Testing and Treating Helicobacter Pylori Infection in Individuals with Family History of Gastric Cancer is Cost-effective

Introduction

Gastric cancer (GC) is a major source of cancer-related morbidity and mortality worldwide.¹ Up to 10% of cases demonstrate familial aggregation, although whether this is due to genetic factors, shared environment with exposures to toxins such as helicobacter pylori infection or tobacco smoke, or a combination of the two is unclear.² Family history of a first degree relative with GC increases lifetime risk by 2.5-fold, as does H. pylori infection.^{3, 4} A recent randomized controlled trial demonstrated a significant reduction in GC incidence when treating patients with a first-degree relative with GC for H. pylori infection, in patients undergoing regular endoscopic surveillance.⁵

Although universal GC screening is not recommended in the United States, previous studies have suggested there may be benefit to screening certain high-risk populations.^{6, 7} The recent American Gastroenterological Association Clinical Practice Guidelines suggest including patient’s family history when making decisions regarding surveillance of intestinal metaplasia.⁸

The goal of this study is to determine whether individuals with a family history of GC benefit from testing for and treating h. pylori infection as compared to not testing. We hypothesize that given the dramatic reduction in GC demonstrated by Choi et al that the screening strategy can be a cost-effective intervention.

Methods

A Markov state-transition model was created in TreeAge (TreeAge Pro 2020, Williamstown, MA) comparing three strategies: (1) No Screening for H. pylori infection, (2) Screening with 13C-Urea Breath Test (UBT), and (3) Screening with Stool Antigen test. Inputs for the model were obtained from published literature (supplemental table 1). The hypothetical cohort for this analysis is comprised of 40-year-old men and women with a first-degree relative with GC in the US. The model follows this cohort for 60 years or until death and has a cycle length of one year. The health states in our model included H. pylori negative, H. pylori positive, GC, and death.

The distribution of the cohort at the beginning of the simulation was based on U.S. H. pylori prevalence rates. Patients with H. pylori infection were treated with 7-day triple therapy and if this regimen failed to eradicate the infection then retreatment with 14-day quadruple therapy (as described in the trial by Choi et al) was given.⁵ The primary outcomes of this study were total cost, quality adjusted life years (QALYs), incremental cost-effectiveness ratios (ICERs) as well as GC incidence and mortality. A willingness to pay (WTP) threshold of $100,000/QALY was used to determine whether a strategy was cost-effective. For further details, please see supplemental methods.

Results

The base-case analysis of this model found that Screening is cost-effective compared to Not Screening (Table 1). Furthermore, the two noninvasive Screening strategies were both less costly and led to greater QALYs than the No Screening strategy. The costs of No Screening were $2,694.09 and resulted in 21.95 QALYs. In contrast, UBT Screening cost $2,105.28 and yielded more QALYs (22.37). Screening with Stool Antigen was comparable to UBT, with costs of $2,126.00 and 22.30 QALYs.

Table 1.

Base-Case Analysis of the Model

Strategy	Cost ($)	QALYs	ICER	GC Incidence	GC Mortality	Risk Reduction
13C-UBT	2,105.28	22.37	Reference	1.59%	1.41%	22.1%
Stool Antigen	2,126.00	22.30	Dominated	1.65%	1.46%	19.1%
No Screening	2,694.09	21.95	Dominated	2.04%	1.82%	Reference

ICER = Incremental Cost-Effectiveness Ratio

GC = Gastric Cancer

All three screening strategies led to a decrease in GC incidence and mortality. In the No Screening strategy, approximately 2.04% of patients developed GC and 1.82% died of GC. In contrast, with Screening, GC incidence fell to 1.59–1.65% and mortality fell to 1.41–1.46%, depending on the strategy. This represents a 19.1–22.0% reduction in risk.

To validate our model, we assessed the 10 year decrease in GC assuming 100% H. pylori prevalence and compared the model results to the trial results. Choi et al found a 55% decrease in gastric cancer at median 9 years by intention-to-treat analysis; among those in whom H. pylori was eradicated the decrease was 73%.⁵ In this sensitivity analysis, our model found a 53.3% to 64.5% reduction in gastric cancer incidence, similar to that found in the trial.

Extensive one way deterministic sensitivity analyses were performed. When evaluating the effectiveness of the strategies as measured by QALYs, the model was most sensitive to the relative risk of family history, utility of H. pylori infection and eradication success of treatment (supplemental figure 1).

Probabilistic sensitivity analysis showed the screening strategies were cost-effective compared to not screening at all WTP thresholds. When cost was removed from consideration at a WTP threshold of $0, UBT was cost-effective 64.6% of the time (data not shown). Stool antigen was a viable alternative and was cost-effective 35.3% of the time.

Discussion

Our study found that screening for and treating H. pylori infection in those at high risk for gastric cancer is cost saving and more effective than not screening. A recent randomized controlled trial showed a significant decrease in gastric cancer incidence by H. pylori treatment,⁵ and this modeling study suggests that even after accounting for costs of treatment, the strategy is beneficial and results in substantial decrease in cancer incidence. The two noninvasive screening strategies had very similar results as measured by costs and QALYs.

Although recent studies and commentaries have called for GC screening for certain groups, the role of testing for H. pylori specifically among those with a family history of GC in the US has not yet been explored. Our study suggests it may be worthwhile in this group.

There are several limitations. As with all modeling studies the model is limited by the published literature that is available. Our study does not incorporate the possible adverse effects of H. pylori treatment, such as bacterial resistance and potential increased acid reflux. Although our model was analyzed from the US perspective, the trial that informed the risk reduction was performed in a Korean population. Finally, important risk factors such as the presence of atrophy, intestinal metaplasia and dysplasia were not incorporated in the model as these were not included in the original trial by Choi et al.⁵

In conclusion, this study suggests that screening and treating patients with a family history of GC for H. pylori is cost-effective and reduces GC incidence. Future clinical studies with sufficient time for surveillance are needed to validate this finding.

Data availability:

no original data was used for this study; all literature is publicly available.