: 1.3-2.5, all p < 0.05) and MHC class II ( HLA-DPA1, HLA-DRB1, HLA-DPB2, TAP1, TAP2, FC: 1.1-3.2, all p < 0.05) genes. PIM3-H PC had increased infiltration of M2 macrophages (8% vs 5.2%) and NK cells (5.2% vs 3.3%), while PIM2-H PC had increased Tregs (1.7% vs 0%) (all p < 0.05). Moreover , PIM1/PIM2/PIM3-H PC had a high T cell inflamed score compared to PIM1/PIM2/PIM3-L PC (66, 80, 56 vs -143, -157, -136, respectively, all p < 0.05). Conclusions: Our data suggest a strong association between PIM expression and increased MAPK activation score, T cell inflamed score, inflammatory, MHC class I and MHC class II gene expression, and differential immune cell infiltration. A better understanding of these differences will provide a rationale for tailored therapeutic approaches for PIM-expressing PC.

Somatic Mutation of PRKAR1A in Four Cases of Sporadic Cardiac Myxoma.

Inactivating mutations of the protein kinase A regulatory subunit 1 alpha (PRKAR1A) gene have been reported in familial cardiac myxoma. However, the role of PRKAR1A mutation in sporadic cardiac myxoma remains unknown. Targeted next-generation sequencing (NGS) was performed to identify mutations with the PRKAR1A gene in seven cases of sporadic cardiac myxomas. Sanger sequencing of DNA from cardiac myxoma specimens and matched peripheral blood samples was performed to verify the identified mutations. Targeted NGS of myxoma DNA revealed 232 single nucleotide variants in 141 genes and 38 insertion-deletion mutations in 13 genes. Six PRKAR1A mutations were identified in four of the seven cardiac myxoma cases, and thus, the PRKAR1A inactivating mutation rate was 57.2% (4/7, 95% CI=0.44-0.58, P<0.05). The PRKAR1A variants identified by Sanger sequencing analysis were consistent with those from the NGS analysis for the four myxoma specimens. All of the pathogenic PRKAR1A mutations led to premature termination of PRKAR1A, except for one synonymous mutation. Moreover, none of the nonsense and missense mutations found in the myxoma specimens were found in the matched peripheral blood samples. Pathogenic mutations of the PRKAR1A gene were identified in tumor specimens from four cases of sporadic cardiac myxoma, and the absence of these mutations in peripheral blood samples demonstrated that they were somatic mutations.

Establishing clinical performance characteristics for Duoseq, a new assay for DNA and RNA sequencing.

e15059 Background: Despite advances in genomics and next generation sequencing (NGS), clinical implementation of NGS remains challenging. This challenge is critical in hematologic malignancies where NGS has become an essential part of the clinical workup. Many clinical labs resort to send-out testing that can delay appropriate treatment and drive up the cost of care. To address these issues, we have developed Duoseq as a unified approach for DNA and RNA NGS. Duoseq assesses mutations (single nucleotide variants [SNVs], insertions/deletions [indels] and splice site variants [SSVs]), structural variants, and gene expression using a panel of over 400 genes. It is supplied as a kit that enables preparation of DNA and RNA sequencing libraries in a single workflow performed at the point-of-care. It includes complete bioinformatics support that enables a clinical laboratory with an NGS instrument to perform the assay with a turn-around time of 2-3 days. Duoseq is optimized to work with formalin-fixed, paraffin embedded (FFPE) samples that are the standard fixation practice. Methods: We evaluated Duoseq on four dimensions critical to clinical performance: reproducibility, precision, analytical sensitivity (limit of detection, LoD), and analytical specificity. Identical experiments were run at clinical laboratories of two different institutions. Analytical validation samples included 4 FFPE cell lines and 2 synthetic DNA samples designed to harbor variants of interest at known allele frequencies. Results: We determined the LoD to be 5% mutant allele fraction (MAF) for SNVs, 10% MAF for indels/SSVs, and 10% tumor purity for SVs. Duoseq achieved 100% precision and reproducibility for each variant class for events at or above the LoD. Analytical specificity was 100%, indicating no instances of cross-contamination of samples within or across runs. We further evaluated Duoseq performance in 54 clinical cases of myeloid and lymphoid malignancies compared against clinical results provided by FoundationOne Heme NGS panel. Collectively, we assessed 119 SNVs and 49 indels/SSVs. Duoseq performed with 100% specificity for both SNVs and indels/SSVs. Duoseq achieved 98% sensitivity for SNVs and 96% for indels/SSVs. For example, Duoseq detected TP53 variants in 6 cases of myeloid neoplasms, which has relevance for the 2022 WHO and ICC classifications. Duoseq also identified therapeutically relevant variants in IDH1/2, EZH2, FLT3, KRAS and TET2. Conclusions: This study demonstrates performance characteristics that suggest Duoseq would perform well as a clinical-grade assay. Local laboratory factors that include staffing, case volumes, and reimbursement affect adoption and turnaround times in the individual clinical laboratory. However, we anticipate that Duoseq implementation will accelerate the diagnosis of hematologic malignancies and facilitate rapid, more cost-effective care to improve patient outcomes.

