Sensitivity analysis: non-MSI-H comparison. (A) Standardized mean difference is shown for features pre- and post- adjustment for the TMB10+ cohort. Propensity-adjusted comparison of ICI (tTMB≥10) vs. taxanes (any tTMB level) is shown for (B) TTNT and (C) OS among the cohort excluding MSI-H.

746 Background: PACC is a rare subtype accounting for ~1% of all exocrine pancreatic neoplasms and remains understudied compared to conventional pancreatic ductal adenocarcinoma (PDAC). Here, we performed a comparative analysis of the genomic landscapes of PACC versus PDAC using the FoundationCORE dataset to identify potential therapeutic vulnerabilities. Methods: Comprehensive genomic profiling using hybrid capture-based next-generation sequencing (NGS) was performed on 196 PACC and 29,493 PDAC tumors to identify pathogenic genomic alterations (GAs). All patients (pts) had stage III/IV disease at the time of profiling. Genomic ancestry, microsatellite instability (MSI), tumor mutational burden (TMB), and homologous recombination deficiency signature (HRDsig) were assessed. Statistical analysis was performed using Fisher’s exact test, with false discovery rate correction applied via the Benjamini-Hochberg method for independent variables or the Benjamini–Yekutieli method for dependent variables. Results: PACC was more common in males than PDAC (67.9% vs 52.7%, p<0.0001) with similar age (median 66 yrs, IQR 58-74 vs 67 yrs, IQR 60-74) and genomic ancestry distributions. A median of 4 GAs/tumor (IQR 2.25-5.75) were identified in PACC. MSI-high status was rare in both cohorts (1.0% vs 0.5%, p=1.0). TMB was higher in PACC than PDAC, driven by greater enrichment of cases in the 10-20 mut/Mb (5.1% vs 0.7%, p<0.0001) and ≥20 mut/Mb (2.0% vs 0.5%, p=0.039) ranges. HRDsig positivity was also enriched in PACC (20.9% vs 4.6%, p<0.0001), consistent with prior reports indicating strong association with germline DNA damage repair mutations. Alterations in canonical PDAC driver genes were significantly less frequent in PACC, including KRAS (6.1% vs 93.6%, p<0.0001), TP53 (17.4% vs 78.9%, p<0.0001), CDKN2A (31.6% vs 58.6, p<0.0001), and SMAD4 (19.9% vs 29.1%; p=0.0054). MTAP loss was also less frequent in PACC (15.8% vs 23.8%; p=0.0097). Conversely, PACC was enriched for mutations in BRAF (11.2% vs 2.0, p<0.0001), BRCA2 (14.3% vs 2.9%, p<0.0001), APC (10.2% vs 1.1%, p<0.0001), CTNNB1 (12.2% vs 0.7%, p<0.0001), and GNAS (12.2% vs 3.2%, p<0.0001). Conclusions: PACC exhibits a distinct genomic profile from PDAC, characterized by fewer canonical drivers and frequent KRAS wild-type status. Enrichment for BRAF mutations highlights opportunities for MAPK- targeted therapies, while increased TMB and HRD signatures suggest potential sensitivity to immunotherapy and DNA repair-directed strategies, respectively.

