Circulating tumor DNA (ctDNA) analysis has emerged as a pivotal minimally invasive tool for early detection, monitoring, and treatment stratification in cancer patients. However, the accuracy and reliability of ctDNA assays are profoundly influenced by preanalytical variables. This review discusses the impact of biological features (circadian rhythm, age, and sex), lifestyle factors (diet, smoking, and physical activity), as well as technical aspects such as hemolysis, leukocyte lysis, and delayed plasma separation on ctDNA integrity and concentration. Fluctuations in ctDNA levels driven by these factors highlight the need for clear guidelines regarding precollection timing, dietary restrictions, and sample processing. Furthermore, the adoption of harmonized protocols is essential to reduce variability and improve reproducibility across clinical and research settings.

Abstract INTRODUCTION: Liver disease occurs on a continuum from steatosis to fibrosis, cirrhosis and ultimately hepatocellular carcinoma (HCC), with a 30% lifetime risk of HCC among those with cirrhosis (LCr). If identified early, steatosis and fibrosis are potentially reversible, and in LCr, surveillance can reduce cancer morbidity and mortality. Despite these benefits, conventional LCr detection modalities are invasive or have limited performance. We previously demonstrated that cost-effective liquid biopsies of genome-wide cell-free DNA (cfDNA) fragmentomes enable early detection of HCC. Here, we use these technologies to detect liver steatosis, fibrosis, and cirrhosis towards improved pre-cancer intervention and HCC surveillance. METHODS: We performed low-coverage, whole genome sequencing of plasma cfDNA from separate Discovery (n=423) and Validation (n=221) cohorts including individuals with no known liver disease (n=397), chronic liver disease and early fibrosis (n=91) including viral hepatitis and metabolic associated steatotic liver disease, or advanced fibrosis/cirrhosis (n=156). We computed genome-wide fragment length, coverage, and repeat element features (DELFI and ARTEMIS), cross-validated a machine learning classifier for fibrosis and LCr detection in the Discovery Cohort and evaluated the locked model in the Validation Cohort. We then performed whole methylome sequencing (n=28) and cell-type deconvolution to reveal mechanisms of change to cfDNA fragmentomes in LCr. RESULTS: Individuals with early liver disease/fibrosis and advanced fibrosis/cirrhosis were detected with high performance (AUC=0.90, 95% CI=0.86-0.95 and AUC=0.95, 95% CI=0.93-0.98, respectively) in the Discovery Cohort. At an 80% specificity locked cutpoint, Validation Cohort sensitivity was 70.8% (90% CI=52.3%-87.5%) for early liver disease/fibrosis and 90.1% (90% CI=84.4%-94.4%) for advanced fibrosis/cirrhosis. The model displayed low cross-reactivity for other fibrotic origin conditions including benign lung nodules or chronic pancreatitis (median scores 0.087 and 0.068 respectively vs. 0.55 for LCr, p<0.0002). The approach outperformed the existing fibrosis index FIB-4, detecting 5.07x (95% CI=3.03-17.35) and 1.2x (95% CI=1.18-1.32) more cases of early liver disease/fibrosis and advanced fibrosis/cirrhosis in simulations. cfDNA methylome deconvolution revealed increased contributions of liver endothelium (p=0.00016) and blood monocytes (p=5.2x10-5) and decreased contribution of hepatocytes (p=0.00035) with shorter fragment lengths in LCr. CONCLUSIONS: A cfDNA fragmentome biomarker enabled early detection of liver disease including LCr and reflected both liver-derived and immune-cell related changes. These analyses may enable accessible early detection of pre-cancer conditions with potential to improve liver disease management and early detection of HCC. Citation Format: Akshaya Vijaya Annapragada, Zachariah Foda, Hope Orjuela, Carter Norton, Shashi Koul, Noushin Niknafs, Sarah Short, Keerti Boyapati, Adrianna Bartolomucci, Dimitrios Mathios, Michael Noe, Chris Cherry, Jacob Carey, Alessandro Leal, Bryan Chesnick, Nic Dracopoli, Jamie Medina, Nicholas Vulpescu, Daniel Bruhm, Sarah Bacus, Vilmos Adleff, Amy Kim, Steve Baylin, Greg Kirk, Andrei Sorop, Razvan Iacob, Speranta Iacob, Liana Gheorghe, Simona Dima, Katherine McGlynn, Manuel Ramirez-Zea, Claus Feltoft, Julia Johansen, John Groopman, Jillian Phallen, Rob Scharpf, Victor Velculescu. Non-invasive early detection of cancer-predisposing liver diseases using genome-wide cfDNA fragmentomes [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 4074.

