21 Background: Monitoring molecular residual disease (MRD) status showed prognostic value for recurrence prediction in colorectal cancer (CRC) patients. However, conventional tissue-informed MRD approaches lack the feasibility where tumor biopsy is not available or rapid turnaround time is necessitate for clinical decision making. We previously developed CanCatch Surf, a highly sensitive tissue-agnostic, methylation-based circulating tumor DNA (ctDNA) assay. Here, we report its clinical application in detecting early relapse for patients with surgical resectable CRC. Methods: MUSETALK-CRC01 is a prospective, multicenter study designed to monitor ctDNA status with CanCatch Surf for patients with clinical stage I–III CRC undergoing complete surgical resection. 88 patients were enrolled and blood samples were collected after surgery for longitudinal ctDNA analysis until recurrence. Computed tomography (CT) imaging was performed every 6 months after surgery. Cell-free DNA from blood samples were extracted and subjected to CanCatch Surf testing. A machine learning-based algorithm was trained to classify each blood samples as MRD-positive or MRD-negative as described previously. Sensitivity and specificity were calculated with reference to the clinical recurrence outcomes for each patient. Additionally, the accuracy of ctDNA quantification estimated by CanCatch Surf was validated and compared with a tumor-informed, mutation-based approach. Results: In the MUSETALK-CRC01 cohort (60% male, mean age 56.53±13.22 years), 50 (57%) patients had stage II CRC, and 24 (27%) had stage III CRC with a median follow-up of 16 months. During the period of surveillance, 19 (21%) patients experienced radiologic recurrence. Serial ctDNA analysis during surveillance of the 88 patients with longitudinal collected blood samples identified relapse with 75% (95%CI: 51-91%) sensitivity and 97% (95%CI: 90-99%) specificity. ctDNA positivity during longitudinal surveillance was associated with a reduced recurrence-free survival compared with ctDNA negativity (HR, 22.72; 95% CI, 8.04-64.24; P < 0.005). For 14 relapsed patients with positive ctDNA results, MRD predicted molecular relapse earlier than radiologically confirmed recurrence with a median lead time of 4.7 months (IQR, 6.5 months; Wilcoxon signed rank test P<0.005). ctDNA estimated by tumor methylated fraction (TMeF) derived from CanCatch Surf demonstrated a strong correlation (Pearson correlation coefficient: >0.95) with results predicted by cTAF, a tumor-informed ctDNA quantification metrics. Conclusions: In this study, we demonstrated the high accuracy of tissue-agnostic ctDNA assay. This approach offers a clinical and practical alternative for molecular residual disease detection, especially when tumor tissue is unavailable or insufficient for genomic profiling.

8031 Background: Curative-intent surgery for early-stage non-small-cell lung cancer (NSCLC) is followed by relapse in approximately 20–30% of patients. We previously reported interim results of the prospective MUSETALK-Lung01 study (multiomics sequencing technique application kick-start), showing that pre-operative, ctDNA positivity predicted increased relapse risk. Here we present the final analysis of the full cohort and incorporate pathological risk factors to conduct a more comprehensive prognostic evaluation for early-stage NSCLC. Methods: The MUSETALK-Lung01 study is a prospective, longitudinal, observational study designed to evaluate the clinical utility of a tumor-naïve ctDNA assay in patients with early-stage NSCLC. Cell-free DNA (cfDNA) was extracted, and subjected to CanCatch Surf test as described previously. The assay algorithm reports both the ctDNA detection status (positive/negative) and the estimated ctDNA levels. Longitudinal data, including vital status, treatment, relapse, and survival status, were collected over 5 years. The study was approved by the institutional review board or ethics committee at each site, with all participants providing written informed consent. Results: The complete cohort of MUSETALK-Lung01 consisted of 455 early-stage NSCLC patients, among whom the median follow-up time was 63 months, with a 2-year recurrence rate of 7.0% and a presurgical ctDNA positivity rate of 6.4%. To identify clinical and biological factors associated with survival, both univariate and multivariate analyses were performed. Presurgical ctDNA positivity was strongly associated with inferior recurrence outcomes (χ 2 p < 1 x 10 -9 ), indicating that tumor burden can be prognostic beyond clinical stage and pathological conditions. Specifically, among clinical stage I lung adenocarcinoma (LUAD) patients, ctDNA detection demonstrated robust prognostic stratification for recurrence risk (2-year RFS: 60% [95% CI: 40%–91%] vs. 96% [95% CI: 94%–98%]; log-rank p < 1 x 10 -7 ). Similar prognostic value of ctDNA was observed in stage II-IIIA LUAD patients, albeit with reduced discriminative power (log-rank p < 0.01). After incorporating pathological information, the performance of prognostic stratification was further improved. Conclusions: By evaluating ctDNA abundance, we provide highly sensitive and specific prognostic assessments and risk stratification for early-stage NSCLC patients. The non-invasive, nature of this test enables precision peri-operative management and selection of patients who may benefit from innovative treatments, with the potential to improve survival.

