Systematic Review and Meta-analysis: Use of Statins Is Associated with a Reduced Incidence of Oesophageal Adenocarcinoma

Oesophageal carcinoma (OC) affects 450,000 people worldwide and is the sixth leading cause of cancer-related mortality in the world [1]. The two most common histological subtypes, squamous cell carcinoma (OSC) and adenocarcinoma (OAC), together make up approximately 98% of this number. Global epidemiological studies show that whilst global incidence of OSC is greater than OAC, the incidence of OAC is now greater than OSC in the developed countries of the West such as the UK, the Netherlands, and the USA [2, 3]. Currently, prognosis for oesophageal cancer remains one of the poorest amongst all cancers, with 5-year survival rates around 15% [3, 4]. This is largely due to the proportion of patients presenting at advanced disease stages. In OAC in particular, this has led to management strategies revolving around monitoring risk factors, most important being Barrett’s oesophagus (BO). To this effect, endoscopic surveillance as well as ablative procedure, principally radiofrequency ablation (RFA), has been developed [5]. Although surveillance and intervention strategies do seem to reduce the incidence of OAC, RFA is not without costs and complications [5].

However, due to the generally asymptomatic nature of BO, most cases of OAC arise outside surveillance programs. This is an area where chemopreventative strategies could make a difference to the general population. In the past 10 years, experimental evidence has shown that statins [hydroxymethlyglutaryl-CoA reductase inhibitors (HMG-CoA inhibitors)] could potentially play a role in reducing progression of BO to OAC [6, 7]. The mechanisms proposed include upregulation of pro-apoptotic proteins such Bad and Bax, as well as reduced cell proliferation and enhanced apoptosis through reduced activation of signalling G-proteins as a consequence of reduced levels of melavonate due to inhibition of the HMG-CoA reductase [6, 7]. Through these mechanisms and possibly other mechanisms, statins attenuate the malignant behaviour of OAC cell lines [6‐10]. Evidence from a variety of clinical studies supports this hypothesis. Four years ago, two separate meta-analyses using slightly different methods came to almost identical conclusions: statin use was associated with a significant 43% reduction in the incidence of OAC in Barrett’s cohorts and a smaller [19%] reduction in the incidence of all oesophageal cancers in population studies [11, 12]. However, these studies highlighted areas of lack of information including the precision of the inverse association with statin use, effects of confounding risk factors, and effects in different populations. Since the publication of these meta-analyses, multiple additional studies have been published and we have performed an updated systematic review and meta-analysis examining the association between statin use and OAC with the aim of further refining the association and in particular defining knowledge gaps that could be beneficially be the target of future studies [12]. The inclusion of more recent data has enabled us to specifically examine the association between statin use and oesophageal cancer in three categories: statin use in relation to malignant progression to OAC in Barrett’s oesophagus cohorts, the association between statin usage and OAC in a population cohorts, and the association between statin usage and all oesophageal cancers on a population scale.

This systematic review was performed according to the Preferred Reporting Items for Systematic reviews and Meta-Analyses Guidelines (PRISMA guidelines) and as previously described [12, 13].

The search strategy identified 110 individual potentially suitable publications. Through the eligibility and screening processes, a total of 20 publications were included in the meta-analysis [14‐33]. This totalled 22 different relevant studies as two publications reported separate data and adjusted odds ratios for different cohorts (either separate data for males and females or separate histological subtype of oesophageal cancer) [27, 32] (Fig. 1). All studies were published in English. The studies identified were clearly split into three categories based upon the specific population being examined: the association between statin use in the community and incidence of OAC, the relationship between statin use in a BO cohort on OAC rates, and the association with all oesophageal cancer on a population scale. Table 1 lists the studies included and their important characteristics. Two of the included studies were only published in abstract format; however, sufficient relevant information necessary for analysis was extracted from them [20, 29]. The other studies were in full peer-reviewed format.

For each planned subgroup analyses, separate meta-analyses were conducted on adjusted and unadjusted data. This was due to the heterogeneity in presentation of risk factor adjustment in the studies being analysed. As in the previous meta-analysis [12], heterogeneity in the reporting of statin dose and duration meant statistical analysis of this data would not be feasible.

The primary objective of this meta-analysis was to examine association of statins with the development of oesophageal cancers, particularly oesophageal adenocarcinoma. Our pooled estimates show that statins have a strong association with a reduction in the risk of OAC (approximately 41% reduction of risk) particularly when used in patients with pre-existing Barrett’s oesophagus. The adjusted data from 11 studies provides a reasonably narrow confidence interval for this figure. These data are similar to those reported in the previous meta-analyses: Singh et al. [11] and Beales et al. [12] adjusted OR = 0.59; 95% CI, 0.45–0.78 and adjusted OR = 0.57; 95% CI, 0.43–0.75, respectively. The current updated meta-analysis has more than doubled the number of studies included (from 5 to 11) and increased the number of OAC cases 3-fold (317 to 1057), and this inclusion of additional studies has allowed a much tighter definition of the likely margin of association with much smaller confidence intervals.

A similar degree of reduction in OAC was associated with statin use in population cohorts. The pooled adjusted odds ratios showed a reduction of 43% in OAC incidence. However, with only three datasets included, it is relatively the most underpowered of all analyses included in this study. However, this is a new finding; previous systematic reviews did not allow statistical analysis of this particular subgroup. Further studies are warranted to investigate this effect more thoroughly especially in light of the rising incidence rates of OAC. It is notable that this subgroup analysis included a specific oesophagogastric junction adenocarcinoma group, whilst there may be some important biological differences between this and more classical OAC [8], it was felt appropriate to include this cohort as the majority of the other studies included in this systematic review included junctional cancers within the cancer cases, making no distinction between the OAC and OGJA for statistical analysis. Two studies [18, 33] specifically excluded Siewert III OGJA cancer.

