Gastrointestinal Cancer

often ignored when PKC is claimed to be able to act as a transforming oncogene. Overall the chapters are written in a fashion easily understood by biologists, pharmacologists and biochemists and by students in these disciplines. Each chapter is comprehensively referenced. The editors include also a very helpful glossary at the end of the book. If the book has weaknesses, they are those inevitably associated with multi-author collections of reviews on highly topical scientific issues: (i) there is considerable overlap between the different contributions, and (ii) findings published after 1991 are not included, which will make the book rapidly outdated. The urge for speedy publication is probably responsible for an unusually large number of trivial spelling mistakes. However these minor objections are easily outweighed by the usefulness of the overview provided. Nevertheless one has to agree with the conclusion drawn in chapter 10. 'In spite of a decade of intense effort since the discovery of PKC our knowledge of the cell biology and cellular functions of this ubiquitous family of enzymes is rudimentary.' This is a multi author book covering the field of 'luminal GI cancer'. The two editors and five of the other ten authors are drawn from the staff of the Memorial Sloan-Kettering Cancer Center in New York. The six chapters included in this volume cover Oesophageal tumours, Stomach tumours, Colonic polyps, Colorectal cancers, Pancreatic and biliary tract cancers and Endocrine tumours of the GI tract respectively. In line with the editors specific chosen field of luminal cancers no attempt is made to cover the primary liver tumours although the management of hepatic secondaries from the GI tract is covered in the individual chapters. The volume is well laid out and each chapter follows a clear uniform format covering epidemiology, aetiology, pathology, presentation, diagnosis, treatment and outcome. Where appropriate palliative surgical techniques and methods of pain relief are fully described as in the section on pan-creatic cancer. Potentially confusing areas such as the differences between Duke's original staging of colorectal cancer and the subsequent more commonly used Astler-Coller and Kirklin-Dockerty modifications are clearly and concisely explained. Each chapter is individually referenced with 73 to 222 papers being cited. The references and the content of each chapter are fully up to date including for

Background and Aim: Assays for circulating tumor DNA (ctDNA) biomarkers have shown utility for cancer detection and management. Methylated BCAT1 and IKZF1 are ctDNA biomarkers that have high sensitivity for colorectal cancer (CRC), but it is not clear if these biomarkers are also associated with other common cancers. The aim of this study was to investigate the sensitivity and utility of detecting methylated BCAT1 and IKZF1 in the ctDNA of patients with breast or prostate cancer. Methods: Blood was collected from patients diagnosed with colorectal, breast, or prostate cancer, before receiving any treatment, at Flinders Medical Centre, South Australia. In addition, blood was collected from healthy controls who had been confirmed to have no evidence of CRC at colonoscopy. Circulating cell-free DNA isolated from plasma was bisulfite converted and assayed for methylation in BCAT1 and IKZF1. Logistic regression was used to assess ctDNA test positivity for BCAT1 and IKZF1 DNA methylation compared with healthy controls. Associations between assay positivity and tumor staging were also assessed by Poisson or logistical regression. To further explore the BCAT1 and IKZF1 DNA methylation levels in these cancers, 450 K DNA methylation array data from The Cancer Genome Atlas (TCGA) were assessed for matched tumor and normal samples from the PRAD (n = 50), BRCA (n = 90), and COADREAD (n = 44) cohorts. DNA methylation was interrogated at three out of 20 CpG sites encompassed in the BCAT1/ IKZF1 ctDNA quantitative polymerase chain reaction assay (cg10764357 and cg07589773/cg18607529, respectively). Logistical regression of BCAT1/IKZF1 methylation for the different cancer types and tumor staging was also performed for all tumor cases with American Joint Committee on Cancer staging data (PRAD,503;BRCA,788;and COADREAD,284). Results: The study included 290 blood samples from patients diagnosed with CRC, 32 with breast cancer, 101 with prostate cancer, and 727 healthy controls. Only 7/101 prostate cancer (6.9%) and 3/32 breast cancer (9.4%) blood samples were positive for BCAT1/IKZF1 DNA methylation. The assay positivity rate was not significantly different to that of healthy controls (8.0% for females vs breast cancer; odds ratio [OR], 0.84; 95% CI, 0.23-3.1; and 7.6% for males vs prostate cancer; OR, 0.86; 95% CI, 0.65-2.1). In patients with CRC, the assay had a 60.3% positivity rate (175/290), which was significantly higher than that in healthy controls (7.7%, 56/727; OR, 18.2; 95% CI, 11.1-29.0). A higher assay positivity rate was observed in more advanced CRC disease, with overall stage IV cases having an assay positivity of 90.7% (39/43) compared with 23.7% (14/59) for stage I (OR,21.9;95% CI,. TCGA data showed that BCAT1 and IKZF1 were significantly hypermethylated in prostate and colorectal tumors compared with matched normal tissues (P < 0.0001) but were no different for breast tumors versus normal tissue (P = 0.854 for BCAT1 and P = 0.172 for IKZF1). There was no correlation between cancer staging and level of tissue DNA methylation in IKZF1 and BCAT1 for breast, prostate, or colorectal cases in the TCGA dataset. Conclusion: Circulating tumor DNA methylated in BCAT1 and IKZF1 is specific for CRC and is not sensitive for detection of breast or prostate cancer, despite a subset of the targeted CpG sites in these genes being hypermethylated in prostate tumors. This supports the use of methylated BCAT1 and IKZF1 for CRC detection and monitoring.

42
Upper gastrointestinal ulceration after transarterial chemoembolization or radioembolization: A retrospective cohort study M RAVAL,* A RAJADURAI,* HJ SLIFIRSKI, † T WORLAND,* L ZORRON CHENG TAO PU,* M EFTHYMIOU,* , † R VAUGHAN,* , † S CHANDRAN* , † *Department of Gastroenterology and Hepatology, Austin Health, † University of Melbourne, Melbourne, Victoria, Australia Background and Aim: Hepatocellular carcinoma accounts for the majority of primary liver cancers, with a small proportion amenable to curative resection. Palliative treatments include transarterial chemoembolization (TACE) and transarterial radioembolization (TARE). TACE and TARE can be complicated by gastrointestinal ulceration. To date, only small case series have reported the incidence of upper gastrointestinal ulcers after TACE or TARE. The aim of this study was to determine the incidence of upper gastrointestinal ulcers following TACE or TARE. Methods: Patients receiving TACE or TARE at a tertiary Australian hospital from January 2012 to April 2020 were retrospectively assessed. Patients were included if they received TACE or TARE for liver malignancy and had an upper endoscopy within 12 months of the TACE/TARE. The presence of upper gastrointestinal ulceration was determined by review of the endoscopy report. Demographic data were retrieved, in addition to medical history and procedural variables.
Results: A total of 472 patients received an endoscopy within 12 months of their TACE or TARE procedure; the timing of endoscopy was 0-3 months in 112 patients (23.7%), 3-6 months in 213 patients (45.3%), and 6-12 months in 146 patients (30.9%). The overall incidence of upper gastrointestinal ulceration was 6.6% (n = 31), with 54.8% (n = 17) detected at 0-3 months after TACE/TARE, 29.0% (n = 9) detected within 3-6 months, and 16.1% (n = 5) detected within 6-12 months. Eighteen patients (58.1%) had gastric ulcers, and 15 patients (48.4%) had duodenal ulcers. There were no esophageal ulcers, and two patients had concurrent gastric and duodenal ulcers. Overall, one case (3.2%) was histologically confirmed to be induced by beads. In 21 patients (67.7%), ulceration was attributed to TACE/TARE as no association was found with Helicobacter pylori or the use of anti-inflammatories or other medications. Most ulcers (26, 83.9%) were Forrest class III, with small numbers of higher-risk ulcers identified. No endoscopic therapy was required. Conclusion: Upper gastrointestinal ulceration is not uncommon in patients undergoing endoscopy after TACE or TARE. While TACE/TARE-induced ulceration was only histologically confirmed in one case, it was suspected as the most likely etiology in two-thirds of patients. Further prospective studies are needed to identify the true incidence of TACE/TARE-induced gastrointestinal ulceration.