Abstract Background: Liquid biopsy is rapidly emerging as a cornerstone of precision oncology, offering sensitive and accurate profiling of tumor-derived information in real-time. Although droplet digital PCR (ddPCR) delivers single-molecule sensitivity, its low-plex design has limited it to interrogating one or a few variants at a time. Here we introduce COMET plus, a 25-plex ddPCR assay that simultaneously detects both genetic and epigenetic changes in a single amplification reaction, enabling highly sensitive detection of circulating tumor DNA (ctDNA) in colorectal cancer (CRC). Methods: Following analysis of over 500 tumor tissue and more than 1,000 cfDNA specimens, we selected 18 methylation variants that simultaneously display (i) robust tumor-specific hyper-methylation, (ii) negligible signal in healthy donor plasma, and (iii) high recurrence across patients. These loci were co-amplified with somatic mutations and microsatellite instability loci (MSI) markers frequently altered in CRC. The assay runs on a 7- or 6-color ddPCR system (D3200 platform, Pilotgene; QX600TM, Bio-Rad). A classifier was trained and subsequently locked to make sample-level calls of ctDNA positivity or negativity using optimized thresholds for each variant. Limit of detection (LoD) for each analyte was determined by Probit analysis of serially diluted contrived cfDNA and cell-line DNA. Limit of blank (LoB) was established from 10 independent replicates of two healthy donor cfDNA pools. Results: In 30 ng of contrived cfDNA or cell-line DNA we detected actionable KRAS/BRAF mutations at 0.08% VAF and MSI-H DNA at 0.1% in an MSS background. Individual methylation markers were called to 0.08-0.2%, while the aggregate 18-marker signature reached 0.01% (cell-line) or 0.02% (cfDNA); no false positives were observed across 20 blank replicates. Conclusions: We developed and validated a high-plex ddPCR assay that sensitively and simultaneously quantifies genetic and epigenetic alterations in colorectal cancer. The multiplexing strategy is compatible with most mainstream ddPCR instruments and readily extendable to other tumor types, offering a simple, streamlined workflow for cancer diagnosis, minimal residual disease (MRD) detection, and therapy selection. Citation Format: Ya Zhou, Yuwei Ni, Xingyu Yang, Qiancheng You, Jing Su, Yunpeng Zhang, Xiaoling Li, Xinyue Kang, Jiayue Xu, Bingsi Li. Ultrasensitive multiplex ddPCR for integrated genomic-epigenomic ctDNA detection in colorectal 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 2596.

Abstract Background: Epigenetic and genetic alterations synergistically drive cancer initiation and progression, yet one blood draw rarely yields enough cell-free DNA (cfDNA) to profile both modalities. Current co-detection strategies either demand high tumor burden or custom chemistry and cannot be retro-applied to existing datasets. We present MMcall, a computational tool that reconstructs the original four-base genome from conventional bisulfite-sequencing reads and simultaneously detects mutations and methylation variants without new benchwork, delivering fully integrated genomic/epigenomic signatures from one single library. Methods: MMcall reconstructs the original four-base genome from standard bisulfite-converted reads by jointly modeling complementary top- and bottom-strand base counts. Because the G nucleotide on the strand opposing a C is unaffected by bisulfite conversion, the strand-specificity principle is used to restore pre-conversion sequence. A machine-learning error-suppression module is then applied to suppress the high technical noise inherent to bisulfite treatment. The algorithm simultaneously outputs methylation-variant allele frequency (MVAF) and somatic-variant allele frequency (SVAF) from the same library, enabling epigenetic and genetic profiling without additional wet-lab steps. Results: Benchmarked against 0 -1% tumor-fraction serial dilutions of Seraseq® ctDNA Reference Material and an in-house standard (OverC Monitor panel; 1000X methylation depth, 20000X mutation depth), MMcall demonstrated near-perfect concordance with expected methylation levels (R2>0.99) and detected mutations down to 0.25%. At 0.5-1% VAF, SVAF measurements by MMcall closely matched those obtained by ultra-deep sequencing (HS-UMI, 35000X), yielding > 99.7% NPA (95% CI: 99.3-99.9%) and > 86.9 % PPA (95 % CI: 77.8-93.3%). No false-positive calls were observed across predefined hotspot loci in any negative control, confirming robust suppression of technical noise. Conclusion: MMcall jointly calls mutations and methylation variants at single-base resolution without additional bench steps. The method offers a cost-effective route to richer molecular information for early cancer detection and minimal residual disease monitoring. Citation Format: Xingyu Yang, Jing Su, Chen Yang, Xiaoling Li, Si Zhang, Xianrong Chen, Bingsi Li. Integrated, base-resolution profiling of genetic and epigenetic signatures in cell-free DNA [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 111.

Abstract Background: Rapid progress in molecular and computational technologies for circulating tumor DNA (ctDNA) analysis is transforming precision oncology, allowing tumor detection, relapse surveillance, and treatment selection without invasive tissue sampling. Among these approaches, methylation profiling of ctDNA offers a sensitive and quantitative measure of tumor burden. However, signal-to-noise ratios varies across cancer types and technologies, underscoring the need for tumor-specific assay optimization and rigorous validation. We previously developed a tissue-free, methylation-based assay and demonstrated its clinical validity in lung, colorectal, and liver cancers. Here we report expanded, larger scale validation across additional tumor types, including pancreatic cancer (PDAC) and bile tract cancer (BTC). Methods: The CanCatch® Surf classifier was trained and subsequently locked, as previously described. The total analytical performance testing dataset comprises >1,000 contrived and clinical samples (10-30 ng input). In this study, we constructed dilution series of PDAC- and BTC-derived cell-line and cfDNA mixtures spanning 0.001-0.5% tumor allele fraction (TAF). The in-silico titration dataset was generated by digitally blending sequencing reads from PDAC and BTC patients with those from cancer-free donors at defined proportions (0.001-0.5%). Probit regression identified the limit of detection (LoD) as the lowest TAF detected with ≥95% probability; the limit of blank (LoB) was the per-sample positivity rate in age-matched cancer-free donors. Results: The analytical sensitivity for PDAC and BTC reached 0.02% for in silico mixtures, cell-line dilutions and cfDNA titrations, with no false positives among 72 age-matched cancer-free donors (0 %; 95% CI 0-5.0%). Evaluation of 176 pre-treatment plasmas (57 PDAC, 50 BTC, 69 controls) demonstrated a sensitivity of 84.2% for PDAC (stage I 75.0 %, II 86.7%, III 83.3%, IV 100%) and 82.0% for BTC (stage I 71.4%, II 66.7%, III 93.3%, IV 92.3%), both at 98.6 % specificity; larger cohorts are in progress, and updated results will be presented as available. Conclusions: CanCatch® Surf’s non-invasive approach to ctDNA detection in solid tumors has demonstrated performance comparable to traditional tissue-based methods, thus holding promise for broader clinical use in monitoring disease recurrence and assessing treatment effectiveness. Citation Format: Zeliang Deng, Xiaoling Li, Xinyue Kang, Jiayue Xu, Jing Su, Xianrong Chen, Qiancheng You, Xingyu Yang, Bingsi Li. Validation of a sensitive, tissue-free blood test for biomarker discovery and tumor burden assessment [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 1137.

