RESEARCH REPORTS Gastroenterology
Proton Pump Inhibitor–Associated Hypomagnesemia: What Do FDA Data Tell Us? Chee Phun Luk, Richard Parsons, Ya Ping Lee, Jeffery David Hughes
P
roton pump inhibitors (PPIs) are a class of medications indicated for the treatment of gastric acid–related diseases such as gastroesophageal reflux disease and peptic ulcer disease, as well as part of Helicobacter pylori eradication regimens and for the prevention and treatment of nonsteroidal antiinflammatory drug–induced gastric injury. PPIs act by inhibiting the proton pump or H+/K+ adenosine triphosphatase found on gastric mucosal parietal cells, therefore preventing both basal and stimulated gastric acid secretion.1 PPIs on the market are omeprazole, esomeprazole, pantoprazole, rabeprazole, lansoprazole, and dexlansoprazole.1,2 Because of their superior efficacy and consistency in gastric acid suppression compared to histamine H2 antagonists, PPIs are widely used in many countries.2,3 In the US, PPIs ranked as the third-largest selling drug class in 2009.4 In Australia,
Author information provided at end of text.
© 1967-2013 Harvey Whitney Books Co. All rights reserved. No part of this document may be reproduced or transmitted in any form or by any means without prior written permission of Harvey Whitney Books Co. For reprints of any article appearing in The Annals, please contact
[email protected]
theannals.com
background: Proton pump inhibitors (PPIs) are a class of medications indicated
for the treatment of gastric acid–related diseases. Hypomagnesemia is a rare but serious adverse effect of PPIs.
objective: To address the association between the use of different PPIs and hypomagnesemia by examining the frequency of occurrence of hypomagnesemia among the reported adverse drug reactions from the Food and Drug Administration (FDA) Adverse Event Reporting System database.
methods: We conducted a cross-sectional study of PPI-associated adverse effect cases reported to the FDA between November 1, 1997, and April 1, 2012. Logistic regression was used to examine the association of sex, age, and different PPIs with hypomagnesemia. χ2 Analysis was conducted to investigate the association of PPIassociated hypomagnesemia with hypocalcemia and hypokalemia. results:
Among 66,102 subjects identified as experiencing 1 or more adverse effects while taking a PPI, 1.0% (n = 693) were reported to have hypomagnesemia. The mean (SD) age of PPI users presenting with hypomagnesemia was 64.4 (12.9) years. Results from logistic regression indicated that, compared with esomeprazole, all other PPIs had a higher rate of hypomagnesemia, with pantoprazole having the highest rate (OR 4.3; 95% CI 3.3-5.7; p < 0.001). The risk of female subjects having hypomagnesemia (OR 0.83; 95% CI 0.71-0.97; p = 0.016) was significantly lower than that of males. Elderly subjects (age >65 years) were at increased risk of PPIassociated hypomagnesemia (OR 1.5; 95% CI 1.2-1.7; p < 0.001). χ2 Analysis showed strong association between hypomagnesemia and both hypocalcemia (p < 0.001) and hypokalemia (p < 0.001). conclusions:
All PPIs were associated with hypomagnesemia, with esomeprazole having the lowest risk and pantoprazole having the highest risk. The risk of PPI-associated hypomagnesemia was higher in males and the elderly population. Hypocalcemia and hypokalemia commonly coexisted with PPI-associated hypomagnesemia. Ann Pharmacother 2013;47:773-80.
