Proton pump inhibitors (PPIs) are substituted benzimidazole derivatives and are the most potent suppressors of gastric acid. Pantoprazole is the prototype proton pump inhibitor which was approved for patient prescription in the late 1980s. Since then PPIs became one of the most widely prescribed classes of medications. Several studies have reported overusage of PPIs in hospital as well as at the time of discharge after hospitalization with most of the indications being inappropriate.[1,2,3, 4,5]

Reports published in the initial decade of the clinical usage of PPIs, have shown that the risk of adverse effects from these drugs is low and most of the adverse events are minor and selflimiting.[6,7,8,910] The relatively good safety profile of this class of drugs could be one of the reasons for the over prescription of these drugs. But polypharmacy in this way can lead to potential undesirable drug-drug interactions and also results in an increase of the health care costs.

Recently several reports have highlighted certain clinically relevant adverse events of a serious note in relation to PPI usage (listed in Table 1). In light of these findings, it is imperative to analyze the data regarding the adverse effects and safety profile of PPIs which will help us in understanding the magnitude of the problem and also rationalize our approach while prescribing medications to the patients.

Table 1: Adverse events of clinical significance reported with PPIs

Effect of PPI – Clopidogrel interaction on thrombotic events

Clopidogrel, a thienopyridine derivative, is a prodrug which is metabolized by hepatic cytochrome P4502C19 to its active metabolite which inhibits platelet aggregation by selectively and irreversibly binding to the purinergic P2Y12 receptors on the platelet surface. PPIs are also metabolized in the liver from an active to inactive state by the same CYP2C19 isoenzyme. Thus PPIs and clopidogrel compete for the same subset of cyt P-450 isoenzymes (depicted in Figure 1).

Gilard et al[11] first reported reduced effectiveness of clopidogrel in patients receiving PPIs by assessing the vasodilator stimulated phosphoprotein (VASP) phosphorylation index. They found significantly higher VASP values in PPI users compared to non-users, higher values indicating a diminished biological action of clopidogrel on platelet function. Subsequently the same research group conducted the OCLA study which was a prospective, doubleblind, placebo-controlled, randomized trial in which they demonstrated a signiûcantly decreased clopidogrel inhibitory effect on platelet P2Y12 as assessed by VASP phosphorylation test in patients receiving omeprazole 20 mg/day for 7 days compared to placebo.[12]

Though all PPIs are metabolized by CYP2C19, the clopidogrel-PPI interaction was not uniform among the different PPIs. Only omeprazole but not pantoprazole or esomeprazole was found to have attenuating effect on the platelet response to clopidogrel in a cross-sectional observational study in which multiple electrode platelet aggregometry (MEA) was used to measure adenosine diphosphate (ADP)-induced platelet aggregation.[13] The influence of omeprazole and pantoprazole on a high 150-mg clopidogrel maintenance dose was assessed in the PACA study which is a prospective, randomized controlled trial and it was found that patients receiving pantoprazole had a significantly better platelet response to clopidogrel as compared to those receiving omeprazole, as assessed by the platelet reactivity index vasoactive stimulated phosphoprotein (PRI VASP).[14] In another study where in vitro platelet aggregation was measured, it was found that omeprazole and rabeprazole but not lansoprazole significantly attenuated the anti-platelet function of clopidogrel in rapid metabolizers of CYP2C19 whereas in decreased metabolizers there was no significant change with any of these of three PPIs used concomitantly with clopidogrel.[15]

Following these mechanistic in vitro studies which have demonstrated conclusively the PPI-clopidogrel interaction, studies were carried out to estimate the relevance and magnitude of this interaction with regards to clinical outcomes. A systematic review of all studies reporting clinical outcomes has been performed and this included 18 studies investigating the PPI-clopidogrel interaction.[16] 13 of these studies were judged to be of a low a quality and 5 studies to be of a moderate or high quality. Out these 18 studies, 10 studies reported an increased risk of thrombotic events/ adverse cardiovascular outcomes with concomitant use of PPIs and clopidogrel. However all these 10 studies were judged to be of a low quality. The 5 studies which were judged to be of a moderate or high quality reported no statistically significant association of PPIclopidogrel interaction leading to an adverse clinical outcome.