799 Background: ERBB3 is often altered ( ERBB3 -alt) in advanced urothelial bladder cancer (aUBC) and its activation may drive tumor growth and treatment resistance, thus it may serve as a putative biomarker and therapy target. We explored the genomic landscape of ERBB3 -alt and ERBB3 -wild type ( ERBB3 -WT) aUBC. Methods: 9,969 cases of clinically aUBC underwent comprehensive genomic profiling (CGP) using a hybrid-capture based FDA-approved assay (FoundationOneCDx) to assess all classes of genomic alterations (GA). MSI status, TMB, HRD signature, genetic ancestry and COSMIC signature were all calculated from sequencing data for research use-only purposes in the context of this study. PD-L1 expression was determined by IHC using Dako 22C3 TPS assay. PD-L1 IHC was performed on a subset of samples for ERBB3- WT (N=136) and ERRB3 -alt (N=9). Results: 636 (6.4%) aUBC cases featured GA in ERBB3 including 90.2% short variant mutations, 8.8% amplifications and 1% had both. Compared to ERBB3 -WT aUBC, patients with ERBB3 -alt were slightly more often male (78.6% vs 74.7%). Both groups had similar genomic ancestry profiles (European range 84.9% to 86.6%). Median TMB was higher in ERBB3-alt aUBC (12.5 vs 6.3 mutations/Mb) as was TMB ≥10 mut/Mb (61.2% vs 33.0) and ≥20 mut/Mb (28.6% vs 11.1) (Table). APOBEC cosmic genomic signature was more frequent in the ERBB3 -alt aUBC (50.0% vs 30.3%). PD-L1 expression was higher in ERBB3 -WT aUBC with high positive (≥50% TPS) 19.1% vs 11.1% (not significantly different). Individual GA more frequent in the ERBB3 -alt cases included AKT2 (6.8% vs 3.5%), ARID1A (38.4% vs 24.4%), ERBB2 (24.2% vs 18.0%) and TERT (87.1% vs 77.5). GA more frequent in ERBB3-WT aUBC included CDKN2A (38.8% vs 27.2%), CDKN2B (31.2% vs 20.1%), FGFR3 (18.1% vs 14.6%) and MTAP (25.7% vs 16.4%). Conclusions: ERBB3 GA in aUBC identify a subset of cases with relatively distinct genomic characteristics, including higher median TMB and frequency of APOBEC signature as well as potential “targetable” pathways. CGP can help identify putative biomarkers for treatment selection and may inform clinical trial designs. Study limitations include retrospective nature, lack of clinical outcomes data annotation, selection and confounding biases. Genomic and molecular features associated with ERBB3 -altered versus ERBB3 -wildtype tumors. ERBB3-WT ERBB3-alt P-value† N 9333 636 Median TMB 6.3 12.5 <.0001 TMB≥20 mut/Mb 11.10% 28.60% <.0001 APOBEC Signature 30.30% 50.00% <.0001 ARID1A 24.40% 38.40% <.0001 CDKN2A 38.80% 27.20% <.0001 ERBB2 18.00% 24.20% 0.0006 FGFR3 18.10% 14.60% 0.0482 KMT2D 23.60% 16.70% 0.0002 MTAP 25.70% 16.40% <.0001 TERT 77.50% 87.10% <.0001 †FDR corrected using Benjamini/Hochberg adjustment.

2022 Background: ERBB2 (HER2) alterations are associated with increased risk of brain metastases in breast cancer and non-small cell lung cancer (NSCLC). However, the relationship between ERBB2 alterations and brain metastasis across other tumor types remains unclear. We hypothesized that ERBB2 -altered tumors demonstrate increased propensity for brain metastases across multiple cancer types. Methods: Using the US-based deidentified Flatiron Health-Foundation Medicine Clinico-Genomic Database (FH-FMI CGDB), we analyzed patients with advanced solid tumors who received ≥1 line of therapy. ERBB2 alterations included pathogenic/likely pathogenic variants and amplifications. Brain/CNS metastases were identified via ICD-9/10 codes and validated against manual abstraction in NSCLC patients. Focusing on patients without brain/CNS metastases at diagnosis, we performed Cox proportional hazards analyses to evaluate time from initial diagnosis to first brain/CNS metastasis, modeling death as a competing risk. Multivariable models were adjusted for tumor types. Results: The analysis included 47,653 patients across 16 tumor types. Excluding breast/NSCLC, there were 29,272 patients: 2,149 ERBB2 -altered (7.3%) and 27,123 ERBB2 -wild type. 653 patients (30% of ERBB2 -altered patients) had mutations and 1644 patients (77%) had amplifications (some had both). In a multivariable analysis adjusting for tumor type and excluding breast/NSCLC, ERBB2 alterations independently predicted accelerated brain/CNS metastasis development (HR 1.55, 95% CI 1.29-1.86, p<0.001). In separate disease-specific Cox models, the strongest associations were observed in gastric (HR 2.33, 95% CI 1.68-3.22, p<0.001) and ovarian cancers (HR 2.52, 95% CI 1.45-4.38, p<0.001) as well as breast (HR 1.70, 95% CI 1.48-1.96, p<0.001) and NSCLC (HR 1.33, 95% CI 1.11-1.60, p=0.002). Conclusions: ERBB2 alterations predict a significantly increased risk of accelerated development of brain/CNS metastases independent of tumor types, with the strongest effects observed in gastric and ovarian cancers in addition to breast and lung cancers.

