Gastric cancer risk stratification and surveillance after Helicobacter pylori eradication: 2020

Nam et al, Take et al, and Kaji et al have all confirmed the hypothesis that H pylori eradication results in a significant reduction in the development of gastric cancer. Two of the studies (ie, Take et al and Kaji et al) assessed cancer risk endoscopically at baseline with the Kimura-Takemoto gastric atrophy scale. Both studies showed that the degree and extent of gastric atrophy was a major determinant of cancer risk.^,  For example, Take et al found that the incidence of gastric cancer at 1 year after eradication was 0% for those with mild atrophy, 0.26% with moderate atrophy, and 0.66% with severe atrophy (P = .01). Persistent infection (HR 3.9), mucosal atrophy (HR 3.9), and age (HR 2.0) were associated with development of gastric cancer. The Take et al study also illustrates the concept that a population defined by average age (eg, average age 50) consists of a heterogeneous group of patients differing in terms of the extent and severity of atrophy and thus in gastric cancer risk. The effect of H pylori eradication is to halt disease progression and also to blunt cancer risk. By contrast, damage and risk are progressive in untreated individuals and in those for whom treatment has failed. The rate of increase depends on the extent and severity of atrophy, and when it becomes exponential, the difference in cancer risk rapidly separates from those whose H pylori infection was eradicated.

The natural history of gastric cancer includes a latent period followed by an upward inflection in which the risk increases exponentially (Fig. 1). This timeline reflects the extent and severity of mucosal atrophy. The inflection point between gradual increase and exponential increase likely represents the age when the population average score progresses to moderate to severe atrophy. Although H pylori eradication results in healing of inflammation and halts progression of damage, eradication is unable to reset the biologic clock to zero but rather tends to halt the increase in risk at the point where the infection was eradicated.

Gastric cancer risk stratification involves directly or indirectly assessing the extent and severity of gastric atrophy. In Western countries, this commonly involves histologic assessment tools such as Operative Link for Gastritis Assessment of Atrophic Gastritis (OLGA) or the Operative Link for Gastritis Assessment of Intestinal Metaplasia (OLGIM) risk assessment systems.^,  Where the expertise is available, the endoscopic Kimura-Takemoto classification system is often used. Both approaches require endoscopy; the first requires targeted gastric biopsies, and the second requires clean stomachs and training in endoscopic classification.

Ideally, one would prefer to have a validated reliable noninvasive marker to predict cancer risk. Nam et al suggested low-serum HDL as a possible biomarker for risk stratification for gastric cancer. Their observation will require further confirmatory studies. A 2015 meta-analysis summarized studies associating serum HDL and cancer and identified an association with high-density lipoprotein cholesterol (HDL-C). However, they noted that only the e4 allele was significantly associated with increased risk for the development of cancer in Asian populations (ie, odds ratio [OR], 1.40; 95% CI, 1.00-1.94) with no significant risk reduction in white populations (OR, 0.92; 95% CI, 0.81-1.03). Gastric cancer was not among the cancers listed as having an increased risk associated with HDL.

Serum pepsinogen levels have a long history of use as a biomarker for detection of atrophic gastritis. However, serum pepsinogen tests generally have a relatively low specificity and sensitivity, most often between 70% and 80%.^,  The clinical interpretation of such tests is greatly affected by the pretest probability (ie, the prevalence of atrophy in the population). For example, in a population with 10% atrophy, with pepsinogen sensitivity and specificity of 80%, the negative predictive value (NPV) of a normal result would be 97.3%, and only 30.8% of those with a positive test result for atrophy (ie, positive predictive value [PPV]) would be correctly identified (Fig. 2A). That same test with 40% atrophy in the population would have NPV 85.7% and PPV 72.7% (ie, only 86% with normal results and 72% with the atrophic results would be correctly characterized) (Fig. 2B). Improving the sensitivity and specificity of the test to 95% would not materially change the results (Fig. 2C).

Although pepsinogen testing is useful for population screening, it typically fails to provide information with sufficient granularity to be of much value for decisions regarding individual patients. Also, the current cutoff values are reliable only before H pylori eradication. In populations with a low prevalence of gastric cancer, most of those with H pylori infections will have nonatrophic stomachs and normal pepsinogen results. In such regions, most patients receive the diagnosis, and eradication is confirmed by noninvasive tests with no further testing. However, subpopulations will exist with a higher risk of gastric cancer in whom risk stratification would be appropriate (ie, high-risk ethnic groups, recent emigrants from high-risk areas, positive family history, or clinical history). The decision to use pepsinogen testing versus endoscopy would depend on the pretest probability of cancer and on the sensitivity and specificity of the available tests. For example, the miss rate (false negative test results) is low with a low prevalence of atrophic gastritis, such that a negative test result could be trusted whereas those with a positive test result would need to undergo confirmatory endoscopy. In high-prevalence populations, the miss rate would be unacceptably high, and endoscopic screening would be preferred. The actual choice would need to balance the costs and accuracy of available risk stratification schemes. In high-risk populations, pepsinogen testing might offer cost savings. Consider a test with sensitivity and specificity of 80% in a population of 1000 patients with 20% (200 patients) at high risk. Overall, 640 of the 800 with nonatrophic stomachs would have normal test results. In 160 with false positive results, unnecessary endoscopy would ensue. Among the 200 with atrophic gastritis, 20% (40) would also be misclassified as normal. Thus, 320 individuals with abnormal values would undergo endoscopy, 50% of whom would have nonatrophic stomachs. However, 680 endoscopies would be averted, of which only 4% (40) of the results were misclassified, resulting in undiagnosed atrophic gastritis. The important variables are therefore the sensitivity and specificity of the test in the specific population, the pretest probability of atrophic gastritis, the costs versus benefits of biomarker screening, and the sensitivity and specificity of endoscopic and histologic risk stratification schemes available in that population.

To date, we do not have a validated noninvasive test with sufficiently high sensitivity and specificity to stratify high and low cancer risk and avoid endoscopy. Considerable interest in finding such a test remains, and many candidates are being considered, including use of circulating microRNAs for both early detection and prognostication of gastric cancer.

The current criterion standard is endoscopy with targeted biopsy. Because endoscopy is expensive, time consuming, and not without hazards, there is little place for “random gastric biopsies,” which cannot provide an assessment of the extent and severity of the gastric mucosal damage. Whenever a gastric biopsy specimen is taken for whatever reason (eg, investigation of a visual finding or to rule out H pylori infection), one should also consider whether clinical interpretation will also require knowledge of the status of the gastric mucosa. In most cases one should take the 5 specimens required for OLGA or OLGIM interpretation (ie, normal-appearing mucosa from the lesser and greater curves of the antrum and gastric angle, and mid greater and lesser curves of the corpus with the antral and corpus biopsy specimens put into separate bottles). Biopsy specimens from abnormal-appearing mucosa should not be mixed with those taken for cancer risk assessment. Thinking ahead allows one to obtain the maximum information at each encounter.