Multiomic circulating tumor DNA (ctDNA) assays offer several benefits over tumor-dependent ones, including being more rapid, less invasive, and cost-effective for monitoring tumor recurrence or therapy response through serial testing. Despite these advantages, identifying ultra-low frequency ctDNA alterations, whether genetic or epigenetic, from limited input material is highly challenging. Furthermore, complex assays are often more susceptible to experimental and computational variables and artifacts. This poses a pressing need for standardized samples to assess the analytical performance and enable unbiased comparisons between different technology platforms. We have developed contrived standards using cell line, tissue, plasma, and PBMC samples that were either commercially purchased or provided with informed consent. The 'ground truth' for tumor fraction (TF) was defined by the dilution ratios and validated through in silico reads-mixing simulations. Subsequently, estimates of tumor allele fraction (TAF) using different biomarkers were analyzed and compared. (i) Mutation-based TAF: Both tumor-agnostic (somatic variants allele frequency, SVAF) and tumor-informed (circulating tumor allele fraction, cTAF) ones were characterized by a validated 520-/168-gene panel (10, 000-35, 000X) or an individualized mutation tracking panel (100, 000X depth) following a WES scan on paired tumors. (ii) Methylation-based TAF: Both tumor-agnostic (tumor methylated fraction, TMeF) and tumor-informed (methylated tumor allele fraction, mTAF) ones were characterized by a validated targeted methylation assay. We have tested >400 contrived and clinical samples across different cancer types. Sample dilutions at pre-determined levels of TAF (0.001%-1%) are suitable for evaluating the limits of detection (LoDs) for most mainstream tumor-informed/-agnostic ctDNA assays. Samples with TAF near the LoD and background plasma were tested with replicates for precision and accuracy studies. Overall, reliable estimates of TAF were observed across different biomarkers, although the measurement of ultra-low fractions (i.e. TAF<0.01%) still remains a challenge. Notably, the number of unique patients/healthy donors is crucial for generating methylation-based standards, as demographic factors such as age, sex, and ethnicity have been shown to influence DNA methylation patterns. With clinical adoption underway, it is critical to understand and evaluate the variables that impact the analytical performance of complex multiomic ctDNA assays. Here, we report a systematic evaluation of the analytical features and clinical relevance of standardized samples, thereby providing a versatile tool for assay optimization, validation, and cross-platform comparison. Qiancheng You, Jiayue Xu, Jing Su, Xingyu Yang, Sa Zhang, Zeliang Deng, Xiaoling Li, Guangyou Li, Xinyue Kang, Si Zhang, Meifang Wu, Bingsi Li. Development and evaluation of standardized samples for multiomic ctDNA assays across multiple cancer types [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 4068.

The lack of quantitative methylation reference datasets (ground truth) and cross-laboratory reproducibility assessment hinders clinical translation of epigenome-wide sequencing technologies. Using certified Quartet DNA reference materials, here we generate 108 epigenome-sequencing datasets across three mainstream protocols (whole-genome bisulfite sequencing, enzymatic methyl-seq, and TET-assisted pyridine borane sequencing) with triplicates per sample across laboratories. We observe strand-specific methylation biases across all protocols and libraries. Cross-laboratory reproducibility analyses reveal high quantitative methylation levels agreement (mean Pearson correlation coefficient (PCC) = 0.96) but low detection concordance (mean Jaccard index = 0.36). Using consensus voting, we construct genome-wide quantitative methylation reference datasets serving as ground truth for proficiency testing. Key technical parameters-including mean CpG depth, coverage, and strand consistency-correlate strongly with reference-dependent quality metrics (recall, PCC, and RMSE). Collectively, these resources establish foundational standards for benchmarking emerging epigenomic technologies and analytical pipelines, enabling robust, standardized quality control in research and clinical applications.