Clinical and Pathological Features of Treatment Assignment by Treatment Assignment

Pre-adjustment and post-adjustment cohort Imbalances. Standardized mean difference is shown for features pre- and post- adjustment for the (A) TMB10+ cohort and (B) TMB < 10 cohort.

This cohort study examines patterns of bone-modifying agent use from 2013 to 2024 among men with metastatic castration-resistant prostate cancer.

Purpose We sought to characterize outcome associations of poly ADP-ribose polymerase inhibitors (PARPi) in patients with ovarian cancer (OVCA) whose cancers contain homozygous loss (deletions) of homologous recombination repair (HRR) pathway genes, an alteration class not uniformly reported on clinical testing and not susceptible to acquired resistance via secondary reversion mutations. Methods This study used US-based deidentified Flatiron Health-Foundation Medicine OVCA Clinico-Genomic Database. Patients with advanced OVCA and genomic profiling by Foundation Medicine were assembled into cohorts evaluating PARPi monotherapy use, and separately, PARPi maintenance therapy use, with outcomes associated with HRR gene alterations: BRCA1/2, PALB2, RAD51B/C/D, BARD1, and BRIP1. Results 1.5% (82/5,404) had HRR gene homozygous loss, with BRCA1 and BRCA2 loss being most common. Among all OVCAs with BRCA1/2 alterations, 7.5% (63/836) were homozygous losses. Most BRCA1 (34/38, 89%) and BRCA2 (18/25, 72%) homozygous losses were intragenic, with breakpoints not spanning entire gene. Maintenance cohort: 22 patients had HRR gene homozygous loss, with strongly improved progression-free survival (PFS) for PARPi versus no PARPi use (HR, 0.06, 95% CI, 0.01 to 0.37). Monotherapy cohort: patients without HRR alterations (n = 198), HRR mutations or rearrangements (n = 93), and HRR homozygous loss (n = 8) had median PFS of 4.9, 10.6, and 25.4 months, and median overall survival of 18.5, 22.5, and 51.8 months, respectively. Conclusion We observe durable PARPi effectiveness for patients with OVCA and HRR gene homozygous loss in routine practice. We demonstrate the importance of detecting and reporting homozygous loss in HRR genes, an aspect not standard to all sequencing assays.

PURPOSE:Immune checkpoint inhibitors (ICI) have shown limited efficacy in unselected patients with metastatic castration-resistant prostate cancer (mCRPC). However, ICIs are approved for biomarker-defined subsets: microsatellite instability-high (MSI-H) and/or high tumor mutational burden (TMB-H). The efficacy of ICIs in TMB-H but not MSI-H disease remains unclear, and limited data exist evaluating ICI outcomes associated with blood-based MSI (bMSI) in mCRPC. EXPERIMENTAL DESIGN:This study used the United States-based deidentified Flatiron Health-Foundation Medicine prostate cancer Clinico-Genomic Database. Patients with tissue-assessed MSI (tMSI) and TMB (tTMB) status by an algorithm supporting an FDA-approved CDx for pembrolizumab were included if treated with single-agent ICI. Separately, outcomes on ICI associated with bMSI were assessed, including if treated with single-agent ICI or taxane. RESULTS:Among 2,965 patients with mCRPC, tMSI-H (3.2%) was nearly always also tTMB ≥10 mut/Mb (4.7%). In 84 ICI-treated patients, time to next treatment (TTNT) and overall survival (OS) were more favorable in tMSI-H with any TMB [TTNT HR, 0.18; 95% confidence interval (CI), 0.09-0.37 and OS HR, 0.32; 95% CI, 0.15-0.66] and tTMB ≥10 without tMSI-H (TTNT HR, 0.18; 95% CI, 0.04-0.48 and OS HR, 0.20; 95% CI, 0.05-0.77) compared with tTMB <10 without tMSI-H group. In intrapatient assessments, patients with tTMB ≥10 had more favorable TTNT with subsequent ICI versus prior taxane. Detection of bMSI-H was associated with more favorable TTNT on ICI (HR, 0.34; 95% CI, 0.14-0.83) and OS (HR, 0.21; 95% CI, 0.06-0.75) when tumor fraction ≥1%. CONCLUSIONS:These findings add support for tTMB and tMSI in predicting ICI monotherapy benefit in mCRPC and provide evidence supporting bMSI testing when tissue is unavailable.

