Introduction

Several animal and human studies support the conclusion that gastric acid secretion can affect the absorption of several nutrients, vitamins, and drugs [1–3]. The effect on absorption of vitamin B12, iron, calcium, and magnesium has received particular attention recently because of the widespread maintenance use of proton pump inhibitors (PPIs) and H+K+ATPase inhibitors (omeprazole, lansoprazole, pantoprazole, esomeprazole, rabeprazole), which are potent acid suppressants [4•, 5–8]. These drugs generate more than $13.5 billion in sales (third largest selling drug class), and in 2009 more than 119 million PPI prescriptions were written in the United States; therefore, PPIs are widely used, and many patients take them for extended periods [9]. This situation is particularly true in patients with gastroesophageal reflux disease (GERD), which occurs monthly in up to 40% of American adults; in the proportion of patients with moderate to severe GERD, long-term maintenance treatment with PPIs is needed to control symptoms [2]. Many studies show the risk of minor adverse effects from PPIs is low, with rates of withdrawal in various studies of less than 1% to 2% [8]. Furthermore, the risk of long-term adverse events in most large reports is low; however, because large numbers of patients take these drugs on a long-term basis, and because of their known effects on nutrient absorption, the possible long-term effects in this area are receiving increasing attention. These studies report conflicting results, particularly the possibility that long-term PPI use increases the occurrence of bone fractures (possibly by decreasing calcium absorption) [6–8, 10–15]; can result in vitamin B12 deficiency [2, 3, 7, 8, 13, 16–18]; and can lead to iron deficiency [2, 3, 7, 8, 19–21]. The recent results in each of these controversial areas are briefly reviewed in this article, concentrating on articles published within the past few years.

Long-Term Use of PPIs: Effects on Vitamin B12 Absorption

It is well established that gastric acid secretion is needed for dietary vitamin B12 (VB12) absorption from foods [2, 3, 22] (Table 1). VB12 is an essential nutrient that must be acquired from the diet and is present in foods bound to protein. The presence of gastric acid is needed for the pancreatic proteases to cleave the VB12 from the protein, allowing its reassociation with intrinsic factor and eventual absorption in the terminal ileum [2, 3, 18, 22]. In short-term studies, various acid suppressants (histamine-2 receptor antagonists [H2RAs], PPIs) were reported to decrease the absorption of VB12 from foods, but not to decrease the absorption of crystalline VB12, which is not protein bound [2, 18, 23–25].

Table 1 Long-term studies of effects of proton pump inhibitors on vitamin B12
Full size table

The most recent study examined 36 long-term care patients, aged 60 to 80 years (17 taking long-term PPIs, 19 not taking PPIs), and the effect of a VB12 nasal spray for 8 weeks on VB12 status (Table 1) [26•]. At baseline, the chronic PPI users had lower serum VB12 levels and higher methylmalonic acid levels (MMA), and a greater percentage were VB12-deficient (75 vs 11%, P = 0.006). After 8 weeks of VB12 nasal spray (500 μg, once weekly), serum VB12 levels were significantly increased in comparison to pretreatment in the chronic PPI users, and the frequency of VB12 deficiency was significantly decreased from pretreatment in the chronic PPI users (75% to 24%, P = 0.012) [26•]. These authors concluded that older individuals who are long-term PPI users are at increased risk of VB12 deficiency, and should be more systematically screened for VB12 deficiency than currently performed in most institutions of chronic care. Limitations of this study include the open study design, no placebo control, and relatively small number of subjects.

In 2008, three studies evaluated the effects of PPIs on VB12 status and reached different conclusions (Table 1) [16, 27, 28]. Two were cross-sectional studies of elderly patients [16, 27] and one was a longitudinal study in patients with Zollinger-Ellison syndrome (ZES) [28]. In the first cross-sectional study, the effects of chronic use (> 6 years) of H2RAs (150 patients), PPIs (141 patients), or neither (251 patients) were examined in elderly patients in nursing homes or at community ambulatory care facilities [27]. Low/marginally low VB12 status was found in 20% of the nursing home patients and 29% of the community care patients, consistent with other studies reporting 25% in such patients, with a range of 3% to 40% [17, 27]. VB12 deficiency can cause neurologic disorders, including neuropathy, spinal cord degeneration, gait disorders (leading to falls), depression, and dementia, which are reversible if diagnosed in time. These results demonstrate VB12 deficiency is a problem in older persons, and thus the importance of identifying any contributing factor. In the first cross-sectional study, PPI use, but not H2RA use, was associated with lower VB12 levels, the percentage decrease in VB12 levels correlated with the time of PPI use, and concomitant use of oral VB12 delayed, but did not prevent, this decrease (Table 1) [27]. In the second cross-sectional study, serum VB12 levels, homocysteine (HCY) levels, and mean corpuscular volume (MCV) were compared in 125 older (> 65 years), long-term PPI users and their partners not taking PPIs (Table 1) [16]. No differences in serum VB12, serum HCY, or MCV were detected. These two studies differed in populations studied, with older patients in the first study (81 vs 73 years) [27]. Other differences were the percentage of female patients (63% vs 50%) and possibly ethnicity. Whether these factors contributed to the differences in results is unclear.