In a meta-analysis which evaluated the effect of PPIs on the clinical outcomes in patients treated with clopidogrel, 159138 patients from 25 studies were included.[17] It was found that the concomitant use of PPIs with clopidogrel resulted in an increase in adverse cardiovascular events but it did not influence the risk of mortality. However there was significant heterogeneity among the studies included indicating that the evidence was biased, confounded or inconsistent.

Data from the COGENT trial is the best quality evidence that is available regarding the clinical outcomes of PPIclopidogrel interaction.[18] In the COGENT trial 5000 patients with an indication for a dual antiplatelet therapy following acute coronary syndrome or stent implantation were planned to be randomly assigned to omeprazole or placebo. The dual primary end points were gastrointestinal (a composite of overt or occult bleeding, symptomatic gastroduodenal ulcers or erosions, obstruction or perforation) and cardiovascular (a composite of death from cardiovascular causes, nonfatal myocardial infarction, revascularization or stroke). The trial was terminated prematurely owing to loss of finances by the study sponsor and only 3873 patients were included. Patients in the omeprazole group had significantly lesser primary GI end points compared to placebo whereas the number of patients reaching the primary cardiovascular end point was similar in both the groups. The authors concluded that among patients receiving aspirin plus clopidogrel, the prophylactic use of a PPI reduced the rate of upper gastrointestinal bleeding, and though there was no apparent cardiovascular interaction between clopidogrel and omeprazole, the results did not rule out a clinically meaningful difference in cardiovascular events due to use of a PPI.

To conclude, there is surrogate laboratory data supporting the biologically plausible PPI-clopidogrel interaction, however the available clinical evidence for the same is questionable with the high quality trials not showing detrimental clinical effects.

PPIs and bone health

Calcium absorption requires an acidic mileu in the stomach and upper small bowel to solubilise dietary calcium salts and facilitate the release of ionized calcium from insoluble calcium salts.[19,20] Physiologic compensatory mechanisms occur in response to impaired calcium absorption leading to secondary hyperparathyroidism which further can lead to increase in osteoclastic bone resorption, increase in skeletal turnover, decrease in bone mass and thence may result in an increase of risk of fractures (depicted in Figure 2).

There is a paucity of long-term studies with regards to the effects of PPIs on calcium absorption. Short-term studies revealed a highly variable effect of gastric acid suppression on calcium absorption and bone turnover.[21,22,23,24] At the level of osteoclasts, omeprazole at high concentrations was found to interact with the vacuolar adenosine triphosphatases on osteoclasts in vitro and in rodents, which has an antiresorptive effect on bones.[25] However, long-term administration to rats resulted in lower bone density.

Following the initial large population health database derived case-control studies published by Yang et al[26] and Vestergaard et al[27] in 2006, various other case-control studies as well as cohort studies have been done investigating the relationship between PPI usage and the risk of developing osteoporosis and bone fractures. No randomized controlled trial has been performed investigating the same. Two metaanalyses have been published recently regarding the risk of fractures with acid suppressive medications.[28,29]

Kwok et al found a significantly increased risk for spine fractures (odds ratio [OR] 1.50, 95% confidence interval [CI] 1.32–1.72), hip fractures (OR 1.23, 95% CI 1.11–1.36) and fractures overall (OR 1.20 95% CI 1.11–1.30) with usage of PPIs. Except for the data regarding spine fractures, substantial heterogeneity was found in the rest of the analyses in this study.[28] Yu et al found an increased risk of hip fractures (relative risk [RR] 1.30, 95% CI, 1.19-1.43),spine fractures (RR 1.56, 95% CI, 1.31-1.85) and any-site fractures (RR 1.16, 95% CI, 1.04- 1.30) among PPI users.[29] They reported similar findings in men and women as well as with duration of PPI usage of less than or more than 1 year. The authors also acknowledged a possibility of residual confounding bias in this analysis.

