Society: AGA

Background: Colonic polyposis, which is a risk factor for colorectal cancer (CRC), can be driven by a pathogenic variant in a polyposis gene. However, in many cases germline genetic testing is negative, leaving the cause of the polyposis unknown. There has been increasing evidence that specific microbiome and metabolomic signatures may be associated with colon polyp development and CRC. However, at this time it remains uncertain if the microbiome plays a role in the development of colonic polyposis in the absence of a known polyposis gene mutation.

Methods: Subjects undergoing a standard-of-care colonoscopy were recruited via an IRB-approved protocol as either control subjects with 2 or fewer lifetime colon adenomas, or polyposis subjects with ≥10 total colonic adenomas and either negative polyposis multigene panel genetic testing or a known familial adenomatous polyposis (FAP) or MUTYH-associated polyposis (MAP). A stool sample was collected prior to the colonoscopy, and during the procedure mucosal biopsies were obtained from the proximal and distal colon. To measure bacterial community composition, we carried out 16S rRNA marker gene sequencing (hypervariable region V1-V2) on the Illumina MiSeq platform. Analysis was performed using QIIME2 and R.

Results: A total of 30 subjects were recruited, including 17 with polyposis (5 with FAP/MAP [gene-positive] and 12 with negative polyposis genetic testing [gene-negative]) and 13 control subjects. Global analysis of the entire microbial community showed that there is a significant decrease in alpha diversity as measured by richness in the gene-positive polyposis group compared to the gene-negative polyposis and control groups (P= 0.047, P=0.02). Faith’s PD was decreased in the gene-positive polyposis compared to the gene-negative polyposis groups (P=0.031). Bacterial communities in the gene-positive and gene-negative polyposis groups were also different by weighted and unweighted UniFrac distances (PERMANOVA test, P=0.02, P=0.01). Although we could not see any statistically significant differences in individual taxa between the groups through linear models, our analysis was limited by the sample size.

Conclusions: This study provides the first comprehensive microbiome characterization and comparison between individuals with colonic polyposis both with and without a known genetic driver, showing these distinct populations have different microbiome compositions. Elucidating microbiome signatures of the gene-negative polyposis cohort is a critical first step for future studies examining the potential causal role of the microbiome in the development of gene-negative polyposis and for the development of microbiome-targeted therapies to aid in decreasing polyp development and CRC risk.

Purpose: HCC incidence is highest in Asians and Pacific Islanders, followed by Blacks, Hispanics, and non-Hispanic Whites, and is rising at an alarming rate in the U.S. Most HCC is diagnosed at an advanced incurable stage, emphasizing the need for accurate biomarkers and early diagnosis. The predictive accuracy in available biomarkers such as alpha-fetoprotein (AFP) is sub-optimal, and these have not been systematically assessed across ethnic and racial subgroups. Here, we report on the performance of a new predictive model for HCC in patients from diverse ethnic/racial backgrounds, using clinical characteristics and biomarker proteins from the TGF-β superfamily signaling pathway that have shown relevance to hepatocarcinogenesis.
Experimental Procedures: We combined human genomic studies using The Cancer Genome Atlas and mechanistic insight from animal models to identify 108 biologically relevant proteins involved in TGF-β superfamily signaling, which were measured using SomaScan proteomic analysis, utilizing 55 µL serum samples from 248 individuals with cirrhosis, 94 of whom had HCC by imaging criteria. Patients were recruited from four participating centers with significant cohorts of ethnic minorities (11% Asian, 18% Black, 5% Hispanic, 4% Hawaiian/Pacific Islander, 54% White patients). In a derivation sample (N=129; 31 with HCC), we tested serum markers using univariable non-parametric analysis, adjusted for false discovery rate, used quintiles to eliminate skewness, and then combined the resulting serum biomarkers with clinical and demographic risk factors to develop a new model for detecting current early-stage HCC on a background of cirrhosis, and tested it in a separate validation sample (N=119; 63 with HCC). We tested heterogeneity by race in the derivation and test samples for area under the ROC curve (AUC).
Results: The new functional model included 4 clinical variables (age, serum AFP, Bilirubin level, HBV y/n) and 3 serum protein biomarkers (AKR1B10, COL15A1, and INHBB). The new model had better AUC than Doylestown or AFP in both the derivation and validation samples. In the derivation sample, AUC for our model was 0.92 in Black patients, 0.90 in White patients, and 0.81 in other race/ethnic groups, all of which were better than AFP alone or the Doylestown model AUC in those samples (Table). In the validation sample, our model outperformed the Doylestown model, but AFP alone performed slightly better in White patients.
Conclusions and Next Steps: This study demonstrates the improved performance of our novel biological markers and model to detect early-stage HCC across cirrhotic patients compared to the Doylestown model or AFP alone in a racially and ethnically diverse population. Next steps are large-scale validation of this biomarker panel in cirrhotic patients across all racial and ethnic subgroups in Phase II/III studies.

