Drusbosky, Leylah

Genomic mutations impact non-small cell lung cancer (NSCLC) biology. The influence of sex and age on the distribution of these alterations is unclear. We analyzed circulating-tumor DNA from individuals with advanced NSCLC from March 2018 to October 2020. EGFR, KRAS, ALK, ROS1, BRAF, NTRK, ERBB2, RET, MET, PIK3CA, STK11, and TP53 alterations were assessed. We evaluated the differences by sex and age (<70 and ≥70) using Fisher’s exact test. Of the 34,277 samples, 30,790 (89.83%) had a detectable mutation and 19,923 (58.12%) had an alteration of interest. The median age of the ctDNA positive population was 69 (18–102), 16,756 (54.42%) were female, and 28,835 (93.65%) had adenocarcinoma. Females had more alterations in all the assessed EGFR mutations, KRAS G12C, and ERBB2 ex20 ins. Males had higher numbers of MET amp and alterations in STK11 and TP53. Patients <70 years were more likely to have alterations in EGFR exon 19 del/exon 20 ins/T790M, KRAS G12C/D, ALK, ROS1, BRAF V600E, ERBB2 Ex20ins, MET amp, STK11, and TP53. Individuals ≥70 years were more likely to have alterations in EGFR L861Q, MET exon 14 skipping, and PIK3CA. We provided evidence of sex- and age-associated differences in the distribution of genomic alterations in individuals with advanced NSCLC.

Abstract Background: Non-small cell lung cancer (NSCLC) makes up ~85% of lung cancers, and its major subtypes are adenocarcinoma (AD) and squamous cell carcinoma (SC). The histology classification of NSCLC has significant therapeutic and prognostic implications. As liquid biopsies are rapid and non-invasive, computational methods to classify NSCLC subtypes using cell-free DNA (cfDNA) are desired. Methods: Using The Cancer Genome Atlas (TCGA) NSCLC methylation data, we developed a method to classify AD and SC subtypes from cfDNA samples. We mapped TCGA methylation data to Guardant Infinity’s promoter regions and transformed them to allow for the transfer of the TCGA trained model to Infinity cfDNA epigenomic data. An advantage of this approach is that TCGA training data originates from the primary tumor, while cfDNA samples have low tumor fraction (TF) and may also contain signals from metastatic tumors. We trained a penalized Logistic Regression model with LASSO. To better fit the model to Infinity cfDNA data, we performed region filtering. Results: Our model effectively classified AD and SC subtypes from TCGA samples (AUC 0.97) (10-fold cross validation; Table 1). When evaluating on cfDNA samples with epigenomic TF ≥ 1%, the final model shows reasonable performance (AUC 0.94) (Table 1); the performance decreases when reduced to TF &gt; 0.05% (AUC 0.89) as expected (Table 1). When evaluating cfDNA from uncategorized lung samples, the predicted AD probabilities have significant association with KRAS mutations (p-value = 0.027). Conclusion: Our results show that the penalized Logistic Regression model trained with tissue TCGA methylation data through mapping, transformation, and region selection can classify NSCLC subtypes from cfDNA samples. This method may offer a fast, non-invasive tool to inform subtype-based clinical decisions, especially in the setting of poorly-differentiated carcinomas, and to identify subtype switches upon resistance to targeted therapies. Table 1: Model evaluation on tissue TCGA samples and cfDNA Infinity samples (different TF cutoffs) Sample type N samples AUC Accuracy Sensitivity Specificity Precision F1 score tissue TCGA 907 0.97 0.93 0.94 +/- 0.02 0.93 +/- 0.02 0.94 +/- 0.02 0.94 +/- 0.01 cfDNA Infinity 17 (TF ≥ 1%) 0.94 0.88 0.83 +/- 0.3 0.91 +/- 0.17 0.83 +/- 0.3 0.83 +/- 0.21 cfDNA Infinity 29 (TF &gt; 0.05%) 0.89 0.83 0.88 +/- 0.23 0.81 +/- 0.17 0.64 +/- 0.28 0.74 +/- 0.2 Citation Format: Laura Tung, Denis Tolkunov, Catalin Barbacioru, Deepika Kale, Scott Shell, Jill Tsai, Leylah Drusbosky, Kaushal Parikh, Han-Yu Chuang. Non-small cell lung cancer (NSCLC) histology classification using DNA methylation data captured from liquid biopsies [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 3499.

