Abstract Background: Advances in cancer treatment and surveillance have increased survivorship, consequently elevating the risk of multiple primary cancers (MPC). However, distinguishing single primaries (SP) and MPC from recurrences or metastases remains a key barrier to studying gene and environmental drivers of second malignancies at scale. Methods: We developed an automated algorithm to classify tumors as distinct primaries versus recurrences/metastases using IARC criteria, curated exceptions, and clinical and molecular data from Memorial Sloan Kettering’s Cancer Data Science Initiative in 91,906 cancer patients. The classifier was validated against an expert-adjudicated dataset. Standardized incidence ratios (SIRs) were computed using age and sex-adjusted SEER-21 reference rates. To assess therapy-related risk, MPCs were stratified by exposure and analyzed for latency and survival. Inherited etiology based on rare germline pathogenic variants (PV) and cancer-specific polygenic risk scores (PRS) was assessed. Results: The classifier achieved 93% concordance in distinguishing MPC from SP tumors. Applied to 91,906 patients in the MSK-IMPACT cohort, the algorithm identified 16,990 (18.5%) with MPC. Among metachronous cases, the median latency to a second primary was 8.2 years. Twenty-one cancer pairs showed elevated SIRs, including four matching known hereditary syndromes. Excess risk persisted in non-carriers of PV, suggesting polygenic or exposure-related causes. Treatment-related pairs included ovary-leukemia (SIR=5.7), breast-leukemia (SIR=4.0), breast-lung (SIR=2.9), breast-uterus (SIR=2.7), and male bladder-lung (SIR=3.3). Therapy exposure significantly modified risk and latency. Tamoxifen exposure conferred a 3.5-fold higher uterine cancer hazard with earlier onset yet improved survival. Among radiotherapy-exposed breast cancer survivors (n=3,482), higher chest irradiation correlated with a 4-fold secondary lung cancer hazard. Platinum exposure for ovarian cancer (n=2,811) increased Acute Myeloid leukemia (AML) hazard by 4.4-fold while alkylating agents for breast cancer (n=9,497) conferred a 2.9-fold AML risk, with shorter latency supporting treatment-related mechanisms. In male bladder cancer survivors, smokers had 9.1-fold increased hazard of subsequent lung cancer (n=1,316). Among smokers, older age at bladder cancer diagnosis predicted shorter latency to lung cancer. Increased SIR persisted in several cancer pairs, even after accounting for rare PV and PRS, suggesting undiscovered genetic and environmental factors and interactions. Conclusions: Automated classification of MPC reveals genetic and exposure-related patterns in secondary cancer risk, timing, and survival. Ongoing work is integrating tumor genomics as well as polygenic risk scores to identify inherited and therapy-related drivers of cancer development and aggressiveness. (Supported by MSK Niehaus Center and BCRF). Citation Format: Johnathan Amsalem, Ying Liu, Aliya Khurram, Yelena Kemel, Andrew Marderstein, Mitul Waghmare, Semanti Mukherjee, Michael Conry, Vignesh Ravichandran, Saibaba Magunta, Ritika Kundra, Matthew Buas, Christopher Fong, Justin Jee, Michael Berger, Jian Carrot-Zhang, Zsofia Stadler, Venkatraman Seshan, Nikolaus Schultz, Kenneth Offit, Vijai Joseph. Multifactorial risks for multiple primary cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5222.

PURPOSE:Germline likely pathogenic or pathogenic variants (LP/PV) in the RET proto-oncogene account for approximately 25% of medullary thyroid cancers (MTCs). Depending on the variant, individuals with RET LP/PV can have > 70% lifetime risk for MTC or C-cell hyperplasia, an MTC precursor. We assessed prevalence and genotype-phenotype correlations of RET LP/PV in a pan-cancer population undergoing agnostic multigene germline cancer genetic testing. METHODS:Patients with cancer at Memorial Sloan Kettering Cancer Center provided informed consent to Memorial Sloan Kettering-Integrated Mutation Profiling of Actionable Cancer Targets, a paired tumor-normal genomic sequencing protocol (ClinicalTrials.gov identifier: NCT01775072) from 2015 to 2023. Consent included multigene germline testing inclusive of RET. LP/PVs in RET were categorized per standardized MTC risk levels. Clinical data, including family history, surveillance results, and pathology, were extracted from medical records. RESULTS:Twenty-one of 31,866 patients harbored an LP/PV in the RET gene, with the most common variant being p.Val804Met (n = 10) of moderate risk. Although six patients had a personal history of MTC, 15 of 21 patients (71%) had neither a personal nor a family history of MTC. Of the 15, nine underwent endocrinology high-risk surveillance, with five patients deciding to proceed with thyroidectomy due to abnormal surveillance findings. Pathologic findings consisted of isolated C-cell hyperplasia only (n = 1), microMTC and C-cell hyperplasia (n = 2), stage I MTC and C-cell hyperplasia (n = 1), and stage IVA MTC (n = 1). CONCLUSION:In this pan-cancer cohort, 71% of RET LP/PV findings were incidental, with no prior personal or family history of MTC. High-risk surveillance and potential thyroidectomy are warranted in patients with an incidental germline RET LP/PV finding due to high rates of precursor lesions and MTC even among these patients.