It remains an area of active research whether statins have beneficial anti-cancer effects merely on progression of non-dysplastic BO or whether they are actually associated with a reduction in the development of BO itself. Whilst most of the in vitro studies have examined the effects of statins on various models of Barrett’s oesophagus, recent data including a meta-analysis have reported that statin use is associated with a reduced incidence Barrett’s oesophagus [35].

Statin use in the population is associated with a reduced incidence (by 18%) of all types of oesophageal cancers. Again, three of the major studies [27‐29] included in the analysis provided very consistent data in describing this effect. The previous meta-analysis [12] suggested very similar reduction of risk at OR = 0.81; 95% CI 0.75–0.88. There was much more heterogeneity in the unadjusted results examining statin use at the population level and all oesophageal cancers than in the other analyses. The heterogeneity was significantly reduced when comparing the odds ratios adjusted for other variables. This probably reflects the wider variety of geographical and health care locations included in this group (certainly compared to the more homogeneous secondary care Barrett’s cohorts) and reinforces the problems with potential confounding by indications and risk factors in studies at that population level. Despite this, there were no obvious outlying studies and the pooled analysis of the adjusted odds ratios showed no significant heterogeneity.

Most of the published data have examined the relationship between statins and OAC; there are much less data concerning squamous cancer. Although there are well-performed laboratory studies providing biological plausibility and exploring the effects of statins on oesophageal adenocarcinoma and non-malignant Barrett’s cells [6, 7, 10], these corroborating studies are lacking as regards oesophageal squamous cancer. Given the contrasting biology and epidemiology between oesophageal adenocarcinoma and squamous cancers, it would not be surprising if the effect of statins differed between the two. The relatively smaller apparent protective effect against all oesophageal cancers combined compared to specifically against OAC may imply that statins have less effect on the development of squamous cancer but further studies are indicated. The relative and absolute numbers of adenocarcinomas and squamous cancers within the studies examined in this population-level data meta-analysis are unknown and, given the disparity of geographical locations studies, are likely to vary between studies, further contributing to the heterogeneity of results in this subgroup analysis. We acknowledge that these population-level, all-oesophagal cancer data are the weakest and least informative of those specifically included in this systematic review, but feel these data should be included for completeness and to stimulate further research into any associations between statin use and squamous cancer. However, the pooled data on all oesophageal cancers are still consistent with statin use being associated with a lower risk of oesophageal adenocarcinoma.

The main limitation to this meta-analysis is that all included studies are observational (i.e. cohort or case-control). As such, regardless of the sample size involved in this study, it is not completely free of bias and despite the generally high quality of the observational studies included in this systematic review, uncorrected bias may still influence the results. However, the consistency of the results, including lack of heterogeneity, from different geographical locations, health care systems, and research methodologies suggests that unadjusted bias is unlikely to have a major effect on our findings, although it would be important to establish what the association is between statin use and cancer risk in relation to specific other risk factors for OAC such as smoking, obesity, and visceral obesity [8]. In the adenocarcinoma-specific analysis, particularly in the Barret’s cohorts, the minimal difference between the pooled adjusted and unadjusted ORs, suggests that confounding by indication and residual bias, are unlikely to be contributing to the results.

It was notable that the number and variety of potential cofounding variables adjusted varied considerably between the studies and this would seem to be a major focus for future work. Despite the evidence that proton pump inhibitors and NSAIDs [33, 36] are also associated with a reduced incidence of OAC, not all studies provided data to enable accurate adjustment. Laboratory studies and our previous meta-analysis showed that the combination of statins and cyclo-oxygenase inhibitors appeared to be additive in both effects on cell lines [6, 7] and in reducing the incidence of OAC (pooled aOR 0.26 (0.1–0.68)) [12]. However, the more recent studies have not examined or reported the data to enable further verification of this association and it is recommended that any further studies specifically examine this.

Probably the weakest aspect of the current study is the lack of data on statin type (lipophilic or hydrophilic), dose, drug, or duration. Previous meta-analyses [11, 12] were similarly limited in this regard but unfortunately despite the increase in the number of studies and subjects included in the systematic review, the data are still inadequate to draw any firm conclusions. Most studies did not specifically examine this aspect of statin use and when it was reported, the categories used were too heterogeneous for statistical analysis. Where reported there seemed to be a tendency for higher doses and longer durations to be associated with a lower incidence of OAC but any future studies should specifically address these issues.

In conclusion, our updated systematic review and meta-analysis show that statin use is associated with a significantly reduced incidence of oesophageal adenocarcinoma in both Barrett’s cohorts (by 41%) and population-based studies (by 43%). There is also a reduced incidence of all oesophageal cancers in population-based studies (by 18%), although it is possible that this is driven mainly through a reduction in the incidence of OAC rather than OSC; that hypothesis requires further investigation. The inclusion of additional studies has enabled much tighter precision of the estimate of the reduction of OAC associated with statin use in Barrett’s cohorts.

The consistency of the results across multiple studies combined with supportive experimental evidence does suggest that statins may have useful chemopreventative effects in Barrett’s oesophagus. Given the relatively low rate of neoplastic progression in BO (approximately 1 in 300 per year [37, 38]), it may not be cost-effective to merely use statins as chemopreventative agents, but given that the main cause of death in patients with BO is not oesophageal cancer but vascular diseases, it may be prudent to ensure that statins that are definitely prescribed to patients with Barrett’s oesophagus were indicated by vascular risk [39].