Methylated DNA biomarkers and incident cancer in the American Cancer Society (ACS) Cancer Prevention Study-3 (CPS-3) cohort.

3004 Background: Multi-cancer early detection (MCED) tests are being developed to complement current single-cancer screening, and are based on next-generation sequencing of cell-free DNA in blood plasma with subsequent analysis of genomic features by machine-learning algorithms. To evaluate the detectability of cancer signals in individuals without known cancer at the time of blood draw, we assessed pre-diagnostic signals using a previously validated locked, methylation-based assay in plasma samples from the ACS CPS-3 cohort. Methods: CPS-3 is a prospective cohort study of cancer incidence and mortality. Participants (N=294,000) donated a blood sample (36 mL) at enrollment (2006-2013) and are followed for cancer outcomes via routine linkages with population-based cancer and death registries. Selected participants diagnosed with cancer within 3 years (Y) of blood donation and an equivalent number of matched non-cancer participants were chosen for processing. Despite differences in sample collection (tube, protocol) and long storage time in freezer, standard requirements for MCED sample quality were applied. Primary outcome: demonstrate that cancer signal was detectable via the MCED test up to 3Y prior to conventional diagnosis. Secondary outcomes: assess detectability based on time from blood draw and other clinical factors. Sampling weights were considered in the statistical analyses. Results: Of 268,726 CPS-3 participants, 2,850 were selected and assayed by cohort sampling, and 2,754 were included in the primary analysis set. CPS-3 cohort demographics and cancer incidence patterns compared to Surveillance, Epidemiology, and End Results suggested that this cohort is younger and healthier than the U.S. population. The test negative rate was 100% among non-cancers. The test positive rate (TPR) stratified by time between blood draw and clinical diagnosis was 2.3% (3Y), 6.0% (2Y), 15.2% (1Y). For eventually fatal cancers, TPR was 5.1% (3Y), 21.6% (2Y), 55.1% (1Y) vs 1.8%, 4.1%, 10.2% for non-fatal cancers. By stage, within 1Y, TPR was 73.7% for Distant at 4-6 months (M) and 94.1% at 1-3M vs 26.7% for Regional at 4-6M and 71.0% at 1-3M. The overall accuracy of predicting cancer signal origin was 95% (1Y). Many cancers were detected 3Y prior to clinical diagnosis: for example, a case of colon/rectum cancer at Regional stage was detected at 991 days. Conclusions: The ACS CPS-3 cohort includes persons at lower-than-average risk of cancer but offered a biospecimen repository with sufficient plasma samples to assess the preclinical detectability of cancer signals by retrospective evaluation. Rapid changes in detectability over time underscore the short preclinical time scales for aggressive cancers to develop, and therefore to be detected. These insights into cancer biology reinforce the importance of pre-diagnostic samples in the study of cancer signals for early detection.

Consensus molecular subtyping of colorectal cancer to demonstrate cetuximab benefit in right-sided CMS2 tumors, and pembrolizumab benefit in MSS CMS1 tumors.

