Capture Panel Research Articles

CAPTVRED: an automated pipeline for viral tracking and discovery from capture-based metagenomics samples

Abstract Summary Target Enrichment Sequencing or Capture-based metagenomics has emerged as an approach of interest for viral metagenomics in complex samples. However, these datasets are usually analyzed with standard downstream Bioinformatics analyses. CAPTVRED (Capture-based metagenomics Analysis Pipeline for tracking ViRal species from Environmental Datasets), has been designed to assess the virome present in complex samples, specially focused on those obtained by Target Enrichment Sequencing approach. This work aims to provide a user-friendly tool that complements this sequencing approach for the total or partial virome description, especially from environmental matrices. It includes a setup module which allows preparation and adjustment of the pipeline to any capture panel directed to a set of species of interest. The tool also aims to reduce time and computational cost, as well as to provide comprehensive, reproducible, and accessible results while being easy to costume, set up, and install. Availability and implementation Source code and test datasets are freely available at github repository: https://github.com/CompGenLabUB/CAPTVRED.git

Chronic Lymphocytic Leukemia IGHV Somatic Hypermutation Detection by Targeted Capture Next-Generation Sequencing.

Somatic hypermutation (SHM) status of the immunoglobulin heavy variable (IGHV) gene plays a crucial role in determining the prognosis and treatment of patients with chronic lymphocytic leukemia (CLL). A common approach for determining SHM status is multiplex polymerase chain reaction and Sanger sequencing of the immunoglobin heavy locus; however, this technique is low throughput, is vulnerable to failure, and does not allow multiplexing with other diagnostic assays. Here we designed and validated a DNA targeted capture approach to detect immunoglobulin heavy variable somatic hypermutation (IGHV SHM) status as a submodule of a larger next-generation sequencing (NGS) panel that also includes probes for ATM, BIRC3, CHD2, KLHL6, MYD88, NOTCH1, NOTCH2, POT1, SF3B1, TP53, and XPO1. The assay takes as input FASTQ files and outputs a report containing IGHV SHM status and V allele usage following European Research Initiative on CLL guidelines. We validated the approach on 35 CLL patient samples, 34 of which were characterized using Sanger sequencing. The NGS panel identified the IGHV SHM status of 34 of 35 CLL patients. We showed 100% sensitivity and specificity among the 33 CLL samples with both NGS and Sanger sequencing calls. Furthermore, we demonstrated that this panel can be combined with additional targeted capture panels to detect prognostically important CLL single nucleotide variants, insertions/deletions, and copy number variants (TP53 copy number loss). A targeted capture approach to IGHV SHM detection can be integrated into broader sequencing panels, allowing broad CLL prognostication in a single molecular assay.

Parallel DNA/RNA NGS Using an Identical Target Enrichment Panel in the Analysis of Hereditary Cancer Predisposition.

Germline DNA testing using the next-gene-ration sequencing (NGS) technology has become the analytical standard for the diagnostics of hereditary diseases, including cancer. Its increasing use places high demands on correct sample identification, independent confirmation of prioritized variants, and their functional and clinical interpretation. To streamline these processes, we introduced parallel DNA and RNA capture-based NGS using identical capture panel CZECANCA, which is routinely used for DNA analysis of hereditary cancer predisposition. Here, we present the analytical workflow for RNA sample processing and its analytical and diagnostic performance. Parallel DNA/RNA analysis allowed credible sample identification by calculating the kinship coefficient. The RNA capture-based approach enriched transcriptional targets for the majority of clinically relevant cancer predisposition genes to a degree that allowed analysis of the effect of identified DNA variants on mRNA processing. By comparing the panel and whole-exome RNA enrichment, we demonstrated that the tissue-specific gene expression pattern is independent of the capture panel. Moreover, technical replicates confirmed high reproducibility of the tested RNA analysis. We concluded that parallel DNA/RNA NGS using the identical gene panel is a robust and cost-effective diagnostic strategy. In our setting, it allows routine analysis of 48 DNA/RNA pairs using NextSeq 500/550 Mid Output Kit v2.5 (150 cycles) in a single run with sufficient coverage to analyse 226 cancer predisposition and candidate ge-nes. This approach can replace laborious Sanger confirmatory sequencing, increase testing turnaround, reduce analysis costs, and improve interpretation of the impact of variants by analysing their effect on mRNA processing.

Performance of non-formalin fixed paraffin embedded samples in hybrid capture and amplicon next-generation sequencing panels.

Genomic profiling using next-generation sequencing (NGS) is fundamental for driving prognostic and therapy in cancer. Formalin-fixed paraffin embedded (FFPE) tissue is the widely used material, whereas non-FFPE may represent an alternative. However, studies comparing the NGS performance of non-FFPE materials to FFPE are still lacking in the literature. The objective of this study was to characterize in non-FFPE preparations the nucleic acid yield and NGS performance on both a capture-based and an amplicon-based NGS platform. NGS quality metrics obtained from non-FFPE preparations were compared to FFPE. We analyzed the cellularity and nucleic acid yield in 111 tumors from non-FFPE preparations. In addition, comprehensive hybrid capture panel sequencing metrics obtained from DNA and RNA libraries were compared between independent non-FFPE and FFPE samples. A paired comparison between non-FFPE and FFPE samples was performed to analyze concordance in mutant allele detection using an amplicon panel. The mean target coverage from DNA libraries was 2× higher in non-FFPE samples than in FFPE. The detection of exogenous DNA was 2.5× higher in non-FFPE than in FFPE. Conversely, a lower performance was observed in non-FFPE RNA libraries in comparison to FFPE DNA libraries with no impact in minimum standard cutoffs. The variant allele detection in non-FFPE was found to be comparable to that of FFPE tumor samples in matched samples. Non-FFPE was demonstrated to be a suitable material for DNA and RNA library preparations using a comprehensive NGS panel. This is the first study reporting library quality metrics according to the TSO500 analysis pipeline.

