Abstract The properties of cancer-associated genetic changes in cell-free DNA (cfDNA) are not fully understood. We performed whole-genome sequencing (WGS) of cfDNA as well as tumor tissue and white blood cells (WBCs) from 1,807 samples of 1,064 patients across eight common cancer types. Characterization of single base substitutions, small insertions and deletions, structural variants (SVs), and phased variants in single cfDNA molecules revealed unique properties of tumor-derived alterations as well as differences in error rates that spanned orders of magnitude. Given the low error rate associated with detection of tumor-specific rearrangement junctions in cfDNA, we hypothesized that these types of changes could enable detection of circulating tumor DNA (ctDNA) without prior knowledge of the alterations in the tumor tissue. As an example of this approach, we scanned each sequenced fragment genome-wide in cfDNA samples from the CheckPAC trial of patients with metastatic pancreatic cancer treated with radiation and immunotherapy to identify putative rearrangement junctions. We identified 22,010,911 such fragments but only 1,572 (0.007%) and 58,339 (0.27%) of these were present in the matched tumor or WBC samples, respectively, with the remaining identified only in cfDNA. We characterized each cfDNA fragment by the SV type, SV size, microhomology and insertion at the breakpoint junction, fragment size, and the location of the breakpoint with respect to the nearest fragment end, identifying differences depending on the origin of the SV. Machine learning analyses of SVs from cfDNA resulted in a high cross-validated performance for detection of tumor-specific SVs with an area under the curve (AUC) of 0.97 (95% CI: 0.97-0.98). After enriching for fragments most likely to be tumor-derived, we found that the number of cfDNA fragments containing SVs was highly correlated with the number obtained using a tumor-informed approach (Pearson correlation coefficient = 0.87, p<0.001), and could recapitulate longitudinal ctDNA levels and clinical outcomes using only low-coverage (∼4x) plasma WGS. The universal nature of tumor-associated sequence and structural alterations in cfDNA may be broadly useful for cancer detection. Citation Format: Daniel C. Bruhm, Carolyn Hruban, Adrianna L. Bartolomucci, Akshaya V. Annapragada, Sarah Short, Shashikant Koul, Kaui P. Lebarbenchon, Julia S. Johansen, Inna M. Chen, Andrei Sorop, Razvan Iacob, Speranta Iacob, Liana Gheorghe, Simona Dima, Katherine A. McGlynn, Manuel Ramírez-Zea, John Groopman, PLCRC-MEDOCC group, Remond J. Fijneman, Gerrit A. Meijer, Zachariah H. Foda, Jillian Phallen, Robert B. Scharpf, Victor E. Velculescu. Sequence and structural DNA alterations in the circulation of patients with cancer [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 2591.

Supplementary Table S1: Training cohort description with plasmaCHORD feature values. Supplementary Table S2: Description of the serial cohort. Supplementary Table S3: Validation cohort description with plasamCHORD feature values. Supplementary Table S4: Description of plasmaCHORD features. Supplementary Table S5: Correlation of cfDNA fragmentomic feature statistics between baseline and follow-up time points in the serial cohort. Supplementary Table S6: Possible CH-origin variants identified on retrospective analysis of 45 patients reviewed at the JH-MTB.

