Addition of genetic information to clinical factors does not increase identification of individuals at risk for esophageal adenocarcinoma, researchers report in the January 2019 issue of Gastroenterology.
New screening and risk evaluation methods are needed to improve early detection of esophageal adenocarcinoma, because most patients receive a diagnosis at a late stage when they have low odds of survival after treatment. Clinical factors, or combinations of genetic, clinical, and demographic/lifestyle factors, can identify individuals at high risk of esophageal adenocarcinoma. A model including patient age, sex, body mass index (BMI), smoking status, and prior esophageal conditions identified individuals who developed esophageal adenocarcinoma within 5 years with an area under receiver operating characteristic curve of 0.80.
However, additional information is needed to identify individuals at sufficiently high risk to warrant endoscopy screening and to help individuals at low risk avoid unnecessary and invasive procedures.
Andrew T. Kunzmann et al analyzed data from the UK Biobank to determine whether the addition of genetic factors to their clinical risk determination system increased their ability to identify individuals at high risk for development of esophageal adenocarcinoma within 5 years. They combined information on clinical factors and genetic variants previously associated with esophageal adenocarcinoma (polygenic risk scores, calculated by summing the positive risk allele counts for 18 genetic variants weighted by their odds ratios from genome-wide association studies). These genetic variants might contribute to development of esophageal cancer.
A total of 329,643 (65.6%) were included in the analysis; 214 individuals were diagnosed with esophageal adenocarcinoma within 5 years. Individuals diagnosed with esophageal adenocarcinoma were more likely to be older, male, and current or former smokers; they also had a higher BMI, an existing esophageal condition, and a higher polygenic risk score than noncases.
The strongest adjusted associations with esophageal adenocarcinoma within 5 years were for single nucleotide polymorphisms (SNPs) at, or near, the genes CFTR (rs17451754; odds ratio, 1.21) and BARX1 (rs11789015; odds ratio,1.20).
The association between polygenic risk score and risk of esophageal adenocarcinoma within 5 years was modest and not statistically significant (adjusted odds ratio for the middle vs bottom tertile 1.09; adjusted odds ratio for the top vs bottom tertile,1.38).
The combination of clinical factors and polygenic risk score identified individuals who developed esophageal adenocarcinoma within 5 years with an area under receiver operating characteristic curve (AUROC) of 0.80—the same value as for clinical factors alone (see Figure). This was true even in a secondary analysis, with 18 SNPs added to the model (AUROC, 0.81).
When the authors compared the combined clinical and polygenic risk score model with the original clinical points-based model, changes in net reclassification index and Youden’s index were modest at any of the cutoff points—modest improvements in sensitivity were offset by modest reductions in specificity.
Kunzmann et al conclude that testing for esophageal cancer-associated germline mutations is unlikely to help identify individuals at risk for this cancer, at a population level. Studies are needed to examine other biomarkers associated with detection of this tumor.