70
Evaluation of the use of screening and risk-reducing surgery for gynecological cancer in patients with Lynch syndrome: A review N LIM University of Melbourne, Melbourne, Victoria, Australia Background and Aim: Lynch syndrome is a hereditary cancer syndrome caused by mutations in mismatch repair (MMR) genes or EPCAM. Risk of multiple cancers, including endometrial cancer and ovarian cancer, is increased in patients with Lynch syndrome. Screening or risk-reducing surgery is used to prevent gynecological cancers. However, practice varies internationally due to limited evidence. The aim of this study was to review the literature on the use of screening and risk-reducing surgery for gynecological cancer in patients with Lynch syndrome. Methods: Searches of the MEDLINE, Embase, and PubMed databases were conducted. Studies reporting the incidence of cancer in screened asymptomatic women, or histopathological results from prophylactic hysterectomy and/or salpingo-oophorectomy, were included. Results: Eighteen studies that screened for endometrial cancer using transvaginal ultrasound, hysteroscopy, or endometrial biopsy showed an incidence of 3.9% at the time of screening. Screening detected 59.5% of endometrial cancers in MMR gene carriers, with the balance detected during intervals or through symptoms. The sensitivity of endometrial screening was 71.0%, and the number needed to screen (NNS) was between four and 27 (median, 6). Twelve studies that screened for ovarian cancer using transvaginal ultrasound and/or CA-125 testing showed an incidence of 1.3% at the time of screening. Screening diagnosed 36% of ovarian cancers in MMR gene carriers. The sensitivity of ovarian cancer screening was 47.4%, and the NNS was between nine and 191 (median, 23). Nine studies reporting prophylactic specimen results after risk-reducing surgery diagnosed endometrial cancer in 6.9% and ovarian cancer in 0.8%. Conclusion: There is limited evidence to support screening for gynecological cancers in patients with Lynch syndrome. Further prospective randomized trials comparing targeted screening methods for gynecological cancers would be useful. At present, risk-reducing surgery remains the mainstay in preventing the development of gynecological cancer. Methods: We first estimated sex-and MMR gene-specific cumulative risk of first CRC without colonoscopic surveillance using an optimization algorithm. We then used the CRC risk estimates in a microsimulation model, "Policy1-Lynch," to model ongoing colonoscopic surveillance in confirmed LS carriers. A total of 126 surveillance strategies, comprising combinations of start age, end age, and surveillance interval, were compared against no surveillance. Results: Considering all strategies, we found the following. Starting surveillance at age 25 years, rather than 20 years, was more cost-effective without changing life-years saved (LYS), and starting 5-10 years later for MSH6/PMS2 carriers further improved the cost-effectiveness. Surveillance end age (70, 75, or 80 years) had a relatively minor effect. Three-yearly surveillance strategies were more cost-effective than 1-or 2-yearly surveillance strategies but prevented three fewer CRC deaths per 1000 LS carriers. Finally, less intensive surveillance after the age of 60 years had minimal impact on LYS, but less intensive surveillance up to the age of 40 years resulted in substantially fewer LYS. Accordingly, the most cost-effective strategy was 3-yearly surveillance from the age of 25 to 70 years (MLH1/MSH2), with delayed surveillance for carriers of MSH6 (age 30-70 years) and PMS2 (age 35-70 years) (incremental cost-effectiveness ratio, A$8833/LYS). Compared with no surveillance, this strategy averted 60 CRC deaths and involved 9206 colonoscopies (153 colonoscopies/death averted) over the lifetime of 1000 confirmed LS carriers. Conclusion: This is the first formal evaluation of MMR gene-specific colonoscopic surveillance. We found that a tailored approach would be effective and cost-effective. Ongoing emerging data will reduce the uncertainty of the lifetime risk estimates and inform ongoing discussions about optimal surveillance. Background and Aim: Colonoscopic surveillance is undertaken at regular intervals (typically 3 or 5 years) to reduce the incidence of colorectal cancer (CRC) in people with an elevated risk (personal or family history of neoplasia). Pathology findings at index and each surveillance colonoscopy determine recall intervals. Due to coronavirus disease 2019, hospital services around the world have been limited, resulting in some surveillance colonoscopies being delayed beyond the recommended time frames. Previous studies suggest that delays to colonoscopy might increase the incidence of advanced neoplasia (advanced adenoma/CRC). However, it is possible that this risk could be reduced by ensuring that individuals are maintained in a CRC screening program with an immunochemical fecal occult blood test (FIT) in the interval between surveillance colonoscopies. Our aim was to determine whether risk of advanced neoplasia increases if surveillance colonoscopy is delayed in people with elevated CRC risk who perform and have a negative FIT result in the interval between colonoscopies. Methods: We performed a retrospective cohort study using data from the Southern Cooperative Program for the Prevention of Colorectal Cancer on people at elevated risk because of family history or personal history of adenoma or CRC. People with at least two consecutive colonoscopies of a 3-or 5-year surveillance interval and who had at least one negative interval FIT result were included in the study. They were stratified based on a previous colonoscopy finding of advanced adenoma, non-advanced adenoma, or no neoplasia. People with early colonoscopies (3 months before the recommended due date), poor bowel preparation, previous CRC, or hereditary CRC syndromes were excluded from the study. Colonoscopy was defined as "delayed" if it did not occur within 6 months after the recommended recall interval and was further subdivided into delays of 6-12, 12-24, and >24 months. The incidence of advanced neoplasia was calculated for all groups. The relative risk (RR) and 95% confidence intervals estimated from a robust multivariable modified Poisson regression were used to assess the association between surveillance colonoscopy delay and risk of advanced neoplasia. Results: A total of 1748 public hospital surveillance colonoscopies (in 1516 participants) were included in the analysis. More than half of the colonoscopies (56.86%, 994/1748) were delayed by at least 6 months because of system and/or patient factors. In people with delayed colonoscopies, the incidence of advanced neoplasia was higher in those with previous advanced adenoma (16.72%, 48/287) and previous non-advanced adenoma (15.23%, 37/243) compared with those with no neoplasia (6.25%, 29/464) (P < 0.001). However, relative to on-time colonoscopy, delay of surveillance colonoscopy was not associated with an increased risk of advanced neoplasia for people who had at least one negative interval FIT result, regardless of previous colonoscopy finding (previous advanced adenoma: RR, 1.01; 95% CI, 0.70-1.46; non-advanced adenoma: RR, 1.41; 95% CI, 0.85-2.33; and no neoplasia: RR, 0.96; 95% CI, 0.55-1.66) ( Table 1). Conclusion: In an elevated-risk cohort undergoing FIT screening between surveillance colonoscopies, delays to colonoscopy did not increase risk of advanced neoplasia. These results suggest that surveillance colonoscopy could be safely extended in people at elevated CRC risk by participating in FIT testing between colonoscopies within a surveillance program.

93
Is it appropriate to stop bowel cancer surveillance at the age of 75 years? M AGACIAK,* K SIMPSON, ‡ C COCK,* , ‡ P BAMPTON,* , ‡ R FRASER,* , ‡ EL SYMONDS †, ‡ *College of Medicine and Public Health, † Flinders Health and Medical Research Institute, Flinders University, ‡ Southern Cooperative Program for the Prevention of Colorectal Cancer, Flinders Medical Centre, Adelaide, South Australia, Australia Background and Aim: Increasing age is a well-known risk factor for colorectal cancer (CRC); however, the 2018 Clinical Practice Guidelines for Surveillance Colonoscopy suggest that surveillance colonoscopy in people aged ≥75 years may be of little benefit. The literature suggests a more complex picture, whereby age is not the only deciding factor, but factors such as comorbidities and life expectancy may also need consideration. By establishing the risk factors for advanced neoplasia (AN; CRC or significant adenoma), surveillance can be targeted to those where the benefits of neoplasia detection and removal will outweigh the risks of colonoscopy. We aimed to determine the incidence and predictors of AN in people aged ≥75 years undergoing surveillance colonoscopy. Methods: This was a retrospective analysis of surveillance colonoscopy results between 2015 and 2020 in people aged ≥75 years who were enrolled in the Southern Cooperative Program for the Prevention of Colorectal Cancer (a surveillance program for people with an elevated risk of CRC, due to a personal history of neoplasia or a significant family history of CRC). Pathology details of the surveillance colonoscopy were recorded, along with demographic details, family and personal history of CRC, total number of colonoscopies, and the previous colonoscopy date and findings. Colonoscopies in patients with inflammatory bowel disease and those with poor quality or indeterminate pathology outcomes were excluded. Pathology  (IQR,(1)(2)(3)(4), and the median number of years since the most recent colonoscopy was 3.13 (IQR, 1.1-5). AN was found in 35.95% of the surveillance colonoscopies, including a finding of CRC in 1.58% (11/698) and significant adenoma in 34.38% (240/698). The primary independent factor for a finding of significant adenoma was a significant adenoma at previous colonoscopy. Previous CRC was the only independent predictor of CRC, with eight of the 11 detected cancers being a recurrence or metachronous CRC. Age, sex, and a family history of CRC were not associated with an increased risk for AN. Conclusion: In individuals aged ≥75 years undergoing surveillance colonoscopy, one-third of individuals were diagnosed with significant adenoma, with risk being greatest if the previous colonoscopy finding was a significant adenoma. Furthermore, a previous CRC increased the risk of a future CRC in this population. These results will assist in decision making for clinicians regarding scheduling of surveillance colonoscopies in patients aged ≥75 years.