Molecular profiling and feasibility using a comprehensive hybrid capture panel on a consecutive series of non-small-cell lung cancer patients from a single centre

Targeted next-generation sequencing (NGS) is recommended to screen actionable genomic alterations (GAs) in patients with non-small-cell lung cancer (NSCLC). We determined the feasibility to detect actionable GAs using TruSight™ Oncology 500 (TSO500) in 200 consecutive patients with NSCLC. DNA and RNA were sequenced on an Illumina® NextSeq 550 instrument and processed using the TSO500 Docker pipeline. Clinical actionability was defined within the molecular tumour board following European Society for Medical Oncology (ESMO) guidelines for oncogene-addicted NSCLC. Overall survival (OS) was estimated as per the presence of druggable GAs and treatment with targeted therapy. Most patients were males (69.5%) and former or current smokers (86.5%). Median age was 64 years. The most common histological type and tumour stage were lung adenocarcinoma (81%) and stage IV (64%), respectively. Sequencing was feasible in most patients (93.5%) and actionable GAs were found in 26.5% of patients. A high concordance was observed between single-gene testing and TSO500 NGS panel. Patients harbouring druggable GAs and receiving targeted therapy achieved longer OS compared to patients without druggable GAs. Conversely, patients with druggable GAs not receiving targeted therapy had a trend toward shorter OS compared with driver-negative patients. Hybrid capture sequencing using TSO500 panel is feasible to analyse clinical samples from patients with NSCLC and is an efficient tool for screening actionable GAs.

Cancer Genomic Profiling and Minimal Residual Disease Monitoring By Cell-Free DNA Sequencing in Pediatric Leukemia

Introduction: Cell-free DNA (cfDNA) sequencing has been reported as a promising non-invasive tool for detecting cancer and monitoring treatment response. We sought to evaluate the potential of a cfDNA-based strategy for detecting diverse classes of somatic mutations relevant to pediatric leukemia and cancers. Methods: Plasma samples collected at diagnosis from 135 acute lymphoblastic leukemia (ALL) and 34 acute myeloid leukemia (AML) patients were evaluated, including 114 B-ALL and 21 T-ALL. We also studied diagnostic plasma samples from 38 solid tumor and 19 brain tumor patients, and plasma collected from 12 patients with B-ALL at days 15, 22 and end of induction chemotherapy (EOI). Plasma cfDNA extraction and library preparation were performed with the NucleoSnap cfDNA kit and ThruPLEX Tag-Seq HV kit, respectively. Next-generation sequencing was performed using a capture panel covering 355 pediatric cancer genes (including noncoding regions covering hotspots of structural variation) and 8,103 heterozygous population polymorphisms. We developed analytic approaches for de novo detection of all types of somatic mutations, including focal copy number alterations, rearrangements, and large chromosomal abnormalities. Whole-genome sequencing (WGS) was also performed for 10 cfDNA samples from leukemia. Data were compared with corresponding whole-genome, exome and transcriptome tumor cell sequencing data at diagnosis. Results: At diagnosis, plasma cfDNA yield was higher for leukemias than for solid and brain tumors (median, 142 ng/ml, 7.16 ng/ml, and 7.07 ng/ml, respectively. The average coverages of leukemia cfDNA sequencing were 790X by capture panel post error suppression and 69X by WGS. We found that 100% cases of leukemia at diagnosis had detectable circulating tumor DNA (ctDNA) in plasma cfDNA, and the median ctDNA purity was 77%, comparable to primary tumor purity from diagnostic sequencing. Overall, our panel covered 86% somatic variants present in the tumor samples. Of those covered events, capture-based cfDNA sequencing detected 98% SNVs/indels, 95% focal copy gains/losses, 100% internal tandem duplications, 95% chromosomal translocations, and 99% aneuploidy. In addition, capture-based sequencing identified cfDNA-specific variants such as NRAS/KRAS SNVs and a IKZF1 deletion that were subclonal in plasma cfDNA. Of those uncovered events, 85% were structural variants with their breakpoints in intronic/noncoding regions, and cfDNA WGS identified all somatic events present in the 10 sequenced tumor samples. To detect minimal residual disease (MRD) in B-ALL patients during induction, we increased cfDNA sequencing depth for low VAF events, and the detection limit was 0.4% for SNV/indel and structural variations. Compared to the MRD results from flow cytometry, somatic mutations were detected in 7/8 MRD-positive and 1/4 MRD-negative patients on day 15, 3/4 MRD-positive patients on day 22, and 0/2 MRD-positive patients at EOI. We also sequenced cfDNA from peripheral blood and bone marrow plasma for 2 B-ALL patients on day 0, day 15, and EOI, with comparable ctDNA purity and dynamics observed between the two compartments. In a patient with concurrent ALL and neuroblastoma (NBL) who experienced NBL relapse four months after the primary diagnosis, cfDNA genomic profiling identified ctDNA derived from ALL (52% of total plasma cfDNA), primary NBL (20%), and relapsed NBL (10%), suggesting that genetic variations of multiple tumors could be detected in a single cfDNA sample. Conclusion: Our study has demonstrated the utility of cfDNA sequencing to identify both sequence and structural somatic mutations in childhood acute leukemia. This approach will be useful to monitor disease, as well as providing a non-invasive diagnostic approach.

Development of Circulating Tumor DNA (ctDNA) for Molecular Measurable Residual Disease (MRD) in Acute Myeloid Leukemia (AML)