Early detection of hepatocellular carcinoma (HCC) in patients with liver cirrhosis (LC) and/or hepatitis virus B/C infection (HVI) improves survival, highlighting the need for accurate, affordable diagnostic tools. Here, 11 methylated DNA markers (MDMs) are identified during marker discovery. In phase I, each selected MDM is validated in 175 plasma samples (HCC, n = 85; LC/HVI, n = 72) by the CO-methylation aMplification rEal-Time PCR (COMET) assay. Of these, 8 MDMs are qualified for phase II study, where a logistic regression model (COMET-LR) is trained and validated with 336 plasma samples (HCC, n = 211; LC/HVI, n = 113; training vs validation, 2:1). In the validation, the COMET-LR achieved 90.0% sensitivity at 97.4% specificity. Notably, sensitivity in patients with TNM stage I, diameter<3 cm, AFP-negative (<20 ng mL-1), PIVKA-II-negative (<40 mAU mL-1) is 82.4%, 77.8%, 88.6%, and 85.7%, respectively. The COMET-LR outperformed multiple protein markers (AFP, AFP-L3, and PIVKA-II) and published scores for HCC screening (GALAD, Doylestown, and ASAP), in terms of both sensitivity and specificity. The assay represents a significant advancement in addressing the unmet need for accurate, non-invasive, accessible, and cost-effective early detection tools for LC/HVI individuals. Further validation in a prospective cohort is warranted.

3045 Background: There is a growing need for risk evaluation and treatment monitoring in cancer care. However, current methods, mainly imaging, can be burdensome for patients over time and prone to variability among readers. Recent research has highlighted the potential of tumor-informed circulating tumor DNA (ctDNA) testing for identifying postoperative minimal residual disease (MRD) due to its high sensitivity by tracking individualized mutations. Nevertheless, its use in early-stage patients prior to surgery is constrained by limited tissue availability and extended turnaround times. MUSETALK-Lung01 (multiomics sequencing technique application kick-start) is a prospective, longitudinal, observational study designed to evaluate the clinical utility of a tumor-naïve ctDNA assay in patients with early-stage non-small cell lung cancer (NSCLC). Methods: Pretreatment plasma samples were prospectively collected from participants with stage I-IIIA NSCLC. Cell-free DNA was extracted and analyzed using a blood assay that interrogates both epigenetic and genetic information. The detection status and the estimated fraction of ctDNA were reported by a machine learning classifier and an independent statistical model, respectively. The calling threshold corresponding to a 99% clinical specificity was verified in a subgroup from the THUNDER study (NCT04820868). Longitudinal data, including vital status, cancer status, and treatment, were collected for up to 5 years. The study was approved by the institutional review board and all participants were required to provide informed consent. Results: A total of 289 participants from the MUSETALK-Lung01 study were analyzed. Of these, 49% (141/289) reached the 5-year follow-up, with a median follow-up duration of 59 months. To assess the prognostic value of preoperative ctDNA levels, relapse-free survival (RFS) and overall survival (OS) were evaluated across different stages and pathological subtypes separately. In stage I LUAD patients (N = 179), ctDNA-positive patients (N = 20) had significantly inferior RFS compared to ctDNA-negative patients (2-year RFS: 70% [95% CI: 46%–88%] vs. 94% [95% CI: 90%–97%]; log-rank p < 0.001). In contrast, no association was found between preoperative ctDNA detection and RFS in stage II-IIIA LUAD or non-LUAD NSCLC, irrespective of the clinical stage. Specifically, among the 179 stage I LUAD patients, 11 relapsed within 2 years, and 6 of these had positive ctDNA test results. This rate was significantly higher than in patients who relapsed between 2 and 5 years (1/12) or never relapsed (13/156; χ 2 test, p < 0.001). Conclusions: These findings indicate that presurgical ctDNA can serve as a prognostic indicator in early-stage NSCLC. Tumor-naive ctDNA testing may enhance the standard workflow by identifying high-risk patients who could benefit from innovative treatments. Clinical trial information: NCT04820868 .

Circulating tumor DNA (ctDNA) has emerged as a promising biomarker of minimal residual disease (MRD) detection and therapy response assessment. However, the low tumor burden in early-stage cancer and interference from non-tumor variants (e.g. clonal hematopoiesis) often necessitate tumor tissue to achieve the desired sensitivity and specificity. This requirement can limit the clinical application, particularly in situations where tissue biopsy is challenging or rapid turnaround times are critical. We previously validated an tissue-free assay (OverC) that combines a novel methylation sequencing method and machine learning for sensitive detection of low-frequency ctDNA in multiple cancer types. Here we describe the expansion of this assay for recurrence monitoring and treatment assessment. DNA extracted from 139 formalin-fixed paraffin-embedded tumor/normal tissues and 249 plasma samples subjected to whole-genome methylation sequencing (5-30X depth) for single-molecule pattern recognition. The targeted methylation panel of the refined version, known as OverC Monitor, covered 70, 546 CpGs (∼1.0 Mb), representing promoters, introns and other epigenetic regulatory elements. The classifier was developed with a larger THUNDER training set and independently validated with 506 pre-treatment plasma samples from patients with cancer and age-matched non-cancer controls. A positive call is made upon the identification of a cancer signal exceeding a locked threshold. In addition, a quantitative measure of tumor allele fraction (TAF) was developed and reported for each sample to assess its prognostic and monitoring potential. The assay performance was tested using plasma samples from 104 patients with lung cancer (LC), 48 colorectal cancer (CRC) and 74 liver cancer (HCC) and 280 age-comparable non-cancer donors. For LC, the sensitivity was 69.2% (Stage I: 30.3%, Stage II: 60.0%, Stage III: 86.7%, Stage IV: 91.7%) at a specificity of 98.6%. The sensitivity for CRC was 81.3% (Stage I: 41.7%, Stage II: 90.0%, Stage III: 95.0%, Stage IV: 100%) at a specificity of 98.9%. For HCC, the sensitivity reached 95.9% (Stage I: 89.5%, Stage II: 100.0%, Stage III: 95.0%, Stage IV: 100.0%) at a specificity of 98.9%. Further validation with larger cohorts and additional tumor types is underway, and updated data will be presented as it becomes available. By leveraging cancer-derived methylation signatures and machine learning, we have developed and validated a highly sensitive assay (OverC Monitor) for detecting trace amounts of ctDNA in blood. This innovative, tissue-free approach has demonstrated performance on par with cutting-edge tissue-dependent mutation tracking methods for certain cancer types, highlighting its potential for broader clinical use in recurrence monitoring and treatment efficacy assessment. Jiayue Xu, Xingyu Yang, Qiancheng You, Jing Su, Zeliang Deng, Xinyue Kang, Guangyou Li, Si Zhang, Bingsi Li. A methylation-based, tissue-independent blood test for recurrence monitoring and treatment assessment [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 4566.