239 Background: In part, TMB guides therapeutic decisions on the use of immune checkpoint inhibitors (ICIs) in aPC. Most patients (pts) with aPC do not have high TMB (TMB-H) at the time of initial assessment. However, TMB changes over time (from treatment pressure) as pts progress through lines of therapy (LOT) has not been extensively explored in aPC. We describe real world data on changes to blood TMB (bTMB) in pts with aPC exposed to several LOT. Methods: Genomic results from pts with aPC who had ≥1 ctDNA test(s) with TMB analyzed (Guardant360, Guardant Health, Inc) as part of clinical care from Oct 2020 – Jun 2022 were retrospectively queried. Clinical factors were extracted from test requisition forms; pts from Mayo Clinic had additional factors, including intervening treatment, extracted from medical records. High bTMB (TMB-H) was defined as ≥13.4 mut/Mb, as previously reported. Proportions were compared using Fisher’s Exact Test. Results: 316 patients with a median age of 72 (range: 44-91) were analyzed. Initial median TMB for the entire cohort was 8.18 mut/Mb (range: 0.01-268.3). On subsequent test, TMB decreased in 122(40%), increased in 173(55%), of whom 29 (17%) became TMB-H; 21 (7%) had unchanged TMB. Median time between tests was 182 days (range: 21-578). 101 patients with 262 tests and a median age of 71 years (range: 46-87) were clinically annotated in the Mayo cohort of whom 58 had >2 consecutive TMB scores: 25 (43%) had TMB increase, of whom 4 (16%) became TMB-H, 23 (40%) had TMB decrease, 10 (17%) had dynamic changes with increases and decreases. Numeric but not statistically significant differences in clinical factors were observed between groups (p>0.05). Conclusions: In this real world cohort, variation in TMB is seen with increasing LOT. Importantly, nearly 1 in 5 patients with increasing tumor mutational burden develop TMB-H, which may be a clinically actionable finding as these pts become eligible for ICI. Findings underscore the importance of repeat genetic testing with time and LOT. [Table: see text]

3601 Background: Two cfDNA assays are FDA-approved for use in advanced cancer pts across solid tumors. While several NCCN guidelines recommend cfDNA as a valid and useful method for molecular testing, the utility varies by cancer type. Despite the growing number of molecular alterations pertinent to first-line treatment in mCRC, current guidelines offer little guidance on the use of cfDNA in mCRC. Here we assess the ability of cfDNA to detect guideline-relevant actionable biomarkers informative for first-line therapy selection and beyond in patients with mCRC. Methods: We queried consecutive samples from advanced cancer pts with a diagnosis of CRC who underwent testing via a commercially available cfDNA assay (Guardant360) from September 2018 to January 2022. Validation and content of this assay has been previously described. To enrich for less heavily treated pts, we limited analysis to those whose first cfDNA test was done during the study period and only included the pt’s first test. The maximum variant allele fraction (maxVAF) for each test was used as a proxy for tumor shed and was compared between cancers with a Mann-Whitney test. The frequency of suspected germline alterations was assessed only in pts who received testing via a panel version with expanded mismatch repair (MMR) gene coverage. Tissue frequencies for select biomarkers were obtained from the MSK-IMPACT Clinical Sequencing Cohort in cBioPortal. Results: In total, 14,345 pts with mCRC received their first cfDNA test during the study period, with 92% of pts (13,190) having >1 cfDNA alteration detected. Median age was 62 and 44% of pts were female. The frequencies of key biomarkers are listed in Table 1. The median maxVAF was 5.6%, compared to 2.5% and 1.8% for pts with advanced gastric and NSCLC, respectively (p = < 0.0001 for both). The average turnaround time (TAT) for tests was 7 calendar days. Conclusions: cfDNA results from pts with mCRC demonstrated detection of biomarkers essential to first-line treatment decisions at a frequency comparable to what has been reported via tissue genotyping. mCRC pts had significantly higher tumor shed compared to other cancer types with guidelines that recommend cfDNA as an option for molecular testing, suggesting a potential gap in coverage. Tissue availability in mCRC does not guarantee comprehensive testing is completed quickly (compared to the 7-day TAT seen for cfDNA here), making it an attractive alternative strategy for first-line therapy selection in mCRC. [Table: see text]