Abstract Hepatocellular carcinoma (HCC) is the third leading cause of cancer death globally and is one of the most rapidly increasing causes of cancer mortality in North America and Europe due to metabolic and emerging risk factors. Here, we show that cell-free DNA (cfDNA) fragmentation profiles reflected underlying disease biology and that a machine learning classifier detected HCC across diverse populations and was enhanced by incorporating patient clinical risk and protein concentrations. Characteristics of the cfDNA fragmentome, including genome-wide chromatin, repeat elements, methylome, and mutational profiles, were altered in patients with HCC, including those with metabolic risk factors and aflatoxin exposure. Novel whole-genome tissue-of-origin deconvolution analyses identified increased representation of cfDNA originating from hepatocytes (p=2.4x10-10) and CD8+ T cells (p=5.8x10-7) and decreased contribution of NK cells (p=3.0x10-6) in patients with cancer. Using a previously locked fragmentome classifier for liver cancer detection, we analyzed 377 individuals with and without HCC from two distinct geographic cohorts. We found that the cfDNA fragmentome classifier detected HCC across all stages and diverse etiologies with a sensitivity of 70% (95% CI=65%-75%) and specificity of 94% (95% CI=90%-97%), outperforming the commonly used alpha-fetoprotein (AFP) biomarker which had a sensitivity of 62% (95% CI=57%-67%) and specificity of 93% (95% CI=88%-96%). A combined approach using cfDNA fragmentomes, AFP, and clinical risk achieved a sensitivity of 74% (95% CI=66%-80%) at a specificity of 85% (95% CI=78%-89%) in transplant curable disease (Milan criteria) in both cohorts and was more sensitive than standard-of-care AFP and ultrasound performance for early-stage disease (reported 63% sensitivity at 84% specificity). This study provides insights into the origins of altered cfDNA and circulating proteins for populations at risk of liver cancer and supports the use of a genome-wide fragmentome approach for non-invasive detection of HCC. Citation Format: Hope Orjuela, Carter Norton, Shashikant Koul, Daniel C. Bruhm, Akshaya V. Annapragada, Sarah Short, Keerti Boyapati, Adrianna Bartolomucci, Vilmos Adleff, Nicholas A. Vulpescu, Kauí Lebarbenchon, Jacob Carey, Carter Portwood, Andrei Sorop, Razvan Iacob, Speranta Iacob, Liana Gheorghe, Simoni Dima, Katherine A. McGlynn, Manuel Ramirez-Zea, Jillian Phallen, Robert B. Scharpf, John Groopman, Victor E. Velculescu, Zachariah Foda. Characterizing the cfDNA fragmentome in patients with hepatocellular carcinoma [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 1125.

Performance of individual features in predicting variant origin in the training cohort. Receiver-operator curves and area-under the curve (AUC) values for individual model features as predictors for CH versus tumor-origin. The top performing individual features are gene heme fraction, length cluster mutant, and length max delta S. However, no individual feature performs as well as the plasmaCHORD model combining all features.

Description of the training cohort. (A) Distribution of cancer types, stage, and patient age in the training cohort. The mean age for patients in this cohort was 65 years old (range 19-87), the most common cancer type was colorectal, and most patients had early stage disease (22 stage I, 101 stage II, 48 stage III, 53 stage IV, and 1 unknown stage). (B) Distribution of WBC only versus WBC and tumor matched sequencing and determined reference variant origin. The proportion of variants with tumor-derived reference origin is slightly higher from samples that had both WBC and tumor sequencing versus WBC sequencing alone. (C) Distribution of variants included in training cohort by gene and reference variant origin. After DNMT3A, TP53 was the second-most commonly mutated gene (98 variants) of which 67 (68%) are tumor-derived and 31 are WBC-derived.