Introduction: Despite advances in therapeutic options, the majority of AML patients experience relapsed or refractory disease. Standard-of-care (SOC) methods for MRD detection (multiparametric flow cytometry and single gene or structural variant molecular assays) are hampered by (1) inadequate lower limits of detection (LOD), (2) paucity of recurrent mutations across AML genomes, (3) insufficient numbers of identified mutations to characterize clonal evolution, and/or (4) need for invasive bone marrow biopsy. Herein we develop and apply non-invasive circulating tumor DNA (ctDNA) liquid biopsies using a customized Cancer Personalized Profiling by Deep Sequencing (CAPP-Seq) approach to simultaneously quantify MRD and characterize clonal dynamics in AML. Methods: To increase the number of tumor-specific mutations per case for MRD detection and improve the LOD for ctDNA-MRD, we performed whole exome sequencing (WES) on purified blasts and paired T cells (latter as germline control) from 24 patients with AML (n = 20 previously untreated, n = 4 with relapsed/refractory disease) and designed personalized hybrid capture panels. We combined the personalized panels with a shared hybrid capture panel of 58 recurrently mutated (‘canonical‘) AML genes covering 225kB as a composite approach to generate patient-specific AML-CAPP-Seq assays. We applied AML-CAPP-Seq to track single nucleotide variants (SNVs) in 468 biological samples from the 24 AML patients and 21 healthy control subjects. We compared the performance of MRD detection in cellular (bone marrow mononuclear cell/BMMC, peripheral blood mononuclear cell/PBMC) and cell-free (BM-ctDNA, PB-ctDNA) compartments against SOC clinical MRD assays as well as patient specific clinical outcomes. Results: We observed a median of 30 SNVs (IQR 16.5-40.25) per patient from WES, compared to only 2 (IQR 1-3.25) if using the AML-focused canonical panel alone (p < 0.0001) ( Fig. 1A). This number of variants enabled sensitive molecular disease monitoring in 96% (23/24) of patients with a LOD of < 0.01% tumor fraction from typical quantities of cell-free DNA. In contrast, when using the canonical panel alone, only 83% (20/24) of patients had 1 or more SNV to allow MRD detection. Notably, only 12.5% (3/24) of patients in this cohort had a feature amenable to SOC molecular MRD methods. Upon comparing quantitative levels of MRD in each biological compartment, we found that tumor burden in PB-ctDNA, BM-ctDNA and BMMC samples were highly correlated (r > 0.92, p < 0.0001) with similar amounts of disease. Interestingly, PB-ctDNA contained quantitatively more tumor DNA than corresponding PBMC samples (p = 0.0002), most pronounced at low tumor burdens (median 8.2-fold enrichment when mean tumor allele fraction in PBMCs was < 1%). These results suggest that PB-ctDNA effectively captures total body BM disease and mutational burden in AML and serves as a superior analyte for MRD monitoring compared to PBMCs. When evaluating the performance of AML-CAPP-Seq against clinical SOC methods, our assay correctly detected molecular disease in 99.6% (237/238) of samples known to contain measurable disease, including 98.2% (54/55) of samples at timepoints of clinical MRD positivity during clinical remission (CR). However, restricting trackable SNVs to only ‘canonical’ genes led to significant degradation of performance. Specifically, only 69.1% (38/55) of samples with positive MRD during CR by clinical SOC assay were detected by ctDNA if canonical genes alone were employed (p = 0.0001). Intriguingly, when using the complete gene panel, AML-CAPP-Seq also identified molecular MRD in 37.5% (30/80) of samples with undetectable MRD by SOC assays. Finally, stratifying patients who achieved CR after 1st treatment line based on PB-ctDNA MRD significantly predicted relapse free survival (p = 0.046, hazard ratio (HR) = 4.5). Moreover, attaining an MRD negative CR in PB-ctDNA at any timepoint was strongly prognostic of overall survival (p = 0.0004, HR = 8.5) ( Fig. 1B). Conclusions: ctDNA is a promising biomarker for ultrasensitive detection and tracking of MRD in AML without need for invasive bone marrow assessment. Tracking mutations beyond canonical AML genes improves performance of such assays. ctDNA analysis has numerous areas of clinical utility, including response and relapse prediction, evaluation of clonal dynamics, and exploration of mechanisms of therapy resistance.

Deregulation of HLA-Class I and Class II Genes in Chip, MDS, and Acute Leukemia and Similarity between Chip and MDS in HLA Profile

Introduction: Human leukocyte antigen (HLA) plays a major role in the interaction between the immune system and oncogenic process in various types of tumors. Emerging data suggests that the HLA system may have profound effects on response to immunotherapy and other types of therapy. Structural defects in HLA genes and variation in methylation have been reported to play a role in the expression of various HLA-I and HLA-II genes. Furthermore, an association between the expression of specific HLA-I or HLA-II genes and prognosis and outcome has been reported in some types of leukemia. Since HLA-I is expressed on all cells while HLA-II is expressed mainly in antigen presenting cells, we hypothesized that profiling the expression levels of both HLA-I and HLA-II in the bone marrow of patients with various myeloid neoplasms may provide information on the interaction between the microenvironment and the leukemic cell in these diseases. We used next generation sequencing to evaluate the expression level of HLA-I and HLA-II genes in myeloid neoplasms and compared these levels to normal bone marrow control. We used targeted next generation sequencing (NGS) for quantification of the RNA expression of 15 HLA-I and HLA-II genes. Methods: RNA was extracted from fresh bone marrow aspiration samples from 115 acute myeloid leukemia (AML), 196 clonal hematopoiesis of indeterminate potential (CHIP), 84 myeloproliferative neoplasm (MPN), 177 myelodysplastic syndrome (MDS), 40 acute lymphoblastic leukemia (ALL), and 279 normal bone marrow cases. RNA sequencing was performed using a targeted hybrid capture panel. The sequenced HLA-I genes were HLA-A, HLA-B, HLA-C, HLA-E, HLA-F, and HLA-H. The sequenced HLA-II genes were: HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DQB2, HLA-DRA, HLA-DRB1, HLA-DRB5, and HLA-DRB6. Salmon v1.4.0 software was used for expression quantification (TPM). To evaluate the depth of the significance in variation of the expression profiles between diseases we tested machine learning algorithm (random forest) incorporating the expression of all 15 genes in blindly classifying diseases. Two thirds of samples were used for training the random forest algorithm and one third was used for testing. Results: Bone marrow samples with CHIP showed no significant difference from normal bone marrow in level of expression of HLA-I genes but showed significantly (P <0.001) lower level of most HLA-II markers (DMB, DQA1, DQB2, DRB1, and DRB5). The same pattern was observed in MDS samples. In contrast, AML showed significant (P < 0.001) increase in the expression of both HLA-1 genes (A, C, and F) and HLA-II genes (DQA1, DPA1, DQB1, DRB6, and DPB1). Bone marrow samples from ALL patients showed significantly (P< 0.001) higher expression levels in DRB1 and DQA1 as compared with normal bone marrow. MPN samples showed no statistically significant difference from normal BM in any of the HLA-I or HLA-II genes, after adjusting for multiple testing. Machine leaning algorithm poorly classified MDS and AML and suggests complex overlap between the two diseases in HLA response. In contrast, the machine learning algorithm showed significant difference between MPN, MDS, and AML, and ability to distinguish all MPN cases with AUC of 0.738 (95% CI : 0.546-0.930) in the testing set (see figure). Conclusions: These findings show that the bone marrow microenvironment immune profile in MDS and CHIP likely plays a major role in the disease. Furthermore, this data suggests that this immune microenvironment is critical in CHIP and perhaps that therapeutic approaches supporting this immune microenvironment may help in preventing CHIP from progressing to MDS. The data also supports that AML is a disease that completely overwhelms the HLA immune system. In contrast, MPN appears to evade the HLA system and perhaps activating this system may help in the therapy of MPN.