BACKGROUND:Approximately 5% of renal cell carcinoma (RCC) occurs in the setting of a hereditary RCC syndrome, but accurate phenotype and cancer risk estimates remain limited by ascertainment bias in reported series. METHODS:We analyzed 32,728 cancer patients who underwent paired tumor-normal sequencing with MSK-IMPACT for germline pathogenic variants (PVs) in RCC hereditary syndrome genes VHL, FLCN, BAP1, MET, SDHB, FH, and proposed RCC risk variants MITF E318K and FH K477dup. We integrated clinical, tumor immunohistochemistry, and genomic data. We performed burden testing across tumor types and, using case-control analysis, validated novel gene-cancer associations with UK Biobank data. RESULTS:Germline PVs diagnostic of hereditary RCC syndromes were identified in 109 of 32,728 patients (0.33%), including 3.6% of RCC cases. Only 61.5% of carriers met clinical criteria for their syndromes and most patients were undiagnosed prior to testing. Using burden testing, we confirm and suggest novel cancer associations, including BAP1 PVs in hepatobiliary cancers and FLCN in colorectal cancer. We independently validated in the UK Biobank the association of FLCN and colorectal cancer, but analysis of BAP1 and hepatobiliary was limited by small numbers. Tumor analyses demonstrated biallelic inactivation in syndromic tumors and supported pathogenic roles for BAP1 in hepatobiliary cancers and FLCN in colorectal cancers. The FH K477dup pathogenic variant appears to lack association with RCC risk and there was weak evidence for association of MITF E318K. CONCLUSIONS:Using large, unselected pan-cancer cohorts and integrated tumor and genomic analysis can help refine the phenotype of rare cancer predisposition syndromes.

Abstract Introduction: Male carriers of BRCA2 pathogenic variants are at elevated risk of prostate cancer (PC), with wide variation in disease frequency observed (88% vs. 34%) for men in the top versus bottom 5th percentile of a 147-SNP prostate cancer polygenic risk score (PRS) (PMID: 34320204). Causal variants and genes at most of these PRS loci remain uncharacterized, limiting our understanding of the molecular mediators and biological mechanisms underlying putative modifiers of BRCA2 risk. Methods: We adopted our published informatics pipeline to prioritize candidate functional variants at PC risk loci using Functional Potential Scores (FPS), assembled from disease-relevant annotations of chromatin accessibility, histone marks, and transcriptional factor (TF) binding. We identified 74 loci with a high-scoring lead SNP or strongly correlated variant (r2>0.80), typically mapping to a predicted enhancer region, and selected several such loci for experimental interrogation: 10q25, 11q13, 12q14, 19p13. Genomic fragments spanning prioritized SNPs were cloned in forward and reverse orientations upstream of a minimal promoter driving Nanoluc luciferase (pNL3.1), and luciferase reporter assays were conducted to evaluate allele-specific enhancer activity in normal prostate (RWPE-1) and PC cell lines (LnCAP, VCAP and 22PC). CRISPR genome editing studies are underway to assess the impact of enhancers and risk alleles on regional gene expression profiles. Results: Three of the four candidate enhancer fragments tested exhibited enhancer activity in at least two cell lines -- 10q25/SNP5, 11q13/SNP10, 19p13/SNP17. Among these, the enhancer at 10q25 exhibited allele-specific activity, with up to ∼two-fold stronger signal observed for allele C versus T in both normal and cancer cell lines. Initial studies also suggested allelic specificity for enhancers at 11q13 (SNP10 G>T) and 19p13 (SNP17 G>A). At 10q25, the functional variant identified maps to the third intron of TCF7L2, a critical gene in the Wnt signaling pathway, in a region bound by several key TFs in prostate including androgen receptor and FOXA1. Functional studies of validated target genes are being conducted with and without genotoxic stressors using human organoid cultures we have established from prostate biopsies of men with BRCA2 mutations. Conclusions: Our findings support the utility of the FPS informatics framework for prioritizing likely functional/causal variants at inherited cancer susceptibility loci and provide a foundation for downstream functional assays using physiologic model systems to investigate causal mechanisms of genetic modifiers of BRCA2-associated prostate cancer risk. (Supported by the Breast Cancer Research Foundation, CureBRCA, the MSK Niehaus Center and the Sabin Foundation) Citation Format: Kenneth Offit, Sanchari Bhattacharyya, Matthew Buas, Brett Carver, Jonathan Fainberg, Yelena Kemel, Catherine Fanjoy, Viaji Joseph, Kathryn Graz, Xu Zhang, Shiv Prakash Verma, Ninghui Mao, Kyrie Pappas, Mitul Waghmare. Toward a functional characterization of polygenic modifiers of BRCA2 associated prostate cancer risk [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 5285.