Multivariable Outcomes by Biomarker Status on Chemo. Adjusting for potential known confounders, multivariable models evaluating (A) TTNT and (B) OS are shown.

Multivariable Outcomes by Biomarker Status on ICI. Adjusting for potential known confounders, multivariable models evaluating (A) TTNT and (B) OS are shown.

684 Background: Validated homozygous loss detection is becoming increasingly important in clinical practice, with BRCA1/2 loss portending durable PARP inhibitor benefit in prostate and ovarian cancers, recent approval of capivasertib ( AKT pathway, PTEN loss) in breast cancer, and ongoing multi-tumor trials of PRMT5 and MAT2A inhibitors (biomarker: MTAP loss). Homozygous losses are challenging to detect and require intentional NGS assay design and validations. We sought to evaluate the most common losses in advanced pancreatic cancer, their prognostic associations with standard of care therapies, and their prevalence pre- and post-treatment. Methods: This study used the nationwide (US-based) de-identified Flatiron Health-Foundation Medicine Pancreatic cancer clinico-genomic database (FH-FMI CGDB), originating from approximately 280 US cancer clinics (~800 sites of care). Patients with advanced pancreatic cancer and tissue-based genomic testing by FoundationOne CDx were included. Logistic regression assessed associations of prior treatment and biopsy site with homozygous losses. Survival outcomes were evaluated with univariable Cox proportional hazards models. Results: Among 1525 specimens, homozygous losses were most frequently found in CDKN2A (47%), CDKN2B (44%), MTAP (34%), and SMAD4 (14%). MTAP loss nearly always co-occurred with CDKN2A loss (100%) and CDKN2B loss (96%), whereas only 72% with CDKN2A and 74% with CDKN2B loss cases showed concurrent MTAP loss. Prior chemotherapy was associated with reduced likelihood of CDKN2A / B loss ( p < 0.05), but not MTAP or SMAD4 . Liver biopsies were more likely to harbor CDKN2A / B , and MTAP loss ( p < 0.01) but not SMAD4 , compared to other sites. The prevalence of MTAP loss was unaffected by prior chemotherapy exposure (p > 0.05). Consistent with the other genes, MTAP loss did not impact first-line (1L) treatment outcomes. Time to next treatment was similar for patients with and without MTAP loss: FOLFIRINOX, 5.4 [95% CI 4.5–6.1] vs. 6.1 months [95% CI 5.4–7]; Gemcitabine regimens, 4.6 [95% CI 3.8–5.3] vs. 4.9 months [95% CI 4.3–5.6]. MTAP loss was more frequent in KRAS -mutated (34.8%, 489/1406) than in KRAS WT tumors (22.7%, 27/119), where it occurred with or without other actionable alterations. Conclusions: Using an algorithm that supports a validated, FDA-approved test, homozygous losses were frequently detected in pancreatic cancer, with MTAP loss found in 34% of cases, defining a subset potentially eligible for MTAP -targeted therapies. MTAP loss prevalence was unaffected by prior chemotherapy, suggesting biopsy timing has minimal impact. Liver biopsies were more likely to harbor homozygous losses in CDKN2A/B and MTAP. No significant association was observed between homozygous losses and 1L treatment outcomes.