Cross-validated test; DELFI score distribution; Annual transitions.

Whole genome cfDNA analyses. Clinical and demographic characteristics of individuals analyzed.

Accessible liquid biopsies, including analyses of genome-wide cell-free DNA (cfDNA) fragmentation, are emerging for early detection of cancer but remain largely unexplored in other diseases. Here, we used whole-genome sequencing to examine cfDNA fragmentomes in 1576 individuals, including those with liver disease or with other morbidities such as vascular, autoimmune, and neurodegenerative conditions. As a prototype for disease-specific cfDNA fragmentomic biomarkers, we developed a machine learning classifier that detected early liver disease, advanced fibrosis, and cirrhosis with high sensitivity in separate discovery (n = 423) and validation cohorts (n = 221) and had limited cross-reactivity for other diseases. Genome-wide fragmentome and methylome analyses revealed liver-derived and immune-mediated changes in cfDNA in the circulation of individuals affected with liver disease. Fragmentomic changes were also observed across a range of other human morbidities and reflected disease-specific changes in the circulation. A machine learning model using cfDNA fragmentomes predicted overall survival in separate morbidity discovery (n = 571) and validation cohorts (n = 231). These analyses demonstrate the connection between cfDNA fragmentomes and an individual's physiologic state and provide previously unrecognized possibilities for cfDNA liquid biopsies across human disease.

Abstract Introduction: Lung cancer is the leading cause of cancer-related mortality worldwide. Accurate histological subtyping to differentiate between lung adenocarcinoma (LUAD), lung squamous cell carcinoma (LUSC), and small cell lung cancer (SCLC) is critical for guiding optimal therapeutic strategies. However, up to 20% of patients lack sufficient tissue for conventional histopathological classification. Liquid biopsies using cell-free DNA (cfDNA) fragmentomics offer a promising non-invasive alternative for cancer characterization when tissue is not available. Methods: We examined 761 patients with newly diagnosed, treatment-naive lung cancer of all stages, including lung adenocarcinoma (n=468), squamous cell carcinoma (n=156), small cell carcinoma (n=42), large cell carcinoma (n=15) and other subtypes (n=80) from the prospective Lung Cancer Early Molecular Assessment trial (LEMA, NCT02894853). Low-coverage whole genome sequencing of cfDNA plasma samples was performed to derive genome-wide fragmentation features. Circulating tumor DNA (ctDNA) burden was estimated from fragmentation using the DELFI-TF method. We developed a machine learning classifier trained exclusively on the tissue-based copy number signatures from the Clinical Lung Cancer Genome Project (CLCGP) and applied it to patient cfDNA samples to predict lung cancer subtypes. Results: This tissue-trained subtyping algorithm was evaluated on all available plasma samples, achieving an AUC of 0.99 (95% CI = 0.98-1.00) for distinguishing NSCLC from SCLC and an AUC of 0.91 (95% CI=0.87-0.95) for differentiating LUAD from LUSC. The model correctly classified 88% of SCLC, 80% of LUAD and 87% of LUSC cases where the tumor fraction was ≥0.3% (n=276). Among a subset of 361 NSCLC patients, integration of five blood protein biomarkers resulted in a multimodal model that differentiated LUAD from LUSC across all tumor fractions with high performance (AUC=0.85, 95% CI=0.80-0.90), an improvement over cfDNA (p<0.01; AUC=0.78, 95% CI=0.74-0.82) or protein-only classifiers (p<0.001; AUC=0.70, 95% CI=0.62-0.78). Conclusions: These findings establish cfDNA fragmentation and protein biomarkers as a viable non-invasive approach for lung cancer subtyping when tissue is unavailable, with potential to expedite subtype-specific treatment selection and improve clinical outcomes Citation Format: Stephen Cristiano, Paul van der Leest, Jamie Medina, Zachary Skidmore, Milou M. Schuurbiers, Garrett Graham, Alessandro Leal, Bryan Chesnick, Kim Monkhors, Nicholas C. Dracopoli, Robert Scharpf, Peter B. Bach, Daan van den Broek, Amoolya Singh, Victor E. Velculescu, Sian Jones, Michel M. van den Heuvel, Lorenzo Rinaldi. Lung cancer subtyping using cell-free DNA fragmentomes and protein biomarkers [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 1135.