143
Trends in obesity-associated gastrointestinal cancers in Australia: An analysis of population-based data, 1985-2014 E DAVIS* , † *Princess Alexandra Hospital, † University of Queensland, Brisbane, Queensland, Australia Background and Aim: International studies have identified increasing incidence rates of cancer, particularly obesity-associated cancers, in young adults. Excess body fat has been associated with increased risk of several gastrointestinal cancers, including esophageal adenocarcinoma and cancers of the gastric cardia, colon, rectum, liver, and pancreas. Increasing rates of colorectal cancer in young adults have been examined in the Australian population, but no studies have investigated age-specific trends for a wider range of obesity-associated gastrointestinal cancers. We aimed to examine temporal trends in incidence rates of six obesity-associated gastrointestinal cancers in the Australian adult population from 1985 to 2014. Methods: We obtained cancer incidence data for 1985 to 2014 from the Australian Cancer Database for people aged 20-84 years, for colon, rectum, liver and intrahepatic bile duct, and pancreatic cancer. Histological subtype data necessary for analyses of esophageal adenocarcinoma and gastric cardia cancer were available only from 1985 to 2012. Joinpoint regression was used to identify trends in age-standardized and age-specific incidence rates per 5-year age group, stratified by sex. We used age-period-cohort modeling to estimate average annual percentage change in incidence rates for each 5-year age group and incident rate ratios (IRRs) by 10-year overlapping birth cohorts and per 5-year periods.
Results: The greatest increase in age-standardized incidence rates over the study period was seen in liver and intrahepatic bile duct cancer (4.9% per year; 95% CI, 4.7-5.0%) and esophageal adenocarcinoma (4.2%; 95% CI, 3.5-5.0%). Age-standardized incidence rates were stable for rectal cancer (À0.1% per year; 95% CI, À0.4% to 0.2%) and colon cancer (À0.3%; 95% CI, À0.6% to 0.0%). We observed significantly increasing age-specific rates of colon cancer in younger age groups (5.2% per year for ages 20-24 years; 95% CI, 4.4-6.0%), but falling rates among those aged 40-70 years. Similarly, age-specific incidence of rectal cancer increased in younger cohorts but remained static in those aged over 50 years. Age-specific rates of liver and intrahepatic bile duct cancer showed the greatest increase in those aged 45-60 years (5.4-6.5% per year) and 75-85 years (5.9-6.1%), while IRRs increased for progressively younger age groups. Age-specific rates of pancreatic cancer rose across all age groups for the study period, with a small increase in IRRs for younger cohorts. There was a small trend towards increased rates in younger age groups for gastric cardia cancer, whereas age-specific rates increased relatively uniformly for esophageal adenocarcinoma. Conclusion: In Australia, rates of liver and intrahepatic bile duct cancer and esophageal adenocarcinoma have significantly increased over the past 30 years. Colon, rectal, liver and intrahepatic bile duct, pancreatic, and gastric cardia cancer all showed a trend towards increased rates in younger cohorts. The obesity epidemic is potentially a significant contributor, although further studies are warranted to investigate these trends. Cases with missing SES data were excluded from the analysis. We used the patients' Socio-Economic Indexes for Areas (SEIFA) scores, which were categorized into quintiles and then reclassified for analysis into "low SES," including SEIFA 1 to SEIFA 3, and "high SES," including SEIFA 4 and SEIFA 5. The chi-squared test was used to analyze etiology of liver disease and the presence of metabolic risk factors and specific comorbidities within low-and high-SES groups. The log-rank test was used to compare survival in low-and high-SES groups. To determine the impact of these variables on survival, we performed Cox univariate and multivariate regression survival analysis in low-and high-SES groups. All statistical analyses were performed with IBM SPSS Statistics 26.
Results: Of 5609 patients with HCC, 5339 were included in the analysis (low SES, 3886; high SES, 1453). The average age at HCC diagnosis was 64.6 years, and most patients (80.2%) were men. There were no differences between low-and high-SES groups in underlying cause of liver disease or demographic variables such as age, sex, ethnicity, metabolic risk factors, and comorbidities, including cardiac disease, chronic kidney disease, and cerebrovascular incidents. However, those with high SES had better survival than those with low SES (median, 6.0 vs 5.0 months; hazard

156
BRAF mutation induces rapid neoplastic transformation in the aged and aberrantly methylated intestinal epithelium L FENNELL,* A KANE,* , †, ‡ C LIU,* , † D MCKEONE,* G HARTEL,* , † C SU,* , † C BOND,* , † M BETTINGTON, †, § B LEGGETT,* , †, ¶ V WHITEHALL* , †, ‡ *QIMR Berghofer, † University of Queensland, ‡ Conjoint Internal Medicine Laboratory, Pathology Queensland, § Envoi Specialist Pathologists, ¶ Royal Brisbane and Women's Hospital, Brisbane, Queensland, Australia Background and Aim: Sessile serrated lesions (SSLs) of the colorectum are common in both young and older individuals, but the BRAF mutant cancers arising from SSLs occur predominantly in older people. Aberrant DNA methylation changes drive progression of SSLs, and these changers are uncommon in SSLs from young patients. We aimed to determine whether DNA methylation changes acquired during the aging process increase the risk of spontaneous neoplastic progression in older people.

Methods:
We used an inducible model of BRAF mutation to direct recombination of the oncogenic BRAF V637E allele to the murine intestine. BRAF mutation was activated in mice at weaning or after aging for a period of 9 months. After a further 5 months of aging in each group, histological, DNA methylation, and gene expression analysis was performed. We used an organoid model to test the impact of DNA methylation inhibition on growth and the acquisition of cytological dysplasia. We also investigated DNA methylation alterations in human SSLs.
Results: Inducing BRAF mutation in aged mice was associated with a 10-fold relative risk of serrated lesions compared with young mice in which we induced the BRAF mutation for the same period of 5 months. Methylation analysis showed extensive differences in age-associated DNA methylation between animals induced at 9 months versus induction at weaning, with relatively little differential BRAF-specific methylation. DNA methylation at WNT pathway genes scaled with age, and BRAF mutation accelerated age-associated DNA methylation. Treating intestinal organoids derived from aged mice with a DNA methyltransferase inhibitor (azacitidine) before inducing the BRAF mutation significantly reduced growth ( Fig. 1a) and prevented organoids acquiring dysplasia (Fig. 1b).
In human SSLs, increased epigenetic age was associated with high-risk serrated colorectal neoplasia. Conclusion: SSLs arising in the aged intestine are at a significantly higher risk of spontaneous neoplastic progression. These findings provide support for a new conceptual model for serrated colorectal carcinogenesis, whereby risk of BRAF-induced neoplastic transformation is dependent on age and may be related to age-associated molecular alterations that accumulate in the aging intestine.