3590 Background: Co-occurring alterations (COAs) in the HER2 and MAPK pathway are rare in mCRC. RAS, HER2 and BRAF alterations have been associated with resistance to EGFR inhibitor therapy (EGFR-mab). Typical co-occurring mutually exclusive gene mutations such as in KRAS and BRAF have been reported. This study aims to evaluate the prevalence of co-occurring alterations in the MAPK pathway or across MAPK/ HER2 and correlate to outcomes in patients with mCRC treated with EGFR-mab. Methods: This is a retrospective study of patients with mCRC within the Mayo Clinic database who have available blood-based NGS Guardant 360 data (September 2017-November 2021) and contained co-alterations in the MAPK pathway or across MAPK/ HER2. MSI-high cases were excluded. Typical-RAS was defined as alterations in codons 12, 13, 59, 61, 117, and 146 of KRAS, HRAS, and NRAS. Atypical-RAS alterations were considered mutations at other codons of these genes. Patient characteristics, specific mutations, and outcome data were assessed. Results: 692 patients (pts) with mCRC were identified and 66 (9.5%) of those had COAs detected. 59/66 pts had clinical data available. Median age was 52 years, 55.9% were male, and 61.0% had left-sided tumors. The frequency of detected alterations was: BRAF-V600E (7.3%), BRAF-non-V600E (7%), BRAF-amplification (3.04%), RAS-typical (39.5%), RAS-atypical (2.1%), HER2 amp (0.5%), HER2-mutation (1.2%) and PIK3CA (13%). COAs are described in the table. Twenty-three pts received EGFR-mab; COAs developed in 19 of them after progression on EGFR-mab. Four pts received EGFR-mab with documented COAs at baseline with none of them responding to treatment. Conclusions: COAs in MAPK and MAPK/ HER2 pathway are not uncommon in pts with mCRC emphasizing on the importance and feasibility of blood based NGS vs. selected gene testing. In pts with prior EGFR-mab treatment, COAs were detected after progression on treatment in the majority of pts, suggesting a mechanism of secondary treatment resistance. Further data including prospective validation is warranted. [Table: see text]