6071 Background: Validated detection of homozygous loss is becoming increasingly important in clinical practice. Examples include targeting the AKT pathway in breast cancer with PTEN loss and ongoing multi-tumor trials of PRMT5 and MAT2A inhibitors, with MTAP loss serving as a key biomarker. Homozygous losses are challenging to detect and require intentional next-generation sequencing assay design and validation. We evaluated the prevalence of the most common homozygous losses in advanced head and neck squamous cell carcinoma (HNSCC) and their associations with HPV status, primary site, and clinical outcomes on standard first-line (1L) therapies. Methods: This study used the nationwide (US-based) de-identified Flatiron Health-Foundation Medicine head and neck cancer clinico-genomic database (FH-FMI CGDB), originating from approximately 280 US cancer clinics (~800 sites of care). Patients with advanced HNSCC who underwent tissue-based genomic profiling with FoundationOneCDx were included. First-line therapy included chemotherapy alone or in combination with cetuximab or immune checkpoint inhibitors (ICIs). Logistic regression assessed the associations of prior treatment, HPV status, and primary disease site with homozygous losses. Clinical outcomes were evaluated with Cox proportional hazards models. Results: Among 969 HNSCC specimens, homozygous losses were most frequent in CDKN2A (22.1%), CDKN2B (17.8%), MTAP (9.4%), and PTEN (5.3%). CDKN2A, CDKN2B and MTAP losses were enriched in tumors arising from the larynx and hypopharynx compared with the oral cavity and oropharynx and were largely mutually exclusive with HPV positivity. PTEN loss was most common in the oropharynx and was enriched in HPV-positive tumors and in metastatic liver biopsies. In univariable analysis, homozygous losses were not associated with outcomes after 1L therapy. In multivariable analysis, HPV negativity and higher ECOG scoring, but not PD-L1 status, were independently associated with worse overall survival. Conclusions: Using an assay that is FDA-approved to detect and report copy-number (CN) losses, homozygous losses of CDKN2A , CDKN2B , MTAP , and PTEN were frequently observed in HNSCC. These alterations were not directly associated with outcomes following 1L therapy. Similar to CDKN2A/B loss , MTAP loss was enriched in HPV-negative tumors and defines a clinically relevant subset of HNSCC (~10%) potentially eligible for emerging MTAP-targeted therapies.

5044 Background: The TP53 Y220C base substitution mutation has become a significant therapeutic target with the development of reactivator drugs that restore TP53 function with a regulatory function in the cell cycle. However, its real-world prevalence and associated genomic landscape is not established in CAPC. Data on concurrent genomic alterations (GA) may guide the development of combinatorial therapeutic strategies. Methods: 26,156 cases of CAPC underwent hybrid capture-based comprehensive genomic profiling (CGP) to study all classes of GA including base substitutions, short insertions, deletions, copy number changes, rearrangements and fusions. Microsatellite instability (MSI) status and tumor mutation burden (TMB) were determined from the sequencing data; comparisons utilized the Fisher Exact method. Results: 144 CAPC (0.6%) of CAPC cases featured TP53 Y220C mutation ( TP53 Y220C+). Patients with TP53 Y220C+ CAPC had slightly higher median age (70.0 vs 68.2; p = 0.020) and numerically lower frequency of TMPRSS2 GA (33.0% vs 39.4%; Not significant (NS)). GA potentially associated with CAPC primary hormonal-based therapy response more frequently found in TP53 Y220C+ cases included SPOP (10.5% vs 1.4%; p < 0.0001) with AR GA (13.6% vs 11.3%; NS) being similar in both groups. GA linked to PARP inhibitor response slightly more frequent in TP53 Y220C+ cases included BRCA2 GA (8.7% vs 3.5%; p = 0.0003) and ATM GA (5.9% vs 4.2%; NS). GA in RAD21 were slightly higher in TP53 Y220C- CAPC (14.8% vs 8.3%; p = 0.029). GA in PIK3CA (6.6% vs 7.0%) and PTEN (32.4% vs 34.5%) were similar in the two groups. Biomarkers likely associated with benefit with anti-PD1/L1 agents were very low in both groups, included higher frequency of CDK12 GA (5.4% vs 0.7%; p < 0.0001) in TP53 Y220C+ cases and higher frequencies of MSI-High status (2.7% vs 0.0%; p = 0.045), mean TMB (3.8 vs 2.1 mutations/Mb; p < 0.0001) and TMB > 10 mutations/MB (4.4% vs 0.0%; p = 0.010) in TP53 Y220C- cases. Conclusions: TP53 Y220C is a rare finding in CAPC, but it may offer a potential therapeutic avenue for these patients whose tumors feature such a mutation. In addition, TP53 Y220C+ cases appear to be genomically relatively distinct from TP53 Y220C- CAPC cases and may feature genomic signatures that could influence treatment selection and trial design. Study limitations include a retrospective, descriptive design, a lack of clinical data annotation, and selection and confounding biases, so our findings are hypothesis-generating. Selective GA in TP53 Y220C+ vs TP53 Y220C- in patients with CAPC. TP53 Y220C+ CAPC (144 cases) TP53 Y220C- CAPC (26,012 cases) P Value Median Age (yrs) 70 (46-89+) 68 (35-89+) TP53 (all) 100.0% 39.7% <0.0001 TP53 (non-Y220C) 7.6% 39.7% <0.0001 SPOP 10.5% 1.4% <0.0001 BRCA2 8.7% 3.5% 0.0003 CDK12 5.4% 0.7% <0.0001 APC 9.1% 4.2% 0.0004 MSI-High 0.0% 2.7% 0.045 Mean TMB 2.1 3.8 <0.0001