The third 2008 study was a longitudinal study of patients with ZES (Table 1) [16]. Surgery offers a long-term cure in only 40% of patients with ZES [29], thus they require lifelong acid antisecretory treatment, for which PPIs are now the drugs of choice [30, 31]. In this study of 61 acid hypersecretors (46 with ZES) receiving long-term treatment (up to 12 years) with PPIs, 10% had low VB12 levels and 31% had normal levels with VB12 deficiency (increased HCY/MMA with normal folate) [16]. No decrease was seen in serum VB12 levels with duration of PPI treatment. This study’s potential deficiencies were lack of clarity regarding whether all patients had all studies yearly, lack of a control group not taking PPIs, and no correction for possible multivitamin use. This study’s results differ from a previous prospective study [18] involving 130 patients with ZES followed for a mean of 4.5 years, in which patients treated with PPIs developed lower levels of VB12, but not lower levels of folate; the lower levels of VB12 correlated with the presence of PPI-induced hyposecretion, and the duration of PPI treatment correlated inversely with VB12 levels (P = 0.013). A limitation of the latter study was that serial HCY/MMA levels were not measured, so the true level of VB12 deficiency may have been higher.

Two older case-control studies dealing with this subject are included in Table 1 [17, 32]. In a case-control study from 2004 among a geriatric population (≥ 65 years), the use of H2RA/PPIs and several other variables were compared in 53 patients with vitamin B12 deficiency to 212 controls (matched for age, gender, multivitamin use, and frequency of Helicobacter pylori infection) [17]. The current chronic use of H2RA/PPIs was associated with a significant increase in risk of VB12 deficiency (OR, 4.45), and no association was found with past or short-term H2RA/PPI use. The authors concluded their results support an association between chronic use of H2RA/PPIs by older adults and the development of VB12 deficiency. In a second case-control study from 2003, involving 10,844 statewide Medicaid patients, 125 patients who had parenteral VB12 supplementation started were identified, and the frequency of chronic H2RA/PPI use and other variables were compared to 500 age- and gender-matched controls (Table 1) [32]. For the patients requiring VB12 supplementation, 18% had been exposed to chronic suppressive acid therapy (> 12 months of H2RA/PPI) compared to 11% of the control group, which was a significant difference (OR, 1.82; CI 1.08–3.09; P = 0.025). These study investigators concluded that an association exists between the need for parenteral VB12 supplementation and chronic suppressive gastric acid therapy.

Many of the studies summarized in Table 1 suggest an association among the prolonged, chronic use of gastric acid-suppressant drugs (particularly PPIs), the development of lower VB12 levels, and an increased frequency of VB12 deficiency, especially in the older population. However, this finding is neither firmly established nor widely acted on, and thus remains controversial. Although needed, randomized trials would be costly and adequate control groups would be difficult to define; thus it is unlikely they will be forthcoming. Based on the available information, it would seem appropriate to evaluate VB12 status at appropriate intervals in long-term users of PPIs, especially in the older population, who may have poorer nutrition and lower body stores initially, and in unique groups of patients requiring lifelong PPI treatments (eg, those with ZES or other gastric acid hypersecretory states).