Published Online, 30 Apr 2013, theannals.com, doi: 10.1345/aph.1R556
The Annals of Pharmacotherapy
Downloaded from aop.sagepub.com by guest on October 11, 2013
■
2013 June, Volume 47
■
773
CP Luk et al.
esomeprazole and pantoprazole are 2 of the top 10 drugs by prescription counts.5 With health practitioners continuing to increasingly prescribe PPIs for acid suppression and the availability of certain PPIs over the counter, there is increasing concern about the potential long-term adverse effects of these drugs.6-8 Reported adverse effects associated with longterm use of PPIs tend to be rare but severe, and include bone fractures, enteric infections, community-acquired pneumonia, Clostridium difficile–associated diarrhea, hypergastrinemia, and nutritional deficiencies such as vitamin B12 deficiency and hypomagnesemia.1-4,6-9 Hypomagnesemia is a rare but serious adverse effect that went undetected during the clinical trial phase of PPIs.10 The first case of PPI-associated hypomagnesemia was reported in 2006, despite the introduction of the first PPI (omeprazole) in 1989.11 Subsequently, a number of other case reports of PPI-induced hypomagnesemia have emerged. To date, 23 case reports and case series of PPI-induced hypomagnesemia have been reported and published.11-33 Patients in these case reports often exhibited low serum magnesium levels accompanied by a range of clinical symptoms, which ranged from vomiting, diarrhea, and cramps, to tetany, confusion, seizures, and QT interval prolongation. Some patients had symptoms that were severe enough that the PPI needed to be discontinued and their serum magnesium levels restored.11-30,33 Often, patients with PPI-associated hypomagnesemia develop other electrolyte disturbances, such as hypocalcemia and hypokalemia, that are refractory to calcium and potassium supplements unless hypomagnesemia is corrected.7,13,34 Two cross-sectional studies have shown that PPI use may be associated with hypomagnesemia, particularly in individuals aged 65 years or older and those with chronic kidney disease.35,36 Both of these studies were observational and cannot provide clear evidence of a causal relationship. In March 2011, the Food and Drug Administration (FDA) issued a safety announcement regarding the association between PPI use and hypomagnesemia.37 In June 2011 the Australian Medicines Safety Update of Therapeutic Goods Administration released a similar alert about the risk of hypomagnesemia with PPI use.38 Building upon the knowledge and information from published case reports, observational studies, and reviews, this study addresses the association between the use of different PPIs and hypomagnesemia by examining the frequency of occurrence of hypomagnesemia among the reported adverse drug reactions from the FDA Adverse Event Reporting System (AERS) database.
methods
study design
This was a cross-sectional study of cases of hypomagnesemia associated with PPI use reported to the FDA and a 774
■
The Annals of Pharmacotherapy
■
review of cases of PPI-associated hypomagnesemia published in the medical literature. data collection from the fda
To obtain data from the FDA, a request was made for all case reports of adverse effects associated with the use of the 6 marketed PPIs. The FDA provided the deidentified records from their AERS database. Case reports included all subjects who were taking any of the 6 PPIs and a list of all the adverse effects that they experienced. The period covered by the data review was from November 1, 1997, to April 1, 2012. Records were scanned for the words hypomagnesemia, blood magnesium decreased, and magnesium deficiency; if any of these were found, the record was marked as hypomagnesemic. Published literature
A literature review was conducted by searching MEDLINE articles discussing the association of PPI use with hypomagnesemia. The search was restricted to English-language articles from 2002 to the present. Keywords used for the search included proton pump inhibitors, hypomagnesemia, magnesium absorption, magnesium deficiency, magnesium, omeprazole, esomeprazole, pantoprazole, lansoprazole, rabeprazole, dexlansoprazole, and other related MeSH terms. Searches were also conducted on other online databases such as Science Direct as well as Google Scholar using the same key words. The reference lists of articles obtained from the online search were reviewed for additional material. statistical analysis