157
Natural history and management of Barrett's esophagus with true low-grade dysplasia: A 12-year Barrett's esophagus referral unit experience T HE,* , † E TSOI,* , † BA HOLT,* , † P DESMOND,* , † A TAYLOR* , † *St Vincent's Hospital Melbourne, † University of Melbourne, Melbourne, Victoria, Australia Background and Aim: Barrett's esophagus with low-grade dysplasia (LGD) is a diagnostic and management challenge, as the natural history  and optimal treatment are not well described.
LGD, as diagnosed by expert gastrointestinal pathologists, is a predictor for progression to high-grade dysplasia (HGD) and/or adenocarcinoma. 1 In this setting, endoscopic resection or ablation of Barrett's esophagus with LGD is becoming more common. 2 However, surveillance also remains an appropriate option for some patients. 3 We describe an Australian Barrett's esophagus referral unit's experience of managing patients with Barrett's esophagus with true LGD and their outcomes, with the aim to define appropriate treatment strategies for this group of patients.
Methods: This is a 12-year retrospective review of a prospectively maintained database (Microsoft Access) of patients with Barrett's esophagus with true LGD. True LGD was defined as confirmation of LGD by expert gastrointestinal pathologist review of referral and expert center histology. Clinical and histopathological data collected included indications for surveillance or endoscopic therapy (endoscopic mucosal resection [EMR], radiofrequency ablation [RFA], or both), procedural complications, interval endoscopy timing, histology, and date of last endoscopic follow-up. Results: Seventy-four patients (median age, 68 years; IQR, 61-75; 84% male) had true LGD. Four patients with referral histology reported as non-dysplastic Barrett's epithelium were upgraded to LGD by expert gastrointestinal pathologist review, and eight patients with referral histology of HGD were downgraded to LGD. In the 74 patients with LGD, the management strategy was surveillance in 10, RFA alone in 26, EMR alone in 11, and combination EMR-RFA in 27. The most common indication for a surveillance strategy was focal LGD (LGD in one biopsy; 8, 80%), while the most common indication for RFAwas multifocal LGD (49, 92.4%), and for EMR, it was visible mucosal abnormality (24, 63.1%). The most common complication after endoscopic therapy was stricture formation, which occurred in 13 patients (17.6%); all resolved with endoscopic dilatation. The median follow-up time for the cohort was 3.6 years (IQR, 2.5-4.2). Four patients (5.4%) had disease progression, to HGD in two, intramucosal cancer (IMC) in one, and T1bN0M0 adenocarcinoma in one, with a median length of follow-up before progression of 8.5 years (IQR, 5-9.5). All patients who had disease progression had undergone prior combination EMR-RFA therapy. Two of these patients had been referred originally with HGD before being downgraded to LGD after expert gastrointestinal pathologist review. The three patients who progressed to HGD or IMC were treated by EMR, with no persistence or recurrence of dysplasia at last follow-up. The one patient diagnosed with T1bN0M0 adenocarcinoma after 8 years of follow-up underwent successful definitive chemoradiotherapy and remained cancer free after 8 months of follow-up. Of the 60 patients who underwent endoscopic therapy (EMR, 11; RFA, 26; and EMR-RFA, 23) and did not have disease progression, only six (10%) had recurrent LGD and 16 (26.7%) had non-dysplastic Barrett's epithelium at last follow-up. Ten patients with LGD had a surveillance strategy, with no disease progression after a median of 2 years (IQR, 1-7). Three patients under surveillance eventually had RFA after a median of 2 years (IQR, 1-7), for multifocal LGD in two and patient preference in one. Conclusion: In our cohort, management of Barrett's esophagus and true LGD was individualized. Focal LGD was the primary indication for surveillance, and multifocal LGD and visible mucosal abnormality were reasons for RFA and EMR, respectively. Disease progression was seen in a small proportion of patients. Potential risk factors for disease progression included HGD on referral histology (although downgraded by a Barrett's referral unit expert pathologist) and disease requiring more intense endoscopic therapy (EMR and RFA), suggesting a more treatment-refractory or multifocal dysplastic group. A minority of patients (13.5%) were suitable for surveillance alone, and no disease progression was seen in this group. Larger studies with longer-term follow-up are required to better guide our understanding of the natural history of Barrett's esophagus with LGD, with the ultimate aim of improving management guidelines.  Background and Aim: Regorafenib is a multikinase inhibitor approved for treating refractory metastatic colorectal cancer. Previous studies have suggested that combining kinase inhibitors with aspirin may improve patient outcomes. We recently found that aspirin and regorafenib interact synergistically to inhibit proliferation of colorectal cancer cells in vitro. Therefore, we aimed to determine the efficacy of aspirin and regorafenib combination treatment in an ex vivo patient-derived organoid model and an in vivo cell line-derived xenograft model. We also aimed to identify potential mechanisms underlying the synergistic antiproliferative effect. Methods: Organoids were established from three human colorectal cancer tissue specimens and treated with 1 mM aspirin and 4 μM regorafenib. Cell viability was measured after 72 h using the Presto Blue assay. For the in vivo study, 5 × 10 6 SW480 colorectal cancer cells were injected subcutaneously into dorsal flanks of female athymic BALB/c-Foxn1 nu mice.
Once average tumor volume reached 200 mm 3 , the mice were treated by oral gavage with aspirin (100 mg/kg/day) and/or regorafenib (10 mg/kg/ day) or vehicle only for 15 days. Immunohistochemical analyses were subsequently performed to measure expression of Ki67 and cleaved caspase 3.
To explore underlying molecular effects, SW480 cells were also treated with aspirin and regorafenib in vitro for RNA sequencing.
Results: Combination treatment with aspirin and regorafenib reduced the viability of patient-derived organoids compared with the untreated control (P < 0.001), to a greater extent than either aspirin (P < 0.05) or regorafenib treatment (P < 0.05) alone. Analysis using the Bliss independence model indicated a synergistic interaction between aspirin and regorafenib. Combination treatment also reduced the growth of SW480-derived xenografts in vivo (P < 0.01), to a greater extent than either aspirin (P < 0.001) or regorafenib treatment (P < 0.001) alone ( Fig. 1). Expression of Ki67 was reduced in tumors from the combination-treatment group compared with the regorafenib single-treatment group (P < 0.05), indicating a greater antiproliferative effect. Expression of cleaved caspase 3 was increased with combination treatment (P < 0.0001), to the same extent as with regorafenib single treatment (P > 0.05), indicating a proapoptotic effect driven by regorafenib. RNA sequencing of cells treated with aspirin and regorafenib in vitro showed downregulation of various proliferative pathways and downregulation of the PI3K-Akt-mTOR signaling pathway.
Conclusion: Aspirin and regorafenib show synergistic antiproliferative effects in preclinical models of colorectal cancer. Inhibition of the PI3K-Akt-mTOR pathway may mediate these effects. However, further molecular studies are required for better characterization of the underlying mechanisms. These findings provide a rationale for investigating aspirin and regorafenib combination therapy as a potential treatment strategy for patients with refractory metastatic colorectal cancer.

163
Influence of birthplace on colonoscopy outcomes after positive fecal occult blood test result in a culturally and linguistically diverse population MA GULIYARA,* , † L WILLMANN,* , † B AL SHIWANNA,* , † D ASTON,* , † S GEORGE,* , † JH KOO* , † *Liverpool Hospital, † University of New South Wales, Sydney, New South Wales, Australia Background and Aim: Colorectal cancer (CRC) screening with the fecal occult blood test (FOBT) improves survival. The influence of birthplace on the outcomes of a patient's colonoscopy after a positive FOBT result in a culturally and linguistically diverse (CALD) population is largely unknown. This study aimed to determine differences in these outcomes, as set by the Australian Colonoscopy Clinical Care Standards, among a CALD population compared with Australian-born residents. Methods: We prospectively collected data from patients having a colonoscopy after a positive FOBT result at Liverpool Hospital from September 2016 to March 2020. Data included birthplace, preferred spoken language, need for interpreter service during clinical assessments, previous colonoscopy, family history of CRC, adequacy of bowel preparation, and the results of the colonoscopy. The primary outcomes were any differences in the prevalence of CRC, advanced adenomas (size >1 cm, villous component, or high-grade dysplasia) and adenoma, adequacy of bowel preparation, cecal/terminal ileum intubation, and complication rates, according to patients' CALD status. Secondary outcomes included demographic differences within these groups. Statistical analysis was performed using SPSS using the chi-squared test for categorical data.

Conclusion:
The primary colonoscopy outcomes of CRC, advanced adenoma and adenoma detection rates, cecal/terminal ileum intubation rate, and complication rate after a positive FOBT result were within recommended guidelines among the CALD population and comparable with Australian or New Zealand-born patients. However, Middle Eastern patients had the poorest bowel preparation and the least female participants. These findings may assist with the implementation and allocation of resources to improve the quality of bowel preparation and the uptake of colonoscopy among selected subgroups of the CALD population. Background and Aim: Family history of colorectal cancer (CRC) is used in Australian screening guidelines to stratify individuals into risk categories warranting more intensive CRC screening than the general population. Although there is evidence that colonoscopy reduces the risk of subsequent development of CRC in average-risk individuals, data regarding enduring risk after colonoscopies with no neoplastic findings at index colonoscopy in people with a family history of CRC are limited. This study aimed to determine the incidence of advanced neoplasia (advanced adenoma or CRC), as well as the associated family history-classified by 2017 National Health and Medical Research Council (NHMRC) categories-in a population undergoing regular surveillance colonoscopies due to family history of CRC. Methods: A retrospective audit was undertaken of individuals enrolled in the Southern Cooperative Program for the Prevention of Colorectal Cancer surveillance program due to a family history of CRC. Individuals with a normal index colonoscopy and at least 10 years of subsequent surveillance colonoscopy follow-up were included in the analysis to determine risk of Figure 1 Mean (+SD) tumor volume of SW480-derived xenografts (n = 20) over 15 days of treatment with aspirin (100 mg/kg/day) and/or regorafenib (10 mg/kg/day). Statistical analysis was performed using a two-way ANOVA and Tukey's post hoc test (**P < 0.01 and ***P < 0.001).
advanced neoplasia under surveillance. Advanced neoplasia included adenocarcinomas, adenomas with features of villous change, size ≥10 mm, high-grade dysplasia, five or more small tubular adenomas, and sessile serrated adenomas with dysplasia or size ≥10 mm. Patients were classified into risk groups, according to the 2017 NHMRC family history screening guidelines, as Category 1 (near average risk), Category 2 (moderately increased risk), or Category 3 (high risk), on the basis of their family history demographics. Risk of advanced neoplasia was compared between risk groups, demographics, and family history characteristics using chi-squared and nonparametric Kruskal-Wallis tests. Hazard ratios were calculated for advanced neoplasia among different demographic and risk categories.
Results: Of 425 patients undergoing surveillance colonoscopies, 30.6% had NHMRC Category 1 risk, 59.8% had Category 2 risk, and 9.7% had Category 3 risk. The incidence of advanced neoplasia in the overall cohort was 14.4% (61/425). The median follow-up time overall was 12.7 years. There was no statistically significant difference between the incidence of advanced neoplasia during the surveillance period between the different family history category risk groups (Category 1 risk group, 15 cases per 1000 person-years; Category 2 risk group, 9 cases per 1000 person-years; and Category 3 risk group, 9.3 cases per 1000 person-years; P = 0.16).
The number of affected family members and the age at diagnosis of CRC in the youngest affected family member did not affect the risk of developing a significant finding in the surveillance population (P > 0.05). Patients aged ≥60 years showed a threefold higher risk (hazard ratio, 2.78; 95% CI, 1.02-7.54) of advanced neoplasia compared with patients aged less than 40 years at first colonoscopy (P = 0.04). The risk of advanced neoplasia after normal surveillance colonoscopies was low: 85.6% of those without neoplasia at index colonoscopy did not develop advanced neoplasia at any time during the surveillance period; 91.2% of those without neoplastic findings at the index and first surveillance colonoscopy did not develop advanced neoplasia at any time during the surveillance period; and 94.1% of those without neoplastic findings at index colonoscopy and two surveillance colonoscopies did not develop advanced neoplasia at any time during the surveillance period. Conclusion: Overall, there was a low risk of advanced neoplasia in the family history cohort, regardless of the type of family history risk category, with age ≥60 years at index colonoscopy being the only significant risk factor for future development of advanced neoplasia. These results suggest that family history may not pose as significant a risk of advanced neoplasia as previously thought and that regular colonoscopies may not be required, particularly if an individual has had at least one normal colonoscopy. Delaying age of onset of colonoscopy screening while providing regular fecal immunochemical testing could be a more efficient use of resources in this population. • health-care providers to manage their patients' participation in the NCSP and NBCSP, including opting out and deferring their screening; • health-care providers to access screening information for participants of both programs (including those who have been invited to participate but have not yet done so); and • general practitioners and gastroenterologists to lodge clinical program forms electronically.
The NCSR platform is integrated with market-leading clinical information systems in the field of primary care, and integrations with vendors providing endoscopy software are planned for later in 2021. Since the transition of the register to the NCSR, the Australian Government has commenced biennial screening, with the register supporting the program to issue home test kits to eligible participants every 2 years. Work is progressing with several laboratories to codesign functionality for capturing bowel histopathology data via HL7 messaging, to improve consistency of reporting and data quality and increase the histopathology outcome data for reporting purposes. Improved data quality and completeness-of-program data will be securely shared electronically via a self-service business intelligence tool accessible by the Department of Health and state and territory governments. Several preconfigured reports will be available, including data on positivity rates, participation in the program, and public versus private colonoscopy.