Abstract Introduction: Patients with stage II colon cancer (CC) not classified as high risk do not receive adjuvant chemotherapy (ACT) according to Dutch guidelines. However, 15-20% of patients with stage II CC experience disease recurrence, highlighting an unmet clinical need to identify patients who could benefit from ACT. Observational studies demonstrate that postoperative circulating tumor DNA (ctDNA) is indicative of minimal residual disease (MRD) and a strong prognostic biomarker for disease recurrence. Aim: The MEDOCC-CrEATE trial aims to assess 1) the proportion of stage II CC patients with detectable postoperative ctDNA accepting ACT, and 2) whether ctDNA-guided ACT reduces 2-year recurrence rate (RR). Methods: MEDOCC-CrEATE is an interventional trial following the ‘trial within cohorts’ design. Participants of the Prospective Dutch Colorectal Cancer cohort with stage II CC and no indication for ACT, and randomized to the intervention arm undergo postoperative tumor-informed MRD testing using PGDx elio® plasma resolve or Labcorp Plasma Detect®. Labcorp Plasma Detect is a highly sensitive ctDNA assay integrating whole genome sequencing (WGS) analyses of formalin-fixed paraffin-embedded tumor tissue, germline, and plasma cell-free DNA. Patients who test ctDNA-positive are offered 4 cycles of adjuvant capecitabine + oxaliplatin. ctDNA-negative patients in the intervention arm and patients in the control arm receive standard of care followup. For all patients, blood is collected every 6 months for 3 years to monitor disease recurrence. The primary endpoint is the proportion of patients with detectable postoperative ctDNA willing to receive ACT. Secondary endpoints include 2-year RR, disease-free and overall survival, quality of life and cost-effectiveness of ctDNA-guided ACT. The study will continue until 10 patients with detectable postoperative ctDNA are treated with ACT in the intervention arm. Results: Logistics for timely multicenter collection of tumor tissue and blood have been optimized across 29 Dutch hospitals. At present, 525 patients have been randomized. Of the 263 patients in the intervention arm, 212 provided consent for immediate postoperative ctDNA analysis (83%). Of the 197 currently available results, 16 (8,1%) had detectable ctDNA of which 11 started ACT. The median time from surgery to blood collection was 15 days (IQR 12-19), with a median turnaround time from surgery to ctDNA result of 46 days (IQR 40-53). Conclusion: Multicenter postoperative tumor informed ctDNA testing for MRD is operationally and technically feasible within the clinically relevant 8-12 week window to start ACT. The observed ctDNA detection rate is in accordance with expectations and the study design. Upon study finalization, the results will be used for health technology assessment to demonstrate the putative clinical utility of ctDNA-guided ACT in stage II CC. Citation Format: Ingrid A. Franken, Suzanna J. Schraa, Steven L. Ketelaars, Carmen Rubio-Alarcón, Teunise Bisschop-Snetselaar, Bregje Adriaans, Miranda van Dongen, Linda J. Bosch, Mirthe Lanfermeijer, Samuel Angiuoli, Amy E. Greer, Ellen Verner, Jennifer B. Jackson, Rebecca A. Previs, Veerle M. Coupe, Daan van den Broek, Gerrit A. Meijer, Jill Phallen, Victor E. Velculescu, Anna J. van Tetering-Houben, Jeanine M. Roodhart, Niels F. Kok, Frederieke H. van der Baan, Mark Sausen, Miriam Koopman, Geraldine R. Vink, Remond J. A. Fijneman, the PLCRC-MEDOCC group. MEDOCC-CrEATE trial update: Feasibility of measuring circulating tumor DNA post-surgery to guide adjuvant chemotherapy in stage II colon cancer patients [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 2 (Late-Breaking, Clinical Trial, and Invited Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(8_Suppl):Abstract nr CT157.