10612 Background: Identifying hereditary cancer predisposition in patients with hematologic malignancies is critical for their clinical management, surveillance, donor selection, family, and reproductive counseling. However, genetic testing for patients with hematologic malignancies has been limited, largely due to the difficulty of obtaining suitable DNA samples for germline analysis. Heme Germline-MSK-IMPACT is a clinically validated assay for germline testing using nail or saliva DNA, depending on the patient’s cancer type, in a non-invasive, rapid, and high-throughput manner, with the matched tumor analysis supporting results interpretation. Methods: Genetic testing was performed on 1300 consecutive patients with a clinical diagnosis or suspicion of a hematologic malignancy on Heme Germline-MSK-IMPACT, targeting 82 hereditary cancer predisposition genes. All individuals receiving somatic testing via paired tumor-normal sequencing on MSK-IMPACT-Heme as part of their clinical care were eligible to consent for germline analysis. Results: Germline pathogenic or likely pathogenic variants (gPVs) in hereditary cancer predisposition genes were identified in 16.7% (217/1300) of patients, with 5.8% (76/1300) having high penetrance autosomal dominant gene variants. gPVs known to be associated with hematologic malignancies were detected in DDX41 (n = 14), POT1 (n = 3), RUNX1 (n = 2), ETV6 (n = 1), TP53 (n = 1). Additionally, four patients had biallelic variants in genes causing autosomal recessive disorders that confer increased risk for hematologic malignancies ( FANCA (n = 2), ATM , MSH6 ). Other high-risk cancer predisposition genes with gPVs identified in multiple patients were BRCA2 (n = 16), BRCA1 (n = 15), MSH6 (n = 4), RTEL1 (n = 4), CDKN2A (n = 3), and PALB2 (n = 3). Moderate penetrance gPVs were detected in 6.2% (81/1300) of patients in genes such as CHEK2 (n = 36), ATM (n = 24), and LZTR1 (n = 8). Twenty-four (1.8%) patients had two or more gPVs identified. Conclusions: Germline testing in a broad cohort of individuals with hematologic malignancies detected hereditary cancer predisposition in a substantial proportion of the patients. This ratio is comparable to the rate of cancer predisposition variants identified in patients with solid tumors in recent studies. Matched tumor data analysis is ongoing to uncover second hits and correlations for improving our understanding of the hereditary contribution to hematologic malignancies.

DNA damage response genes (DDRG), implicated in several cancers as both predisposing risk factors as well as biomarkers for aggressiveness, have not been fully explored in multiple myeloma (MM). Herein, we analyzed disease associations of pathogenic variations in nine putative candidate genes using 3 446 MM cases and 323 233 cancer-free controls. Increased MM risk was found to be associated with inherited rare pathogenic mutations in TP53, ATM, CHEK2, KDM1A, and ARID1A, with an enrichment of these variants among individuals with early onset or family history of MM. Individuals with TP53 or ATM germline mutations are also likely to have worse overall survival. Our results suggest expansion of the phenotypic spectrum of some of these DDRG to include MM. The identification of these germline predisposition genes opens the avenue for targeted screening of higher risk individuals especially those with young-onset or a family history of plasma cell gammopathies.