TPS2673 Background: The main challenge for developing CAR T therapies for solid tumors is the lack of targets that distinguish tumor from normal cells, resulting in on-target, off-tumor toxicity. Tmod logic-gated CAR T therapy addresses this challenge by incorporating 2 CARs on the same T cell: an activator targeting a marker on both tumor and normal cells, and a blocker targeting HLA-A*02 that inhibits CAR T activity against normal cells while allowing activation against tumor cells (with HLA-A*02 LOH), improving tumor selectivity and decreasing toxicity. Early safety results from 3 ongoing phase 1/2 clinical trials of logic-gated, Tmod CAR T therapy (EVEREST-1, EVEREST-2, and DENALI-1) have demonstrated manageable safety and tolerability in patients with advanced solid tumors (Grierson et al, SITC , 2024; Ward et al, SITC 2025; Specht et al, SABCS , 2025). Early efficacy results include the first ever reported complete response in a patient with non-small cell lung cancer following treatment with a CAR T-cell therapy (A2B694). A2B543 is an autologous Tmod CAR T therapy that contains the same Tmod construct as A2B694 with an added membrane-tethered IL-12 (memIL12) booster. Specifically, A2B543 is comprised of autologous Tmod cells transduced with 2 lentiviral vectors: one expressing both the HLA-A*02-targeted blocker and the mesothelin-targeted CAR activator; and a second expressing the memIL12 booster. Interleukin 12 (IL-12) is a potent, pro-inflammatory cytokine that plays a crucial role in inducing antitumor immune responses; however, systemic IL-12 can be prohibitively toxic (Jia et al, Front Immunol , 2022). In A2B543, expression of the memIL12 cassette is under the control of an NFAT promoter and is induced during antigen engagement or T cell activation. This inducible memIL12 is designed to reduce the toxicity associated with systemic IL-12 while enhancing the long-term potency and persistence of Tmod (Zhang et al, J Immunother Cancer , 2025). Methods: EVEREST-2 (NCT06051695) is a phase 1/2, open-label, nonrandomized study evaluating the safety and efficacy of A2B543 in adults with recurrent/metastatic mesothelin-expressing cancers with tumor-associated HLA-A*02 LOH, including mesothelioma, colorectal, non-small cell lung, pancreatic, or ovarian cancer. Patients are enrolled through BASECAMP-1 (NCT04981119), a master prescreening study that identifies patients with HLA LOH via next-generation sequencing and cryopreserves leukapheresis product. Upon progression, A2B543 is manufactured and then administered after lymphodepletion. The phase 1 primary objective is to evaluate the safety and tolerability of A2B543 and identify a recommended phase 2 dose (RP2D). The phase 2 primary objective is to assess overall response rate. Clinical trial information: NCT06051695 .

The optimal treatment sequence in non-small cell lung cancer harboring class I BRAF mutations and atypical BRAF variants remains unclear. To better characterize therapeutic strategy, we retrospectively evaluated a multi-institutional cohort of BRAF-mutant NSCLC patients (n = 97) and an independent clinico-genomic database (n = 342), performed structural modeling, and conducted chemical screens of BRAF-mutant cell lines. Patients with class I BRAF mutation treated with BRAF–MEK inhibitors at any line of therapy had significantly greater median overall survival compared to those who did not receive BRAF–MEK inhibitors (40 vs 10 months, Log-rank p = 0.043). There, however, was no significant survival difference between patients treated with immune checkpoint inhibitors versus those not treated. Tumors with class II or III BRAF variants were significantly more likely to harbor concurrent MAPK pathway alterations relative to class I (Chi-Square p < 10−4). Cell line studies identified genetic dependency on BRAF in class II cell lines without sensitivity to BRAF inhibitors, and dependency on EGFR in class III cell lines.