Abstract Introduction: Genomic profiling of circulating tumor DNA (ctDNA) through liquid biopsies has become an important diagnostic method in clinical oncology. However, detection of variants related to clonal hematopoiesis (CH) is a major confounder that impairs the clinical utility of liquid biopsies. Strategies that reduce biological noise from CH in plasma NGS include deep sequencing of matched WBC DNA and/or tumor tissue sequencing. While these methods effectively distinguish most CH variants, the need for extra biospecimens and sequencing raises costs and limits feasibility. Methods: Using a training cohort of 426 variants identified in ctDNA NGS from 225 patients with stage I-IV solid tumors, we developed plasmaCHORD, a machine learning model (MLM) that includes fragmentomic, variant, and patient-level features to distinguish between tumor- and CH-origin for mutations detected by fixed gene panel hybrid capture NGS. Model performance was assessed by comparison to the reference origin of each plasma variant determined from matched WBC and tumor NGS. Following locking the model parameters, we applied plasmaCHORD to an independent validation cohort of 1,412 plasma variants detected in 114 patients with metastatic cancers, as well as to cfDNA NGS from patients enrolled in a prospective liquid biopsy-informed clinical trial (NCT05585684). Results: PlasmaCHORD predicted tumor versus CH-origin in the training set with high accuracy (cross-validated AUC=0.94), outperforming individual features such as variant allele frequency and canonical CH genes. Model performance remained robust when restricted to mutant DNA fragments supported by 3-5 mutant reads (AUC = 0.84). plasmaCHORD was locked for evaluation using a score of 0.5 as cutoff for distinguishing tumor- versus CH-origin variants. In the independent validation cohort, the locked model maintained similar overall accuracy (AUC=0.9) with a sensitivity of 82%, specificity 80.3% and accuracy of 80.2%. Our approach was shown to be highly reliable in classifying variant origin in clinically actionable genes not canonically associated with CH, including AKT1, ATM, BRCA1, BRCA2, and EGFR, as well as adjudicating cellular origin for TP53 mutations that are encountered in both solid and hematologic malignancies. Performance was consistent across cancer types, sequencing platforms, mutation classes, and a wide range of allele fractions. When applied to clinically challenging cases in the context of a precision oncology clinical trial, plasmaCHORD precisely determined variant origin, preventing mismatches with genotype-targeted therapies. Conclusions: plasmaCHORD, a multi-feature machine-learning classifier, can significantly enhance the ability to identify bona fide tumor variants in routine plasma-only NGS, addressing a critical need in implementing liquid biopsy-guided therapy by minimizing misinterpretation caused by CH. Citation Format: Daniel J. Rabizadeh, Jenna VanLiere Canzoniero, Ilias Ziakas, Jaime Wehr, Archana Balan, Amna Jamali, Blair V. Landon, Susan Combs Scott, Gavin Pereira, Vincent K. Lam, Christine L. Hann, Christine M. Lovly, Jessica Tao, Patrick M. Forde, Joseph C. Murray, Mark Sausen, Gerrit A. Meijer, Geraldine Vink, Remond J. A. Fijneman, Victor E. Velculescu, Jillian Ayn Phallen, Robert Scharpf, Valsamo Anagnostou. PlasmaCHORD- A machine learning method for identifying clonal hematopoiesis variants in liquid biopsies [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 95.

Abstract Background: Timely assessment of response to immune checkpoint inhibition (ICI) is critical but often limited by the heterogeneity of radiographic responses. Mutation-based analyses of cell-free DNA (cfDNA) circumvent these challenges, but are, in turn, prone to artifacts arising from clonal hematopoiesis. Plasma cfDNA fragmentome analyses using low-pass whole genome sequencing (WGS) may enable a scalable approach to evaluate systemic tumor burden in a tumor- and mutation-naïve manner. Methods: cfDNA fragmentome analyses were performed following low-pass WGS of 244 plasma samples from 62 patients treated with pembrolizumab +/- radiotherapy (NCT02492568). A locked instance of the DELFI-TF, a random forest regression model based on genome-wide fragmentation patterns and aneuploidy, was applied to determine cfDNA fragmentome-based estimates of tumor fraction. The 95th percentile of tumor fraction in a non-cancer reference set established the limit of blank (LOB). Fragmentome-based landmark molecular response was defined as ctDNA below LOB at 6 weeks. Baseline tumor samples (n=24) patients were analyzed by RNA sequencing to characterize transcriptomic profiles stratified by cfDNA fragmentome profiles. Clinical outcomes were evaluated by RECIST 1.1 (at 6 and 12 weeks), progression-free survival (PFS), and overall survival (OS). Results: At baseline, DELFI-TF values were correlated with radiographic tumor burden (R=0.32, P=0.017). Notably, tumors from patients with high DELFI-TF showed an enrichment in gene sets related to cell cycle, DNA replication and repair (adjusted P<0.05), suggesting that fragmentome TF accurately captured cellular turnover. At 6 