Background and Aim: Gastrointestinal (GI) adenocarcinomas account for a significant amount of newly diagnosed cancer cases and cancer-related mortality globally. There are currently no clinically available blood biomarker tests that are sensitive and specific for GI adenocarcinomas. Methylated BCAT1 and IKZF1 are ctDNA biomarkers that have high sensitivity and specificity for detection colorectal cancer (CRC). However, it is not clear if these biomarkers are also sensitive for detection of other GI adenocarcinomas. This study aimed to investigate if the ctDNA assay for methylated BCAT1 and IKZF1 is sensitive as a pan-GI cancer blood test for colorectal, gastric, oesophageal and pancreatic adenocarcinomas.
Methods: Blood was collected from patients diagnosed with colorectal (n=312), gastric (n=31), oesophageal (n=92) and pancreatic (n=24) adenocarcinoma, prior to any treatment commencement at Flinders Medical Centre (South Australia). Blood was also collected from control participants (n=607) with no evidence of CRC or precursor adenomas at colonoscopy. Circulating cell-free DNA isolated from plasma was bisulphite-converted and assayed using a multiplex qPCR assay for simultaneous detection of methylation in BCAT1 and IKZF1. ACTB was measured as a reference gene. Logistic regression and chi-square tests were employed to assess ctDNA BCAT1 and IKZF1 DNA methylation test sensitivity in GI adenocarcinomas compared to controls, test positivity rates between each of the cancer types, and test positivity between AJCC (American Joint Committee on Cancer) tumour stages for each GI adenocarcinoma.
Results: Overall test sensitivity for all GI adenocarcinomas combined was 56.0% (95% confidence interval (CI) 51.3 – 60.6%) and 8.1% (95%CI 5.9 – 10.3%) for controls. The proportion of cases with positive ctDNA for each type of GI adenocarcinoma was significantly higher compared to the control group (Table 1). The ctDNA positivity rate ranged between 48.4% for gastric adenocarcinoma and up to 57.1% for CRC. There were no significant differences in test positivity rates between any of the GI cancer types (P=0.746). Late AJCC stage (III/IV) was associated with a higher ctDNA test positivity rate for CRC (P<0.001), gastric (P<0.001), and oesophageal (P<0.001) adenocarcinoma compared to early stage (I/II) disease, but not for pancreatic adenocarcinoma (P=0.408) (Figure 1).
Conclusions: Assessment of ctDNA methylated in BCAT1 and IKZF1 is sensitive for detection of multiple GI adenocarcinomas, particularly for patients diagnosed with more advanced disease. Further investigation of the efficacy of this pan-GI cancer test as a tool to monitor for treatment response and disease recurrence across these GI adenocarcinomas is warranted.

Introduction
Methylated DNA markers (MDMs) assayed from plasma cell-free DNA (cfDNA) are a promising detection tool in multiple cancer types. Management of pancreatic neuroendocrine tumors (PNETs) requires metabolic imaging for staging at diagnosis, and at serial timepoints during post-operative surveillance and for monitoring of treatment response in metastatic disease. This current treatment paradigm is expensive and reliable blood-based biomarkers are limited. We have previously discovered and validated in tissue a panel of highly discriminatory MDMs that distinguish PNETs from normal pancreatic tissue. In this study we aimed to explore the diagnostic performance of these MDMs in plasma.

Methods
A panel of 29 candidate MDMs previously identified in tissue were assayed from 4 ml of plasma in 131 cases with biopsy-proven treatment naïve PNET and 74 age- and sex-balanced healthy controls. cfDNA was isolated and libraries prepared using enzymatic conversion. DNA pools were used to enrich target sequences in hybrid capture reactions using probes specifically designed to target each candidate MDM. Mean CpG methylation values across the target region for each MDM were used as classifiers in a random forest (rFor) model for distinguishing PNET cases from controls; area under the ROC curve (AUC), sensitivity, and specificity were summarized for the rFor model after in silico cross validation. Relative MDM importance within the rFor model was assessed by randomly permuting marker levels within the cross validation to estimate the mean decrease in accuracy. Fisher’s exact test compared positivity rates of the rFor model between patient subgroups.

Results
Methylated RASSF3 and SRRM3 had the highest relative importance within the rFor model with individual AUCs of 0.76 (95% CI 0.69-0.82) and 0.72 (95% CI 0.65-0.79), respectively, for distinguishing PNET cases from controls. The cross-validated AUC for the rFor model incorporating all 29 MDMs was 0.76 (95% CI 0.7-0.83). At a 90% specificity cutoff for the rFor model prediction, the cross-validated sensitivity was 85% (95% CI 73-93%) for Stage IV disease and 57% (48-66%) for all stages combined. Among the non-metastatic PNETs, the sensitivity of the MDM panel was higher in tumors >2cm compared to smaller tumors (49% (95% CI 33-65%) vs 13% (95% CI 4-31%); p=0.0022) and in those with regional lymph node involvement (56% (95% CI 31-78%)) vs in those without (26%(95% CI 15-40%), p=0.0416).

Discussion
A target capture-MDM panel can detect metastatic PNETs in plasma with high accuracy. Further analysis of methylation patterns within target capture regions are under evaluation with the aim to enhance sensitivity of early-stage detection. These results are promising and can be further explored in the context of blood-based multicancer detection tests and for applications in treatment monitoring and surveillance of PNETs.

<b>Figure 1:</b><br /> Boxplots demonstrating the distribution of <i>RASSF3</i> and <i>SRRM3 </i>methylation levels among healthy controls and pancreatic neuroendocrine tumor cases across the different stages

Figure 1:
Boxplots demonstrating the distribution of RASSF3 and SRRM3 methylation levels among healthy controls and pancreatic neuroendocrine tumor cases across the different stages

Saurabh Baheti

Appearances