Long-Term Use of PPIs: Effects on Calcium Absorption/Metabolism and Bone Fractures

The effect of PPIs on calcium absorption/metabolism has received much attention recently and is even more controversial than its possible effect on VB12 status. In a 2006 nested case-control study using the General Practice Research database from the United Kingdom, Yang et al. [6] reported that long-term PPI therapy, particularly at high doses, is associated with an increased risk of hip fractures (Table 2). The study cohort consisted of PPI users and non-users of acid suppression drugs who were aged older than 50 years, and included all patients with an incident hip fracture between 1987 and 2003 (Table 2) [6]. PPIs (OR 1.44; 95% CI, 1.3–1.59) and H2RAs (OR 1.23; 95% CI, 1.09–1.40) taken for longer than 1 year were associated with an increased risk of hip fractures, and the risk was significantly greater with PPI use than with H2RAs (adjusted OR [AOR] for >1 year of use 1.34; 95% CI, 1.14–1.38). The adjusted rate of hip fractures was significantly higher in patients who were prescribed long-term, high-dose (> 1.75-fold average daily dose) PPIs (AOR, 2.65), and the risk progressively increased with the duration of PPI treatment [6]. The positive association with hip fractures and PPI therapy was stronger in men (OR 1.78; 95% CI, 1.42–2.22) than in women. These results were consistent with a 2006 case-control study by Vestergaard et al. [33] in a Danish population, which showed that PPI use was associated with an increased risk of hip fractures (OR, 1.45; 95% CI, 1.28–1.165) (Table 2). In contrast to the study by Yang et al. [6], the study by Vestergaard et al. [33] observed neither a dose-response effect nor a duration-response effect of PPI use with hip fractures. This finding may be related to the difference in follow-up in the two studies, with a follow-up of 15 years in the study by Yang et al. [6] and only 5 years in the study by Vestergaard et al. [33]. Several subsequent studies showed differing results, which are discussed in this section (Table 2).

Table 2 Long-term studies of effects of proton pump inhibitors on calcium absorption/metabolism and/or bone fractures
Full size table

Limited animal and human studies show that gastric acid secretion can facilitate calcium absorption and that acid suppressants (eg, PPIs) can decrease calcium absorption and decrease bone density [4•, 6, 13, 34–38]. An acidic environment in the stomach facilitates the release of ionized calcium from insoluble calcium salts, and calcium solubilization is thought to be important for calcium absorption [4•, 6, 13, 34–38]. Several conditions that cause hypochlorhydria or achlorhydria in humans (eg, gastrectomy, pernicious anemia, and atrophic gastritis) are associated with an increased occurrence of osteoporosis and bone fracture; it is assumed that these conditions are secondary to the effect of low gastric acid levels on calcium absorption [4•, 6, 13, 34–36, 39]. Limited experimental evidence indicates that PPIs also may potentially influence bone resorption by inhibiting the osteoclastic proton transport system, thus ameliorating the negative effect of PPIs increasing the occurrence of osteoporosis by decreasing calcium absorption [6, 39–41].

The large, case-control studies by Yang et al. [6] and Vestergaard et al. [33] led to considerable interest in the possibility that chronic PPI use could lead to an increase in bone fractures, and resulted in speculation about the possible mechanisms [5, 9, 10, 13, 42–46]. These studies and others reviewed below aroused sufficient attention that in May 2010, the US Food and Drug Administration (FDA) issued a warning of the “possible increased risk of fractures of the hip, wrist, and spine with high doses or long-term use of a class of medications called proton pump inhibitors. The product labeling will be changed to describe this possible increased risk” (US FDA News Release, May 25, 2010).

Since 2006, five studies [14••, 33, 37, 43, 47] reported that long-term use of PPIs is associated with an increased occurrence of bone fractures, whereas two studies [12, 15] reported no association. Only Roux et al. [48••] prospectively studied the effect of PPIs on bone fractures, whereas the others were either nested control studies, cross-sectional studies, or cohort studies (Table 2). The investigation by Roux et al. [48••] was confined to 1211 postmenopausal females who were studied at baseline and at the end of a 6-year period for vertebral fractures assessed by radiographs and correlated with PPI use (Table 2). The age-adjusted rate for vertebral fractures was 3.1-fold higher in chronic PPI users compared to nonusers (1.89 vs 0.60 per 100 person-years). This increase is larger than the 18% risk for any fracture, 45% increase for hip fracture, and 60% increase for spine fracture with chronic PPI use reported by Vestergaard et al. [33]; the 23% increase reported for hip fractures by Yang et al. [6]; the 34% increase reported for women in nonspinal fractures by Yu et al. [9]; the 92% increase reported by Targownik et al. [43]; the 30% increase reported by Corley et al. [47]; and the 47% increase reported for spine and 26% increase for total fractures reported by Gray et al. [14••] (Table 2).

