MODELING THE RISK OF COLORECTAL LESIONS BASED ON A MIXED CHINESE COHORT

Purpose:

Remarkable progress has been made in understanding colorectal cancer (CRC) progression. This, in turn, has led to a more rational and effective clinical approach to the prevention of CRC, which is considered as one of the most preventable cancers in the Western world. However, it is not the case in resource-deprived countries like China, which starts to suffer a pandemic of CRC as well as many other gastrointestinal diseases. In all cases of precancerous colorectal lesions, early disease recognition is essential to devise suitable cancer prevention strategies. This research marks the first attempt in China to systematically analyze colonoscopy testing data for both asymptomatic and symptomatic populations. The hypothesis is no pronounced difference exists between the two populations in using causal variables for assessing the risk of lesions.

Method:

Colonoscopy testing results (whether having lesions) were recorded from 375 symptomatic and 484 asymptomatic subjects in a Beijing hospital. For each subject, the record also contains information of demographic (gender, age, height, weight, etc.), lifestyle (smoking, alcohol assumption, daily diet, etc.), and family history. For both populations, we performed correlation analysis to select causal variables influencing on the lesion incidence. We then developed a logistic regression model based on the selected factors.

Result:

The proportions of having lesions were 58% and 42% in asymptomatic and symptomatic groups, respectively. This difference was shown significant with a Chi-square test (p < 0.05). Factors that were significantly correlated with the risk of lesion differed between the two groups (see Table 1). Gender, age, and height were influential in both groups. However, smoking was only influential in the symptomatic group whereas dietary habit was only influential in the asymptomatic group.

Conclusion:

Contrary to the natural belief, this cohort analysis suggests the lesion risk among asymptomatic people may be higher than in the symptomatic group. This implies that at least equal attention be given to asymptomatic people for early detection of gastrointestinal diseases. Meanwhile, slightly different factors should be taken into consideration for the asymptomatic group when assessing the lesion risk. Overall, the proposed models will help devise risk-based screening strategies to improve the cost-effectiveness of CRC early detection. Multicenter data and sophisticated analysis are necessary to explore the cohort differences and for implementing a risk-based CRC screening strategy in China.