weeks, 73% (8 out of 11) of patients with radiographic response attained fragmentome molecular response, while the subset of patients with radiographically stable or progressive disease was more heterogeneous in their fragmentome molecular response (24 out of 50, 48%). Fragmentome molecular response was more concordant with best overall response (BOR) at 12 weeks (Fisher’s exact P=0.018). Analysis of baseline tumors from patients achieving fragmentome molecular response revealed an inflamed tumor microenvironment (adjusted P-value <0.001). Among patients with stable or progressive disease at the first radiographic evaluation, fragmentome response predicted longer PFS (logrank P=0.0096) and OS (logrank P=0.012). Similarly, fragmentome molecular response predicted PFS (logrank P=7.4e-5) and OS (logrank P=0.00028) across the entire cohort. Conclusions: Plasma cfDNA fragmentome-derived tumor fraction reflects cellular turnover and lung cancer biology within the context of immunotherapy, while also enabling reliable, cost-effective, and scalable molecular response evaluations. Citation Format: Noushin Niknafs, Lavanya Sivapalan, Bahar Alipanahi, Gavin Pereira, Amna Jamali, Jaime Wehr, Daniel Rabizadeh, Christopher Cherry, Bryan Chesnick, Nicholas C. Dracopoli, Jamie Medina, Stephen Cristiano, Willemijn S. Theelen, Robert Scharpf, Lorenzo Rinaldi, Victor E. Velculescu, Valsamo (Elsa) Anagnostou. Cell-free DNA fragmentomes capture response to immuno-radiotherapy in metastatic non-small cell lung cancer [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 1134.

Abstract Introduction: Patients with stage II colon cancer not classified as high risk (pT4 microsatellite stable) do not receive adjuvant chemotherapy (ACT) according to Dutch guidelines. However, 15-20% of patients with stage II colon cancer experience a disease recurrence, indicating that there is an unmet clinical need to identify patients who could benefit from adjuvant treatment. Observational studies demonstrate that postoperative circulating tumor DNA (ctDNA) is indicative of minimal residual disease (MRD) and a strong prognostic biomarker for disease recurrence. Aim: The MEDOCC-CrEATE trial aims to assess 1) the proportion of stage II colon cancer patients with detectable postoperative ctDNA accepting ACT and 2) whether ctDNA-guided ACT reduces 2-year recurrence rate (RR). Methods: MEDOCC-CrEATE is an interventional trial following the ‘trial within cohorts’ design. Participants of the Prospective Dutch Colorectal Cancer cohort with stage II colon cancer and no indication for ACT randomized to the intervention arm undergo postoperative tumor-informed ctDNA testing using next-generation sequencing of 33 genes through the PGDx elio™ liquid biopsy assay. If ctDNA is detected, they are offered 4 cycles of adjuvant capecitabine plus oxaliplatin. Patients in the intervention arm who test ctDNA negative and patients in the control arm receive standard of care follow-up. For all patients, blood is collected every 6 months for 3 years to monitor disease recurrence. The primary endpoint is the proportion of patients with detectable postoperative ctDNA willing to receive ACT. Secondary endpoints include 2-year RR, disease-free and overall survival, quality of life and cost-effectiveness of ctDNA-guided ACT. The study is powered for 2-year RR, randomizing 660 patients to each study arm in order to treat 30 patients in the intervention arm with detectable postoperative ctDNA, assuming 5% ctDNA detection and 10% noncompliance. Results: Logistics for timely multicenter collection of tumor tissue and blood have been optimized across 28 Dutch hospitals. At present, 240 patients have been randomized. Of the 119 patients in the intervention arm, 98 provided consent for immediate postoperative ctDNA analysis (82%). Of the 92 currently available results, 5 (5 %) had detectable ctDNA. The median time from surgery to blood collection was 15 days (IQR 11-19), with a median turnaround time from surgery to ctDNA result of 47 days (IQR 39-54). Conclusion: Multicenter postoperative tumor-informed ctDNA testing for MRD is operationally and technically feasible within the clinically relevant 8-12 week window to start ACT. The observed ctDNA detection rate is in accordance with expectations and the study design. Upon study finalization, the results will be used for health technology assessment to demonstrate the putative clinical utility of ctDNA-guided ACT in stage II colon cancer. Citation Format: Ingrid Franken, Suzanna Schraa, Karlijn van Rooijen, Dave van der Kruijssen, Carmen Rubio-Alarcón, Steven Ketelaars, Sietske van Nassau, Teunise Snetselaar, Bregje Adriaans, Jillian Phallen, Sam Angiuoli, Amy Greer, Ellen Verner, Veerle Coupé, Helena Verkooijen, Miranda van Dongen, Linda Bosch, Mirthe Lanfermeijer, Daan van den Broek, Gerrit Meijer, Victor Velculescu, Mark Sausen, Miriam Koopman, Remond Fijneman, Geraldine Vink, on behalf of the PLCRC-MEDOCC group. MEDOCC-CrEATE trial: Feasibility of measuring circulating tumor DNA after surgery to guide adjuvant chemotherapy in stage II colon cancer patients [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 5022.