Clonal Hematopoiesis and Inflammation in the Vasculature (CHIVE), a Prospective, Longitudinal Clonal Hematopoiesis Cohort and Biorepository

Introduction Clonal hematopoiesis (CH), the expansion of a clonal population of hematopoietic cells, is an age-associated phenomenon leading to increased risk of both hematologic malignancy and non-malignant organ dysfunction. There have been significant advances in our understanding of the underlying pathophysiology of CH in recent years. However, additional studies are needed to better understand clinical trajectories and optimal care of patients with CH. To address this gap, the prospective, longitudinal CHIVE (Clonal Hematopoiesis and Inflammation in the VasculaturE) registry and biorepository was created to support multidisciplinary CH clinics. Methods Adults without active hematological malignancy who were clinically identified to have CH or who were at risk for CH were sequentially enrolled utilizing a multidisciplinary referral approach between October 2020 to April 2023 at Vanderbilt University Medical Center (VUMC). Research samples were procured at the time of routine clinical blood draws or bone marrow biopsies and stored in a newly established biorepository, equipped to store DNA, plasma, and viably cryopreserved cells to enable a spectrum of downstream applications. Specimens were sequenced on a highly cost efficient (~$6/sample) targeted next generation sequencing capture panel that identified mutations residing in the 24 most frequently mutated CHIP genes.The VUMC CH clinic, which serves as a referral pool, were included in Arm A of CHIVE (n=57). We also sought out a population at risk for CH which included patients over 18 years of age with a history of solid tumor, cardiovascular disease, renal disease, rheumatologic disease, or diabetes, who were included in Arm B (n=122). Patients from Arm A or B were included in the CH-positive (CH+) group if they met criteria for CH on the research assay at any time during their enrollment in the study. All other genotyped patients who were not found to have CH by the research assay, regardless of their clinical CH status at study enrollment, were included in the CH-negative (CH-) group.Anthropometric data and clinical assessments, including laboratories and cardiovascular studies obtained in the course of routine clinical care were extracted from the electronic health record, and recorded and maintained in the CHIVE database. Results 261 patients were approached for consent to enroll in CHIVE and 179 were ultimately enrolled. More than one half (54%) of all participants had a CH mutation (CH+) and of those, 49.5% were female. Median age of CH+ patients was 71.9 (IQR 64.0-77.5). DNMT3A (n=44) and TET2 (n=39) were the most commonly mutated genes, while variant allele fractions (VAF) ranged from 2.1 - 79.8% (Figure 1). 246 samples were serial samples collected from 89 patients at regular intervals. Significantly more CH+ patients had a diagnosis of coronary artery disease compared to CH- patients (55.7% vs 32.9%, p=0.004). CH+ patients also had increased rates of hypertension and heart failure compared to the CH- group (p<0.001, 0.035, respectively). Laboratory evaluation showed CH+ patients to have a lower median white blood cell count, hemoglobin, and platelets, and elevated blood urea nitrogen values compared to CH- patients, differences that did not meet statistical significance. 8.2% of the CH+ patients progressed to frank hematologic malignancy over the course of the 2.5-year study (Table 1). Eight of the 97 CH+ and 2 of the 82 CH- patients died over the course of the study period resulting in a trend toward decreased survival in CH+ patients (p=0.112). Conclusions We demonstrate the feasibility of a prospective, observational study of CH patients utilizing a robust referral network to support both clinical care and translational research. A well-genotyped and phenotyped cohort of CH patients will be critical to future translational research efforts and clinical trials - a key function of our study design. Early clinical findings from our cohort recapitulate large scale retrospective datasets where CH patients are at an increased risk of development of hematologic malignancy, end organ damage, and all-cause mortality. Scaling this resource in collaboration with other centers is underway and will ultimately enable the development of clinical guidelines and treatment strategies for this increasingly recognized patient population.

Lymphoma Diagnosis and Classification Using Next Generation Sequencing of 30 CD Markers and Machine Learning As an Alternative to Immunohistochemistry

Introduction: Lymphoma diagnosis and classification requires pathologist interpretation of morphology and large numbers of immunohistochemistry (IHC) stains of various CD markers. This process is subjective and requires a significant amount of tissue. In contrast, RNA quantification of the same CD markers used in IHC using next generation sequencing (NGS) requires little tissue and is less influenced by the antigen retrieval process used in IHC. However, IHC staining and microscopic examination allows evaluation of the expression in various subpopulations and makes diagnosis possible. In contrast, when total RNA is evaluated by NGS, distinguishing between subpopulations is lost. Machine learning algorithms are capable of multi-marker normalizing and compensate for the loss of subpopulation analysis. To confirm this, we explored the capability of using RNA quantification of 30 CD markers by NGS from FFPE tissue along with machine learning in the clinical diagnosis and classification of various types of lymphoma. Methods: Formalin-fixed paraffin-embedded (FFPE) tissue from 130 diffuse large B-cell lymphoma (DLBCL), 70 mantle cell lymphoma, 92 T-cell lymphoma, 48 follicular lymphoma, 36 Hodgkin lymphoma, and 52 marginal zone lymphoma samples were used for extracting mRNA. The studied samples were consecutive without selection and included mainly lymph node excisional biopsies or core biopsies. RNA sequencing was performed using a targeted hybrid capture panel that included CD1A, CD2, CD3D, CD3E, CD3G, CD4, CD5, CD7, CD8A, CD8B, CD10, CD14, CD19, CD20, CD22, CD33, CD34, CD38, CD40, CD44, CD47, CD68, CD70, CD74, CD79A, CD79B, CD81, CD138, CD200, and CD274 genes. Salmon v1.4.0 software was used for expression quantification (TPM). Random forest machine learning algorithm was used for predicting diagnosis. Randomly selected two thirds of samples were used for training and one third was used for testing. Results: In some cases, diagnosis can be made by simply inspecting the RNA levels of various CD markers. However, machine learning shows remarkably high sensitivity and specificity in the diagnosis of most lymphoma subclasses. Area under the curve (AUC) was at 1.00 (95% CI: 1.000-1.00) for DLBCL vs. T-cell lymphoma, Hodgkin vs. T-cell, Hodgkin vs. DLBCL, mantle vs. DLBCL, and Follicular lymphoma vs. marginal lymphoma with 100% sensitivity and specificity in the testing set. AUC was at 0.974 (95% CI: 0.920-1.000) for marginal lymphoma vs. mantle cell lymphoma with sensitivity of 88% and specificity of 100%. The AUC was at 0.887 (95% CI: 0.776-0.999) for follicular lymphoma vs. DLBCL with sensitivity of 81.3% and specificity of 83.7%. Conclusions: This data demonstrates that NGS quantification of RNA from 30 CD markers when combined with machine learning is adequate for reliable classification of various types of lymphoma. This approach can provide valuable information to distinguish between difficult diagnoses, and if trained adequately has the potential to expand to more borderline cases. More importantly, this technology can be automated and less susceptible to human errors. RNA quantification using NGS has the potential to replace the need for IHC and can be applied when samples are limited such as in needle aspiration or core biopsies.

