AG

AA

bst

ract

sTu1900

PREMM1,2,6 Model Has Limited Sensitivity for Pre-SymptomaticIdentification of Individuals At Risk for Young Onset CRCCaitlin Foor-Pessin, Erika S. Koeppe, Elena M. Stoffel

Background: Incidence of colorectal cancer (CRC) in young individuals is rising and casesdiagnosed at age<50 years raise suspicion for possible genetic predisposition. ThePREMM1,2,6 model is a validated risk assessment tool for estimating the probability thatan individual carries a mutation in one of the DNA mismatch repair (MMR) genes associatedwith Lynch Syndrome (LS). Our aim was to examine the sensitivity of PREMM1,2,6 foridentifying individuals at risk for young onset CRC. Methods: We conducted chart reviewsof all patients diagnosed with CRC at age< 50 years referred for genetic evaluation at anacademic cancer center. Two sets of PREMM1,2,6 model scores were calculated for eachsubject, one including and another excluding their own diagnosis of CRC, with a score of>5% designated as "high risk." Results: We identified 151 individuals diagnosed with CRCat age <50 (mean age at diagnosis 40 years, range 16-49 years). Males and females wereequally represented, with the majority of cancers located distal to the splenic flexure (66.2%)and diagnosed at advanced stages (III and IV, 52.3%). Only 40 (26.5%) subjects reporteda family history of CRC diagnosed in a first degree relative. Clinical genetic testing identifiedgermline alterations in genes associated with cancer predisposition in 39 (25.8%). Pathogenicmutations in DNA MMR genes associated with LS were detected in 24 (15.8%), with variantsof uncertain significance in 7 (4.6%). Mutations in APC and MutYH (biallelic) were foundin 5 (3.3%) and 3 (2.0%) subjects, respectively. PREMM1,2,6 scores calculated includingsubjects' personal history of CRC exceeded the threshold of 5% in 124 (82.1%) cases andmean scores were significantly higher for individuals with pathogenic MMR gene mutationscompared with those without mutations (37.7 vs 11.1, p<0.0001). Use of PREMM1,2,6with cutoff of >5% identified all cancer-affected carriers of mutations in MMR genes, APCand MutYH. However, PREMM1,2,6 scores calculated excluding the proband's own diagnosisof CRC would have classified only 21 (13.9%) subjects as high risk and would have missed12 (50%) individuals with LS. Conclusion: While the PREMM1,2,6 model threshold of >5%identified all young onset CRC patients with germline mutations, this approach to riskassessment would have failed to detect half of cancer-unaffected individuals with geneticpredisposition to CRC. Additional strategies are needed to improve pre-symptomatic identifi-cation of young individuals at high risk for CRC.

Tu1901

Clinical Characteristics and Natural History of Multiple Colorectal AdenomaPatients Without Germline APC or MYH/MUTYH MutationsAlan Tieu, Daniel Edelstein, Jennifer Axilbund, Katharine Romans, Cherryl Blair,Elizabeth A. Wiley, Linda Hylind, Francis M. Giardiello

Background: Patients with multiple colorectal adenomas (MCRA) but without known geneticcause (i.e. familial adenomatous polyposis/ MYH associated polyposis) are increasing beingdiagnosed. The clinical characteristics and natural history of this condition are not wellstudied. Method: 27 patients (70% male) with MCRA, defined as having cumulatively 10-99 colorectal adenomas and without deleterious mutations of the APC or MYH gene werestudied. The results of 93 colonoscopies with mean time of follow-up of 4.9 years (range0-27) between colonoscopy were evaluated. Findings from EGD were analyzed. Also, extraco-lonic manifestations were assessed. Results: The mean age at polyp development and MCRAdiagnosis was 47.8 ± 13.1yrs. (range 21-72) and 50.4±14.6 yrs (range 21-72), respectively.In 6 of 27 pedigrees (22%) another family member had MCRA. At first colonoscopy, themean number of adenomas was 35.0±35.9 (range 0-99). One patient had 2 hyperplasticpolyps with multiple adenomas; none had sessile serrated adenomas and one had colorectalcancer. Cumulatively, at last colonoscopy the mean number of adenomas was 53.0±32.2(range 10-101), representing a 12% increase in adenomas per year. EGD was done in 19of 27 patients (70%) at mean age 48.7±12.7 years. Of these, 9 patients (47%) had uppertract findings on EGD including, 4 (21%) with duodenal adenoma, 3 (10%) with fundicgland polyps, 2 (10%) with duodenal adenoma and fundic gland polyps. Patients with uppertract findings were diagnosed with MCRA at significantly younger age (40.3 years) comparedto those without findings (50.6), p<0.05. 18 patients (67%) underwent colectomy at meanage 51.4±15.9 years with mean time from diagnosis of MCRA of 3.1±1.3years. After surgery,2 of 4 surveyed patients developed recurrent adenomas in retained colorectum. 9 patients(33%) had extracolonic cancers including nonmelanoma skin cancer (4), melanoma (3),and breast, bladder and prostate in one apiece. Conclusions: MCRA patients have a clinico-pathological phenotype similar to the known syndromes of attenuated polyposis presentingwith polyps at middle age, continuous development of adenomas, lack of serrated polyphistology, presence of upper tract findings in many patients, family history of the disorderin some patients, and the need for colectomy in the majority. Consequently, the managementof these patients and families should parallel the treatment of those with FAP and MAPincluding routine surveillance of the upper tract. MCRA patients lacking upper tract polypsmay represent a different subset of the disorder which deserves further investigation.