Thus, with the current available evidence there appears to be an increased risk of fractures with PPI usage, the risk being modest at best. However, with concerns regarding the trials’ methodology, properly designed prospective studies are needed to exactly quantify the risk of bone disease with PPI usage.

PPIs and pneumonia

Gastric acid kills ingested bacteria and thus prevents bacterial colonization (other than Helicobacter pylori) of upper gastrointestinal tract. Gastric bacterial overgrowth has been documented in situations of profound acid suppression.[30,31,32] This may further lead to development of respiratory colonization and pneumonia following microaspiration during episodes of gastroesophageal reflux especially in patients with compromised oropharyngeal protective reflexes.

Laheij et al have performed a population based cohort study to examine the association between the use of gastric acid– suppressive drugs and community-acquired pneumonia and it is the first large scale study of such kind.[33] They found that the incidence of pneumonia was about 4.5 times higher in patients who received acid-suppressive drugs (both PPIs and histamine-2-receptor antagonists [H2RA]) compared to those who did not receive them. Subsequently several other observational studies investigating the association between PPIs and risk of community-acquired pneumonia have been reported.

A meta-analysis has been performed recently examining the association between risk of community-acquired pneumonia and PPI use in adults.[34] They found a modest increase in the risk of community-acquired pneumonia with PPI use (OR 1.36, 95% CI 1.12–1.65). However there was significant heterogeneity among the studies included which hindered the investigators in arriving at definitive conclusions. In an exploratory subgroup analysis, shorter duration of PPI use was associated with an increased odds of communityacquired pneumonia [OR 1.92, 95% CI 1.40–2.63] compared to chronic usage [OR 1.11, 95% CI 0.90–1.38]. This is against the proposed theory of strong acid suppression leading to gastric bacterial colonization as the first step in the pathogenesis of pneumonia.

Hospital-acquired pneumonia has also been reported to occur at a higher incidence in patients exposed to acidsuppressive drugs than in the unexposed group (4.6% vs 2.0%) with the adjusted odds ratio for pneumonia being 1.3 (95% CI 1.1–1.4) in the exposed group.[35] This association was significant for PPIs but not H2RAs. However this study was conducted in non-ventilated patients. There is paucity of data in this regard in ventilated patients especially with the currently available potent intravenous PPIs.

To summarize, though biologically plausible, the overall evidence is suggestive of a modest increase in the risk of pneumonia with use of PPIs. However definite conclusions cannot be made because of the quality of evidence available.

PPIs and enteric infections

Clostridium difficile is a gram-positive, anaerobic sporeforming bacillus and is one of the most common causes of nosocomial diarrhoea and also has an increasing presence as a cause of community-acquired pathogen. Traditional risk factors for C.difficile associated diarrhoea (CDAD) are exposure to antibiotics, increased underlying disease severity, old age, immunodeficiency states and prolonged hospital stay.[36,37]

There are several recent studies which have examined the relationship between PPI usage and CDAD. A positive association between PPI usage and nosocomial[38,39,40,41] or community-associated[42,43] CDAD has been reported in some studies, but there have been some studies which reported no such association.[44,45,46,47] In a recent meta-analysis which included 12 studies evaluating 2,948 patients with C.difficile, a significant relationship between PPI use and C diffcile infection was found, with a pooled odds ratio of 1.96 (95% CI 1.28– 3.00).[48] However there was significant heterogeneity between the studies included. Though the data analyzed showed an association between PPI usage and CDAD, it was insufficient to establish causality.

Howell et al conducted a pharmaco epidemiologic cohort study investigating the relationship between gastric acid suppression and the risk of nosocomial CDAD.[49] They found a significant dose-response with more aggressive acidsuppression associated with higher odds ratios of developing CDAD. This evidence of a dose-response effect provided support for the potentially causal nature of iatrogenic acid suppression in the development of nosocomial CDAD.

