Helicobacter pylori infection and mortality from ischaemic heart disease: negative result from a large, prospective study

Introduction

Mendall et al showed an association between Helicobacter pylori infection of the stomach and ischaemic heart disease in 1994-5,1 2 and 18 studies have reported on the relationship over the subsequent two years.3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 The position remains uncertain, an assessment supported by a recent review.21 Since the risk of acquiring H pylori infection in childhood increases with socioeconomic deprivation and overcrowding,3 22 the association may be indirect; H pylori infection and ischaemic heart disease are both related to social class. Some studies have shown little or no excess risk,4 5 6 7 8 9 but others reported a fourfold to fivefold increased risk without adjustment for measures of social class and other risk factors for ischaemic heart diseases10 11 or a twofold to threefold increased risk after such adjustment.1 2 3

We report the results of an investigation using a prospective study of the determinants of major chronic disease. This study is well suited to determine whether there is an independent relationship between H pylori infection and ischaemic heart disease. With 648 deaths from ischaemic heart disease it is much the largest study to report on the association. The likelihood of an indirect association arising through social class differences is minimised by the homogeneity of the study population: the subjects were all professional men attending for a routine medical examination. Random error and systematic error are therefore likely to be small.

Methods

The British United Provident (BUPA) study is a prospective study of 21 520 professional men aged 35–64 who attended the BUPA medical centre in London between 1975 and 1982 for a routine medical examination. Serum cholesterol concentration and other risk factors for ischaemic heart disease were measured23 24 and serum samples were stored at −40°C. The cohort consisted of men resident in Britain, so all records could be flagged at the NHS central register in Southport, permitting automatic notification of death (with the certified cause) and cancer incidence by the Office of Population Censuses and Surveys (now the Office for National Statistics). Further information on the causes of death was obtained by writing to each certifying doctor.

This analysis is based on follow up to the end of December 1994; the average follow up was 15.6 years. In this period 648 men died from ischaemic heart disease (international classification of diseases, 9th revision, codes 410-414) but had no history of ischaemic heart disease on entry. For each case, two controls (who did not die of ischaemic heart disease and did not have a history of ischaemic heart disease on entry) were selected; they were matched for age and duration of storage of the serum sample, each to one year.

In 1997 the frozen serum samples were retrieved. IgG antibody titres specific to H pylori were measured with an enzyme linked immunosorbent assay (ELISA) kit (HelicoG2; Shield Diagnostics, Dundee). This quantified the results in arbitrary units; we designated a result of 10 units/ml as positive, as suggested by the manufacturer of the test. In our laboratory the within-assay coefficient of variation was 2.6% and the between-assay coefficient of variation was 10.0%. Linear models were used for all the variables except triglycerides, which were log transformed to correct for skewness. Blood pressure was somewhat skewed, but analyses using log transformed and untransformed data yielded almost identical results. Cox's proportional hazards models (with strata to identify each case with its two controls) were used to estimate the independent association of risk factors with ischaemic heart disease; the statistical analysis takes account of both the matching and the survival time.

Results

Table 1 shows characteristics of the men at the time of entry to the prospective study. The established risk factors for ischaemic heart disease were more common or more extreme in the men who subsequently died of ischaemic heart disease than in those who did not, as reported previously.23 The prevalence of H pylori infection did not differ significantly between the two groups (47.5% in cases, 45.9% in controls; P=0.50). The study indicated no association between H pylori infection and ischaemic heart disease.

Table 2 summarises risk factors for ischaemic heart disease in controls with and without H pylori infection. Risk factors were almost identical in the two groups, confirming that the study population was ideal in that it avoided the opportunity for confounding.

Table 3 shows the odds of death from ischaemic heart disease in men infected with H pylori relative to that in men not infected. Adjustment for potential confounding factors that may have been associated with H pylori infection (those with P values <0.2 in table 1—height, smoking, and father dying of heart disease)—had negligible effect. The adjusted odds ratio was 1.06 (0.86 to 1.31), indicating no material association. Further adjustment for the other risk factors shown in table 1 did not affect this estimate.

Discussion

This study is the largest published study on H pylori and ischaemic heart disease. It avoids confounding by socioeconomic status because the study population came from the same socioeconomic group; this is confirmed by the minimal effect of adjustment for potential confounding factors. The results show no significant association, and the upper confidence limit indicates that any excess risk is unlikely to exceed 30%. Reports of positive associations have come mainly from small retrospective studies, and it seems likely that these were due to a combination of chance and confounding.

Our negative result cannot be explained by error in the assay determining H pylori status because the same assay showed a strong association between H pylori infection and stomach cancer. With 50 cases and three controls per case (updating an earlier report25), 68% and 37% respectively were positive for H pylori (odds ratio=4.0 (1.9 to 8.2); P<0.001). The assay also confirmed the expected increasing prevalence of H pylori infection with age (30%, 42%, 52%, and 70% in control men who were aged 35-44, 45-54, 55-64, and ≥65 on entry). These rates are higher than might be expected today because they relate to blood samples collected about 18 years ago.

One of the studies originally linking H pylori infection with ischaemic heart disease suggested that the infection increased plasma fibrinogen,2 a plausible mechanism for the association with heart disease. We were unable to measure fibrinogen concentrations in the samples from the men who died of heart disease and their controls (because serum, not plasma, was stored), but it was measured in 206 people who attended for a routine medical examination in 1996. The mean plasma fibrinogen (measured by the Medical Research Council Epidemiology and Medical Care Unit, using the Clauss fibrinogen assay26) was virtually the same in the 55 people who were positive for H pylori and in the 151 who were negative (2.64 and 2.62 g/l respectively, age adjusted); the 95% confidence interval on the difference was −0.16 to 0.19 g/l (P=0.84). Other studies have also shown no relation between H pylori infection and plasma fibrinogen.7 11 12 13 27

Implications

Our finding no relationship between H pylori and ischaemic heart disease strengthens the evidence that the relationship between H pylori and stomach cancer is one of cause and effect: the absence of confounding by social class in the association with heart disease indicates that such confounding will not affect the association with stomach cancer.

The negative result is disappointing in that it indicates that screening people for H pylori infection and treating those found to be infected will do little to prevent ischaemic heart disease. It does, however, reinforce the evidence on stomach cancer and indicates that eradicating H pylori infection may greatly reduce one of the most common cancers worldwide.