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.

Circulating tumor DNA (ctDNA) analyses are informative as an early indicator of immunotherapy response in advanced non-small cell lung cancer (NSCLC); however, the clinical value of ctDNA molecular response requires further validation. As part of a prospective clinical protocol ( NCT05995821 ), we conducted targeted error-correction sequencing of ctDNA (n=328) and matched WBC DNA (n=109) from 109 patients with metastatic NSCLC who received anti-PD-(L)1 either as monotherapy or in combination. Following cellular origin resolution of 2,818 variants, landmark molecular response (mR) was defined as undetectable ctDNA within 3-9 weeks of treatment initiation. Pre-treatment ctDNA burden, but not blood tumor mutation burden, predicted survival. Implementing a tumor-naïve WBC DNA-informed approach increased the number of evaluable cases without compromising the overall accuracy of landmark ctDNA molecular responses. A direct comparison of single-timepoint on-therapy ctDNA assessment with ctDNA dynamics from baseline to the 3-9-week interval, along with an analysis of heterogeneity in molecular response within the 3-9-week window, showed that undetectable ctDNA at the landmark timepoint can effectively predict survival outcomes. A significant enrichment in landmark ctDNA mR was noted among patients with progression-free survival (PFS) ≥6 months with immunotherapy (p=2.5e-05) and chemo-immunotherapy (p=0.02). Patients in the landmark mR group had longer progression-free (p=1.6e-06) and overall survival (p=2.5e-05) than those with molecular progression. Landmark ctDNA molecular response provides a real-time, accurate approach for monitoring immunotherapy clinical outcomes. Although not currently validated for regulatory use, these findings demonstrate the potential utility of ctDNA as an early endpoint in clinical trials. Employing circulating tumor DNA (ctDNA) dynamics as an early indicator of immunotherapy response requires a roadmap for the next-generation sequencing approach, definition of molecular response and establishment of its clinical sensitivity. In this study, we introduce the concept of a landmark ctDNA molecular response, determined 3-9 weeks after initiation of immunotherapy, that maximizes the number of evaluable patients without sacrificing the specificity of the approach. Notably, when evaluating heterogeneity in ctDNA detection within the landmark 3-9-week window and assessing the impact of landmark interval dynamics on survival, we found that a single ctDNA assessment performed similarly to multiple ctDNA measurements within the landmark window (most notably, regardless of whether the timepoints were concordant or discordant). Our findings demonstrate that a single assessment of early on-therapy landmark ctDNA molecular response, can identify patients at risk of disease progression and enable future intervention and therapy optimization.