3535 Background: Consensus Molecular Subtypes (CMS) of colorectal cancer (CRC) were first developed in 2015 using microarray-based assays but are not widely used clinically. We developed a Caris CMS classifier on whole transcriptome sequencing data (WTS) with high concordance with the previously established CMS pipeline (Guinney et al 2015), and applied to a large clinic-genomic database of CRC to investigate the utility of CMS classification in identifying patients that may respond well to therapies commonly used in CRC. Methods: Next-generation sequencing (NGS) of DNA (592-gene or whole exome) and WTS was tested on CRC patient samples (n = 12,788) at a CLIA-certified lab (Caris Life Sciences, Phoenix, AZ). Caris CMS classifier was trained against the original CMS datasets using a classic SVM model and cross-validated for optimization of the SVM parameters. Possible overtraining was evaluated by predicting CMS from an independent blinded dataset (TCGA, N = 512) with an accuracy of 88.3%. Real-world overall survival was obtained from insurance claims and calculated from tissue collection to last contact (OS); time on treatment (TOT) was from first to last of treatment time. Kaplan-Meier estimates were calculated for molecularly defined cohorts. Significance was determined as p of &lt; 0.05. Results: Among all patients, CMS1 was seen in 16%, CMS2 in 32%, CMS3 in 17% and CMS4 in 35%. MSI-H/MMRd (31%) and BRAF mut (33%) were most prevalent in CMS1 with KRAS mt the highest in CMS3 (66%). CMS2 was associated with the longest mOS (33m; 95% CI: 31m-35m), followed by CMS4 (29m; 28-31m), CMS3 (27m; 25-29m) and CMS1 (22m; 20-23m). In the microsatellite stable (MSS) tumors treated with pembrolizumab, CMS1 (N = 22, mTOT: 4.2m; 2.8-9.1m) had longer TOT than CMS2 (N = 45, mTOT: 2.1m; 1.4-3.1m), CMS3 (N = 21, 2.1m; 1.4m-3.0m) and CMS4 (N = 40; 2.1m; 1.4m-2.8m); CMS1 vs. CMS2-4 in MSS (HR: 0.58; CI: 0.34-0.97, p = 0.035). When investigating cetuximab, CMS2 had the longest mTOT (N = 189, mTOT: 6.3m; 5.4m-7.7m) among the four groups. Interestingly, although among all Ras WT tumors left-sided CRC showed longer mTOT on cetuximab (n = 457, mTOT: 5.4m; 4.7m-6.3m) than right-sided (n = 151, mTOT: 3.8m; 3.3m-4.5m), CMS2 showed similar mTOT from the left (n = 128) and the right (n = 16, mOT: 6.3m vs. 5.6m, HR = 0.896, 95% CI: 0.501-1.604, p = 0.722). Notably, CMS2 comprises 49% of left- and 15% of right-sided tumors, potentially underlying the TOT difference seen between left and right. Conclusions: A WTS based CMS classifier allows for investigation in a large real-world clinic-genomic database. We found that MSS CMS1 CRC’s may derive benefit from immunotherapy. Additionally, CMS2 subgroup of right-sided tumors may derive benefit from cetuximab. Routine CMS subgrouping of CRC provides important treatment associations that should be further investigated.

Association of CDK4 amplification with duration of response to bevacizumab in glioblastoma.