BACKGROUND:Treatment decisions in patients with unresectable colorectal liver metastases (CRLM) are largely guided by radiological response to induction systemic therapy. However, radiological assessment alone provides an imprecise estimate of underlying tumour biology or treatment response. Circulating tumour DNA (ctDNA) is an emerging biomarker that can support clinical decision-making. This study evaluated the independent prognostic value of radiological tumour burden and DELFI-TF, a tumour tissue- and mutation-independent cell-free DNA (cfDNA) fragmentome-based ctDNA assay. METHODS:We analysed 202 plasma samples and CT scans collected at baseline and following induction systemic therapy from 101 patients with unresectable, liver-limited CRC enrolled in the phase-III CAIRO5 trial (NCT02162563), treated with FOLFOX/FOLFIRI plus bevacizumab. Total tumour volume (TTV) was centrally quantified via semi-automated segmentation of liver metastases. ctDNA was measured using the DELFI-TF score. Associations with overall survival (OS) and early recurrence were evaluated using multivariable Cox regression models. FINDINGS:At baseline, TTV (median = 139 mL, IQR = 23-497 mL) strongly correlated with DELFI-TF (median = 0.29, IQR = 0.13-0.41; Spearman's ρ = 0.70). DELFI-TF showed a more pronounced reduction than TTV on-treatment (-97.6% vs -49.9%). Baseline levels and on-treatment changes of DELFI-TF (P = 0.001; P = 0.012) and TTV (P = 0.002; P = 0.002) were independently associated with OS in the multivariable model; their combination improved prognostic performance (Uno's C-statistic 0.78 vs 0.73; P = 0.036). Baseline (P = 0.016) and on-treatment DELFI-TF (P = 0.001) also predicted early recurrence after local therapy. INTERPRETATION:Following further validation, integrating cfDNA fragmentome-based testing with radiological tumour volume may provide complementary and clinically meaningful insights for prognostication and treatment response in patients with unresectable CRLM. This exploratory study supports a multimodal biomarker approach to guide personalised treatment strategies. FUNDING:German Research Foundation (DFG, 513004649), Heidelberg Medical Faculty, Dutch Cancer Society/KWF Kankerbestrijding (10438), PPP Allowance via Health ∼ Holland (LSHM22027), Dr. Miriam and Sheldon G. Adelson Medical Research Foundation, Stand Up To Cancer (SU2C)in-Time Lung Cancer Interception Dream Team Grant, SU2C-Dutch Cancer Society International Translational Cancer Research Dream Team Grant (SU2C-AACR-DT1415), Gray Foundation, Commonwealth Foundation, Cole Foundation, Delfi Diagnostics (research grant), US National Institutes of Health (CA121113, CA233259, CA271896).