Accurate monitoring of treatment response in patients with metastatic colorectal cancer (mCRC) is essential for optimizing therapeutic strategies. Current response evaluation relies on imaging-based assessment of tumor size changes. However, this approach is limited by suboptimal sensitivity for detecting lymph node and peritoneal metastases, as well as inter-reader variability. Circulating tumor DNA (ctDNA) is indicative of the number of neoplastic cells and could have clinical value to CT imaging for assessment of treatment response. A mutation- and tumor-independent ctDNA assay was recently developed, demonstrating its potential for longitudinal assessment of treatment response: the DELFI-tumor fraction (DELFI-TF) score. The DOLPHIN study aims to investigate the added clinical value of the DELFI-TF score compared to CT imaging for treatment response monitoring in patients with mCRC. DOLPHIN is a prospective, observational, multi-center substudy of the Prospective Dutch ColoRectal Cancer cohort (PLCRC). Clinical data, images, and blood samples from 400 patients receiving systemic therapy in 11 hospitals in the Netherlands are being collected. Blood samples are drawn longitudinally in conjunction with routine CT imaging. The plasma cfDNA fragmentomes will be assessed using the DELFI-TF. DELFI-TF is a locked and validated machine learning model that quantifies tumor burden uses using cell-free DNA (cfDNA) fragmentomes data derived from low-coverage whole genome sequencing. Droplet digital PCR (ddPCR) ctDNA testing will be used as a reference for patients with confirmed RAS/BRAF mutations. The primary objective is to evaluate the association between ctDNA changes and clinical response. Secondary objectives include analyzing the association between ctDNA changes and RECIST criteria (I), correlation with serum carcinoembryonic antigen (CEA) levels at specified points during systemic therapy (II), lead time of ctDNA-testing versus CT imaging for detecting disease progression (III), the prognostic value of longitudinal ctDNA-testing (IV). Until November 2024, 385 patients have been enrolled, with the anticipated inclusion target of 400 patients expected to be reached by December 2024. A comprehensive overview of the DOLPHIN study population will be presented at the conference. Application of DELFI-TF and ddPCR ctDNA testing on the initial cohort of collected samples is scheduled to commence in early 2025. The DOLPHIN study will assess the clinical validity of the DELFI-TF in monitoring treatment response and investigate whether longitudinal ctDNA-testing can complement or partially replace imaging-based treatment response monitoring. These findings could pave the way for ctDNA-guided decision-making, supporting and enhancing accurate assessment of therapeutic effectiveness and therapeutic decision-making for patients with mCRC. Denise E. van Steijn, Jamie Medina, Lorenzo Rinaldi, Adria Closa, Lana Meiqari, Erica Peters, Alissa Konicki, Frederieke H. van der Baan, Mariska Bierkens, Haoyue Wang, Marjolein J. Greuter, Birgit I. Lissenberg-Witte, Veerle M. Coupé, Marie V. Coignet, Victor E. Velculescu, Daan van den Broek, Gerrit A. Meijer, Max J. Lahaye, Manon N.G. Braat, Jeanine M. Roodhart, Nicholas C. Dracopoli, Geraldine R. Vink, Niels F. Kok, Remond J. Fijneman. Cell-free DNA fragmentomes for treatment response monitoring in patients with metastatic colorectal cancer: the DOLPHIN study [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 3241.