Abstract Background: Early detection of primary liver cancer (PLC) in individuals with liver cirrhosis (LC) or chronic hepatitis virus infection (CHVI) improves survival. Timely, effective, and affordable tools with high sensitivity are urgently needed in clinical practice. Methods: Tissue and/or plasma samples from 159 healthy individuals and 89 PLC, LC, or CHVI patients were sequenced by a targeted methylation panel (~70,000 CpGs) to screen candidate methylated DNA markers (MDMs). In phase I, the performance of each selected MDM was validated in 175 plasma samples (PLC, n=101; LC/CHVI, n=74) by a CO-methylation aMplification rEal-Time PCR (COMET) assay. Logistic models were then trained and validated in phase II with 310 plasma samples (hepatocellular carcinoma [HCC], n=212; combined hepatocellular-cholangiocarcinoma [cHCC-CC], n=12; LC/CHVI, n=106; training vs. validation, 2:1). Results: The methylation levels of eleven selected MDMs with top performance showed a significant increase in PLC compared with LC/CHVI in both tissue and plasma samples (P<0.05). In phase I, eight of the above eleven MDMs with an area under the curve (AUC) over 0.80 to differentiate PLC and LC/CHVI were chosen for further investigation. In phase II, the MDM-based logistic model achieved sensitivity of 87.2% (95% confidence interval [CI], 80.8%-92.4%) and 88.0% (78.4%-94.4%), at respective specificity of 97.1% (90.1%-99.7%) and 100% (90.3%-100%) in the training and validation sets. In the validation set, sensitivity in patients with BCLC stage 0, diameter<3 cm, AFP-negative, and PIVKA-II-negative was 90.0% (55.5%-99.7%), 88.9% (65.3%-98.6%), 80.6% (64.0%-91.8%), and 81.3% (54.4%-96.0%), respectively. The classifier achieved similar sensitivity in patients with or without HBV/HCV infection (88.2% vs 90.0%). Additionally, our model detected 19 of 24 (79.3%, 57.8%-92.9%) intrahepatic cholangiocarcinoma. Combining AFP and PIVKA-II, the model achieved higher sensitivity of 93.3% (85.1%-97.8%) and specificity of 100.0% (90.3%-100%). Conclusion: In conclusion, the present study demonstrated the feasibility of using an easy-to-implement cfDNA methylation assay (COMET) to discriminate early-stage liver cancer from other liver diseases, with superior accuracy over AFP and PIVKA-II. Notably, the combination of the COMET, AFP, and PIVKA-II increased the overall sensitivity, indicating the potential benefit of integrating these approaches in clinical practices. Moving forward, further investigation is needed to validate these findings in larger prospective clinical studies. Keywords: primary liver cancer, early detection, liver cirrhosis, chronic hepatitis virus infection, methylated plasma DNA marker. Citation Format: Yang Tian, Yanan Wang, Mingxin Pan, Lei Zhang, Qiancheng You, Bingsi Li, Shangli Cai, Feng Shen, Guoyue Lv. An effective cfDNA methylation-based assay in discriminating primary liver cancer from cirrhosis or chronic hepatitis virus infection: marker discovery, phase I pilot, and phase II clinical validation [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 985.

Background The feasibility of DNA methylation-based assays in detecting minimal residual disease (MRD) and postoperative monitoring remains unestablished. We aim to investigate the dynamic characteristics of cancer-related methylation signals and the feasibility of methylation-based MRD detection in surgical lung cancer patients. Methods Matched tumor, tumor-adjacent tissues, and longitudinal blood samples from a cohort (MEDAL) were analyzed by ultra-deep targeted sequencing and bisulfite sequencing. A tumor-informed methylation-based MRD (timMRD) was employed to evaluate the methylation status of each blood sample. Survival analysis was performed in the MEDAL cohort ( n = 195) and validated in an independent cohort (DYNAMIC, n = 36). Results Tumor-informed methylation status enabled an accurate recurrence risk assessment better than the tumor-naïve methylation approach. Baseline timMRD-scores were positively correlated with tumor burden, invasiveness, and the existence and abundance of somatic mutations. Patients with higher timMRD-scores at postoperative time-points demonstrated significantly shorter disease-free survival in the MEDAL cohort (HR: 3.08, 95% CI: 1.48–6.42; P = 0.002) and the independent DYNAMIC cohort (HR: 2.80, 95% CI: 0.96–8.20; P = 0.041). Multivariable regression analysis identified postoperative timMRD-score as an independent prognostic factor for lung cancer. Compared to tumor-informed somatic mutation status, timMRD-scores yielded better performance in identifying the relapsed patients during postoperative follow-up, including subgroups with lower tumor burden like stage I, and was more accurate among relapsed patients with baseline ctDNA-negative status. Comparing to the average lead time of ctDNA mutation, timMRD-score yielded a negative predictive value of 97.2% at 120 days prior to relapse. Conclusions The dynamic methylation-based analysis of peripheral blood provides a promising strategy for postoperative cancer surveillance. Trial registration This study (MEDAL, ME thylation based D ynamic A nalysis for L ung cancer) was registered on ClinicalTrials.gov on 08/05/2018 (NCT03634826). https://clinicaltrials.gov/ct2/show/NCT03634826 .