8074 Background: Liquid biopsies provide an opportunity for non-invasive lung cancer detection and tumor subtyping when tumor tissue is not available. Here we evaluate a blood-based liquid biopsy approach and its relationship to clinical and tumor subtype characteristics of lung cancer cases using a cohort of 578 individuals from a prospective clinical trial (LEMA, NCT02894853). Methods: Pre-treatment plasma samples were processed using the DELFI assay, a cell-free DNA (cfDNA) approach using a genome-wide fragmentomics based machine learning classifier. Clinical data, including overall cancer stage (I=164, II=59, III=133, IV=184), tumor stage, histologic subtypes, lymph node invasion, comorbidities, medications, smoking history, treatment type, and overall-survival (OS) data were collected for all patients. Tissue molecular profiling was performed to identify actionable alterations in driver oncogenes (ALK, BRAF, EGFR, ERBB, KRAS, ROS1, RET, MET) and cancer-specific protein levels (CEA, CA153, CA125, CYFRA, HE4) were measured in the plasma collected from 445 cancer cases. Results: DELFI scores were significantly higher with increasing tumor stage. T2 cases had a 1.3-fold increase in mean scores compared to T1 (p<0.001, Wilcoxon rank-sum), while T4 cases had a 16.2-fold increase (p<0.0001). A similar trend was observed with node staging, with N2 cases having an 11.3-fold higher mean scores compared to N0 (p<0.0001, Wilcoxon rank-sum), while N3 stage cases had a 27-fold increase (p<0.0001). Lung adenocarcinoma (ADC) displayed lower DELFI scores compared to squamous cell carcinomas (SCC) (p<0.01, Wilcoxon rank-sum), while small-cell lung cancer cases had the highest scores among all subtypes (p<0.0001, Wilcoxon rank-sum). cfDNA fragmentome changes in patients with ADC and SCC reflected chromosomal alterations observed in TCGA cohorts (ADC n=518; SCC n=501). The combination of DELFI cfDNA fragmentome characteristics with plasma protein measurements were used to train and cross-validate a classifier that could differentiate ADC from SCC (AUC for stage I=0.71, II=0.85, III=0.85, IV=0.82). Patients with low DELFI scores (below the median) had longer overall-survival (OS) compared to patients with high DELFI score (low DELFI score=18.51 months; high DELFI score=6.58 months; p<0.01, log-rank). DELFI scores were unaffected by underlying patient comorbidities, tumor-specific mutations, or medication status. Conclusions: Overall, this study revealed that DELFI scores are related to tumor burden, predict survival outcomes, and that cfDNA fragmentome analyses can be used to identify lung cancer subtypes. These results suggest future opportunities for subtype-specific treatments in lung cancer based on non-invasive plasma-only analyses. Clinical trial information: NCT02894853 .

The Supplementary Figures file consists of Supplementary Figures 1 to 15 belonging to the manuscript: "Metastatic colorectal cancer treatment response evaluation by ultra-deep sequencing of cell-free DNA and matched white blood cells"

Liquid biopsy approaches provide an opportunity for both cancer detection and clinical assessment in a non-invasive manner. Here we apply a blood-based lung cancer screening test to 578 individuals diagnosed with lung cancer through a real-world diagnostic prospective trial (LEMA, NCT02894853). Pre-treatment plasma samples were processed through the DELFI assay, a validated cell-free DNA (cfDNA) test based on a locked machine learning analysis of genome-wide fragmentation patterns. Clinical data, including tumor stage (I=165, II=60, III=147, IV=206), histologic subtypes, lymph node invasion, comorbidities, medications, smoking history, treatment type and overall-survival data were collected from all patients. Tissue molecular profiling was performed to identify actionable alterations in driver oncogenes. Genome-wide cell-free DNA fragmentome analyses revealed that LEMA patients had similar chromosomal gains (1q, 3q, 5p, 8q) and losses (3p, 4q, 5q, 10q, 13q) to lung cancer patients collected in TCGA (LUAD and LUSC) and L101 trial (NCT04825834), suggesting the landscape of cfDNA alterations are consistent with other lung cancer studies in Europe and the United States. DELFI fragmentation scores were significantly different amongst both tumor stage (p<0.0001, kruskal-wallis) and lymph node invasion (p<0.0001, kruskal-wallis), exhibiting progressively higher scores with increasing disease burden. Lung adenocarcinoma displayed lower DELFI scores compared to squamous cell carcinomas, while small-cell lung cancer cases presented the highest scores among all lung cancer subtypes. Stage III patients not eligible for surgery with curative intent (n=96) had significantly higher DELFI scores compared to surgery-eligible stage III patients (p=0.02, Wilcox rank-sum).Localized cancer cases (I-III) experiencing cancer relapse after complete surgical resection also showed significantly higher DELFI scores compared to patients who did not experience relapse (p<0.01, log-rank). Metastatic cancer patients, treated with chemotherapy, immunotherapy or targeted therapy had a median overall survival of 8.7 months. For all these treatment types, patients with low DELFI score at baseline (below the median) had longer overall-survival (OS) compared to patients with high DELFI score (above the median p<0.01, log-rank). Neither systemic comorbidities nor medications were found to significantly alter DELFI scores amongst groups (q>0.05, Wilcox rank-sum) and cancer cases driven by oncogenic alterations showed equivalent DELFI scores to cases without any driver mutation detected by tissue profiling. Overall, this study reveals that DELFI fragmentomes can be used as prognostic biomarkers, similarly in lung cancer patients from US and European populations. DELFI scores increase with stage and tumor burden, predict survival, and are not altered by mutational status, comorbidities, or medications. Milou Schuurbiers, Zachary L. Skidmore, Paul van der Leest, Jamie E. Medina, Garrett Graham, Tony Wu, Jacob Carey, Alessandro Leal, Bryan Chesnick, Kim Monkhorst, Nicholas Dracopoli, Robert B. Scharpf, Victor E. Velculescu, Daan van den Broek, Michel van den Heuvel, Lorenzo Rinaldi. Analysis of lung cancer clinical characteristics using cell-free DNA fragmentomes [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 1961.