185
Examining microsatellite instability as a biomarker Background and Aim: Microsatellite instability (MSI) is implicated in 10-15% of colorectal cancers (CRCs) and arises as a result of defective DNA mismatch repair (MMR) systems. It is well established that the mainstay of CRC chemotherapy, 5-fluorouracil (5-FU), shows little to no survival benefit in patients with MSI tumors. However, novel in vitro findings from the Conjoint Gastroenterology Laboratory in Queensland have suggested that MSI could be used as a biomarker for sensitivity to a chemically similar derivative of 5-FU, trifluridine-tipiracil (TAS-102). In this project, we aimed to further explore TAS-102 sensitivity and hypothesized that this drug would induce a more potent antitumor effect in preclinical models of MSI CRC, for which 5-FU is predicted to be ineffective.
Methods: Several CRC cell lines were used to investigate TAS-102 sensitivity in vitro. The CRISPR Cas9 system was used to induce MSI in otherwise microsatellite-stable (MSS) HT29 and SW480 cell lines. A sequence-specific guide RNA was used in conjunction with a CRISPR Cas9 nuclease to knockout the MLH1 MMR gene, which is silenced in most MSI CRCs. Sanger sequencing, whole-exome sequencing, Western blotting, and immunohistochemistry were used to validate the MLH1 knockout and the subsequent induction of MSI. Response to TAS-102 and 5-FU was compared across both MSI (CRISPR Cas9 MLH1 knockout HT29 and SW480) and syngeneic MSS (wild-type HT29 and SW480) cell lines. Drug response was first measured by MTS assay and further characterized by flow cytometry. Ki67 and annexin V stains were used to examine the drugs' effects on cell proliferation and apoptosis, respectively. Results: MLH1 was successfully knocked out of both HT29 and SW480 cell lines. Sanger sequencing identified both heterozygous and homozygous mutations in the MLH1 gene in clonal populations of each cell line after CRISPR Cas9 transfection. Relative MLH1 protein expression decreased to 20% in heterozygous knockouts and 6% in homozygous knockouts. MTS assays indicated that HT29 CRISPR Cas9 MLH1 knockout cell lines maintain greater TAS-102 sensitivity over syngeneic wild-type cell lines (HT29 MSI IC50, 4.0 μM; HT29 MSS IC50, 23.8 μM; P < 0.05). Conclusion: These preclinical data indicate that cell-line models of MSI CRC are more sensitive to TAS-102 than 5-FU. This novel in vitro finding has generated important preclinical data that suggest TAS-102 could provide an alternative treatment for MSI CRC that is not likely to respond to 5-FU-based treatments. Thus, there is the potential to use MSI as a biomarker for sensitivity to TAS-102. The next phase of this project will progress to in vivo analyses of TAS-102 sensitivity in MSI and MSS cell line-derived xenograft murine models. Background and Aim: Australians are dying at a rate of 10 people per day from colorectal cancer (CRC). Rapid strides continue to be made with immunotherapy drugs and combination treatments for a variety of cancers. That these approaches have largely failed in improving outcomes for the majority of patients with advanced microsatellite-stable CRC highlights that a new approach is needed. The microbiome plays a key role in health and disease and response to immunotherapies. Although most of our understanding of this microbial ecosystem in the gut is formed from sequencing fecal samples, there is now also recognition that tissues and different tumor types have their own distinct microbiota. Over the past decade, convincing preclinical studies have also shown that, when administered to mice, certain bacteria, such as the probiotic Escherichia coli Nissle 1917 (EcN), selectively colonize tumors rather than normal tissue. However, this has yet to be investigated in humans. Together with synthetic biology experts in our team, we are developing EcN via genetic circuit engineering as an early CRC biodetector or for localized delivery of therapeutic payloads to treat cancer. Our aim is to determine whether the marked, selective bacterial growth of EcN in neoplastic tissue is preserved in humans-a critical first step in the translation of this approach to patients. Methods: Thirty-eight patients undergoing colonoscopy or surgical resection for primary CRC at the Royal Adelaide Hospital or St Andrew's Hospital were recruited and assigned to take either two tablets (10 9 CFU) daily of non-genetically modified EcN (Mutaflor) or placebo for 14 days before their procedure. Mucosal biopsy samples (from colonoscopy) or surgical resection samples from normal and neoplastic tissue were collected and homogenized, and microbes were enriched by overnight liquid culture before EcN-specific polymerase chain reaction (PCR) detection assay to quantitate EcN load per tissue.
Results: After first determining that the EcN nested high-sensitivity quantitative PCR assay designed for use in mouse studies was not suitable for use in patient samples, due to detection of false-positive signals from non-EcN-administered patients, we redesigned, optimized, and validated a human sample-compatible assay to specifically detect EcN. We found that enrichment of EcN via bacterial culture before analysis also increased sensitivity of the quantitative PCR detection assay from human tissue samples. In the final set of samples from 10 patients, the pan-microbial marker 16S was detected as expected across all samples, but EcN was only detected in patients who had been administered with EcN. Furthermore, we found EcN was significantly enriched in tumor samples compared with matched normal adjacent tissue (Fig. 1).

Conclusion:
The probiotic EcN homes to tumor tissue after oral administration in a small cohort of patients with CRC. ANZCTR trial registration: ACTRN12619000210178. Acknowledgments: EcN (Mutaflor) was provided by Ardeypharm (Corinna Wolff, Dr Michael Schakermann) and Evidence Based Probiotics/Natural Therapy Imports (Aare Prinz).