The study by Yang et al. [6] presented evidence of both an increasing effect of higher doses of PPIs on hip fractures and an increasing effect with longer durations of chronic PPI use (Table 2). Similar results were reported by Corley et al. [47]. In the 2008 study by Targownik et al. [49], a time-dependent effect was seen: PPI use for 6 years or less was not associated with an increased occurrence of all fractures; however, PPI use longer than 6 years was associated with a 92% increase in all fractures, and PPI use longer than 5 years was associated with a 62% increase in hip fractures (Table 2). In contrast, no dose effect was seen in the study by Vestergaard et al. [33]. These results raise the possibility that a factor contributing to the variability in the different studies (other than different study methods, populations, and methods of assessing fracture rate) is the failure to clearly define the daily amount of PPI and the duration of its use in the patients studied. This issue is a particular problem now that PPIs are available over the counter, and can be missed as a patient medication even with careful questioning.

Although most of the above studies support the conclusion that chronic PPI use is associated with an increased occurrence of bone fractures, the likely mechanism of this effect is not clear [5, 12, 13, 37, 38, 42, 45, 48••]. The most widely assumed mechanism is that long-term PPI use leads to decreased intestinal absorption of calcium, resulting in negative calcium balance, increased osteoporosis, development of secondary hyperparathyroidism, increased bone loss, and increased fractures [5, 12, 13, 37–39, 42, 45, 48••]. Each of these findings is controversial, even the effect of PPIs on calcium absorption. Calcium is thought to be absorbed in ionized form primarily in the upper small intestine, and the ionization is facilitated by an acidic medium to release calcium from its salt form or food complex [4•, 6, 13, 34–38]. Animal studies show that PPIs, H2RAs, and achlorhydria can reduce calcium absorption and/or acid increased calcium absorption [34, 42]. Short-term studies in humans have provided conflicting results. In some studies, PPIs, H2RAs, or achlorhydria were shown to decrease calcium absorption [35, 39, 50, 51], with omeprazole causing a 41% decrease in one study [35], whereas in other studies they have not decreased calcium absorption (Table 2) [38, 39, 52–55]. The 2010 study by Wright et al. [38] is particularly well done: dual, stable-isotope, state-of-the-art methods were used to assess changes in serum and urinary calcium in a placebo-controlled, double-blind, cross-over study in 12 healthy volunteers, with or without treatment for 3 days with omeprazole (20 mg, twice daily) (Table 2). In this study, neither calcium absorption nor urinary calcium levels differed between PPI treatment periods and placebo treatment, despite a marked inhibition of acid secretion in the PPI-treated group [38]. The factors contributing to these markedly different results are unclear, although one important variable in these studies may be whether they are done fasting or in a fed state [38]. Furthermore, the effects of PPIs on skeletal metabolism have not been well studied, and the available studies give differing results. In some investigations, markers of bone turnover have been altered by PPI treatment in humans, suggesting that PPIs alter bone resorption, whereas in other studies, no effect on bone turnover was seen [39, 40, 56].

The ability of PPIs to alter bone mass and/or cause osteoporosis is also unclear. In animal studies, the PPI omeprazole reduced bone density [57]. In contrast, some human studies report no effect of PPIs to cause alterations in bone turnover and/or bone density, as well as no effect in causing osteoporosis. Four studies reviewed in Table 2 also investigated effects of PPIs on bone mineral density, with differing results [11, 12, 14••, 37]. In the 2010 study by Gray et al. [14••], the use of PPIs for more than 3 years was associated with a marginal decrease in hip bone mineral density (BMD), but not at other sites. In a study of patients with chronic renal disease who were on dialysis, those who chronically used PPIs had lower BMDs at all sites (P = 0.01–0.04). In the study by Yu et al. [37], women, but not men, chronically using PPIs had lower hip and femur BMD (P < 0.01), and no increased rate of bone loss was seen in PPI/H2RA users (P = 0.09). In the 2010 study by Targownik et al. [12], chronic PPI use was not associated with having osteoporosis of the hip or lumbar spine, and in the longitudinal part of the study, PPIs did not cause a significant change in BMD in either the hip or spine. These investigators concluded that the association between PPI use and hip fracture was probably related to factors independent of osteoporosis [12].

Other possible explanations for an effect of PPI on bone fractures and bone metabolism include PPI-induced hypergastrinemia resulting in parathyroid hyperplasia/hypertrophy and increased parathyroid hormone secretion [42, 56]; the ability of PPIs to alter the osteoclastic-based vacuolar proton pump may contribute to alterations in bone turnover and thus to fracture risk [33, 58]; an increased occurrence of comorbidities (eg, thiazide use, different body mass index, different alcohol intake) that contribute to the development of bone fractures, in PPI users in some studies [12, 48••]; low VB12 levels, which have been associated with a lower BMD, may be caused by the PPIs [48••, 59, 60]; PPI users may have a different diet because of intolerances secondary to gastritis [33]; or the bone alterations may be related to PPI aggravation of gastric disease, particularly from H. pylori [33].