165 Background: A prior study of patients with advanced cancers treated with immune checkpoint inhibitors (ICI) associated improved overall survival with tissue tumor mutational burden above the 80th percentile in each histology (tTMB-H). TMB can also reliably be calculated via liquid biopsy (bTMB), and trends higher than tTMB in analogous tumors. Here, we generate a plasma-informed benchmark for bTMB-H in advanced prostate cancer and report preliminary data from an ongoing multi-institutional case series of patients treated with ICI. Methods: Circulating tumor DNA (ctDNA) next generation sequencing (NGS) results from 3,504 patients with advanced prostate cancer who received a Guardant360 liquid biopsy within a 12-month period were analyzed. The 80th percentile of TMB was defined as the benchmark for bTMB-H in this cohort, and prevalence of microsatellite instability (MSI-H) was determined. Subsequently, clinical data were collected from patients across five institutions who received pembrolizumab based on bTMB scores or other tumor characteristics (i.e. CDK12 mutation). Results: Overall, mean and median bTMB were calculated as 12.34 mut/Mb and 8.61 mut/Mb, respectively (range: 0 – 1164.75 mut/Mb). The 80th percentile of bTMB was 13.4 mut/Mb and the 90th percentile was 16.6 mut/Mb. 2.6% of patients were both MSI-H and bTMB-H. No patients with bTMB < 13.4 mut/Mb (TMB-L) were found to be MSI-H. Ten patients from five institutions received pembrolizumab (8 monotherapy) after a median of 4.5 lines of prior therapy (Table). Four patients were MSI-H. The 3 patients in the cohort with PSA decreases > 50% had both MSI-H and bTMB ≥ 13.4 mut/Mb. Per RECIST criteria, disease was controlled (CR, PR, SD) in 6 of 7 (86%) patients with bTMB scores ≥ 13.4 mut/Mb. Conclusions: These data continue to affirm that ctDNA NGS is a practical and cost-effective means of assessing potential predictors of ICI activity in patients with advanced prostate cancer. Preliminary data from this growing cohort of patients who have received ICI guided by bTMB scores and other molecular parameters identified via ctDNA NGS, including MSI-H, demonstrate a clinical benefit from therapy. Further studies are warranted for contextualizing individual tumor bTMB scores with established, pathology-specific benchmarks.[Table: see text]

9108 Background: Non-small cell lung cancer (NSCLC) remains a leading cause of cancer-related death in the U.S. The median age at diagnosis is 70 years, and NSCLC is uncommon among younger individuals ( < 50 years).Overall, outcomes in NSCLC have improved significantly with targeted therapy. A prior study demonstrated patients < 50 are more likely to have targetable alterations including EGFR, ALK, ERBB2, and ROS1. Another study reported an increased prevalence of EGFR mutations in females and KRAS mutations in males with NSCLC. The comprehensive genomic landscape of NSCLC patients in different age groups and genders remains largely unknown. In our study, we aim to investigate the genomic alterations in patients with advanced NSCLC according to age and sex. Efforts that are focused on identifying targetable alterations in NSCLC will likely help personalize treatment and improve outcomes. Methods: We performed a retrospective review of de-identified data from the Guardant Health database from March 2018 through October 2020. We reviewed 34,237 profiles from patients with NSCLC who underwent molecular profiling using the plasma-based circulating-tumor DNA (ctDNA) Next-Generation Sequencing (NGS) assay Guardant360. Single nucleotide variants (SNV), fusions, indels and copy number variations (CNV) of up to 83 genes were analyzed. We assessed for genomic differences among patients with advanced NSCLC by both sex and age (≥70 and < 70). We conducted two-tailed tests of equality of proportions comparing males to females and ≥70 to < 70. Results: Of the 34,237 profiles reviewed, somatic alterations were seen in 81.7% (n = 27,972) of the patients. The median age was 70 (range 16-102) and 55% were female. Our study demonstrated that the most common genomic alterations in both age groups and genders were TP53, EGFR, KRAS, ATM, and MET. Patients ≥70 were more likely to have ATM (21% versus 14%, p < 0.0001) and MET (12% versus 10 %, p < 0.0001) mutations than those < 70. Patients < 70 were more likely to have EGFR (30% versus 27%, p < 0.0001), STK11 (14% versus 11%, p = 0.0056), and KRAS (26% versus 24%, p < 0.0001) alterations. EGFR was seen more frequently in females (33% versus 26%, p < 0.0001). ATM (11% versus 6%, p < 0.0001) and MET (8% versus 5%, p = 0.0050) were seen more frequently in males. Conclusions: Significant differences in the distribution of targetable genomic alterations were identified among different age groups and genders in patients with advanced NSCLC. These findings highlight the importance of taking personalized approaches to diagnostic testing and treatment of advanced NSCLC.

