Agne Krilaviciute, Christian Stock, Marcis Leja, Hermann Brenner

Regular screening for gastric cancer (GC) is based on invasive upper gastrointestinal endoscopy and is limited to few high-incidence countries. As GC is a major cause of cancer death worldwide, a non-invasive, simple screening test is of value. We assessed the prevalence of preclinical GC and the corresponding numbers needed to screen (NNS) to detect GC cases both without and with preselection using breath tests from the literature in various populations. Using age- and sex-specific GC incidence data and rates of transition from preclinical to clinical GC, we estimated the prevalences of preclinical GC worldwide in populations aged 50-74 years, and we evaluated the accuracy of breath testing for GC detection based on published studies. We then derived the expected positive predictive values for breath testing in populations with different preclinical GC prevalences. Four studies reporting the sensitivity and specificity of breath tests were identified, and summary estimates of 83% sensitivity and 91% specificity were derived by meta-analysis. The estimates of the overall prevalence of preclinical GC were <0.5% in men and <0.2% in women aged 50-74 years across different regions of the world. The positive predictive values, the prevalence among breath test positive people, were approximately nine-fold higher in all populations, resulting in an approximately nine-fold lower NNS to detect one GC case when breath tests were used for preselection for screening. Given the low prevalence of preclinical GC, non-invasive breath tests show promise for making screening more efficient. Further validation of breath tests and evidence on the rates of transition from preclinical to clinical GC are needed to validate the breath test approach.