A Genomic Approach for High-Grade B Cell Lymphoma with MYC and BCL2 and / or BCL6 Rearrangements: Two Clusters Stood out on Next Generation Sequencing Results

Introduction High-grade B cell lymphomas with double or triple hit rearrangement (HGBL DH/TH), which appeared on the 4th revised edition WHO classification of lymphoid neoplasms, are aggressive mature B-cell lymphomas with high-grade cytology (Burkitt-like, blastoid or DLBCL-like) that have a MYC (8q24) rearrangement in combination with a BCL2 (18q21) and / or a BCL6 (3q27) rearrangement. There are currently no treatment guidelines for this subtype, and the usual immunochemotherapy leads to few complete remissions. While the NGS technology has been used in diffuse large B cell lymphoma, there has been no specific study on additional genetic events which would predict the clinical outcome in HGBL DH/TH. Here, we report our preliminary results using a targeted DNA capture panel approach (NGS) to identify subgroups with different clinical outcomes among HGBL DH/TH. Methods We conducted a retrospective study on 44 patients diagnosed with DH or TH HGBL and treated at Centre Hospitalier Lyon Sud or Centre Léon Bérard starting in 2013. We collected the medical information regarding the treatment, time to response to treatment or relapse, progression free survival, and overall survival. They all benefited from a DNA-targeted NGS analysis covering 54 genes. Results The cohort included 20 females and 24 males, with a median age of 65 (31-87). Among them, 22 patients were MYC/ BCL2 DH (50%), 12 were MYC/ BCL6 DH (27,3%) and 10 were TH (22,7%). Eleven patients had B symptoms (25%). The Ann Arbor stage was high (III-IV) in 97.5%. The WHO Performance Status was low (0-1) for most of the patients (70%). The majority of them presented with an IPI above to 2 (82,5%) and an elevated LDH level (80%). The Ki67 index was above 70% (33/42) in 78,5% of the cases and above 90% (19/42) in nearly half of them. The major finding of the NGS analysis was the identification of 2 groups based on the TP53/ CDKN2A_p14ARF mutational status. The CDKN2A locus encodes for 2 alternative open reading frames leading to 2 unrelated proteins p14ARF and p16INK4A. P14ARF is involved in the regulation of the TP53 pathway whereas p16INK4A is not. We therefor defined 2 groups : the first group included patients with either TP53 or CDKN2A_p14ARF-only alterations (group A, n=10) while the second group had no alteration in these two coding sequences (group B, n=34) although 12 patients had a CDKN2A_p16INK4A loss or mutation affecting both CDKN2A transcripts. Most of TP53 mutations were missense mutations mostly in TP53 DNA-binding domain and 2 cases appeared to be single hit mutations with an allele frequency below 50%. The patient with a p14ARF only mutation had a start codon mutation p.(M1*) with an allele frequency of 96% without CDKN2A copy number loss. The median overall survival (OS) (Figure 1) was 6,2 months for group A, and 55,5 months for group B (p = 0,0076). The median follow-up time was 14 months (7,5 months in the group A, 18 in group B). Only 2 patients in group A survived and are still in complete remission, one was DH BCL6 and the other DH BCL2 Conclusion Our preliminary results indicate a strong negative prognostic impact of either CDKN2A_p14ARF or TP53 gene alterations on the OS of HGBL DH/TH patients. The biological basis for the difference between groups would be in favour of the alteration of the p14ARF-TP53 tumor suppressive pathway as major contributor of chemoresistance. A more detailed analysis is on-going to further identify mutation profiles of double hit and triple hit lymphomas and compare our finding with existing data. Our results confirm the more favourable outcome of DH BCL6 patients (data not shown), which is consistent with the 5 th revised OMS classification that removed that specific entity. Our results also indicate that routine NGS analysis of HGBL DH/TH patients offers new perspectives for patient stratification. Regarding their survival, patients with TP53 pathway alterations should probably not be treated with the standard RCHOP chemo-immunotherapy. Instead, a more intensive immunochemotherapy regimen or novel strategies such as first-line bispecific antibodies or CARTCells might be considered.

Using Next Generation Sequencing of Flow Cytometry CD Markers and Machine Learning As a Replacement to Flow Cytometry Analysis for the Diagnosis of Hematologic Neoplasms