In addition to CDAD, an association between other enteric infections and PPI usage has also been observed. This relationship has been evaluated in a recent meta-analysis of 11,280 patients from 6 studies.[48] The diagnosis of enteric infection was established either by microbial isolation of the pathogen (Salmonella, Campylobacter, Escherichia coli, C difficile, Shigella) or a clinical definition based on acute selflimited diarrheal disease. All 6 studies associated an increased risk of acute bacterial enteric infection with use of PPIs (pooled OR 3.33, 95% CI 1.84–6.02), but there was significant heterogeneity among the studies included. Though an association between enteric infections and PPI usage has been demonstrated, the available evidence is insufficient to establish causality.

The mechanism by which PPIs promote C. difficile infection is not clear. Though it has been widely postulated that acidic pH of the gastric contents may kill C. difficile spores and PPIs nullify this effect, recently evidence has been generated against this hypothesis. Nerandzic et al demonstrated survival of C. difficile spores in nonbilious gastric contents with acidic pH.[50] This raises the question whether the association found in some studies between CDAD and PPI usage is due to some unidentified confounders or whether the mechanism of interaction between PPI and CDAD is true but unknown yet. PPIs and pregnancy

Omeprazole is listed under class B risk category whereas other PPIs are listed under class C risk category for usage in pregnancy. In a recent meta-analysis 7 observational studies were included and data from 134,940 patients of whom 1,530 were exposed and 133,410 not exposed to PPIs, was analyzed.[51] No association was found between PPIs and major congenital birth defects (OR 1.12, 95% CI 0.86-1.45), spontaneous abortions (OR=1.29, 95% CI: 0.84-1.97), or preterm delivery (OR=1.13, 95% CI: 0.96-1.33). Secondary analysis with omeprazole alone also revealed no association with any major congenital malformations (OR 1.17, 95% CI 0.90-1.53).

In a subsequently published study, Pasternak and Hviid assessed the association between exposure to PPIs during pregnancy and the risk of major birth defects evident in a large cohort of live births over a 13 year period.[52] They found no significant association between exposure to PPIs in the first trimester of pregnancy and any major birth defects (OR 1.10, 95% CI 0.91 to 1.34). Similar results were found in secondary analyses of exposure to different individual PPIs. They also found that women who had received PPIs within 4 weeks before conception but not enough doses to have a theoretical chance of continued exposure beyond conception, were at increased risk of having infants with major birth defects (adjusted prevalence OR 1.53; 95% CI, 1.22 to 1.92). However, among women who received enough doses within 4 weeks before conception to have a theoretical chance of exposure in the first trimester, the exposure to PPIs was not significantly associated with major birth defects (adjusted prevalence OR 1.12; 95% CI, 0.94 to 1.35). The authors opined that the incongruent results regarding the risk of birth defects in women who were exposed to PPIs within 4 weeks before conception could either be unmeasured confounding or may just be by chance.

PPIs and acute interstitial nephritis (AIN) Several reports have implicated PPIs in causing AIN. The exact pathogenesis of PPI-induced AIN is not yet known though it is thought to be triggered by a hypersensitivity mmune
reaction to the drug or one of its metabolites. Usually patients present with non-specific symptoms like fatigue, nausea, malaise, weight loss and urinary examination usually reveals pyuria, eosinophiluria or proteinuria. Sierra et al performed a systematic review regarding PPI associated AIN and they found 64 reported cases of PPI related AIN, 59 of which were biopsy proven.[53] All of the PPIs, except for dexalansoprazole (which was not available at that time), were implicated, suggesting a class effect. They found that the available evidence may suggest a low-prevalence association , however it is insufficient to establish a causal relationship.