Abstract Introduction: Recently, methylation sequencing has drawn significant attention as it has shown great promise for cancer detection and relapse surveillance. However, as with any low-input sequencing method, quality control is critical for implementation into the clinic, and it is difficult to obtain a sustainable source of patient samples for validation activities. To address these unmet needs, here we describe a design of fit-for-purpose cfDNA Methylation Reference Standards (MRS) to evaluate the performance of cfDNA-based methylation tests for oncology. Methods: We analyzed the genome-wide methylation landscape of 6 human cancer cell lines, 4 normal human lymphocyte cell lines, and 24 healthy donor cfDNA samples via whole genome bisulfite sequencing (WGBS), the gold standard of methylation analysis. Using cancer cell line genomic DNA cleaved by double-stranded DNA endonuclease (Zymo), we developed a range of cfDNA MRS spanning clinically relevant tumor fractions. DNA fragmentation were validated using the Labchip system (Perkin Elmer) and tumor fractions were independently verified by droplet digital PCR (Biorad). Results: The MRS series represent surrogates for lung, colorectal, liver, ovarian, pancreatic, and esophageal cancers. They display a size distribution comparable to cfDNA extracted from plasma, and represent a broad range of tumor fractions from 10% to 0.01%. In addition, the assembly accuracy was confirmed by digital droplet PCR (ddPCR) and ultra-deep mutation sequencing. Conclusions: These results demonstrated that the analytical feature and clinical relevance of MRS provides a versatile tool for assay optimization, validation, and cross-platform comparison. Citation Format: Bingsi Li, Jing Su, Jiayue Xu, Guangliang Zhang, Xiaoling Li, Ya Zhou, Hao Wang, Shuai Fang, Zhihong Zhang. Development of cfDNA reference standards for methylation-sequencing tests [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5109.

10544 Background: The development of ultra-sensitive genomic and epigenomic assays enables the early detection of multiple cancers in parallel. However, large-scale prospective clinical validation data are rare. Here, we report clinical validation data from the THUNDER (The Unintrusive Detection of EaRly-stage cancers, NCT04820868) study, which evaluates the performance of ELSA-seq among 6 cancer types in lung, colorectum, liver, esophagus, pancreas and ovary, which account for 50% of cancer morbidity and 62% of cancer mortality. Methods: This prospective case-control study consists of four stages: marker discovery, model training, validation, and independent validation. A customized panel covering 161,984 CpG sites was established using public data and in-house data. Cancer patients were pre-specified into the training and validation sets, and healthy controls (HC) were age-matched. Two multi-cancer detection blood test (MCDBT-1 and MCDBT-2) models with different cut-offs were established from the retrospectively collected training set and tested in the validation set. An independent validation set was enrolled prospectively, matched by age, and tested with the locked MCDBT-1/2 models. An interception model was then applied based on the model performance and China cancer incidence data to infer potential positive predictive value (PPV) and clinical utility in real-world practice. Results: In total, the training set consisted of 399 cases and 626 HC; the validation set consisted of 301 cases and 123 HC; and the independent validation set consisted of 505 cases and 505 HC. In the training set, the specificities were 98.9% (95% confidence interval [CI], 97.7%‒99.5%) and 99.7% (98.8%‒100.0%) for MCDBT-1/2 models, respectively. In the independent validation set ( n = 856, the rest to be sequenced and reported), MCDBT-1 and MCDBT-2 yielded sensitivity of 76.2% (72.0%‒80.0%) and 70.2% (65.8%‒74.4%) in 6 cancers with specificity of 96.3% (93.9%‒97.9%) and 99.3% (97.8%‒99.8%), respectively. Stage I‒III sensitivity was 68.5% (63.1%‒73.5%) and 60.8% (55.3%‒66.2%) for MCDBT-1/2 models, respectively. The prediction accuracy of top predicted origin was 79.1% (74.5%‒83.2%) and 83.0% (78.4%‒87.0%) for MCDBT-1/2, respectively. The interception model projected an estimated PPV of 3.1% and 12.4% for MCDBT-1/2 models, respectively. MCDBT-1 could reduce 5-year cancer mortality by 20.3%‒24.6% and reduce late-stage incidence by 51.7%‒61.7%, and MCDBT-2 could reduce 5-year cancer mortality by 15.9%‒19.9% and reduce late-stage incidence by 40.9%‒49.2%. Conclusions: cfDNA methylation-based MCDBT-1/2 models can effectively identify multiple cancers simultaneously at early stages with promising sensitivity, specificity, and accuracy of predicted origin. Their performances are to be further validated in a prospective interventional study (NCT05227534).