BACKGROUND:Urothelial cancer typically affects older adults, yet early-onset urothelial cancer (EO-UC, age ≤ 45 years) cases are rising and remain poorly characterized. We hypothesized that EO-UC exhibits distinct clinical and genomic features that reflect a different biological etiology from standard-onset UC (SO-UC). METHODS:We compared clinical, pathologic, and genomic characteristics of patients with EO-UC versus SO-UC evaluated at Memorial Sloan Kettering Cancer Center from 2000 through 2024. Clinical data included 9,221 patients, with tumor and germline genomic profiling by MSK-IMPACT in 2,753 patients. RESULTS:EO-UC accounted for 335/9,221 (3.6%) cases. EO-UC patients were more likely never-smokers (43% vs 26%, p < 0.001), female (30% vs 25%, p = 0.034), and of Asian race (7.4% vs 3.1%, p < 0.001). EO-UC tumors were more often low-grade (41% vs 23%, p < 0.001) and of pure non-urothelial histology (4.0% vs 1.7%, p = 0.003). EO-UC showed marked enrichment for HRAS mutations, independent of histologic subtype, and HRAS-mutated tumors had reduced APOBEC-associated mutation signatures. Germline pathogenic variants were detected in 21.0% of EO-UC vs 17.2% of SO-UC patients (p = 0.43), driven mostly by mismatch repair and MUTYH gene alterations. Rare cases of somatic mosaicism in HRAS and ERCC2 were observed in patients with very early-onset disease. CONCLUSIONS:EO-UC represents a biologically distinct subset characterized by HRAS-driven oncogenesis, reduced APOBEC mutagenesis, frequent germline variants, and, rarely, somatic mosaicism. These findings suggest developmental or genetic mechanisms underlying EO-UC, and supporting age- and biology-informed approaches to risk assessment, genetic counseling, and targeted therapy development.

The NCCN Clinical Practice Guidelines in Oncology (NCCN Guidelines) for Genetic/Familial High-Risk Assessment: Breast, Ovarian, Pancreatic, and Prostate are intended to serve as a resource for health care providers to identify individuals who may benefit from cancer risk assessment and genetic counseling and testing; help guide decisions related to genetic testing; and facilitate a multidisciplinary approach in the comprehensive care of individuals at increased risk for hereditary breast, ovarian, pancreatic, and prostate cancer. The current guidelines focus primarily on assessment of pathogenic and likely pathogenic (P/LP) variants associated with increased risk of breast, ovarian, pancreatic, and prostate cancer and recommended approaches to genetic counseling/testing and care strategies in individuals with these P/LP variants associated with increased risk of these cancers. These NCCN Guidelines Insights summarize the panel's most recent recommendations regarding screening for prostate cancer and pancreas cancer, as well as testing criteria for nonepithelial ovarian cancer.

PURPOSE:Early-onset endometrial cancer (eoEC) is increasing, and germline drivers may be enriched in younger patients. We sought to define germline pathogenic variants (gPVs) in those with EC by age. METHODS:We identified patients with EC who underwent clinical tumor-normal sequencing from December 2014 to June 2021 and collected clinical variables. Logistic regression models evaluated associations between age at EC diagnosis and presence of gPV, biallelic inactivation, and Lynch Syndrome (LS). Age categories were defined as early-onset (eoEC, EC < 50 years) and late-onset (EC ≥ 70 years) and were compared with those diagnosed ages 50-69 years. RESULTS:Among 1,625 patients with EC, the median age at diagnosis was 63 (range, 24-96) years. We observed gPV in 28 (16%) of 170 patients with eoEC, 152 (14%) of 1,066 patients diagnosed age 50-69 years, and 36 (9%) of 389 patients with late-onset EC (P = .016). LS was enriched in eoEC, with 6.5% of patients diagnosed age <50 years having LS. In multivariable models compared with those with EC diagnosed age 50-69 years, eoEC was more likely to exhibit biallelic inactivation (odds ratio, 3.34 [95% CI, 1.44 to 7.35]) and be associated with LS (hazard ratio [HR], 3.49 [95% CI, 1.63 to 7.01]). Among early-onset EC, 14 (50%) of 28 gPV were high penetrance and 14 (50%) of 28 exhibited biallelic inactivation. However, heterogeneity was observed, and rates of gPV were 8.9% and 19%, biallelic inactivation was 0% and 11%, and LS was 2.2% and 8% in those diagnosed age <40 years and 40-49 years, respectively. CONCLUSION:Rates of gPV, biallelic inactivation, and LS differ across age groups for EC, with high-penetrant genes driving tumorigenesis enriched in younger patients. However, very-early-onset EC may have different drivers and necessitates more research.