1112 Background: Pembrolizumab (pembro) with chemotherapy has shown survival benefit in PD-L1+ (CPS10+) mTNBC, but many responses are not durable, and patients with PD-L1 negative/low tumors do not benefit from the combination. Data from GeparNuevo and TONIC suggest that induction therapy can remodel the tumor immune environment and improve responses. We have conducted a trial with two run-in cohorts and mandatory serial tissue and blood collections in 50 mTNBC patients, comparing pembro vs. nab-paclitaxel (nab-P). Methods: Single-arm, single institution phase II study (NCT02752685) to evaluate safety and clinical activity of nab-P+pembro in PD-L1 unselected mTNBC, 0-2 prior lines of chemotherapy allowed. Patients (n = 50) were enrolled sequentially into two cohorts: chemotherapy run-in (cTNBC, nab-P before nab-P+pembro) and immunotherapy run-in (iTNBC, pembro before nab-P+pembro). Serial tumor biopsies assessed by IHC (Dako 22C3), quantitative multiplex immunofluorescence (qMIF), and gene expression (NanoString). Overall response rates assessed using irRECIST. Tumoral T- and myeloid-cell phenotypes, peripheral lymphocyte-to-neutrophil ratio (LNR), and monocyte-to-lymphocyte ratio (MLR) were correlated with overall response rate (ORR) and survival outcomes. Results: 50 patients enrolled and completed treatment, for 80% of patients: treatment was 1L for metastatic disease. Median follow-up is 19.9 months, clinical results for cTNBC and iTNBC cohorts shown in table. Across both cohorts higher LNR was associated with improved OS ( R = 0.37, p = 0.0075), conversely, higher MLR was associated with poorer OS ( R = -0.46, p = 0.00087). Tumor immune cell subpopulations showed no significant differences between iTNBC and cTNBC at baseline. Analyses of on-treatment samples will be presented at the meeting. PD-L1 expression, while not different at baseline, remained unchanged in cTNBC but increased significantly in iTNBC ( p < 0.02), possibly reflecting pembrolizumab-driven immune modulation. With the caveat of comparing sequential cohorts, the iTNBC cohort showed a trend for higher ORR (47% vs. 23%, p = 0.08) and longer median PFS (8.4 vs. 5.5 months, HR = 0.68, 95%CI: 0.37-1.24), with significantly longer OS (25.8 vs. 18 months, HR = 0.50, 95%CI: 0.26-0.98, p = 0.043) compared to cTNBC. Conclusions: Timing of pembro administration may influence PD-L1 expression and clinical outcomes in mTNBC. We show that the immunotherapy run-in strategy converts more PD-L1-negative/low into PD-L1-positive tumors, possibly rendering more patients eligible for chemoimmunotherapy and improving outcomes. Clinical trial information: NCT02752685 . cTNBC (n=30) iTNBC (n=20) PD-L1 CPS >/=10 5/23 (22%) 2/16 (12%) PD-L1 CPS conversion (CPS<10 to CPS>/=10) 2/14 (14%) 4/13 (31%) Confirmed ORR (CR+PR) 7/30 (23%) 9/19 (47%) mPFS (months) 5.5 8.4 mOS (months) 18.0 25.8