OBJECTIVES:Ovarian cancer is a fatal gynecological cancer due to the lack of effective screening strategies at early stage. This study explored the utility of DNA methylation profiling of blood samples for the detection of ovarian cancer.METHODS:Targeted bisulfite sequencing was performed on tissue (n = 152) and blood samples (n = 373) obtained from healthy women, women with benign ovarian tumors, or malignant epithelial ovarian tumors. Based on the tissue-derived differentially-methylated regions, a supervised machine learning algorithm was implemented and cross-validated using the blood-derived DNA methylation profiles of the training cohort (n = 178) to predict and classify each blood sample as malignant or non-malignant. The model was further evaluated using an independent test cohort (n = 184).RESULTS:Comparison of the DNA methylation profiles of normal/benign and malignant tumor samples identified 1272 differentially-methylated regions, with 49.4% hypermethylated regions and 50.6% hypomethylated regions. Five-fold cross-validation of the model using the training dataset yielded an area under the curve of 0.94. Using the test dataset, the model accurately predicted non-malignancy in 96.2% of healthy women (n = 53) and 93.5% of women with benign tumors (n = 46). For patients with malignant tumors, the model accurately predicted malignancy in 44.4% of stage I-II (n = 9), 86.4% of stage III (n = 59), 100.0% of stage IV tumors (n = 6), and 81.8% of tumors with unknown stage (n = 11). Overall, the model yielded a predictive accuracy of 89.5%.CONCLUSIONS:Our study demonstrates the potential clinical application of blood-based DNA methylation profiling for the detection of ovarian cancer.

Background Aberrant DNA methylation may offer opportunities in revolutionizing cancer screening and diagnosis. We sought to identify a non-invasive DNA methylation-based screening approach using cell-free DNA (cfDNA) for early detection of hepatocellular carcinoma (HCC). Methods Differentially, DNA methylation blocks were determined by comparing methylation profiles of biopsy-proven HCC, liver cirrhosis, and normal tissue samples with high throughput DNA bisulfite sequencing. A multi-layer HCC screening model was subsequently constructed based on tissue-derived differentially methylated blocks (DMBs). This model was tested in a cohort consisting of 120 HCC, 92 liver cirrhotic, and 290 healthy plasma samples including 65 hepatitis B surface antigen-seropositive (HBsAg+) samples, independently validated in a cohort consisting of 67 HCC, 111 liver cirrhotic, and 242 healthy plasma samples including 56 HBsAg+ samples. Results Based on methylation profiling of tissue samples, 2321 DMBs were identified, which were subsequently used to construct a cfDNA-based HCC screening model, achieved a sensitivity of 86% and specificity of 98% in the training cohort and a sensitivity of 84% and specificity of 96% in the independent validation cohort. This model obtained a sensitivity of 76% in 37 early-stage HCC (Barcelona clinical liver cancer [BCLC] stage 0-A) patients. The screening model can effectively discriminate HCC patients from non-HCC controls, including liver cirrhotic patients, asymptomatic HBsAg+ and healthy individuals, achieving an AUC of 0.957(95% CI 0.939–0.975), whereas serum α-fetoprotein (AFP) only achieved an AUC of 0.803 (95% CI 0.758–0.847). Besides detecting patients with early-stage HCC from non-HCC controls, this model showed high capacity for distinguishing early-stage HCC from a high risk population (AUC=0.934; 95% CI 0.905–0.963), also significantly outperforming AFP. Furthermore, our model also showed superior performance in distinguishing HCC with normal AFP (< 20ng ml −1 ) from high risk population (AUC=0.93; 95% CI 0.892–0.969). Conclusions We have developed a sensitive blood-based non-invasive HCC screening model which can effectively distinguish early-stage HCC patients from high risk population and demonstrated its performance through an independent validation cohort. Trial registration The study was approved by the ethic committee of The Second Xiangya Hospital of Central South University (KYLL2018072) and Chongqing University Cancer Hospital (2019167). The study is registered at ClinicalTrials.gov(# NCT04383353 ).

Introduction: Detection of cancer is its early stages can potentially improve therapeutic effectiveness. Previously, we demonstrated that ELSA-seq, a machine-learning-aided methylation profiling test, can detect and locate multiple cancers with high accuracy in plasma samples. Here, we evaluated the analytical performance of a refined test version of ELSA-seq, including analytical sensitivity, specificity, repeatability/reproducibility, and robustness. Methods: The classification algorithm and cut-offs for the ELSA-seq test were established in the THUNDER study (NCT04820868). Here, we describe the analytical performance using both plasma samples from the THUNDER study and DNA blends from cell lines. Analytical sensitivity was established by defining the limit of detection (LoD) using in-house cfDNA Methylation Reference Standards (MRS) with lowest DNA input for the assay. In brief, fragmented genomic control DNA (NA24385, Coriell Institute) was used as a diluent to contrive human cancer cell lines with defined mixing ratios (tumor fractions), which was further verified by digital droplet PCR (ddPCR). The LoD was determined by the lowest tumor fractions at which accurate DOC and TOO was reported in at least 95% of replicates. As variant allele frequency (VAF) is widely used as a surrogate measurement for tumour fractions, we also used ultra-deep mutation sequencing to attain VAF as an independent piece of evidence. Analytical specificity was assessed by the true negative rate in 120 plasma samples from age-matched healthy donors. A batch-to-batch repeatability/reproducibility study was carried out using 168 clinical samples processed across multiple reagent lots, instruments, and operators. Robustness was evaluated using common interfering substances that potentially could be present in plasma samples such as hemoglobin, bilirubin, triglycerides, and genomic DNA. Testing was performed using 24 cancer and 24 non-cancer samples, with or without interfering substances. Results: At 5ng input mass (approximating from 1ml plasma), the LOD95 was estimated down to 0.05% (tumor fractions) or 0.02% (VAF) among 6 cancer cell lines. 2 false positives were detected in 120 age-matched healthy donor samples, yielding a specificity of 98.3% (95%CI: 93.5-99.7%). All test results were concordant across multiple reagent lots, instruments, and operators (100% repeatability and reproducibility), and high Pearson correlation coefficients (>99%) were observed in the pair-wise comparison. In addition, none of the substances tested interfered with the assay. Conclusions: These results suggest that ELSA-seq is a highly sensitive test for detection of the trace amount of tumor-derived methylation signals in plasma samples. The performance is highly reproducible and robust, which is critical for clinical implementations. Citation Format: Bingsi Li, Jing Su, Guangliang Zhang, Jiayue Xu, Jianlong Peng, Ya Zhou, Fujun Qiu, Shuai Fang, Xiaofang Wen, Guoqiang Wang, Jing Zhao, Hao Wang, Shangli Cai, Zhihong Zhang. Analytical performance of ELSA-seq, a blood-based test for early detection of multiple cancers [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2022; 2022 Apr 8-13. Philadelphia (PA): AACR; Cancer Res 2022;82(12_Suppl):Abstract nr 5116.

