Description of the Evidence
Evidence of Benefit Associated With Screening
Several screening techniques, including barium-meal photofluorography, gastric endoscopy, and serum pepsinogen have been proposed as screening methods for the early detection of gastric cancer. No randomized trials evaluating the impact of screening on mortality from gastric cancer have been reported.[12,19] Even in very high-risk areas, the positive predictive value (PPV) of the screening tests may be very low. In a screening program of 17,647 men aged 40 to 60 years in Wakayama City, Japan, the PPV of combined serum pepsinogen and barium meal with digital radiography over the 7-year period was 0.85%. The positive test rates were 19.5% for serum pepsinogen and 22.5% for radiography, with a cancer detection rate of 0.28%. Over the 7-year period, there was no reduction in gastric cancer mortality compared with an age-matched surrounding population.
Barium-meal gastric photofluorography
A national program of population-based screening for gastric cancer using barium-meal photofluorography has been ongoing since the 1960s in Japan. Participation rates have been in the range of only 10% to 20%.[12,20] Although there has been a coincident decrease in mortality from gastric cancer in Japan, mortality rates have been decreasing in many developed countries despite the lack of screening programs. Case-control studies from Japan show decreases in gastric mortality in people who have undergone screening, but results from prospective studies were not consistent.[12,19]
A pilot study of community-based photofluorography was conducted in Costa Rica using the same techniques as those used in Japan's national program (with consultation from Japanese experts). People were invited by letter from a population registry to attend two rounds of screening, and a total of 6,200 eligible screened participants (of a planned 12,000) were analyzed. Their gastric cancer mortality from 2 to 7 years after screening was compared with four control groups that had not been invited to be screened, and the relative risk was about 0.5 (no p-value reported). The study was, however, prone to strong biases, including selection bias, and likely differential exclusion of people with previously diagnosed gastric cancer favoring the screened population. In addition, unlike the community controls, patients diagnosed with gastric cancer through the screening program were treated at a single referral center. The PPV of a suspicious fluorograph was 3%; the specificity in the two rounds was 67% and 80%; and the positivity rates were 34% and 20%. Despite the authors' belief that their results provided substantial evidence that routine screening would decrease gastric cancer mortality, they concluded that the costs of screening with photofluorography would be far too high in their country.