Abstract Breast cancer in women with germline BRCA1/2 pathogenic variants (gBRCA1/2) are generally treated with platinum-based therapies and PARP inhibitors (PARPi) with resistance commonly emerging. As the tumor microenvironment (TME) in gBRCA1 triple-negative breast cancer (TNBC) is enriched with tumor-infiltrating lymphocytes (TILs) and CD8 T cells, treatment trials have been done combining PARPi and immune checkpoint inhibitors (ICIs) in BRCA1 TNBC. This combination has not been shown to be more effective than PARPi alone. Evaluating the TME in gBRCA1/2 TNBC may help identify tumors most likely to benefit from PARPi/ICI therapy. We performed a detailed spatial proteomic analysis to characterize tumor-immune cell interactions in patients with gBRCA1/2 and wild-type (WT) TNBC with spatial tissue multiplexing (PhenoCycler) in 101 gBRCA1, 24 gBRCA2, and 30 WT TNBCs with matched RNAseq for 34 gBRCA1, 8 gBRCA2, and 16 WT TNBCs. A 43-plex antibody panel was developed featuring markers of DNA damage and repair, immune subtypes and exhaustion. We detected single tumor cells (PANCK+) in S/G2 phase (Geminin+) with double-stranded DNA breaks (yH2AX+) and DNA repair capacity (RAD51+) across all three cohorts. gBRCA1/2 TNBC patients exhibited a significantly lower proportion of tumor cells with homologous recombination proficiency (HRP) (gBRCA1 p = 0.006; gBRCA2 p = 0.007) compared to WT TNBC. CD4 & CD8 T cells, and CD20 B cells had intact DNA repair in WT and gBRCA1/2 TNBC. The frequency of CD8+ T (p=0.016) and CD20 B (p=0.003) cells was significantly higher in gBRCA1 compared to WT TNBC; BRCA2 and WT TNBC showed no differences. A detailed characterization of CD8 T cells revealed significantly increased numbers of potentially dysfunctional CD8 T cells in BRCA1 (TOX, p<0.0001; LAG-3, p=0.028; PD-1, p=0.033) and BRCA2 (LAG-3, p=0.033) compared to WT TNBC. We observed two types of TMEs in gBRCA1 TNBC: 1) CD8 low (mean<9.38%) with 1.4-fold increased immune checkpoint (PD-1) expression (mean: 15.4%) and high DNA damage in tumor cells; and 2) CD8 high (>9.38%) with reduced PD-1 and low DNA damage in tumor cells. Our findings suggest that although gBRCA1/2 variants lead to DNA damage and impaired repair in tumor cells, T cells (CD4, CD8) and B cells (CD20) retain intact DNA repair mechanisms. We also found that gBRCA1/2 TNBCs exhibit higher levels of immune checkpoint proteins LAG-3 and PD-1 on CD8 T cells compared to WT TNBC. This finding suggests the potential utility of additional ICI (LAG-3, PD-1) beyond PD-L1 blockade. Importantly, patients with gBRCA1-associated TNBC exhibit two different TMEs, suggesting that the response to ICI- and DNA-damaging-based therapies may differ between tumors, and anticipated prior to treatment. Defining treatment-naïve TME is crucial for designing personalized, targeted ICI strategies for individuals with BRCA-mutated TNBC. Citation Format: Dana Pueschl, Danielle Bragen, Jia-Ren Lin, Anupma Nayak, Derek A. Oldridge, Kate Bennett, Victoria Fang, kConFab Investigators, Kenneth Offit, Andrew K. Godwin, Paul A. James, Phuong L. Mai, Soo Hwang Teo, Antonis Antoniou, Georgia Chenevix-Trench, E. John Wherry, Susan M. Domchek, Katherine L. Nathanson. A single-cell spatial proteomic analysis of the TNBC microenvironment defines genotype-specific features [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 4966.

