ObjectiveExplore the influence of family history of upper gastrointestinal (UGI) cancer on UGI cancer death, based on the Linxian Dysplasia Nutrition Intervention Trial (NIT) cohort.MethodsFamily history of UGI cancer was defined as at least one first-degree relative (parent, child, or sibling) had a history of esophageal or gastric cancer. Cancer death was carried out by ICD-10 code. Family history information was collected at baseline and cancer deaths were assessed at each annual follow-up. The COX proportional risk model was used to estimate the hazard ratio (HR) and 95% confidence interval (95% CI). We compared the positive family history group with the negative to determine the risk of family history on UGI cancer death. The effect of category of relatives, number of relatives with UGI cancer, and diagnosis age of relatives on the UGI death risk were further analyzed. Interaction and stratification analyses were done to see the subgroup effects. Sensitivity analyses were also conducted by exclusion of individuals who were followed up less than three years. We considered controlling of covariates including: gender, age (continuity), community, education level, number of siblings (continuity), BMI (continuity), smoking, alcohol use, fresh fruit intake, fresh vegetable intake, hot beverage intake, edible oil intake, meat intake, and moldy staple food intake. All food intake variables were converted into categorical variables.ResultsFrom1985 to2015, we followed up total 3,318 individuals with 898 UGI cancer deaths (537 from ESCC, 77 from GNCC, and 284 from GCC). In a single factor analysis, family history of UGI cancer increased the risk of death of esophageal squamous cell carcinoma (ESCC) by 27% (HR=1.270, 95%CI1.072-1.504). No associations were observed in gastric cardia carcinoma (GCC) and gastric non-cardia carcinoma (GNCC). After adjusting for multi-factor, a family history of UGI cancer risk of death increased by 31.9% from ESCC (HR=1.319,95%CI:1.110-1.567). Subgroup analysis of different types of relatives with UGI cancers, UGI cancers in the mother (HR=1.457,95%CI:1.200-1.768), brother (HR=1.522,95%CI:1.117-2.073), and sister (HR=1.999,95%CI:1.419-2.817) were independent risk factors for ESCC death, while the father was not. In addition, 2 relatives with UGI cancer (HR=1.495, 95%, CI:1.110-2.013) and ≥3 relatives with UGI cancer (HR=2.836, 95%CI:1.842-4.367) significantly increased the risk of ESCC death, and the trend test was statistically significant (P<0.001). Relatives’ diagnostic age of 51-60 years (HR=1.322, 95%CI:1.046-1.672) and 41-50 years (HR=1.442, 95%CI:1.078-1.930) were the risk factors for ESCC death, with statistical significance in the trend test (P=0.010). No statistically significant result of the family history effect on the risk of death from GCC or GNCC was found. Sensitivity analysis of 80% of subjects, randomly selected, did not change the results.ConclusionA family history of UGI cancer may predict the risk of death from ESCC but not from GCC or GNCC. UGI cancer in the mother may predict the risk of death from ESCC, but not father, which indicates gender differences. Gender and smoking are the interaction items with family history in a similar extent. In the subgroup, the risk of ESCC death is more distinct by family history in younger, female, and better-lifestyle individuals, which indicates the unique role of genetic factors.
Read full abstract