Introduction: Flow cytometry performs multi-parameter analysis of cells and analyzes surface and intracellular markers for accurate phenotypic characterization of a cell population. Flow cytometry is used extensively in the diagnosis and classification of various hematologic neoplasms. However, analysis of the generated data is time consuming and remains subjective, requiring special skill and experience. Furthermore, some diagnostic classes, such as myeloproliferative neoplasms (MPN) and myelodysplastic syndrome (MDS), are difficult to diagnose using flow cytometry. The RNA levels of the CD markers used in flow cytometry can be reliably quantified using next generation sequencing (NGS). However, when all cells are jointly sequenced, studying subpopulation of cells is lost, which hinders accurate diagnosis. However, machine learning algorithms are capable of multi-marker normalizing and compensate for the loss of subclonal analysis. To validate this assumption, we explored the potential of using the RNA levels of 30 CD markers along with a machine learning algorithm in the differential diagnosis between various types of hematologic neoplasms. Methods: RNA was extracted from fresh bone marrow and peripheral blood samples from 172 acute myeloid leukemia (AML), 369 normal control, 68 MPN, 218 MDS, 93 acute lymphoblastic leukemia (ALL), 74 chronic lymphocytic leukemia (CLL), 38 mantle cell lymphoma, and 83 multiple myeloma cases. The samples were consecutive and collected without selection. RNA sequencing was performed using a targeted hybrid capture panel that included CD1A, CD2, CD3D, CD3E, CD3G, CD4, CD5, CD7, CD8A, CD8B, CD10, CD14, CD19, CD20, CD22, CD33, CD34, CD38, CD40, CD44, CD47, CD68, CD70, CD74, CD79A, CD79B, CD81, CD138, CD200, and CD274 genes. Salmon v1.4.0 software was used for expression quantification (TPM). Machine learning algorithm (random forest) was used for classifying diseases. Two thirds of samples were used for training the random forest algorithm and one third was used for testing. Results: While frequently a diagnosis can be made by simply inspecting the RNA levels of various CD markers, machine learning is needed when the fraction of the neoplastic cells is low. Using machine learning (random forest), diagnosis of most hematologic neoplasms was achieved with high sensitivity and specificity in the testing set. Area under the curve (AUC) was at 0.972 (95% CI: 0.950-0.994) for AML vs. normal, 0.936 (95% CI: 0.898-0.974) for normal vs. MM, 0.965 (95% CI: 0.909-1.00) for mantle vs. CLL, 0.962 (95% CI: 0.907-1.00) for CLL vs. ALL, 0.935 (95% CI: 0.866-1.00) for CLL vs. normal, and 0.964 (95% CI: 0.927-1.00) for AML vs. ALL. Diseases that are difficult to diagnose by routine flow cytometry were diagnosed by RNA expression and machine learning at acceptable accuracy. For example, AUC was at 0.761 (95% CI: 0.689-0.834) for MDS vs. normal, 0.831 (95% CI: 0.762-0.901) for MDS vs. AML, 0.888 (95% CI: 0.822-0.954) for MDS vs. MPN, and 0.785 (95% CI: 0.698-0.872) for MPN vs. normal. Conclusions: This data demonstrates that NGS quantification of RNA from 30 CD markers when combined with machine learning is adequate for reliable diagnosis of various types of hematologic neoplasms. This approach can provide valuable information to distinguish between MPN, MDS, and normal bone marrow that flow cytometry cannot provide. Furthermore, this technology can be automated and less susceptible to human error and practically can be used as a replacement to routine flow cytometry analysis.

Polygenic HLA-Class I and HLA-Class II Models Define Various Types of Lymphoma

Introduction: Human leukocyte antigen (HLA) plays a major role in the interaction between the immune system and oncogenic process in various types of tumors including lymphoma. Mono-allelic and bi-allelic loss of the HLA-I has been reported in diffuse large B-cell lymphoma (DLBCL) and other types of lymphoma. Poor outcome was also reported with the loss of the HLA-I system in lymphoma. Since HLA-I and HLA-II are both critical for healthy immune response and both can be expressed in the lymphoma cells as well as in the lymphoma responding microenvironment, analyzing both systems is crucial for deciphering the role of the HLA system in lymphoma. Toward this goal, we evaluated the expression level of various HLA-I and HLA-II genes and compared these levels between subtypes of lymphoma. We quantified RNA expression levels of 15 HLA-I and HLA-II genes using next generation sequencing (NGS) and used machine learning algorithm (random forest) to evaluate the depth of the significance in variation of the expression profiles between various types of lymphoma. Methods: RNA was extracted from formalin-fixed paraffin-embedded (FFPE) tissue from 99 diffuse large B-cell lymphoma (DLBCL), 79 T-cell lymphoma, 40 follicular lymphoma, 28 mantle cell lymphoma, and 14 Hodgkin lymphoma samples. RNA sequencing was performed using a targeted hybrid capture panel. The sequenced HLA-I genes were HLA-A, HLA-B, HLA-C, HLA-E, HLA-F, and HLA-H. The sequenced HLA-II genes were HLA-DPA1, HLA-DPB1, HLA-DQA1, HLA-DQB1, HLA-DQB2, HLA-DRA, HLA-DRB1, HLA-DRB5, and HLA-DRB6. Salmon v1.4.0 software was used for expression quantification (TPM). Random forest machine learning system was used for predicting subtypes. Two thirds of samples were used for training the random forest algorithm and one third was used for testing. Results: RNA levels of individual HLA-I and HLA-II genes varied mildly between various B-cell lymphoma subtypes without specific pattern, especially after adjusting for multiple testing. However, T-cell lymphoma showed overall significantly (P< 0.0001) higher HLA-I (B, C, and F) and lower HLA-II (DRB1 and DRB5) expression levels as compared with B-cell lymphoma. In contrast, using machine learning algorithm to evaluate combination of HLA-I and HLA-II expression profiles showed discrete and significant differences between various types of lymphoma. The levels of expression of HLA-I and HLA-II genes were adequate to distinguish between DLBCL and follicular lymphoma with AUC of 0.778 (95% CI: 0.46-1.00), and between DLBCL and T-cell lymphoma with AUC of 0.833 (95% CI: 0.624-1.00). The number of Hodgkin lymphoma and mantle cell cases was small but the random forest distinguished between DLBCL and Hodgkin lymphoma with AUC of 0.833 (95% CI: 0.294-1.00), and between DLBCL and mantle cell lymphoma with AUC of 0.778 (95% CI: 0.416-1.00). Conclusions: This data confirms the polygenic effects of the HLA-I and HLA-II on various types of lymphoma. The HLA-I and HLA-II combined expression profiles are significantly different between various types of lymphoma. More importantly, this data suggests that immune modulating therapy should consider the polygenic effects of both the HLA-I and HLA-II systems.

Haplotype-based noninvasive prenatal diagnosis of methylmalonic acidemia and the discovery of a recurrent pathogenic haplotype associated with c.609G>A.

Early diagnosis and intervention are crucial for the prognosis of methylmalonic acidemia (MMA). However, research focused on early prenatal diagnosis of MMA is limited. A 161.89kb capture panel was designed for selectively enriching highly heterozygous SNPs. Fetal genotypes were inferred using relative haplotype dosage (RHDO) and Bayes factor, followed by invasive prenatal diagnosis (IPD) for validation. A core pathogenic haplotype associated with c.609G>A was identified based on the frequency differences between pathogenic and normal haplotypes. We recruited 41 pregnancies at risk of MMA with a median gestational age of 8+2 weeks. The assay success rate of NIPD-MMA for maternal variants was 92.7% (38/41), and after incorporating the paternal result, the overall assay success rate reached 100% (41/41). All NIPD results were concordant with IPD. Notably, a core haplotype (hap_2), comprising 28 SNPs, demonstrates significant enrichment within pathogenic haplotypes bearing the c.609G>A variation. On average, c.609G>A carriers had 22.38 heterozygous loci within these 28 SNPs. NIPD-MMA presents a viable choice for early, accurate, and safe prenatal diagnosis. Furthermore, the discovery of the recurrent core pathogenic haplotype provides a novel approach for haplotype phasing and has the potential for realizing proband-independent NIPD in the future.

