Abstract 3383: Clinical and genomic utility of smoking pack years in a real world cohort

Abstract Introduction: In this work, we employ pack years (PYs) to characterize the relationship between smoking exposure and clinical and genomic features in a diverse clinical genomic cohort of lung (NSCLC) and bladder cancer (BLCA) patients. Methods: Clinical annotations including sex, age and stage at diagnosis were extracted from electronic health records (EHR) for 6, 441 NSCLC and BLCA patients who underwent clinical genomic sequencing at Memorial Sloan Kettering Cancer Center after 2014. PYs were extracted from EHR using regular expressions and manually curated. To account for the non-parameteric distribution of PYs we split PYs into four groups: 0(0), 1(0-20), 2(20-40), and 3 (>40). Age at diagnosis was split into 4 groups: Under 45, 46-65, 66-80, and Over 81. Categorical variables such as age, sex, and stage were evaluated by wilcoxon rank-sum test. While continuous variables such as tumor mutational burden (TMB) and fraction genome altered (FGA) were tested by linear regression controlling for age, sex, and stage. Overall survival (OS) was measured from time of sequencing to last follow up and evaluated with a Cox proportional hazards model (CoxPH). Genetic ancestry was determined from MSK-IMPACT sequencing data using reference populations from the 1000 Genomes project, including European (EUR), African (AFR), East Asian (EAS). Ancestry was assigned based on proportion greater than 80%. Results: Our cohort consisted of 4, 953 NSCLC and 1, 488 BLCA patients including 3, 975 former/current smokers and 2, 466 never smokers. There was a significant difference among age and PYs in groups 46-65 and 66-80 in NSCLC (Mdn=1.5 v 20;p <0.0001) and BLCA (Mdn=0 vs 6;p <0.0001). Male patients also tended to have a higher PYs in comparison to female patients in NSCLC (Mdn=9 vs 23;p<0.0001) and BLCA (Mdn=0 vs 6; p<0.05). We did not observe a significant association between FGA and PYs in NSCLC or BLCA patients. However, we did notice a positive relationship between PYs and TMB in NSCLC patients (LR:1.77; CI[1.62-1.92] ; p<0.0001). We also evaluated the relationship between ancestry and PYs where patients with EUR ancestry had a higher PYs than patients with AFR in NSCLC (Mdn=20 vs 3; p<0.0001). Interestingly, we identified a novel genomic association between PYs and STAG2 in BLCA (OR:1.21, CI[1.08-1.36], p=0.001). STAG2 which could be a potential therapeutic target. Lastly, we performed a CoxPH to investigate the relationship between smoking exposure and OS. PYs were associated with an incremental decrease in OS among NSCLC patients where those in PYs group 3 had the highest risk (HR:1.50; CI:[1.34-1.66, ];p<0.0001) and outperform TMB (HR:1.17; CI:[1.06-1.28]; p=0.0012). PYs also predicted OS in BLCA (HR:1.90; CI:[1.49-2.42]; p<0.0001). Conclusion: By utilizing PYs we were able to identify clinical and novel genomic associations such as STAG2 in BLCA and lower OS in NSCLC. PYs can serve as an alternate clinical variable to smoking history alone. Citation Format: Tejiri E. Agbamu, Kelly Pekala, Xinran Bi, Xuechun Bai, David Ma, Nicholas Pickersgill, Michele Waters, Jian Carrot-Zhang. Clinical and genomic utility of smoking pack years in a real world cohort [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2025; Part 1 (Regular Abstracts); 2025 Apr 25-30; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2025;85(8_Suppl_1):Abstract nr 3383.