Standard quantitative statistical methods were used to summarize the main features of the FDA data file. The data were received from the FDA as a set of unformatted text files, each of which contained details for each adverse event including a unique record identifier, age, sex, all adverse events experienced, and all drugs being taken, along with the manufacturer of the drugs, route of administration, and dose. One file was provided for each of the PPIs. The files were assembled into a single file, and then separated into a register (1 record per person), a drug file (possibly several records per person), and a file of adverse effects (possibly several records per person). The sex and age of all subjects in the register were summarized using standard descriptive statistics. The file of adverse effects was searched for the words hypomagnesemia, blood magnesium decreased, magnesium deficiency, hypocalcemia, blood calcium decreased, calcium deficiency, hypokalemia, blood potassium decreased, and potassium deficiency; if they were found, the corresponding record in the register was marked. Tables were then generated to describe the frequency of each of these 3 adverse effects
2013 June, Volume 47
theannals.com
among subjects taking each of the PPIs. The possible association between hypomagnesemia with hypocalcemia and hypokalemia was assessed using the χ2 statistic and the Φ coefficient. A multivariate logistic regression model was fitted to the data, with the presence of hypomagnesemia as the dependent variable, and PPI, age, and sex as independent variables. Because many cases were missing data for age, a separate category for this group was created so that they could remain in the analysis. Analyses were carried out using SAS statistical software version 9.2 (SAS Institute Inc.). For all statistical tests, p < 0.05 was the indicator of a significant association. results
A total of 66,102 subjects were identified from the FDA database as experiencing 1 or more adverse effects while taking a PPI. Among them, 57.3% were female (n = 37,854) and 38.2% were male (n = 25,254). Sex was not recorded for a small proportion of the records (4.5%). While age was not recorded for almost one third of the records in the file, the mean (SD) age of the remaining two thirds of PPI users pre-
FDA Data on PPI-Associated Hypomagnesemia
senting with adverse effects was 60.6 (17.7) years. For all 6 PPIs, there was a higher frequency of female subjects than male subjects reported to have adverse effects. The largest number of adverse effect reports came from esomeprazole users, followed closely by the omeprazole users. Table 1 shows the age and sex distribution of the individuals who experienced PPI adverse effects. Table 2 shows the age and sex distribution of the subjects who experienced hypomagnesemia. The sex distribution was approximately equal across the different PPI groups, except for esomeprazole and rabeprazole, for which there was a predominance of females (68.8% and 81.5%, respectively). The predominance of females was similarly observed in the sex distribution of the whole group (Table 1). The mean age of PPI users who experienced hypomagnesemia was 64.4 (12.9) years. Among the 66,102 subjects reported to have experienced adverse effects with PPI use, 1.0% (n = 693) presented with hypomagnesemia. Hypomagnesemia represented a small proportion of the adverse effects for all PPIs. In the analyses comparing the presence of hypomag-
table 1. Age and Sex Distribution of Proton Pump Inhibitor Users with Adverse Effects Based on FDA Adverse Effect Reporting System drug
male, n (%)
sex
female, n (%)
Esomeprazole
7,312 (31.8)
14,990 (65.3)
Pantoprazole
2,842 (36.9)
4,234 (55.0)
Lansoprazole Omeprazole
Rabeprazole
Dexlansoprazole TOTAL
4,069 (45.8)
9,271 (41.6)
1,656 (42.1)
104 (30.2)
25,254 (38.2)
FDA = Food and Drug Administration.
4,426 (49.8)
unknown, n (%)
22,969
60.9 (16.0)
626 (8.1)
7,702
61.0 (18.4)
392 (4.4)
1,208 (5.4)
232 (67.4)
8 (2.3)
37,854 (57.3)
mean (sd) age, years
667 (2.9)
11,790 (52.9)
2,182 (55.5)
total, n
8,887
22,269
93 (2.4)
3,931
344
2,994 (4.5)
66,102
61.3 (18.0)
59.9 (19.0)
61.8 (16.5)
58.3 (14.5)
60.6 (17.7)
table 2. Age and Sex Distribution of Proton Pump Inhibitor Users Who Experienced Hypomagnesemia Based on FDA Adverse Effect Reporting System drug
Esomeprazole
Lansoprazole
Omeprazole
Pantoprazole
Rabeprazole
Dexlansoprazole TOTAL
male, n (%)
female, n (%)
41 (56.9)
31 (43.1)
28 (30.1)
64 (68.8)
181 (49.2)
184 (50.0)
1 (50.0)
1 (50.0)
65 (49.6) 5 (18.5)
321 (46.3)
FDA = Food and Drug Administration. a SD not determined, as age was reported in only 1 case.