5053 Background: Androgen receptor (AR) antagonists such as enzalutamide (enza) are standard therapy for mCRPC. AR alterations such as amplification (amp) and ligand binding domain (LBD) point mutations (PMs) can cause resistance to hormonal therapies such as enza. The most prevalent AR PMs are T878A and L702H. Preclinical evidence suggests that AR T878A retains sensitivity to enza compared to AR L702H which confers resistance by increasing activation by glucocorticoids. Clinical data evaluating the efficacy of enza in patients with these AR PMs is limited. Here, we analyzed real world data from patients with AR T878A compared to AR L702H, AR amp, or no detected AR mutations ( AR no alt), as identified by circulating tumor DNA (ctDNA), to assess enza efficacy in patients with AR T878A versus AR L702H. Methods: We used GuardantINFORM, a real-world database that combines genomic data from deidentified patients tested via ctDNA with clinical data taken from commercial-payer health claims. Adult mCRPC patients treated with enza who had baseline ctDNA testing and at least 2 claims post ctDNA testing were included. Patients with ≥2 AR PMs were excluded. Matched cohorts were used to assess real-world overall survival (rwOS), time to treatment discontinuation (rwTTD), and time to next treatment (rwTTNT). Propensity score matching was conducted using age and NCI comorbidity index, race, ethnicity, testing location, and enzalutamide line-of-therapy, and was evaluated using Wilcoxon tests. Results: 1,316 mCRPC patients met inclusion criteria. 59 had AR T878A, 56 had AR L702H, 231 had AR amp, and 970 had AR no alt. T878A was compared to L702H, amp, and no alt. Patients with T878A demonstrated significantly improved rwTTD (median 8.0 vs 3.5 mo, P =0.001) and rwTTNT (median 15.8 vs 4.3 mo, P =0.003), but not significantly different rwOS (19.1 vs 13.6 mo, P =0.066) relative to L702H. Patients with T878A demonstrated significantly improved rwTTD (7.7 vs 4.8 mo, P =0.022), but not statistically improved rwTTNT (10.4 vs 6.4 mo, P =0.059) or rwOS (20.2 vs 14.9 mo, P =0.14) relative to AR amp. Patients with T878A demonstrated significantly shorter rwOS (median 19.2 vs 43.6 mo, P =0.03), but no difference in rwTTD (7.7 vs 7.8 mo, P =0.88) or rwTTNT (10.4 vs 14.4 mo, P =0.423) relative to AR no alt. Conclusions: To our knowledge, this is the largest study assessing outcomes of mCRPC patients with AR PMs subsequently treated with enza. Using real-world evidence, we show that AR T878A patients have longer time on therapy with enza relative to patients with AR L702H. rwOS following enza was numerically longer for T878A versus L702H. These findings suggests that AR T878A is relatively more sensitive to enza compared to other resistance mutations. Future work in larger prospective cohorts comparing hormonal treatments in AR -altered patients will help confirm the clinical significance of different AR alterations.

3608 Background: Effective monitoring for relapse is a critical component of post-surgical care in colorectal cancer (CRC), particularly for patients who remain at risk of recurrence despite curative-intent treatment. Circulating tumor DNA (ctDNA) is a powerful biomarker for detecting minimal residual disease (MRD) and predicting relapse with high sensitivity and specificity. However, its predictive value varies over time, with negative results closer to surgery being less reliable than results obtained later. Here we introduce a novel time-weighted approach to ctDNA monitoring using a tumor-informed assay (Signatera), assigning greater predictive power to negative results collected further from surgery. Methods: A Bayesian logistic regression model using 1,246 Signatera serial measurements was developed to predict recurrence risk across 167 patients with early-stage colon cancer, with time-weighted ctDNA dynamics as the primary predictive feature. Negative ctDNA values were assigned greater predictive power based on their temporal distance from surgery. Secondary covariates included clinical stage and adjuvant treatment status. Time-weighted ctDNA was calculated as the product of the ctDNA level at each timepoint and an inverse time factor (1/( t +1)), where t represents the weeks since surgery. The weighted values were aggregated for each patient to compute cumulative and average time-weighted ctDNA levels, which served as inputs to the model. Survival analysis was performed to evaluate recurrence-free survival (RFS), stratified by MRD status. Results: Tumor-informed ctDNA levels were measured longitudinally, with a median of 7 timepoints per patient (range, 2-16) collected over a median follow-up of 2.5 years. Stage distribution was 50% stage III (n = 83), 44% stage II (n = 74), and 6% stage I (n = 10). Mismatch repair deficiency was observed in 22 patients (13.2%). Adjuvant chemotherapy was administered in 102 patients (61.1%), and 16 patients (9.6%) experienced recurrence. Survival analysis revealed a significantly worse recurrence-free survival for MRD-positive patients compared to MRD-negative patients (HR = 4.2, 95%CI:2.8–6.4, Log-rank p < 0.0001). The Bayesian logistic regression model demonstrated robust predictive performance, with a posterior probability of recurrence < 5% for patients with three consecutive negative ctDNA results obtained > 6 months after surgery. Conversely, the model assigned a > 90% probability of recurrence for patients with persistent ctDNA positivity beyond the initial 3-month post-operative window. Conclusions: Time-weighted ctDNA dynamics demonstrated promising predictive capability for CRC recurrence. Our findings suggest that incorporating the temporal context of ctDNA measurements and leveraging the increasing reliability of negative results over time could refine risk stratification and improve personalized care strategies for CRC patients.