4130 Background: Circulating cell-free tumor DNA (ctDNA) genomic profiling is an emerging tool for pancreatic cancer. The impact of detected genomic alterations in tumor response to systemic treatments and outcomes is under investigation. Methods: Patients with advanced pancreatic cancer and ctDNA collected at time of initial diagnosis were retrospectively evaluated. Results of ctDNA analysis were correlated with patients’ demographics, systemic treatment response, progression-free survival (PFS) and overall survival (OS). Results: A total of 104 patients were included in the analysis. The mean age was 70.5 years (SD: 8.3), 50% were male, 37% with locally advanced disease and 63% with metastatic disease. Somatic alterations were detected in 84.6 % of the patients, no genetic alterations were detected in 15.4%, and were associated more with locally advanced pancreatic cancer as opposed to metastatic, p = 0.025. 60.6 % of the cohort had ≥ 2 genomic alterations detected. 28% were treated with FOLFIRINOX and 63% with gemcitabine plus nab-paclitaxel as first-line systemic treatment. Patients with any detectable genomic alterations when compared to patients with no detectable variant had worse median PFS (6.2 versus 15.3 months, p = 0.005) and patients with ≥ 2 detectable genomic alterations had worse median PFS (5.6 versus 11.0 months, p < 0.001) and worse median OS (11.5 versus 24.2 months, p = 0.001). KRAS was detected in 62.5% of the patients and was associated with PD to systemic treatments (80.4% vs 19.6%, p = 0.006), worse median PFS (5.8 versus 13.0 months, p < 0.001) and worse median OS (11.5 versus 26.3 months, p = 0.002). TP53 was detected in 60% of patients and was associated with worse median PFS (5.9 versus 10.9 months, p = 0.02) and worse median OS (13.5 versus 24.2 months, p = 0.001). CCND2 was detected in 14% of the patients and was associated with worse median PFS (3.6 versus 8.2 months, p = 0.004). Conclusions: Our study showed that initial detection of ctDNA may identify different genomic alterations that help predict disease outcomes, confirmation of these findings in larger studies are warranted.

3040 Background: Pembrolizumab was recently FDA approved across solid tumors for TMB scores ≥ 10mut/Mb as assessed by next-generation sequencing (NGS) of tissue (tTMB). A prior study of advanced cancer patients treated with immunotherapy found that higher somatic TMB, as defined by the 80th percentile in each histology, was associated with better overall survival. Previously, bTMB assessed by ctDNA from patients with newly diagnosed advanced NSCLC at a score of 16 mut/MB correlated with a tTMB score of 10 mut/MB. TMB levels vary by cancer type, line of treatment, and therapy received; the distribution of bTMB scores across solid tumor types has not been well characterized. Here we report the distribution of bTMB scores in patients with advanced malignancies. Methods: We queried 5,610 samples from patients with different cancer types undergoing clinical cell-free DNA testing (Guardant360; Redwood City, CA) and assessed bTMB scores from October 2020 - January 2021. bTMB score was derived via a previously described computational algorithm examining the total number of synonymous and non-synonymous SNVs and indels across a 1.0MB genomic footprint. We assessed the success rate of bTMB evaluation, overlap with microsatellite instability (MSI) status, and defined the distribution of bTMB levels across indications in this dataset. Results: bTMB score was successfully assessed in 4,275/5,610 (76.3%) samples (Table). The majority of samples (58%) were tested at disease progression as compared to initial diagnosis (42%). The median turnaround time from sample receipt to clinical reporting was 11 days and decreased to 9 days over the course of the study. For the majority of cancer types the 80th percentile TMB was ≥ 16 mut/MB tissue equivalency. Conclusions: Our analysis demonstrates the feasibility of measuring bTMB using a commercially available liquid biopsy assay. bTMB scores trended higher than tTMB previously reported in these cancer types, reflecting the ability of ctDNA to better capture tumor heterogeneity. cfDNA may allow for exploration of bTMB evolution throughout treatment. TMB should be interpreted in the context of disease, treatment, and method; these data establish a pan-cancer benchmark for bTMB which will serve as a resource for further studies.[Table: see text]