203
Improving the early detection and prevention of bowel cancer with a simple digital tool: A randomized controlled questionnaire study T OW,* , † E TSE,* , † P BAMPTON, ‡ G IYNGKARAN § *Royal Adelaide Hospital, † University of Adelaide, ‡ Flinders University, Adelaide, South Australia, § Royal Melbourne Hospital, Melbourne, Victoria, Australia Background and Aim: National guidelines for bowel cancer screening and surveillance were updated in 2017 and 2019 and summarize the best evidence-based approach based on contemporary literature. Adherence to guidelines may improve quality and consistency of patient care, while improving resource allocation and efficiency. Rates of concordance with national guidelines have, however, previously been reported to be as low as 30%, limiting their potential benefit. Given the relatively increased complexity of new guidelines compared with previous iterations, adherence is unlikely to improve without efforts to support their administration. There is a paucity of data concerning adherence with the current iteration of the recommendations. We sought to evaluate adherence to current national screening and surveillance guidelines and the potential impact of a purpose-built medical application to assist with determination of guideline-concordant recommendations among primary care doctors and specialists.
Methods: We invited primary care physicians and those normally involved in bowel cancer screening and surveillance to participate in a randomized controlled study comparing the use of a digital application against the currently available resources (the web-based National Health and Medical Research Council guideline publication) in arriving at guideline-concordant recommendations. The web-based and mobile application was developed using an open-source platform (OpenAsApp), along with hypothetical case scenarios from which screening and surveillance recommendations could be derived. Case scenarios were designed in pairs, balanced for difficulty, and focused on recommended intervals that had been updated in the latest guidelines. The order in which participants used either the digital application or standard resources and paired case scenarios was block randomized. Baseline demographics and level of concordance with national guidelines were measured and analyzed according to medical practitioner specialty and order of intervention. Usability of the application was assessed with a standardized usability questionnaire and summarized with a score out of 100. Ethics approval was obtained for the study at all participating sites. Results: Seventy medical practitioners participated in the study: 37 general practitioners (52.9%) and 33 specialists. Primary care physicians attended a short 30-min tutorial providing an overview of the current guidelines. The median rate of guideline-concordant decisions with standard resources was 55.6%. The use of the digital application significantly improved the relative rate of guideline concordance by 45% (P < 0.00001). The improvement in guideline-concordant decisions was independent of the order in which participants were asked to use the application or standard resources, excluding a coaching effect. There was no significant difference in the rate of guideline concordance between specialists and primary care practitioners with either standard resources or use of the digital application. The median usability score of the application was 83.9 (IQR, 71.1-88.95). Usability was moderately positively correlated with increased guideline concordance (r(66) = 0.48, P = 0.00004). Conclusion: Even under ideal conditions, decision making in concordance with recently revised guidelines is limited. The provision of a simple digital tool to assist medical practitioners was associated with a significant improvement in the degree of guideline concordance. Specialist training provides no significant advantage in determining guideline-concordant decisions, possibly owing to the complexity of current algorithms. Usability is an important factor in ensuring the successful application of the digital tools to clinical use. Further understanding how such tools could be integrated into routine clinical practice would greatly support the adoption of national surveillance guidelines.

204
The 2019 World Health Organization diagnostic criteria may overdiagnose serrated polyposis syndrome in older patients J MCCARTHY,* S KOSTALAS* , †, ‡ *Port Macquarie Base Hospital, † Port Macquarie Gastroenterology and Endoscopy, ‡ University of New South Wales, Port Macquarie, New South Wales, Australia Background and Aim: Serrated polyposis syndrome (SPS) has emerged as the most common colonic polyposis syndrome in Australia. SPS is characterized by the development of numerous serrated lesions and is associated with an increased risk of colorectal cancer. In the absence of a reproducible genetic abnormality, SPS remains a clinical diagnosis based on endoscopic findings. The 2019 World Health Organization guideline defines SPS diagnosis as when there are either five or more serrated lesions proximal to the rectum, all being 5 mm in size, with two being at least 10 mm in size, or more than 20 serrated lesions of any size distributed throughout the colon, with five being proximal to the rectum. These criteria are applied on the cumulative lifetime count of serrated polyps. Using this clinical definition, many patients are diagnosed with SPS at an older age, with few polyps but enough serrated lesions to just satisfy the WHO definition. Patients of an older age at diagnosis are more likely to have sporadic serrated lesions that accrue in number over time and therefore fulfill the WHO diagnostic criteria. We hypothesize that the WHO diagnostic criteria overdiagnose SPS, particularly in patients aged >55 years. By increasing the diagnostic clinical cutoff from five to 10 serrated polyps proximal to the rectum (with two or more larger than 10 mm), we hypothesize that the older patients who barely fulfill the criteria for diagnosis will be eliminated, while patients with true SPS will continue to be identified. Methods: Patients with SPS were identified prospectively from 2014 to 2019. Clinical and endoscopic data, including age at diagnosis, polyp site, and polyp number, were recorded. Patients fulfilling the first criterion of the WHO diagnostic guideline were included, while those only fulfilling the second criterion or the previously defined criterion of family history with any serrated lesion were excluded. We applied the proposed new clinical diagnostic criteria of 10 serrated polyps proximal to the rectum to this cohort and describe the difference between these groups after the application of our proposed criteria.
Results: Of 205 patients in this cohort, three were excluded because they only fulfilled the previously defined 2010 WHO diagnostic criterion of a family history of SPS and one or more serrated lesions. Of the remaining 202 patients, 31 (15%) were aged ≤55 years and 171 (85%) were older than 55 years. The mean number of serrated polyps was 7.58 in those aged ≤55 years and 8.81 in the older group, which was not a statistically significant difference (P = 0.26). After applying our proposed diagnostic criterion, 25 patients (81%) in the ≤55-year age group were reclassified as not having SPS, and 122 patients (71%) in the older group had their diagnosis withdrawn. In the group redefined as not having SPS, one patient in the younger cohort and 15 patients in the older group had a diagnosis of colorectal cancer. Conclusion: Diagnostic criteria for SPS may require consideration of age at diagnosis to avoid overdiagnosis. However, simply increasing the number of serrated lesions to define SPS, even in older patients, will result in false negatives and expose some patients to increased risk of developing colorectal cancer with less aggressive surveillance. Further studies are needed to identify which older patients with serrated lesions are at risk of dysplasia and ultimately colorectal cancer. Adopting refined clinical criteria in the future that account for background serrated lesion rate may reduce the burden of yearly surveillance colonoscopy.

217
Treatment-refractory checkpoint inhibitor-induced colitis requiring colectomy: A case series review A ELFORD,* , † T WANG, ‡ M SELEEM, ‡ N SHACKEL, †, ‡ S FANNING †, ‡ *Monash Health, Melbourne, Victoria, † University of Tasmania, ‡ Launceston General Hospital, Launceston, Tasmania, Australia Introduction: Immune checkpoint inhibitor (ICI)-induced colitis is the most common adverse effect of ICI therapy. A proportion of patients develop severe colitis requiring biological therapy. There is a paucity of data in the literature about how to treat ICI-induced colitis when biological therapy fails. We present three similar cases of severe ICI-induced colitis that was treated with total colectomy after corticosteroids and biological therapy failed. Case reports: Three men (aged 47, 63, and 47 years) who had metastatic melanoma developed colitis with a history of inflammatory bowel disease. Patient 1 was treated with ipilimumab and nivolumab, Patient 2 with pembrolizumab, and Patient 3 with nivolumab (all standard dosing regimens). Patients 2 and 3 went into remission with their ICI therapy before colitis developed. Patient 1 had disease progression. They were treated for 1, 9, and 6 months, respectively, before developing ICI-induced colitis. All three patients presented with Common Terminology Criteria for Adverse Events (CTCAE) 3 symptoms, with computed tomography evidence of colitis and markedly raised levels of inflammatory markers. They had negative stool cultures. Flexible sigmoidoscopy showed features consistent with Mayo 2-3 colitis in all three patients. Histology showed classical features of ICI-induced colitis, including neutrophilic crypt micro-abscesses, cryptitis, and increased crypt epithelial cell apoptosis. High-dose intravenous corticosteroids (≥80 mg methylprednisolone daily) failed in all three patients, and they received rescue infliximab as salvage therapy on initial presentation. They partially responded to this, with decrease in stool frequency, absence of bloody stools, and improvement of albumin and C-reactive protein levels. They all underwent dose frequency and dose-escalated infliximab regimens (compared with standard 5 mg/kg at 0, 2, and 6 weeks), as well as cessation of ICI therapy. Despite treatment, all three patients relapsed within 6 weeks of infliximab therapy, in terms of symptoms, fevers, raised levels of inflammatory markers, low albumin levels, and Mayo 2-3 endoscopic findings. Patients 1 and 2 each received one dose of vedolizumab 300 mg intravenously. Nonetheless, they continued to deteriorate, and subtotal colectomy was performed. Consideration was made for using vedolizumab in Patient 3, but this was withheld as Clostridium difficile toxin was detected on stool polymerase chain reaction testing and treated with 14 days of oral vancomycin. On repeat flexible sigmoidoscopy, marked erythema, absent vascular pattern friability, erosions, ulcerations, and spontaneous bleeding were still present, without evidence of pseudomembranes. Pathology showed ICI-induced colitis changes, and as he had more than 10 bowel actions a day and fevers, a decision was made to perform subtotal colectomy. Colectomy histology showed classical pathological findings of ICI-induced colitis throughout the colon in all three patients, with diffuse severe active acute-on-chronic ulcerative colitis-like changes across the entire length of bowel and crypt abscesses and extensive lymphoplasmacytic expansion, not just confined to the colon but also in the distal terminal ileum. Patient 3 additionally had positive cytomegalovirus (CMV) immunohistochemistry from the surgical specimen, raising the possibility of superimposed CMV colitis. All patients recovered well after colectomy, with a speedy return of quality of life. The first two patients developed recurrence of their metastatic melanoma and recommenced nivolumab and pembrolizumab, respectively, receiving more than 6 months of therapy at standard dosing regimens; they have not developed ICI complications. Conclusion: Severe ICI-induced colitis can be refractory to medical therapy in rare cases. There are few treatment options described in the literature. Diverting ileostomy, as well as fecal microbial transplant, has been described as a successful treatment option. A literature search of PubMed and Google Scholar did not identify any reported cases of subtotal colectomy for ICI-induced colitis. Our cases suggest that colectomy can be a good definitive treatment option, return quality of life to the patient, and may allow recommencement of ICI therapy without recurrence of clinically significant disease in the remaining rectum or terminal ileum.

