IMPACT OF LIFESTYLE FACTORS ON SCREENING-DETECTED COLORECTAL NEOPLASIA

Background and aims: The detection and removal of precancerous lesions through colorectal cancer (CRC) screening, and the intervention on modifiable risk factors for CRC - such as smoking habits, physical activity, red meat consumption and alcohol intake - represent the two possible ways for reducing CRC incidence and mortality. The aim of this project was to investigate whether lifestyle factors, gender, family history and daily low-dose Aspirin use are important factors in predicting endoscopy findings at a first round screening level and whether they can have a significant impact on the natural history of the disease in screened patients during their follow-up (second round screening level). Patients and methods: Me and my work team identified and selected a study population of 870 men and women of age 50-74 years who underwent a screening colonoscopy at the European Institute of Oncology (IEO) between the years 2007-2009 after a positive Fecal Occult Blood Test (FOBT+). We set up a telephone questionnaire in order to retrieve information on smoking habits, BMI, physical activity, diet, alcohol consumption, family history and usage of low-dose Aspirin at the time of the first colonoscopy. All patients were then interviewed by me by telephone. Ninety-five individuals were not interviewed for various reasons, making the final population size n=775. Patients who could answer the questionnaire were similar to the unreached individuals in terms of outcome of the first colonoscopy. Results: At first colonoscopy, we observed 415 patients presenting with a high-risk neoplasia (i.e. 3 or more adenomas or at least one adenoma bigger than 10 mm / with villous component / with high-grade dysplasia or invasive tumor). At the univariate analysis, gender, family history, physical activity, smoking habits, alcohol intake, fruit and vegetable intake and daily low-dose Aspirin were associated with the prevalence of high-risk neoplasia. Using a “Spike at zero function”, we showed that light drinkers (<5 grams per day) seemed to have a lower risk of high-risk neoplasia compared to non-drinkers. We concluded that a proportion of non-drinkers might avoid alcohol because of some health conditions linked to the endpoint of interest. At a multivariable level, all those factors remained statistically significantly associated with the outcome of interest. We therefore combined the information of lifestyle factors, gender, family history and daily low-dose Aspirin use to obtain a reliable individual risk score (i.e. linear predictor) and build a nomogram. The second colonoscopy visit date was fixed in advanced at the time of first colonoscopy, based on the outcome of the first colonoscopy, following a typical example of Doctor’s care scheme of examinations. After adjusting for the severity of the outcome of the first colonoscopy and for the time from first to second colonoscopy, we obtained a statistically significant association between the linear predictor and the risk of high-risk neoplasia detected at the second colonoscopy. We then applied homogeneous Markov Models to simultaneously model the disease process over time. The effect of the linear predictor on the transitions – from one disease stage to the other – resulted statistically significant. Moreover, as the linear predictor increased, the probability of getting better decreased. In other words, the worse the lifestyle, the lower the probability for the intestinal mucosa to heal. On the other hand, the estimated parameter for the effect of linear predictor on the aggravation transition resulted positive: the worse the lifestyle, the higher the probability to find new high-risk polyps. Conclusions: Lifestyle should be considered in the planning of population CRC screenings, because the identification of different risk groups can lead to more tailored screening policies, and accordingly to more efficient and cost-effective interventions.