e16153 Background: Biliary tract cancer (BTC) is an aggressive malignancy with poor prognosis and limited treatment options. The epigenome-associated multi-dimensional studies for BTC are limited. Here, we proposed a DNA methylation-based classification scheme and investigated the associations between methylation and multi-dimensional data including clinicopathological features, genetic aberrations, and prognosis. Methods: Multi-dimensional data concerning mutation, DNA methylation, and clinical data from 105 BTC patients who received surgical resection (gallbladder cancer [GBC], n=48, cholangiocarcinoma [CCA], n=57) were analyzed. Differentially methylated blocks (DMBs) in GBCs and CCAs were identified by comparing tumors and adjacent tissues. Methylation-based subtyping was performed via non-supervised consensus clustering. Results: The differentially methylated blocks (DMBs) in GBCs and CCAs are highly overlapping. Based on the common DMBs of GBCs and CCAs, the classifier yielded high sensitivity of 95.2% and specificity of 96.0%. Hypomethylated genes were enriched in pathways of transmemberane receptor and ion binding, while hypermethylation occurred in genes concerning DNA binding transcription activity. By non-supervised clustering of common DMBs, 6 clusters with different degrees of methylation change were identified. Higher methylation alteration (cluster risk-high) was associated with more copy number variation and shorter OS. By LASSO regression, a 12-gene prognostic model was trained and validated (Table). As for immune characteristics, the methylation of CD8A gene was lower in the cluster risk-high group, and this association was validated in the TCGA-cholangiocarcinoma cohort. In addition, lower methylation change was associated with higher BCR/TCR diversity, immune cell infiltration, and PD-L1 and CTLA4 mRNA expression. Conclusions: In BTC, methylation patterns may serve as a robust indicator of immune-related features and prognosis. Our integrative analysis highlights the association between genomic and epigenomic features, and provides insights into the molecular heterogeneity, and the potential therapeutic significance in biliary tract malignancies. Included methylation sites in the LASSO model.[Table: see text]

Background Plasma cell-free DNA (cfDNA) methylation has shown promising results in the early detection of multiple cancers recently. Here, we conducted a study to investigate the performance of cfDNA methylation in the early detection of esophageal cancer (ESCA). Methods Specific methylation markers for ESCA were identified and optimized based on esophageal tumor and paired adjacent tissues (n = 24). Age-matched participants with ESCA (n = 85), benign esophageal diseases (n = 10), and healthy controls (n = 125) were randomized into the training and test sets to develop a classifier to differentiate ESCA from healthy controls and benign esophageal disease. The classifier was further validated in an independent plasma cohort of ESCA patients (n = 83) and healthy controls (n = 98). Results In total, 921 differentially methylated regions (DMRs) between tumor and adjacent tissues were identified. The early detection classifier based on those DMRs was first developed and tested in plasma samples, discriminating ESCA patients from benign and healthy controls with a sensitivity of 76.2% (60.5-87.9%) and a specificity of 94.1% (85.7-98.4%) in the test set. The performance of the classifier was consistent irrespective of sex, age, and pathological diagnosis (P > 0.05). In the independent plasma validation cohort, similar performance was observed with a sensitivity of 74.7% (64.0-83.6%) and a specificity of 95.9% (89.9-98.9%). Sensitivity for stage 0-II was 58.8% (44.2-72.4%). Conclusion We demonstrated that the cfDNA methylation patterns could distinguish ESCAs from healthy individuals and benign esophageal diseases with promising sensitivity and specificity. Further prospective evaluation of the classifier in the early detection of ESCAs in high-risk individuals is warranted.

The low abundance of circulating tumour DNA (ctDNA) in plasma samples makes the analysis of ctDNA biomarkers for the detection or monitoring of early-stage cancers challenging. Here we show that deep methylation sequencing aided by a machine-learning classifier of methylation patterns enables the detection of tumour-derived signals at dilution factors as low as 1 in 10,000. For a total of 308 patients with surgery-resectable lung cancer and 261 age- and sex-matched non-cancer control individuals recruited from two hospitals, the assay detected 52–81% of the patients at disease stages IA to III with a specificity of 96% (95% confidence interval (CI) 93–98%). In a subgroup of 115 individuals, the assay identified, at 100% specificity (95% CI 91–100%), nearly twice as many patients with cancer as those identified by ultradeep mutation sequencing analysis. The low amounts of ctDNA permitted by machine-learning-aided deep methylation sequencing could provide advantages in cancer screening and the assessment of treatment efficacy.