218
Risk of advanced colorectal neoplasia in Australasian populations with sessile serrated lesions Y TRINH,* D TANI,* M APPLEYARD,* , † T RAHMAN,* , † C GAN,* , † B LEGGETT* , † *Royal Brisbane and Women's Hospital, † Prince Charles Hospital, Brisbane, Queensland, Australia Background and Aim: Sessile serrated lesions (SSLs) have been shown to increase the risk of developing colorectal carcinomas. These polyps have a distinct carcinogenesis pathway that differs from conventional adenomas (CAs). The serrated pathway is thought to account for up to 20% of sporadic colorectal cancers. Several studies have shown that individuals with SSLs have a higher risk of harboring advanced neoplasia compared with those with CAs in surveillance colonoscopy. However, studies to date have been in the setting of relatively low SSL detection rates, and we hypothesize that if the colon is adequately cleared of polyps, including SSLs, at the time of the index colonoscopy, this higher risk may be mitigated. We aimed to compare the risk of advanced neoplasia in individuals with SSLs against those with isolated CAs in the setting of high detection rates for both adenomas and SSLs. Methods: We examined the colonoscopy data for consecutive patients who underwent the procedure between December 2010 and June 2012 at The Prince Charles Hospital, Queensland. Patients were excluded from the study if: polyps were absent at the initial colonoscopy; surveillance colonoscopy occurred at <1 year; they had a high-risk colorectal cancer syndrome (Lynch syndrome, familial adenomatous polyposis, and serrated polyposis syndrome); they had inflammatory bowel disease; they had incomplete colonoscopy without cecal intubation; they did not have surveillance colonoscopy; they had unsatisfactory bowel preparation; or they had colonic resection. Advanced neoplasia at surveillance was defined as patients with three or more CAs or CA ≥ 10 mm or with tubulovillous histology or high-grade dysplasia. The patients were divided into two groups, with the first group comprising patients with SSLs (with or without CAs) and the second group having isolated CAs. Using logistic regression analysis, the risk (expressed as odds ratio [OR]) of advanced colorectal neoplasia at surveillance was estimated, adjusted for age and sex. Statistical significance was defined by a P value of ≤0.05. All statistical analyses were performed with STATA 13.1. Results: Of 3908 patients, 1718 had polyps at initial colonoscopy (SSLs with or without CAs in 416 and CAs only in 1302). The CA and SSL detection rates were 42.4% and 8.6%, respectively. After application of the exclusion criteria, 849 patients were eligible for the study. The most common cause for exclusion was the absence of polyps at initial colonoscopy (n = 2190). There were 452 men and an age range from 20 to 91 years. Based on initial colonoscopy findings, 210 patients had SSLs with or without CAs, and 639 had only CAs. There were no significant differences in the likelihood of developing advanced neoplasia between these two groups ( Background and Aim: Interactions between pancreatic cancer (PC) cells and pancreatic stellate cells (PSCs), which produce PC stroma, promote tumor growth and metastasis. One of the pathways that mediate cancer cell-PSC interactions is the urokinase plasminogen activator (uPA) pathway. High serum and tumor uPA levels are associated with poor survival in patients with PC. We hypothesize that targeting of the uPA system will inhibit local growth and distant metastasis of PC. We therefore aimed to compare effects of targeted inhibition of uPA versus gemcitabine (chemotherapy) on PC progression using in vivo and in vitro approaches.
Methods: For the in vivo approach, we used an orthotopic model of PC, involving implantation of human PC cells (AsPC-1) plus cancerassociated human PSCs (CAhPSCs) into the pancreas of nude mice. Seven days later, mice were randomly assigned to receive vehicle control, gemcitabine 75 mg/kg intraperitoneally biweekly, and uPA inhibitor (amiloride analog 26) 3 mg/kg (U3) or 10 mg/kg (U10) intraperitoneally daily (five mice per group). At the end of 28 days of treatment, tumor size, histology, and metastasis were assessed. For the in vitro approach, we used co-culture of AsPC-1 cells and CAhPSCs in the presence of gemcitabine or uPA inhibitor and assessment of AsPC-1 apoptosis and epithelialmesenchymal transition (EMT).
Results: In the in vivo study, uPA inhibition was comparable to gemcitabine in reducing tumor volume. Importantly, uPA inhibition was significantly superior to gemcitabine in reducing metastasis, with U10-treated mice showing no evidence of metastasis (Fig. 1A,B). Vimentin expression was unchanged, but E-cadherin expression was significantly elevated in U10-treated mice tumors compared with the other groups, suggesting reduced EMT of cancer cells (Fig. 1C,D). In the in vitro study, uPA inhibition reversed PSC-induced inhibition of apoptosis and caused a trend towards a decrease in EMT of AsPC-1 cells (Fig. 1E,F). Conclusion: Targeting PSC-tumor interactions with a uPA inhibitor significantly reduced tumor growth and eliminated metastasis in vivo. These effects are likely due to increased apoptosis and decreased EMT of cancer cells. Our findings support uPA inhibition as a promising therapeutic approach for PC.

234
Low rates of lymph node metastases in early gastric cancer at gastrectomy: Endoscopic resection criteria appear safe in the Australian population E YOUNG,* , † A RAJANDRAN,* A RUSZKIEWICZ, ‡ R SINGH* , † *Department of Gastroenterology, ‡ SA Pathology, Lyell McEwin Hospital, † Faculty of Health and Medical Sciences, University of Adelaide, Adelaide, South Australia, Australia Background and Aim: In the context of gastric cancer, endoscopic submucosal dissection (ESD) was initially indicated only for the resection of intramucosal (T1a) well-differentiated adenocarcinomas <2 cm in diameter without ulceration. These are now labeled the absolute criteria, as evidence supports the efficacy of ESD for a broader range of early gastric cancers (EGCs), referred to as the expanded criteria. Expanded-criteria lesions include all differentiated T1a EGCs > 2 cm in diameter without ulcer/scar, differentiated T1a lesions <3 cm with ulcer/scar, undifferentiated EGCs < 2 cm, and differentiated lesions <3 cm with superficial submucosal invasion <500 μm. Current evidence suggests significant heterogeneity among different populations with respect to rates of lymph node metastases (LNM) in expanded-criteria lesions. In these lesions, North American studies have found LNM rates of up to 13.6%, compared with less than 1% in studies from East Asian populations. With such heterogeneity between populations and a paucity of local data, proceduralists performing ESD in Australia are unable to quantify local risks of LNM when considering ESD for lesions within absolute or expanded criteria. The aim of this study was to assess the risk of LNM in EGC using local Australian data. Methods: Pathology reports were retrieved and retrospectively reviewed for all patients undergoing gastrectomy for neoplasia in South Australia between 2017 and 2021. Data were collected for those with gastric adenocarcinoma to establish whether each lesion met the absolute or expanded criteria for resection (including size, depth of invasion, ulceration, and differentiation), as well as whether metastatic adenocarcinoma was present in resected lymph nodes. Results: Reports were reviewed for 191 gastrectomy specimens, of which 128 involved gastric adenocarcinoma. Patients had a median age of 70 years (IQR, 61-79), and 60% were male. Seventeen adenocarcinomas (13%) met the criteria for endoscopic resection, of which six (4.7%) met absolute criteria and 11 (8.6%) met expanded criteria. There were no positive lymph nodes detected in any patients with lesions meeting absolute or expanded criteria (Table 1). Regarding T-staging, 42 lesions (32.8%) were T1 or T2, seven (5.5%) were T1a, 19 (14.8%) were T1b, and 16 (12.5%) were T2. Interestingly, there were no positive lymph nodes in any of these T1/T2 lesions (including those outside endoscopic resection criteria) in the absence of poor differentiation or lymphovascular invasion (LVI). Notably, all lesions with LNM in the absence of LVI were poorly differentiated T2 lesions more than 4 cm in diameter.
Conclusion: This retrospective study shows low rates of LNM in EGCs, with no positive lymph nodes associated with lesions meeting accepted Figure 1 In vivo: (A) effects of gemcitabine (G) and uPA inhibitor amiloride analog A26 3 mg (U3) and 10 mg (U10) on endpoint primary tumor volume in an orthotopic pancreatic cancer model. Mice received G 75 mg/kg bodyweight twice weekly or daily intraperitoneal injections of U3 or U10 (3 or 10 mg/kg bodyweight) for 28 days. Data represent the mean ± SEM (five mice per group). # P < 0.0001, G versus control (Ctrl) and U10 versus Ctrl; *P < 0.002, U3 versus Ctrl (five mice per group). (B) Number of mice with histologically confirmed liver metastases. # P < 0.008, U10 versus Ctrl; *P < 0.05, U3 versus Ctrl and G versus U10 (five mice per group). (C) Immunostaining for the mesenchymal marker vimentin. (D) Immunostaining for the epithelial marker E-cadherin. *P < 0.03, U10 versus Ctrl (five mice per group). In vitro: (E) effect of 10 nM uPA inhibitor A26 (U) and 300 μg/mL G on AsPC-1 cell apoptosis. *P = 0.05, AsPC-1 cells + CAhPSCs versus Ctrl; # P < 0.05, U versus AsPC-1 cells + CAhPSCs (n = 3). (F) Effect of 10 nM uPA inhibitor A26 (U) and 300 μg/mL G on epithelial-mesenchymal transition (EMT) in AsPC-1 cells (n = 3). endoscopic resection criteria. In addition, even when including lesions with lamina propria invasion, there were no patients with LNM in the absence of poor differentiation or LVI. Although more data are required to confirm our findings, absolute-and expanded-criteria EGCs appear to be very low risk for LNM in the Australian population and should therefore be considered suitable for endoscopic resection.