Tu1902

The Association of TGFβ Signalling Pathway Gene Polymorphisms WithColorectal Cancer Risk: A Meta-AnalysisJoshua L. McGuire, Mark McPhail, Arun Rajendran, Kevin J. Monahan

Background: Approximately 35% of colorectal cancer risk is due to heritable factors. Todate, a large fraction of this heritability remains unexplained. The TGFβ signalling pathwayhas an increasingly implicated role in colorectal carcinogenesis, with highly penetrant-germline mutations of BMPR1A, SMAD4 and GREM1 causing known polyposis syndromes.We propose that common, low penetrance variation of TGFβ signalling genes may accountfor much of the unexplained heritability of colorectal cancer, underlining the importanceof this signalling pathway in the aetiology of colorectal cancer. Aim: A meta-analysis of theassociation of TGFβ signalling pathway gene single nucleotide polymorphisms (SNP) withlow penetrance colorectal cancer risk. Methods: A systematic literature search of Medline

S-868AGA Abstracts

and Embase was performed. Data was extracted from eligible studies, according to pre-specified criteria. RevMan software, version 5.2, was used to generate pooled odds ratios(OR) to estimate the risk attributed to each variant. In addition to this, subgroup analysesfor ethnicity, gender and tumour site were performed to investigate these as sources ofheterogeneity. Results: Between 9,854 and 27,641 cases were meta-analysed for each SNP.Of the 10 SNPs discovered in a review of the literature, 8 were significantly associated withan increased risk of colorectal cancer in this study. These SNPs were located within BMP4,GREM1, CDH1, SMAD7, RHPN2 and BMP2, the largest effect was for rs10411210 withinRHPN2 (OR=1.15; 95% CI 1.09- 1.22, I2 50%). Subgroup analyses revealed gender as apossible source of heterogeneity, but no preferential associations for any of the SNPs withtumour site or ethnicity were detected. Furthermore, determination of inconsistency betweenstudies, demonstrated by an I2 of <50% for 8 of 10 SNPs, indicated that overall studyheterogeneity was not a common source of bias. Discussion: Whilst 8 out of 10 variantsshowed significant association, the estimates of risk were small with all OR <1.15. This mayresult from suboptimal methods of estimating risk, as well as unknown disease heterogeneity.This process is constrained by a lack of knowledge of the true risk alleles tagged by theSNPs studied. Conclusions: The results of this analysis underline the integral role of theTGFβ signalling pathway in colorectal carcinogenesis. Knowledge of the function of taggedrisk alleles is required to elucidate and accurately estimate the risk attributed to polymor-phisms in this pathway. Key words: COLORECTAL CANCER, TGFβ SIGNALLING,LOW PENETRANCE,

Tu1903

Pancreatic Cancer and Sensory Nerves: Characterization of the Role of TRPV1Leading to Axonal and Cancer Growth Using a Novel Microfluidic DualChamber SystemSmrita Sinha, Ya-Yuan Fu, In Hong Yang, Subhash Kulkarni, April M. Lee, Min G. Joo,Nitish V. Thakor, Michael G. Goggins, Pankaj J. Pasricha

BACKGROUND: Pancreatic ductal adenocarcinoma (PDA) has the highest reported incidence(80-100%) of perineural invasion (PNI) among cancers. Recent studies suggest that reciprocalmolecular signaling between PDA and nerve cells may enhance PNI and tumor growth.AIMS: To use a novel dual chamber microfluidic co-culture system to 1) investigate theeffect of axonal growth on PDA cell proliferation 2) identify whether axonal growth ispromoted in the presence of PDA cells versus non-PDA cells 3) study the role of the transientreceptor potential cation channel subfamily V member 1 (TRPV1) in mediating cancer-neuronal communication. METHODS: Neurons were harvested from the dorsal roost ganglia(DRG) of four-week old wild-type C57BL/B6, Wnt1-cre;tdTomato or TRPV1-/- mice andcultured onto one chamber of the dual chamber system. PDA cells A6L (developed from aliver metastasis from a patient at Johns Hopkins Hospital), CaCo-2 cells (human colorectaladenocarcinoma) or HPDE cell (benign human pancreatic ductal epithelium) were platedin the second chamber. A6L cell proliferation was quantified with S-phase analysis usingnucleoside analog 5-ethynyl-2-deoxyuridine (EdU) uptake assay (Life Technologies). Axonalgrowth was quantified using Avizo 7.1 image reconstruction software (VSG, Burlington,MA, USA) to project and analyze confocal images and determine nerve length density. Therole of TRPV1 on axonal growth was studied using the TRPV-1 inhibitor capsazepine(10μM)in the culture medium. RESULTS: EdU uptake in A6L cells was increased by 20% whenco-cultured with neurons (p=0.006). Axonal growth was significantly increased by about70% in the presence of A6L cells versus CaCo-2 cells or HPDE cells. Capsazepine (10uM)in the culture medium decreased axonal growth by 30% in the presence of A6L cells. Co-culture with DRG neurons from TRPV1-/- mice resulted in 18% decrease in axonal growth.CONCLUSIONS: We have developed a novel model system to co-culture sensory neuronsfrom the mouse DRG with human PDA cell lines (Figure 1). Separation of the neuronal andPDA cell bodies more accurately reflects in-vivo conditions within the pancreas. In our co-culture model, axonal growth resulted in increased A6L cell proliferation as measured byEdU uptake. A6L cells promoted axonal growth to a significantly greater degree whencompared to CaCo-2 and HPDE cells indicating that neurotropism may be a unique featureof PDA cells. Neuronal TRPV1 expression may be an important factor in promoting axonalgrowth in the presence of cancer cells.