Long-Term Use of PPIs: Effects on Iron Absorption

Relatively few studies have assessed directly the long-term effect of chronic PPI use on iron absorption. The results of the available studies are controversial.

Numerous animal and human studies support the conclusion that the absorption of iron is affected by gastric acidity [2, 24, 61]. Dietary iron is present in food as either non-heme (66%) or heme iron (32%); absorption of non-heme iron is markedly improved by gastric acid. Gastric acid helps the food sources containing non-heme iron to dissociate and to solubilize the iron salts, allowing their reduction to the ferrous state, which allows the formation of complexes with ascorbate, sugars, and amines, in turn facilitating absorption [2, 24, 61]. Numerous clinical conditions associated with achlorhydria/hypochlorhydria (atrophic gastritis, pernicious anemia, gastric resections, vagotomy) have been shown to be associated with decreased iron absorption and/or iron-deficiency anemia [2, 24, 61]. In rats, PPI treatment decreased iron absorption in animals taking a low-iron diet [61]. In some studies of patients with long-term PPI use, evidence was found for decreased iron absorption, which was attributed to the PPI (decreased ferritin, iron levels, iron deficiency anemia) [20]; however, no effect was seen in other studies [21, 62]. The former study [20] was a case report of two anemic patients who failed to respond to oral iron treatment while taking a PPI, but whose iron status improved when the PPI was withdrawn. One patient tested on the PPI demonstrated decreased iron absorption, leading the authors to attribute the failure to respond to oral iron replacement to malabsorption secondary to PPI use [20]. In contrast, in a study involving 109 patients with ZES who require lifelong PPI treatment (and who had continuous PPI treatment for at least 6 years), over a 4-year period, no evidence was found of iron deficiency, decreased absorption, or decreased iron stores, although decreased levels of VB12 were found in many of these patients [18]. Patients with hereditary hemochromatosis are treated with frequent phlebotomies, which increase intestinal iron absorption [63•]. In a study of seven such patients [63•], PPI administration for 7 days decreased non-heme iron absorption from a meal, and long-term PPI use resulted in a significant reduction (P < 0.01) in the yearly volume of blood that needed to be removed to keep body iron stores at the appropriate level [63•].

Long-Term Use of PPIs: Effects on Magnesium Absorption

Hypomagnesemia has been reported with PPI use in fewer than 25 cases [64–67, 68•, 69–71]. In a recent review of 10 cases [68•], the patients had been taking PPIs a mean of 8.3 years; they presented with severe symptomatic hypomagnesemia (≤ 0.54 mmol/L); and morbidity was significant (fatigue, unsteadiness, paresthesia, tetany, seizures, cardiac arrhythmias, hospitalizations). The hypomagnesemia resolved when PPI therapy was stopped and recurred if PPI therapy was re-introduced [64, 65, 68•]. In some cases, hypomagnesemia was accompanied by hypokalemia and/or hypercalcemia [66, 68•, 69].

At present, the mechanism of PPI-induced hypomagnesemia is not clear. One study tested the hypothesis that it occurs in poor metabolizers of PPIs, but that was not the case [69]. The study investigators concluded that hypomagnesemia is not specific to a given PPI, but is a generic problem with the PPI class of drugs, because it recurs even when PPIs are changed from one to the other [69]. It was proposed that PPI-induced hypomagnesemia is likely caused by gastrointestinal magnesium loss, although this is unproven at present [65, 68•, 69].

Conclusions

The data reviewed here support the importance of long-term investigations of the possible effects of chronic PPI treatment on absorption of important nutrients, including calcium, vitamin B12, iron, and magnesium. In general, the studies in each area have led to differing conclusions, but when examined systematically, several studies show consistent results supporting the conclusion that long-term adverse effects on these processes can have important clinical implications. In addition to studies of bone fractures, more prospective studies are urgently needed for each of these nutrients. Furthermore, whereas the clinical implications in many cases are much better defined, in almost all cases, the mechanisms of the observed clinical effects are unclear. Therefore, detailed, careful studies of the long-term effects of PPIs on the absorption of these nutrients (vitamins, minerals) and studies of PPI mechanism(s) for inducing clinical problems potentially related to these processes (fractures, anemia, VB12 deficiency manifestations, hypomagnesemia) are critically needed.