2026 Background: Bevacizumab remains the standard second line therapy in glioblastoma (GBM) with reported improved progression free survival, but not overall survival. We investigated a large clinico-genomic database of GBM patients treated with bevacizumab for molecular alterations associated with treatment outcome. Methods: Molecular profiles of GBM were tested by next-generation sequencing (NGS) of DNA (592 genes, NextSeq or whole-exome sequencing, Novaseq) and RNA (whole transcriptome sequencing, NovaSeq) at Caris Life Sciences (Phoenix, AZ). Gene amplification was determined by NGS with a threshold of &gt; = 6 copies. Real-world survival information was obtained from insurance claims data. Time-on-treatment (TOT) of bevacizumab was calculated from start to finish of treatment while post-bevacizumab overall survival (bevOS) from start of bevacizumab to last day of contact. Short-term (ST) and long-term (LT) responders were defined as those with TOT ≤6 months and ≥1 year, respectively. Kaplan-Meier estimates were calculated and significance was determined as p values of &lt; 0.05. For molecular comparisons, Fisher’s exact tests and Mann-Whiney U were used when appropriate. Results: Among the total of 383 ST and 107 LT patients identified, no significant difference in gender or age were seen. The LT cohort had significantly longer bevOS compared to the ST cohort (HR 0.27, [95% CI: 0.28-0.35], p &lt; 0.0001). Interestingly, CDK4 amplification was seen in 23% (18/78) of LT patients but only 7% (21/300, p = 0.0002) of ST. When investigating the total cohort of 498 GBM treated with bevacizumab, CDK4 amplified patients demonstrated significantly improved TOT compared to non-amplified (HR: 0.639, [0.48-0.85, p = 0.002]). Conversely, the ST cohort had significantly more EGFR amplifications than the LT cohort (43%, 130/300 vs. 19%, 15/79, p &lt; 0.0001). In the total bevacizumab-treated cohort, EGFR-amplified GBM had significantly worse TOT on bevacizumab than non-EGFR amplified GBM (HR 1.387, [1.15-1.67], p &lt; 0.001). No associations were seen in other alterations explored including PIK3CA mutation (HR = 0.81 [0.587-1.113], p = 0.2), EGFR mutation (HR: 1.209, [0.977-1.495], p = 0.1) or RAF1 (HR: 0.647, [0.318-1.317], p = 0.2). Conclusions: Using a large clinical genomic database with GBM subjected to comprehensive molecular profiling, we demonstrated that amplification of CDK4 and EGFR were associated with long-term and short-term responses to bevacizumab, respectively. This warrants further investigation in independent cohorts controlled for age and other prognostic factors. If confirmed, a genetic basis for treatment optimization may provide meaningful clinical outcomes.

Molecular and immune profiling of TP53-mutated ovarian cancers with non-BRCA1/2 homologous recombination gene alterations.

5579 Background: Molecular alterations in homologous recombination (HR) pathway genes lead to HR deficiency (HRD) in ~50% of high grade serous ovarian carcinomas (HGSOC). The immune microenvironment of BRCA1/2-mutated HGSOC has been extensively characterized, in contrast to tumors with mutations in other HR genes. We evaluated a cohort of epithelial ovarian cancers (EOC) enriched for HGSOC and compared molecular and immune signatures of tumors with BRCA1/2 mutations ( BRCAm) to those with non-BRCA1/2 HR gene mutations (NBHRD). Methods: We characterized 173 EOC by comprehensive genomic and immune profiling (CGIP) using panel-based next-generation DNA (523 genes) and RNA sequencing (384 genes), respectively. We enriched for HGSOC by selecting for TP53 mutations and excluding tumors with non-serous histology or any mutation in POLE, MLH1, PMS2, MSH2, MSH6, or nucleotide excision repair genes. Tumors lacking BRCA1/2-mutations were classified as NBHRD if a reportable mutation was present in canonical HR genes. Mutational burden (TMB, mut/Mb) was estimated from detected SNV/indels. We calculated immune-related expression signatures, including tumor immunogenicity score (TIGS), cellular proliferation (CP), and cancer testis antigen burden (CTAB). PD-L1 expression was assessed by tumor proportion score (TPS) following 22C3 immunohistochemistry. Differences in TMB, TIGS, CP, CTAB, PD-L1 and gene expression rank were tested for using Wilcoxon rank-sum test. Results: Of 173 EOC, 84 cases had complete CGIP, including 11 (13%) BRCAm, 7 (8%) NBHRD (Table), and 66 (79%) lacking HR gene alterations. On average, NBHRD tumors harbored lower TMB (4.1±2.3 vs. 6.6±2.3 vs. p=0.03) and elevated TIGS (59.1±16.2 vs. 41.7±16.2 p=0.044) compared to BRCAm tumors. We found no difference in PD-L1 (p=0.21), CP (p=0.26), or CTAB (p=0.48). Expression changes were identified in 32 immune genes, of which 31 (97%) were substantially up-regulated (1.3-3.3 fold) in NBHRD compared to BRCAm tumors. Conclusions: Our findings showed our cohort of NBHRD tumors were more immunogenic but less mutagenic than BRCAm tumors, implying enhanced immunogenicity unrelated to neoantigen load. Given that immune checkpoint inhibitors have shown limited efficacy in ovarian cancer, our data suggest NBHRD tumors may have stronger and possibly distinct immune escape mechanisms compared to BRCAm tumors. [Table: see text]