Polygenic risk scores (PRSs), which quantify inherited susceptibility to complex traits and diseases, have emerged as valuable tools for risk stratification and precision medicine. Despite their promise, PRS developed on European cohorts often demonstrate substantially reduced predictive accuracy in non-European populations, due to differences in genetic architecture. The disproportionate representation of European ancestry cohorts in genome-wide association studies (GWAS) leads to inequitable deployment of PRS technologies across diverse populations. Here, we introduce PRANA (Polygenic Risk Adaptation via Neural-network Architecture), a deep learning framework that adapts an existing PRS developed on one population to other ancestries. Unlike methods that require large-scale GWAS in the target population, PRANA leverages pre-trained PRS models derived from European cohorts and adapts them using modestly sized cohorts from the target population. We evaluated PRANA on seven complex traits in South Asian, East Asian and Ashkenazi Jewish populations, as well as in selected smaller East Asian subpopulations where the scarcity of training data poses a particular challenge. PRANA mostly improved predictive performance of the baseline PRS models by 5%-20% in terms of effect size (β) and Nagelkerke's R2, and, in most cases, outperformed existing cross-ancestry multi-PRS approaches. These results highlight PRANA as a scalable and practical strategy to reduce disparities in genomic risk prediction and advance the equitable application of PRS in diverse populations.

The co-occurrence of germline and somatic oncogenic alterations is frequently observed in breast cancer, yet their combined influence on tumour evolution and therapy resistance remains poorly defined. Through an integrated clinicogenomic analysis of more than 5,800 patients, we show that germline (g) pathogenic variants dictate the evolutionary trajectory of acquired resistance. We specifically find that gBRCA2-associated tumours are uniquely predisposed to develop acquired RB1 loss-of-function alterations, resulting in poor outcomes on standard-of-care frontline CDK4/6 inhibitor (CDK4/6i) combinations. This vulnerability is driven by a dual mechanism: baseline RB1 hemizygosity (heterozygous loss resulting in a single functional RB1 allele), which lowers the evolutionary barrier to biallelic inactivation, and ongoing homologous recombination deficiency, which promotes acquisition of RB1 loss-of-function alterations under the selective pressure of CDK4/6i. Preclinical models from gBRCA2 carriers showed near-uniform resistance to CDK4/6i, with consistent post-treatment Rb loss. Across multiple independent models and in our clinical data, PARP inhibition consistently outperformed CDK4/6i. Our findings suggest that prioritizing PARP inhibition in gBRCA2 carriers may intercept RB1-loss trajectories and delay resistance. More broadly, we establish a predictive framework for forecasting drug-resistant trajectories based on pre-treatment allelic configuration and mutational signatures.

Improvements in cancer survival have increased the burden of subsequent primary malignancies. We developed and validated a programmatic classifier of multiple primary cancers (MPC) to derive second cancer phenotypes at scale. Among 81,175 cancer patients, we identified 56 first-second cancer pairs, 22 of which exceeded SEER primary cancer incidence rates. Even after accounting for various known risk factors, substantial elevated risk persisted, even in established hereditary cancer pairs (breast-ovary, breast-pancreas, prostate-pancreas), suggesting that current screening protocols do not adequately account for MPC susceptibility. To address this limitation, we built machine-learning models integrating rare germline variants, polygenic risk scores, treatment exposures, and demographic features to predict site-specific second primaries in breast and prostate cancer survivors. These models accurately predicted second ovarian and pancreatic cancers across a long follow-up period (15-year time-dependent AUC 0.70). This is the first systematic, pan-cancer integration of clinicogenomic factors for early prediction of second-primary malignancies. Our framework enables individualized risk estimation, enhanced targeted surveillance, and cancer prevention amongst a growing population of cancer survivors. Statement of Significance:We identified second cancers that occurred more often than expected among survivors. Predictive models using genetic, lifestyle, and clinical factors accurately identified patients at higher risk of second hereditary cancers. Such predictions can enable cost-effective, selective surveillance in a growing population of cancer survivors, reducing cancer burden.

Bladder cancer is the ninth most common cancer worldwide, caused by genetic and environmental risk factors. Here, we report the findings of a multi-population meta-analysis of genome-wide association studies, including 32,470 individuals with and 1,753,462 without bladder cancer. We identify 70 independent risk loci, of which 43 are novel. Using a 70-marker polygenic risk score (HR = 1.63 per standard deviation), we increase the area under the curve from 0.71 (baseline risk model) to 0.75. Integrative analyses reveal the enrichment of the associated variants within accessible chromatin regions, and of the prioritized genes within pathways for xenobiotic metabolism and smoking behavior. Specifically, we show that the 15q25.1 variant rs71581744-ACCCC/A co-localizes with tissue-specific CHRNA3 expression, modulates mRNA stability, and associates with risk of muscle-invasive bladder cancer among current smokers. Together, these findings substantially expand the known genetic architecture of bladder cancer risk and highlight the germline regulation of smoking behavior as a mechanism driving bladder cancer susceptibility.