245
Dysbiosis in head and neck cancer: Determining optimal sampling site for oral microbiome collection D PANDEY,* , † M SZCZESNIAK,* , † J MACLEAN,* , ‡ I COOK,* , † P GRAHAM,* , § E EL-OMAR,* , ¶ H YIM,* , ¶ F ZHANG,* , ¶ X-T JIANG,* , ¶ P WU* , † *University of New South Wales, † Department of Gastroenterology and Hepatology, ‡ Department of Speech Pathology, § Department of Radiation Oncology, ¶ Microbiome Research Centre, St George Hospital, Sydney, New South Wales, Australia Background and Aim: While chemoradiotherapy has a high cure rate for head and neck cancer (HNC), it frequently causes temporary and permanent debilitating complications. Recent research suggests that dysbiosis of oral microbial community is associated with HNC. Whether this dysbiosis causes these chemoradiotherapy-related complications remains unclear. To answer this question, we must first address the methodological challenge of determining the optimal site of collection and the effect of risk factors, such as smoking and excessive drinking. We aimed to determine the optimal site for oral sample collection (the site with highest microbiome diversity and temporal stability), the impact of risk factors such as smoking and drinking on the microbiome of the selected site, and whether presence of HNC is associated with an altered oral microbiome. Methods: In each of 21 patients with newly diagnosed but untreated HNC (mean ± SD age, 64 ± 12.2 years) and 27 age-matched healthy controls (mean ± SD age, 62 ± 11.1 years), six microbiome samples were collected. Saliva and swabs from the buccal area, tongue, hard palate, and faucial pillars, as well as a single swab from all four mucosal sites, were taken. In healthy controls, two collections were performed 4-6 weeks apart. Microbial DNA was extracted and amplified, and 16S V3 and V4 regions were sequenced using MiSeq paired-end 300 bp. Microbial DNA sequence data were filtered to remove low-quality sequences and then denoised with set parameters, and host DNA contamination was removed. The influence of sample location, smoking and drinking risk factors, and presence of HNC on microbial diversity (alpha within sample and beta between samples) was investigated. Results: In healthy controls, analysis of observed taxonomic units (OTU) detected differences in alpha and beta diversity between sampling sites (P < 0.0001 and P = 0.001, respectively). Saliva was found to have the highest intra-community microbial species diversity (Fig. 1a) and lowest temporal variance (Fig. 1b), while microbiome from the hard palate was found to have the lowest diversity and temporal stability. Saliva was also found to have the highest diversity in patients with HNC, but differences between sites were not statistically significant. In healthy controls, salivary microbiome of smokers tended to have lower alpha diversity than in nonsmokers (P = 0.057), while no difference between excessive drinkers and non-excessive drinkers or non-drinkers could be discerned. In patients with HNC, microbiome diversity was not influenced by smoking and drinking behavior. Across all sites, patients with HNC had lower alpha diversity than healthy controls (P < 0.00001) (Fig. 1c). Beta-diversity analysis showed the microbiome of patients with HNC to be compositionally distinct from that of healthy controls (P = 0.01). This pattern was confirmed when salivary microbiome was considered alone. Conclusion: Patients with HNC exhibit reduced diversity of oral microbiome. Salivary samples show temporal stability, have the richest diversity, and are sufficient to detect perturbation due to presence of HNC. Hence, they can be used as a representative oral sample for microbiome study in patients with HNC.

250
Metastatic melanoma with unknown primary tumor and rare gastric involvement: A case report P SELVANATHAN,* , † P KAAZAN,* , † A TAYLOR* , † *Princess Alexandra Hospital, † University of Queensland, Brisbane, Queensland, Australia Introduction: Melanoma is the third most common cancer in Australia, with one person diagnosed every 30 min. Index presentation with metastatic melanoma is only seen in 4% of cases but carries a poor prognosis, with a 5-year survival rate of less than 20%. Up to 10% of metastatic cases present without a primary cutaneous lesion. This is thought to occur due to either regression of the primary cutaneous tumor or malignant transformation of ectopic melanocytes; there are currently no biomarkers to assist in differentiating these processes. Although melanoma is the most common metastatic malignancy of the gastrointestinal tract, gastric melanoma only comprises 20% of these cases. We describe a case of multiple gastric melanomatous lesions in a patient with no known history of cutaneous disease.
Case report: An 82-year-old man presented to the emergency department with a 1-month history of hemoptysis. No gastrointestinal symptoms were noted, and the patient had replete hemoglobin and iron stores. He had a history of ischemic heart disease, osteoporosis, previous cigarette smoking, and multiple non-melanomatous skin cancers. Computed tomography imaging of the chest showed patchy lung consolidation in the left lower lobe, as well as bilateral hilar and mediastinal lymphadenopathy. A positron emission tomography scan identified fluoro-2-deoxy-D-glucose (FDG) avidity of the enlarged hilar and mediastinal lymph nodes, and areas of focal FDG avidity in the stomach, hepatic flexure of the colon, gall bladder, left lower lobe of the lung, and proximal right femur. Upper gastrointestinal endoscopy identified multiple diminutive to large (>1 cm) pigmented sessile polypoid lesions with a tubular-like pit pattern throughout the cardia, fundus, and body of the stomach (Fig. 1a-c). A similar diminutive lesion was noted in the duodenal bulb (Fig. 1d) and a further lesion at the appendiceal orifice during colonoscopy (Fig. 1e). Histology from biopsies of the gastric polypoid lesions and appendiceal orifice lesion was consistent with BRAF wild-type metastatic melanoma, as were biopsies from a left main bronchus tumor and left paratracheal lymph node taken during bronchoscopy. Palliative immunotherapy was initially considered, but due to the patient's rapid functional decline, he was managed with palliative radiation therapy. Conclusion: We present a rare case of melanoma with unknown primary tumor and multiple gastric metastases. Interestingly, the patient did not have any gastrointestinal symptoms despite the extensive involvement seen endoscopically. It remains unclear whether this patient's disease arose from a primary cutaneous lesion that has regressed or if this is a case of non-cutaneous melanoma originating in the gastrointestinal tract. We present the endoscopy images to highlight the potential appearances of melanoma in the gastrointestinal tract. Background and Aim: Pancreatic ductal adenocarcinoma (PDAC) is one of the leading causes of cancer death in Australia and the world, and patients have a poor 5-year survival rate. Most patients present with unresectable disease at diagnosis. There is no clinically available biomarker to assess prognosis in patients with PDAC. However, KRAS mutation is present in more than 90% of PDAC cases. This study aimed to investigate the role of detecting the KRAS mutation and subtype in patients with PDAC to determine prognosis across all stages of disease. Methods: We performed a retrospective cohort study to assess KRAS status in patients presenting with PDAC between 2016 and 2020. Tissue for genetic testing was obtained either by endoscopic ultrasound fine-needle aspiration (EUS-FNA) or from resection specimens. The presence of KRAS mutations was assessed by multiplex polymerase chain reaction (PCR) (STRIP assay), next-generation sequencing (NGS) (TSO-500 gene panel), or digital droplet PCR. We reviewed patient records to determine age, sex, date of diagnosis, date of death, National Comprehensive Cancer Network (NCCN) clinical stage for resectability, and type of treatment received (surgery, chemotherapy and/or radiotherapy without surgery, and supportive care only). Survival was measured from date of EUS-FNA or resection histological diagnosis until death or the end of the study period.