AGA Abstracts Table 1. Baseline Characteristics Figure 1. Kaplan-Meier survival curves for patients of each age groups. (A) Adjusted overall survival and adjusted hazard ratio of each age groups. (B) Overall survival of SD-CRC and non-SD-CRC group of each age groups, and adjusted hazard ratio of SD-CRC group. Su1662 TIME TO DIAGNOSIS OF EARLY-ONSET VS. "LATE-ONSET" COLORECTAL CANCER: IS THERE A DIFFERENCE? Thomas F. Imperiale, Laura Myers, Kayla L. Chin, Carrie J. Ballard, Jessica Coffing, Joanne K. Daggy Background: Early-onset colorectal cancer (EOCRC) presents with more advanced disease than does CRC in persons aged ≥50 years. While reasons remain unclear, one suggested factor is a delay in diagnosis in younger persons. Using a single-center cancer research database, Chen et al (Clin Gastro & Hepatol 2017) showed a longer median time to diagnosis for EOCRC than for older patients 128 vs 79 days). Whether and how well these findings generalize is unclear. Objective: Compare with EOCRC to older patients with CRC for presentation interval (PI - time from symptom onset to first health care encounter for the symptoms) and diagnostic interval (DI – time from first encounter to CRC diagnosis), and to identify factors associated with a longer DI. METHODS: From an on-going VA-funded, system-wide study on risk factors for sporadic EOCRC, we randomly selected 30 EOCRC cases per year from 2008 to 2015 who were 35-49 years old at diagnosis and matched them 1:1 by year of diagnosis to controls with CRC diagnosed at ages 50-75 years. Subjects with high-risk or whose CRC was discovered by screening were ineligible. From the VA's EMR and national VA databases, we extracted demographic information, duration and type of symptoms, tumor location and stage, and type of initial care setting (primary care, emergency department [ED], other). Patient characteristics between cases of EOCRC and controls were compared with Chi-square tests, t-tests, or Wilcoxon rank-sum tests. Cox proportional hazards regression was used to identify factors associated with longer DI. RESULTS: The 240 EOCRC cases (mean age, 45.2 ± 3.6 years, 62% White) were more likely than the 240 controls (mean age, 63.7 ± 5.8 years, 76% White) to be black (32% vs. 19%; P=0.003), present with specific symptoms (77% vs. 66%; P=0.015), have advanced stage (III-IV) disease (57% vs 49%; P=0.045), and have initial ED contact (61% vs 39%, P=0.007). PIs and DIs, respectively, were present for 74% and 97% of cases and for 66% and 98% of controls. There was no difference in PI between cases and controls (median 30 days for both groups). Cases had a shorter, not longer, DI than controls (median [IQR]: 32[8-76] vs 45[12-101] days; P=0.053). Factors associated with a longer DI were initial contact with a PCP (HR= 1.92; CI, 1.54-2.38), black vs white race (HR=1.42; CI, 1.14-1.78), and early disease stage at diagnosis (HR=1.30; CI, 1.07-1.58). CONCLUSION: Among this national sample of Veterans with CRC, there were no statistical differences in PI and DI between EOCRC cases and controls. EOCRC cases had a numerically shorter DI than did controls. Race, disease stage, and type of initial care setting were independently associated with DI. S-602 AGA Abstracts Su1663 WATER EXCHANGE (WE) ENHANCED DETECTION OF THE MOST ROBUST MARKER OF MALIGNANT POTENTIAL (^10 MM ADENOMA). Felix W. Leung, Hui Jia, Joseph W. Leung Introduction: The prevalence of advanced adenomas ranged from 3.4% to 7.6% [NEJM 2016; 374:1065]. A recent report suggested that identification of non-advanced adenoma might not be associated with increased colorectal cancer (CRC) risk [JAMA 2018;319:2021]. The water exchange (WE) method [GCNA 2013;42:507] embodies the standard features of water infusion to guide intubation to the cecum, gas exclusion to avoid lengthening of the colon, and near-complete removal of the infused water and debris to minimize distension and maximize salvage cleaning during insertion. Gas (air or CO 2 ) is used during withdrawal. 4 recent network meta-analyses each searched multiple data bases to retrieve WE and related randomized controlled trials (RCT). Possibly due to similar reports being analyzed, these meta-analyses consistently showed that WE was superior to water immersion (WI) and gas insufflation (abbreviated as AI) in increasing overall ADR (proportion of patients with at least one adenoma of any size). The impact of WE on ≥10 mm ADR was not addressed in these reports. Since size (≥10 mm) is a more robust marker of CRC risk than histological (villous or dysplasia) features [GCNA 2002;12:1], here we assessed the impact of WE on ≥10 mm ADR. Aim: To highlight the clinical relevance of WE, the hypothesis that WE significantly increased ≥10 mm ADR compared to AI was tested. Method: Data from 8 RCTs (identified in 4 recent network meta-analyses) that compared AI to WE with data on ≥10 mm ADR in the entire or proximal colon were pooled. Seven had data on ≥10 mm ADR; one, only advanced ADR (inclusive of villous or dysplasia features, treated as ≥10 mm ADR in the current report).The primary outcome is combined ≥10 mm ADR. Results: Individual study demographic and procedure-related parameters are shown in Table 1. 6248 patients were randomized to AI (3129) or WE (3119). Compared to AI, WE showed significantly higher ≥10 mm ADR (primary outcome) (4.6% vs. 5.9%, p<0.0198, Fisher's exact test) (Table 2). Discussion: Pooling of the data overcame type II errors in individual RCT of small sample size. The significantly higher ≥10 mm ADR established WE to be superior to AI. The 1.3% increase, albeit numerically small, is a clinically significant 28.3% relative increase. 15 million colonoscopies were performed in the US in 2012. A 1.3% increase would represent 195,000 additional patients with ≥10 mm adenomas detected, if WE were used, with the potential to alter surveillance intervals leading to plausible reduction of CRC incidence in these patients. Limitation: FWL either conducted or trained investigators in WE in all of these RCT. Conclusion: The proof-of-principle analysis showing WE significant increased detection rate of ≥10 mm adenoma (the most robust marker of CRC risk) confirmed WE to be clinically relevant. It is appropriate to include WE in CRC prevention programs. Su1664 HIGH YIELD OF TOTAL AND RIGHT-SIDED COLORECTAL NEOPLASIA BY MULTI-TARGET STOOL DNA TESTING IN AVERAGE RISK PATIENTS IRRESPECTIVE OF PRIOR SCREENING Jason Eckmann, Derek Ebner, Jamie Bering, Allon Kahn, Eduardo A. Rodriguez, Mary E. Devens, Kari L. Lowrie, Karen Doering, Sara Then, Kelli N. Burger, Douglas W. Mahoney, David O. Prichard, Michael B. Wallace, Suryakanth Gurudu, David A. Ahlquist, John B. Kisiel Introduction: Numerous studies have shown that the impact of screening colonoscopy on the incidence and mortality of colorectal cancer (CRC) is substantially lower for right-sided than left-sided disease. A reliable method to enhance detection of right-sided colorectal neoplasia (CRN) could complement colonoscopy with the potential to improve CRC screening outcomes. Aims: To measure the yield of multi-target stool DNA (MT-sDNA) testing for all CRN, right-sided CRN, and CRC in average risk patients, and any association with exposure to prior screening colonoscopy. Methods: In a large multi-setting practice, we identified all patients with positive MT-sDNA tests between 10/1/2014 and 12/31/2017. Patient demo- graphics, personal and family history of CRN, and presence or absence of conditions placing