PPIs and B12 deficiency

Gastric acidity is important for the absorption of vitamin B12. Gastric acid along with pepsin facilitates the release of the dietary vitamin B12 which is tightly bound to protein.[54] PPIs may interfere with this process by increasing the gastric pH. Accordingly PPIs have been demonstrated to decrease absorption of protein-bound vitamin B12 in the majority of studies,[55,56,57] however there are a few studies which differed.[58] Investigations regarding the association between long-term PPI use and vitamin B12 deficiency revealed more inconsistent results.[57,59,60,61]

Valuck and Ruscin performed a case-control study to investigate the association between gastric acid suppression and B12 deficiency in older adults aged 65 years or more.[59] They found that chronic current but not past or current shortterm use of H2RA/PPI was associated with a significantly increased risk of vitamin B12 deficiency (OR 4.45; 95% CI 1.47- 13.34). Termanini et al have performed a prospective cohort study of 131 patients who received long-term PPI/ H2RA therapy for Zollinger–Ellison syndrome, and found a significant decrease of serum vitamin B12 levels in patients who were on omeprazole with the mean duration of treatment being 4.5 years.[60] The decrease occurred primarily among those with omeprazole-induced, sustained hyposecretion or complete achlorhydria, indicating that it was a specific effect of acid suppression. However in another long term prospective study by den Elzen et al in 125 patients aged 65 yr and above who received PPIs for more than 3 yr, no differences in mean vitamin B12 levels were observed between the long-term proton pump inhibitor users and their partners.[61] Schenk et al evaluated the absorption of protein-bound and unbound cyanocobalamin before and during treatment with omeprazole, and cobalamin levels in patients on long-term treatment with omeprazole, the mean treatment duration being 56 months.[57] Though they found a decrease in the protein-bound vitamin B12 absorption among subjects on short-term PPI therapy, there were no significant differences in mean serum levels of vitamin B12 levels with long-term omeprazole therapy.

PPIs and hypomagnesaemia

Magnesium is a predominantly intracellular cation which has an important role in various biological processes like membrane stabilization, neuromuscular excitability, oxidative phosporylation and nucleic acid synthesis. It is absorbed in the intestine mainly by passive diffusion through the paracellular route, an alternative active transport mechanism has also been identified recently.[62] Renal handling of magnesium plays an important role in regulating body magnesium status.

Recently several reports have thrown light of an association of hypomagnesaemia with PPI use, irrespective of the type of drug used indicating a class effect.[63,64] Prompt resolution of the hypomagnesaemia usually occurred within days of stopping the incriminated PPIs and also recurrence of hypomagnesaemia was demonstrated with rechallenge of PPIs. The true prevalence of PPI related hypomagnesaemia or the risk of inducing hypomagnesaemia with PPI usage is not known as it was never studied in a systematic fashion. Patients usually have non-specific symptoms such as anorexia, nausea, tremor, apathy, depression, agitation, confusion with milder cases of hypomagnesaemia being usually asymptomatic. One should be alert to investigate for hypomagnesaemia in conditions of difficult to correct hypokalaemia and hypocalcaemia.

The exact mechanism of hypomagnesaemia in patients with PPI usage is not known. Renal magnesium handling was found to be normal with PPI usage. No apparent difference was found in short term studies of intestinal absorption of magnesium with PPI usage.[65,66] However, there is no data to make conclusions regarding magnesium absorption with long term usage of PPIs as well as regarding fecal losses of magnesium with usage of PPIs.

PPIs and iron absorption

Dietary iron is present in food either as haem iron or non-haem iron, with the latter supplying about 2/3rds of the dietary iron requirements. Majority of the non-haem iron is in the ferric form and gastric hydrochloric acid plays an important role in its optimum absorption by reducing ferric iron to the more soluble ferrous form. In an in vitro study investigating the capacity of gastric juice produced by pentagastrin challenge to release radioiron through solubilisation from biosynthetically labelled bread, it was shown that below pH 2.5 there was a linear relation between pH and percentage solubility of dietary non-haem iron.[67] Similarly close correlation between iron absorption and the capacity of gastric juice to release dietary iron was demonstrated in an in vivo study.[67] However, longterm use of PPIs has not been reported to compromise iron status in normal subjects[68] as well as in patients with Zollinger- Ellison syndrome.[69] In another study done in fully treated patients of hereditary hemochromatosis, it was found that shortterm administration of PPIs resulted in a significant reduction in the absorption of non-haem iron and also led to a significant reduction in annual phlebotomy requirements in patients on long-term PPIs.[70]