Intronic Germline DICER1 Variants in Patients With Sertoli-Leydig Cell Tumor.

Germline pathogenic loss-of-function (pLOF) variants in DICER1 are associated with a predisposition for a variety of solid neoplasms, including pleuropulmonary blastoma and Sertoli-Leydig cell tumor (SLCT). The most common DICER1 pLOF variants include small insertions or deletions leading to frameshifts, and base substitutions leading to nonsense codons or altered splice sites. Larger deletions and pathogenic missense variants occur less frequently. Identifying these variants can trigger surveillance algorithms with potential for early detection of DICER1-related cancers and cascade testing of family members. However, some patients with DICER1-associated tumors have no pLOF variants detected by germline or tumor testing. Here, we present two patients with SLCT whose tumor sequencing showed only a somatic missense DICER1 RNase IIIb variant. Conventional exon-directed germline sequencing revealed no pLOF variants. Using a custom capture panel, we discovered novel intronic variants, ENST00000343455.7: c.1752+213A>G and c.1509+16A>G, that appear to interfere with normal splicing. We suggest that when no DICER1 pLOF variants or large deletions are discovered in exonic regions despite strong clinical suspicion, intron sequencing and splicing analysis should be performed.

Targeted High-throughput Sequencing for Hematological Malignancies: A GBMHM Survey of Practice and Cost Evaluation in France.

The objective of this study was to assess the clinical impact and financial costs of next-generation sequencing (NGS) in 5 categories of pediatric and adult hematological cancers. NGS prescriptions were prospectively collected from 26 laboratories, with varied technical and reporting practice (all or only significant targets). Impact was defined by the identification of (1) an actionable mutation, (2) a mutation with prognostic and/or theranostic value, and/or (3) a mutation allowing nosological refinement, reported by local investigators. A microcosting study was undertaken in 4 laboratories, identifying the types and volumes of resources required for each procedural step. Individual index prescriptions for 3961 patients were available for impact analysis on the management of myeloid disorders (two thirds) and, mainly mature B, lymphoid disorders (one third). NGS results were considered to impact the management for 73.4% of prescriptions: useful for evaluation of prognostic risk in 34.9% and necessary for treatment adaptation (actionable) in 19.6%, but having no immediate individual therapeutic impact in 18.9%. The average overall cost per sample was 191 € for the restricted mature lymphoid amplicon panel. Capture panel costs varied from 369 € to 513 €. Unit costs varied from 0.5 € to 5.7 € per kb sequenced, from 3.6 € to 11.3 € per target gene/hot-spot sequenced and from 4.3 € to 73.8 € per target gene/hot-spot reported. Comparable costs for the Amplicon panels were 5-8 € per kb and 10.5-14.7 € per target gene/hot-spot sequenced and reported, demonstrating comparable costs with greater informativity/flexibility for capture strategies. Sustainable funding of precision medicine requires a transparent discussion of its impact on care pathways and its financial aspects.

Baseline mutations and ctDNA dynamics as prognostic and predictive factors in ER-positive/HER2-negative metastatic breast cancer patients.

Prognostic and predictive biomarkers to CDK4/6 inhibitors are lacking. Circulating tumour DNA (ctDNA) can be used to profile these patients and dynamic changes in ctDNA could be an early predictor of treatment efficacy. Here, we conducted plasma ctDNA profiling in patients from the PEARL trial comparing palbociclib+fulvestrant versus capecitabine to investigate associations between baseline genomic landscape and on-treatment ctDNA dynamics with treatment efficacy. Correlative blood samples were collected at baseline (C1D1) and prior to treatment (C1D15). Plasma ctDNA was sequenced with a custom error-corrected capture panel, with both univariate and multivariate Cox models used for treatment efficacy associations. A pre-specified methodology measuring ctDNA changes in clonal mutations between C1D1 and C1D15 was used for the on-treatment ctDNA dynamic model. 201 patients were profiled at baseline, with ctDNA detection associated with worse PFS/OS. Detectable TP53 mutation showed worse PFS and OS in both treatment arms, even after restricting population to baseline ctDNA detection. ESR1 mutations were associated with worse OS overall, which was lost when restricting population to baseline ctDNA detection. PIK3CA mutations confer worse OS only to patients on the palbociclib+fulvestrant treatment arm. ctDNA dynamics analysis (n=120) showed higher ctDNA suppression in the capecitabine arm. Patients without ctDNA suppression showed worse PFS in both treatment arms. We show impaired survival irrespective of endocrine or chemotherapy-based treatments for patients with HR+/HER2- MBC harbouring plasma TP53 mutations. Early ctDNA suppression may provide treatment efficacy predictions. Further validation to fully demonstrate clinical utility of ctDNA dynamics is warranted.

Abstract B016: Liquid biopsy biomarker analysis of a phase II trial of sacituzumab govitecan in castrate resistant metastatic prostate cancer

