Helicobacter pylori Infection and Gastric Cancer—For Want of More Outcomes

In 1994, Helicobacter pylori was declared a type 1 carcinogen—a definite cause of human cancer—by the International Agency for Research on Cancer (IARC).¹ This conclusion, which was based largely on epidemiological data that were later substantiated,² left behind unsettling questions. First, some latent concern lingered that H pylori was merely a marker for other exposures. Despite subsequent studies in animals, humans, and tissue cultures indicating plausible mechanisms for H pylori–induced carcinogenesis, naysayers have clung to the small, residual probability that H pylori is merely a confounder for the true cause of disease. Because H pylori infection is so closely linked to socioeconomic status, such confounders are not difficult to find.

More important, however, the IARC evaluation left a dilemma. The organization provided no recommendations on how knowledge of a causal association between H pylori infection and cancer should inform clinical or public health decision making. Subsequent consensus conferences left similarly vague recommendations on gastric cancer prevention. The European consensus conference in Maastricht concluded that population screening for H pylori should not be undertaken but that individuals with a high risk of cancer (known preneoplastic conditions, prior treatment for gastric cancer, or a strong family history) should be offered testing and, if positive, treatment.³ The Canadian consensus conference similarly recommended against mass screening for H pylori but then recommended treating anyone in whom H pylori was found.⁴ In the United States, no consensus statement about H pylori has been compiled since the IARC publication, and decisions about which patients to treat are made by physicians based on individual preferences, prejudices, and anecdotes.⁵

Stomach cancer is an important disease—the second leading cause of cancer death worldwide.⁶ Helicobacter pylori infection is curable with relatively simple courses of antibiotics. Moreover, economic models indicate that mass screening and treatment of H pylori to prevent cancer could be cost-effective, even if this approach prevented only a minority of gastric tumors.^{7 - 8} What is lacking is definitive proof that H pylori eradication truly can prevent cancer. Although some studies even show that a subset of preneoplastic conditions can regress with H pylori treatment, these may not be the lesions that would progress to cancer.⁹ Whether treatment would actually prevent cancer and, more to the point, when such treatment would work are not known with certainty. To prevent H pylori–related cancer, is it necessary to prevent infection entirely (in which case a vaccine or other prevention strategy is the only option) or to treat infection early in its course (in which case screening and treatment in childhood might be necessary) or just before the first dysplastic cell appears in the mucosa (in which case screening and treatment of adults might be effective)? Screening programs are huge undertakings. In the absence of definitive proof through randomized clinical trials that H pylori eradication prevents cancer, population screening is unlikely ever to be justified.

In this issue of THE JOURNAL, Wong and colleagues¹⁰ present results from the first randomized clinical trial of H pylori eradication with cancer as an outcome. For those who have long awaited such a study, there will be a sense of anticlimax. On first blush, the results are discouraging. In their study of 1630 Chinese participants, Wong and colleagues found that H pylori eradication did not prevent cancer in 7.5 years of follow-up. The authors identified cancer in 0.86% of the treatment group compared with 1.35% of the control (placebo) group (P = .33). This statistically insignificant difference, however, represented a 37% relative decrease in cancer incidence—a large proportion for any cancer intervention. Unfortunately, the study was underpowered to detect such a difference.

It is unclear why the study was stopped when it was, just at the border of the study's predicted power. Given a few more years of follow-up, a few more outcomes might have resulted in a more definitive result. With the findings as they stand, readers will interpret the numbers to fit their preconceived notions. "True believers" will claim that had a 1-tailed test been used instead of a 2-tailed test (H pylori eradication certainly was not expected to worsen gastric cancer outcome), the findings would have been statistically significant. Moreover, some might cite the "per protocol" and the subgroup analyses, rather than the intention-to-treat analysis, and report significant protection against cancer if H pylori was successfully eradicated or if preneoplasia was absent. The "naysayers" will correctly claim that the main results of the study were negative. Moreover, they will argue that the study was not completely blinded (participants knew the group they were in) and that this may have potentially biased the findings toward significance. To borrow from the old adage, for want of a rider, the battle was lost. In this instance, some might suggest that for want of more outcomes, the truth was lost.

The post hoc subgroup analysis of this study provides some tantalizing information. These data indicate that any benefit of H pylori eradication is restricted to patients with gastritis only, rather than those with defined preneoplastic conditions. This finding is somewhat surprising. One of the underpinnings of the current understanding of gastric cancer is that it is a slow process, developing over decades from normal stomach tissue through chronic atrophic gastritis, intestinal metaplasia, dysplasia, and, finally, cancer.¹¹ That some individuals appear to progress to cancer without having had underlying preneoplasia 7 to 8 years previously indicates that for a subset of gastric cancer patients, the biology of their disease is not fully understood. The risk of cancer in this group, however, is consistent with that observed by Uemura et al in Japan,¹² who found a remarkably similar risk in H pylori–infected individuals without atrophy or intestinal metaplasia.

Unfortunately, the findings of Wong et al present a dilemma for screening and treatment initiatives. If the subgroup of patients with the lowest risk of disease (ie, those without preneoplasia) comprises the subpopulation with preventable cancer, cost-effectiveness of screening and treatment programs is sure to decline dramatically. Indeed, current studies indicate that treatment of patients at younger ages, when preneoplastic conditions are unlikely to yet be extant, is much less likely than later treatment to be cost-effective; the discounted costs are high and the number of patients who would require treatment to prevent 1 case are excessive.⁸ Moreover, the concept of "test and treat" typically requires a relatively low-cost intervention and avoids expensive, invasive screening procedures.

Whereas identification of H pylori–infected individuals can be undertaken inexpensively, identification of those infected without preneoplasia would be considerably more problematic. Serologic tests, such as pepsinogen and gastrin assays, have sensitivities and specificities that may be too low to be broadly applied in screening.¹³ Selecting to treat those with normal pepsinogen or gastrin assays would not only bypass individuals with intestinal metaplasia—the lesion deemed by Wong and colleagues to be unresponsive to treatment—but also bypass those with atrophic gastritis, a condition that might be responsive; the study does not provide useful information in this group. Identifying individuals who would benefit from treatment by adding endoscopy to screening programs would likely make costs prohibitive in parts of the world where gastric cancer is most common. Thus, the authors' conclusion that endoscopy, biopsy, and H pylori eradication for patients without preneoplasia should be considered in high-risk areas requires much further study and evaluation in local geographic contexts before it can be undertaken.

Almost a decade after the IARC evaluation, the report by Wong et al provides the first experimental evidence in humans that H pylori infection causes cancer. If the subgroup analysis is correct, then H pylori may be considered as the preeminent cause of gastric adenocarcinoma. The question of whom to treat, however, remains unanswered. Those in public health are unlikely to be convinced by this study to start the development of H pylori screening programs. It is encouraging that the authors plan to follow up their patients further. With careful follow-up over time, this cohort could reveal an increasing wealth of information. Other trials currently under way, as well as additional cost-effectiveness studies, will supplement these findings. Thus, in 5 to 10 years, the answers to many of these issues may be clear. In the meantime, many patients will come to their physicians asking for screening and myriad others previously diagnosed as having H pylori infection will request treatment. Until more outcomes are available, individual clinicians will continue to do what we have for the past decade: use the very best informed judgment and hope for the best.