PPIs and risk of neoplasia

PPI therapy leads to parietal cell inhibition and acid suppression. The physiological negative feedback effect of acidic pH on gastric secretion is thus lost leading to hypergastrinemia. Gastrin has been found to have trophic effect on epithelial cell growth and proliferation in stomach, colon and pancreas.[71,72,73,74,75,76,77,78] These are mainly animal studies and studies in tumour cell lines. However no human clinical studies have clearly established an increased risk of cancer at any of these sites due to hypergastrinemia occurring due to PPI use.

Hyperplasia of enterochromaffin-like cells latter on leading to carcinoid formation has been reported in rats in response to hypergastrinemia occurring in relation to profound acid suppression.[79] However, carcinoid formation in the setting of PPI use has not been reported in humans and only diffuse, linear, or micronodular hyperplasia of enterochromaffin-like cells was observed in 10–30% of chronic PPI users, particularly in H pylori positive patients with moderate to severe inflammation of body mucosa often with atrophic changes and more markedly increased gastrin levels.[80] Sporadic fundic gland polyps (FGPs) have been reported to occur with long term PPI maintenance therapy. FGPs occur due to parietal cell hyperplasia causing cellular protrusion into the glandular lumen leading to narrowing of the glandular lumen and cystic dilation of the gland. Thus it was proposed that FGPs occur due to mechanistic impairment of glandular flow.[81] Dysplasia is very rare to occur in these lesions. However the incidence and natural history of these lesions has not been studied properly.

In community health record based studies investigating the relationship between PPI usage and the risk of gastric cancer, an increased risk of gastric cancer associated with PPI use beyond the first year of treatment was reported.[82,83] However analysis of these data indicated that the observed risk of gastric cancer was because of confounding by indication rather than a causal relationship with PPI use. In another prospective, population-based cohort study, the risk estimates for gastric cancer among PPI users was found to be close to unity after incorporating a 1-year lag time within the diagnosis of gastric cancer, in contrast to a substantially increased overall incidence of gastric cancer among PPI users in analysis without lag time.[84] Confounding by indication was again found to be one of the main limiations of this study. In view of these limitations, the relationship between PPI usage and development of gastric cancer is not clear and larger studies of long-term PPI usage are needed to clarify this issue.

Though hypergastrinemia was found to be associated with an increased risk of colorectal carcinoma in a nested case control study,[85] several subsequently performed population based studies demonstrated no association between the long term usage of PPIs and the risk of development of colorectal cancer.[86,87,88] Regarding pancreatic cancer, no specific studies have been done to investigate the relationship between chronic usage of PPIs and the risk of development of pancreatic cancer.

Conclusion

PPIs in general have been reported to have low risk of serious adverse events. There have been concerns regarding some clinically important side effects recently and these have been reviewed in detail. Though there is a biologically plausible explanation for a possible interaction of PPIs with clopidogrel which can lead to potential thrombotic risks, such a negative interaction causing clinically meaningful outcomes has not been consistently demonstrated in clinical studies especially in prospective studies. A modest increase in the risk of osteoporotic fractures and the risk of infections such as C difficile diarrhoea, other enteric infections and pneumonia has been found. There has been reassuring data regarding the safety concerns of usage of PPIs in pregnancy. The results regarding interference of PPIs with absorption of B12 and iron have been inconsistent. There have been sporadic reports of association of PPIs with acute interstitial nephritis and hypomagnesaemia. Hypergastrinemia due to acid suppression by PPIs has been shown to be associated with occurrence of benign fundic gland polyps, however the risk of gastric cancer is not clear and no association has been found with colon cancer.

References