Abstract Background: Sacituzumab Govitecan (SG) is an antibody-drug conjugate targeting Trop-2 and has been approved for use in breast and bladder cancers. Trop-2 has been shown to be overexpressed in multiple solid tumors including prostate cancer. Here we report an interim analysis of longitudinal liquid biopsies collected in a phase II clinical trial of SG to identify biomarkers of response and resistance. Methods: We collected longitudinal blood samples from 20 patients for circulating tumor cell (CTC) and cell-free DNA (cfDNA) analysis as part of an open label Phase II trial of SG in patients with mCRPC (ClinicalTrials.gov Identifier: NCT03725761). Eligibility criteria for the trial included rising PSA and radiographic progression on abiraterone acetate or enzalutamide in the mCRPC setting. Liquid biopsies were collected at baseline, C1D8, and then every 3 cycles until disease progression. cfDNA was sequenced using a custom 822 gene targeted capture panel (IDT). CTCs were isolated with antibodies to either EpCAM or Trop-2 followed by parallel enumeration, EpCAM and Trop-2 protein phenotyping and RNA extraction for AR/NEPC marker transcriptional profiling. Results: Although no PSA50 responses were observed, the 6-month rPFS rate was 45% (64% in chemotherapy naïve patients). For this analysis, these patients will be termed responding patients. The number of Trop-2-postive CTCs at baseline was significantly higher in responding (24.3 ±13.4 per 7.5 mL blood) than non-responding (10.0±3.8) patients. During treatment the number of Trop-2 positive CTCs in responders significantly decreased by C3D1 to 10.0 ±3.8. Interestingly, while the number of Trop-2 positive CTCs continued to decline to 2.8±1.99 cells/7.5 mL, EpCAM-positive CTCs increased from 12.5±10.52 cells/7.5 mL to 23.2±19.49 at end of trial. Gene expression analysis demonstrated high expression of genes specific to adenocarcinoma including AR-splice variants, TMPRSS2, KLK2 and KLK3 in the majority of patients, with conversion to AR-variant negative CTCs identified in responders. Preliminary cfDNA analysis identified one patient with a mutation acquired in the gene encoding topoisomerase 1 (TOP1) upon progression on SG, with additional analysis ongoing. Conclusions: Patients who responded to SG had a more Trop-2 positive CTCs with adenocarcinoma characteristics with expression of AR pathway genes. Longitudinal analysis of liquid biopsy biomarkers identified multiple mechanisms of acquired SG resistance. Decline in Trop2-CTC population with parallel expansion of EpCAM-CTC population in multiple patients suggests loss of Trop2 surface protein expression as resistance mechanism, while identification of a cfDNA TOP1 mutation in one patient at time of progression is consistent with acquired resistance to the topoisomerase inhibitor payload as has been reported in metastatic triple negative breast cancer. Citation Format: Jamie M. Sperger, Amy K. Taylor, Yue Shi, Charlotte N. Stahlfeld, Kyle Helzer, Matthew Bootsma, Katherine R. Kaufmann, Marina N. Sharifi, Christos E. Kyriakopoulos, Susan Slovin, Scott Tagawa, Scott Dehm, Shuang G. Zhao, Joshua M. Lang. Liquid biopsy biomarker analysis of a phase II trial of sacituzumab govitecan in castrate resistant metastatic prostate cancer [abstract]. In: Proceedings of the AACR Special Conference: Advances in Prostate Cancer Research; 2023 Mar 15-18; Denver, Colorado. Philadelphia (PA): AACR; Cancer Res 2023;83(11 Suppl):Abstract nr B016.

Impact of panel size on minimum residual disease (MRD) assay performance.

e15022 Background: Sequencing circulating tumor DNA is a promising method for monitoring cancer treatment response and detecting recurrence, but sensitivity at the low tumor fractions typical of early-stage, post-treatment, and early recurrent tumors is limited by the small number of variants targeted in commercially available assays. Methods: We developed a tumor-informed minimum residual disease (MRD) assay using tumor-normal whole-genome sequencing (WGS) followed by interrogation of select somatic variants in cell-free DNA (cfDNA). We describe theoretical limits on panel designability as a function of tumor purity and sequencing depth using simulations and a discovery cohort of 31 patient samples. cfDNA assay performance was tested by creating serial dilutions of patient plasma before somatic enrichment and sequencing. Simulations and in-silico downsampling of target sites were used to test sensitivity as a function of panel size. A bioinformatics pipeline incorporating somatic calling, copy number inference, error assessment, and tumor-fraction estimation was developed to analyze results. Results: Across a range of tissues, tumor-normal WGS identified from hundreds to tens of thousands of somatic variants to inform an MRD capture panel. 97% of samples tested returned at least 300 high-confidence somatic variants, and targeted sequencing demonstrated over 95% positive predictive value in somatic variant calls. In cfDNA dilution experiments, we observed > 99% sensitivity at a tumor fraction of 0.01%. Expanding the panel size from 16 to 300 target variants increased sensitivity at 0.005% tumor fraction from approximately 20% to > 95%. Our tumor fraction model returned accurate quantification down to the parts per hundred thousand (0.001%) range, even when only a single tumor molecule could be observed at each target site. Conclusions: Increasing the number of targeted variants in an MRD assay increases sensitivity at low tumor fractions, and WGS-driven panel design ensures a sufficient number of interrogated variants while minimizing the failure rate due to panel designability. High sensitivity MRD has the potential to enable earlier recurrence detection while also giving researchers and clinicians new tools for monitoring patient treatment responses.

Ultra-sensitive baseline-informed MRD assay to predict prognosis outcomes in patients with resectable hepatocellular carcinoma.

e16228 Background: The detection of minimal residual disease (MRD) has been shown to be a potential biomarker for the prognosis and post-surgery treatment of resectable hepatocellular carcinoma (HCC). However, the application of MRD in HCC has been limited by the traditional requirement of carrying out MRD over 30 days post-surgery, which is often incompatible with the time window for adjuvant treatment decisions. Additionally, MRD results may be impacted by post-surgery imbalances if assayed too soon after surgery. Methods: This retrospective study enrolled 255 early-stage HCC patients who received surgical resections and had complete information on relapse and survival. Matched tumor and normal tissue samples were collected at the time of surgery and plasma samples were collected as early as 7 days post-surgery. An ultra-sensitive, baseline-informed MRD assay (PredicineBEACON) was used to evaluate the correlation between peri-surgical MRD status and prognosis outcomes. Somatic mutations were identified through PredicineWES+ whole exon sequencing, and up to 50 somatic mutations were selected for the design of a patient-specific capture panel, in addition to a fixed panel of 500 actionable variants. Results: The study has completed MRD tests for 58 patients, of which 8 were called MRD positive with ≥0.005% tumor fractions. The tumor fraction proportion of MRD-positive patients ranged from 0.008% to 5.653% (median 0.315%). The relapse-free survival (RFS) and overall survival (OS) were correlated with MRD status: the median RFS of MRD-positive patients was 13.9 months compared to not reached after 44.4 months for MRD-negative patients (p = 0.1), and the median OS of MRD-positive patients was 27.2 months compared to not reached after 55.5 months for MRD-negative patients (p = 0.03). Conclusions: Our results demonstrate the feasibility of using an ultra-sensitive, baseline-informed MRD assay to detect strong correlations between MRD status and prognosis outcomes in peri-surgical plasma samples collected as early as 7 days post-surgery in resectable HCC patients. This study highlights the clinical utility of MRD testing in peri-surgical settings.