Tumor mutational burden (TMB) and PD-L1 are established biomarkers for guiding immune checkpoint inhibitor (ICI) therapy in advanced non-small cell lung cancer (NSCLC). As ICI use expands into early-stage disease, we explored the feasibility of using TMB, which can be determined via a comprehensive genomic profiling assay along with EGFR and ALK genomic alterations, as a biomarker for outcomes in both neoadjuvant and adjuvant settings. TMB-high status (≥10 mut/Mb) showed a numerically higher, but not statistically significant, rate of pathological complete response among patients receiving neoadjuvant ICI and significantly associated with more favorable time to recurrence in patients receiving adjuvant ICI, particularly among patients with PD-L1 expression <50%. TMB should be considered in future early-stage NSCLC ICI clinical trials to further validate these results.

PURPOSE There are few established prognostic biomarkers in metastatic hormone-sensitive prostate cancer (mHSPC). Disease volume and timing of metastases are prognostic and predictive factors but may be inadequate due to heterogeneity. Circulating tumor DNA (ctDNA) provides both circulating volume (tumor fraction [TF]) and genomic data. There are limited data on ctDNA in mHSPC. METHODS This was a multicenter, prospective study of ctDNA testing in mHSPC. Presented here are the results before androgen-deprivation therapy initiation. The primary objective was to evaluate the association between TF and overall survival (OS) and time to mCRPC (TTCRPC). Secondary analyses included evaluating outcomes in subgroups of interest and key genomic subtypes. RESULTS Between 2018 and 2024, 85 patients were enrolled, of whom 72 (25% Black) had evaluable baseline ctDNA samples and are included herein. Baseline TF was positive in 46 patients (64%). Compared with patients with negative ctDNA TF, most patients with positive ctDNA TF had de novo (87% v 39%, P < .001) and high-volume disease (80% v 46%, P = .01). At a median follow-up of 20.8 months, median OS was not reached (NR) with those with negative ctDNA TF and 33 months with positive ctDNA TF (hazard ratio [HR], 3.33 [95% CI, 1.1 to 9.9]; P = .03). However, a negative TF at baseline was associated with an undetectable 7-month prostate-specific antigen, a validated OS surrogate. Median TTCRPC was NR versus 13 months (HR, 3.43 [95% CI, 1.5 to 7.9]; P = .004). ctDNA TF was also potentially prognostic in high-volume disease and those who received doublet therapy. CONCLUSION In this diverse cohort, a positive ctDNA TF at baseline was associated with worse outcomes in mHSPC and may potentially complement current further risk stratification tools.