Supplementary Data 1 shows an overview of ctDNA dynamics, treatments, and radiological response measurements per patient.

Determining response to therapy for patients with pancreatic cancer can be challenging. We evaluated methods for assessing therapeutic response using cell-free DNA (cfDNA) in plasma from patients with metastatic pancreatic cancer in the CheckPAC trial (NCT02866383). Patients were evaluated before and after initiation of therapy using tumor-informed plasma whole-genome sequencing (WGMAF) and tumor-independent genome-wide cfDNA fragmentation profiles and repeat landscapes (ARTEMIS-DELFI). Using WGMAF, molecular responders had a median overall survival (OS) of 319 days compared to 126 days for nonresponders [hazard ratio (HR) = 0.29, 95% confidence interval (CI) = 0.11–0.79, P = 0.011]. For ARTEMIS-DELFI, patients with low scores after therapy initiation had longer median OS than patients with high scores (233 versus 172 days, HR = 0.12, 95% CI = 0.046–0.31, P < 0.0001). We validated ARTEMIS-DELFI in patients with pancreatic cancer in the PACTO trial (NCT02767557). These analyses suggest that noninvasive mutation and fragmentation-based cfDNA approaches can identify therapeutic response of individuals with pancreatic cancer.

AbstractPurpose: Although immunotherapy is the mainstay of therapy for advanced non–small cell lung cancer (NSCLC), robust biomarkers of clinical response are lacking. The heterogeneity of clinical responses together with the limited value of radiographic response assessments to timely and accurately predict therapeutic effect—especially in the setting of stable disease—calls for the development of molecularly informed real-time minimally invasive approaches. In addition to capturing tumor regression, liquid biopsies may be informative in capturing immune-related adverse events (irAE). Experimental Design: We investigated longitudinal changes in circulating tumor DNA (ctDNA) in patients with metastatic NSCLC who received immunotherapy-based regimens. Using ctDNA targeted error-correction sequencing together with matched sequencing of white blood cells and tumor tissue, we tracked serial changes in cell-free tumor load (cfTL) and determined molecular response. Peripheral T-cell repertoire dynamics were serially assessed and evaluated together with plasma protein expression profiles. Results: Molecular response, defined as complete clearance of cfTL, was significantly associated with progression-free (log-rank P = 0.0003) and overall survival (log-rank P = 0.01) and was particularly informative in capturing differential survival outcomes among patients with radiographically stable disease. For patients who developed irAEs, on-treatment peripheral blood T-cell repertoire reshaping, assessed by significant T-cell receptor (TCR) clonotypic expansions and regressions, was identified on average 5 months prior to clinical diagnosis of an irAE. Conclusions: Molecular responses assist with the interpretation of heterogeneous clinical responses, especially for patients with stable disease. Our complementary assessment of the peripheral tumor and immune compartments provides an approach for monitoring of clinical benefits and irAEs during immunotherapy.