Abstract Background: Immunotherapy-containing therapies have shown promise for patients with unresectable diffuse pleural mesothelioma (DPM), yet there are no reliable strategies to monitor therapy response. By expanding the compendium of cancer-associated alterations profiled, including genome-wide fragmentation patterns, cell-free DNA (cfDNA) whole genome sequencing (WGS) enables tracking of tumor dynamics. In tandem, while understudied, T cell clone dynamics may be informative in capturing early immunotherapy response. Methods: Using 314 tumor and peripheral blood biospecimes, we analyzed serial plasma cfDNA samples (n=135) from 55 patients with unresectable DPM, who received durvalumab with platinum-based chemotherapy (NCT02899195). Following cfDNA extraction and genomic library preparation, cfDNA at baseline (C1D1), Cycle 2 Day 1 (C2D1), and Cycle 5 Day 1 (C5D1) underwent low-coverage (1-2x) whole genome sequencing. Tumor- and mutation-naive estimates of tumor fraction were derived by applying the DELFI tumor score (DELFI-TS) model, which integrates genome-wide fragmentation patterns and chromosomal arm aneuploidy. In parallel, we performed TCR Vβ CDR3 next-generation sequencing on tumor (n=43) and peripheral serial blood (n=136) samples. The TCR repertoire was characterized using clonality, TCR clonotype/cluster dynamics, and the Morisita-Horn similarity index to assess repertoire similarity across samples. Clinical outcomes were assessed by radiographic response, progression-free (PFS), and overall survival (OS). Results: Patients with distant metastasis (M1) had higher DELFI-TS levels (p=0.038). At baseline, DELFI-TS levels were numerically higher for radiographic non-responders (SD/PD) vs responders (CR/PR). Using the 92th percentile of DELFI-TS in non-cancer cfDNA control samples to determine the limit of blank, patients with detectable baseline DELFI-TS (ctDNA+) had shorter PFS and OS (log-rank p<0.001). Patients that attained a radiographic response had more clonal peripheral TCR repertoires at both baseline and on-therapy timepoints compared to non-responders (p=0.037 at C1D1; p=0.007 at C2D1; p=0.042 at C5D1). Morisita-Horn similarity between C1D1 and on-therapy was lower in non-responders than responders (p=0.019 for C1D1 vs C2D1; p=0.036 for C1D1 vs C5D1). In contrast, a more diverse intra-tumoral TCR repertoire was noted for patients with an OS of 12 or more months (p=0.018). Conclusions: Our findings provide proof-of-concept that cfDNA fragmentomic analyses can quantify pre-treatment cfDNA tumor fraction that may capture clinical outcomes with chemo-immunotherapy response for patients with DPM. Longitudinal analyses of peripheral TCR repertoires can further differentiate responding from non-responding DPM, supporting the notion that joint analyses may more accurately capture immunotherapy response. Citation Format: Jinny Huang, Jennifer Li, James R. White, Shashikant Koul, Gavin Pereira, Nisha Rao, Jennie Yao, Julie R. Brahmer, Robert B. Scharpf, Rachel Karchin, Victor E. Velculescu, Zhouxin Sun, Suresh S. Ramalingam, Patrick M. Forde, Noushin Niknafs, Valsamo (Elsa) K. Anagnostou. Integrative analyses of the cfDNA fragmentome and TCR repertoires capture chemo-immunotherapy response in diffuse pleural mesothelioma [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 109.

PURPOSE:Targeted next-generation sequencing (NGS) of cell-free DNA (cfDNA) enables comprehensive molecular profiling and can guide the selection of genotype-targeted therapies. However, the detection of variants derived from clonal hematopoiesis (CH) is a significant confounder in liquid biopsies. EXPERIMENTAL DESIGN:Using a training cohort of 426 variants identified in cfDNA NGS from 225 patients with stage I to IV solid tumors, we developed plasma Clonal Hematopoiesis ORigin Detection (plasmaCHORD), a machine learning model that includes fragment-, variant-, and patient-level features to distinguish between tumor and CH origin for each variant detected by liquid biopsies. Model performance was assessed by comparison with the reference origin for each plasma variant determined from matched white blood cell and tumor NGS. Following the locking of the model parameters, we applied plasmaCHORD to an independent validation cohort of 1,418 plasma variants detected in 114 patients with metastatic cancers, as well as to cfDNA NGS from patients enrolled in a prospective clinical trial (NCT05585684). RESULTS:plasmaCHORD predicted tumor origin versus CH origin in the training set with high accuracy (AUC = 0.94). In the independent validation cohort, the locked model maintained similar overall accuracy (AUC = 0.9) and demonstrated significant improvement in accuracy for clinically significant genes. When applied to clinically challenging cases in the context of a precision oncology clinical trial, plasmaCHORD precisely determined variant origin, preventing mismatches with genotype-targeted therapies. CONCLUSIONS:plasmaCHORD, a multifeature machine learning model, can significantly enhance the ability to identify bona fide tumor variants in routine plasma-only NGS, addressing a critical need for implementing liquid biopsy-guided therapy by minimizing misinterpretation caused by CH.

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.