theannals.com
sex
64 (48.9)
22 (81.5)
366 (52.8)
unknown, n (%)
total, n
1 (1.1)
93
0
72
3 (0.8)
368
2 (1.5)
131
6 (0.9)
693
0
27
0
The Annals of Pharmacotherapy
2
■
mean (sd) age, years
61.4 (12.5)
61.5 (12.4)
65.0 (13.4)
66.0 (13.0)
64.7 (5.2)
60.0a
64.4 (12.9)
2013 June, Volume 47
■
775
CP Luk et al.
nesemia across the different drugs, esomeprazole was used as the reference for odds ratio calculation. Dexlansoprazole was not included in the analysis, as there were only 2 cases of hypomagnesemia associated with it. All other PPIs were associated with a higher risk of hypomagnesemia compared to esomeprazole (all ORs >1), with pantoprazole and omeprazole having the highest risk (OR 4.3; 95% CI 3.3-5.7: p < 0.001; and OR 3.8; 95% CI 3.0-4.8; p < 0.001, respectively) (Table 3). While the OR for hypomagnesemia associated with rabeprazole was also higher than 1 (1.5), it was not statistically significantly different from the rate for esomeprazole (p = 0.077). The risk of female subjects having hypomagnesemia (OR 0.8; 95% CI 0.7-0.97; p = 0.016) was significantly lower than for males. While comparing the risk of hypomagnesemia according to the age group (with above 65 years as the older age group), the older age group showed a higher risk of hypomagnesemia (OR 1.5; 95% CI 1.2-1.7; p < 0.001). In case reports, hypocalcemia and hypokalemia commonly coexisted in patients with hypomagnesemia. Therefore, a χ2 test was conducted to investigate the association of hypomagnesemia with these electrolyte disorders. A strong association was found between hypomagnesemia and both hypocalcemia and hypokalemia (p < 0.001 for both). While the association was clear for both of these electrolytes, the correlation between hypomagnesemia and hypocalcemia was stronger (Φ = 0.68) compared to the correlation between hypomagnesemia and hypokalemia (Φ = 0.27). Four hundred twenty-nine of the 693 (61.9%) subjects with hypomagnesemia also had concurrent hypocalcemia, compared with only 134 of 65,409 (0.2%) nonhypomagnesemic subjects. Hypokalemia occurred in 173 of the 693 (25.0%) hypomagnesemic subjects and 406 of 65,409 (0.6%) nonhypomagnesemic subjects.
The combined online search strategies yielded 23 articles related to PPI-induced hypomagnesemia.11-33 Twenty of the articles found were case reports and small case series.11-14,16,17,19-29,31-33 Two of the articles were available only in abstract form,15,18 and a French article contained an abstract written in English.14 From the case reports and case series, 40 individual cases were identified. Table 4 summarizes the age and sex of the patients from these cases. PPI-associated hypomagnesemia was reported in 27 patients on omeprazole, 7 patients on esomeprazole, 3 patients on lansoprazole, 2 patients on pantoprazole, 1 patient on rabeprazole, and none with dexlansoprazole. Symptoms ranged from nausea, vomiting, and diarrhea to seizures, paraesthesia, confusion, and tetany. The severity of the symptoms did not consistently reflect the serum magnesium level. Nineteen of the 40 patients (48%) with PPI-associated hypomagnesemia also experienced hypokalemia and hypocalcemia. However, the potassium level was not reported in approximately half of these patients and calcium level was not reported in 5 of the patients. discussion
The prevalence of PPI-associated hypomagnesemia is unknown.8,39 Our research would suggest that hypomagnesemia represents a small proportion of all reported PPI-related adverse effects. Data from the FDA indicated only 1.0% of all cases reporting some adverse effects included hypomagnesemia. To date, case reports of only 40 patients with PPI-associated hypomagnesemia have been published. The larger number of reported adverse effects associated with esomeprazole and omeprazole might be the result of more patients receiving these 2 PPIs. Omeprazole was the first PPI available on the market (released in 1989), predating the second available PPI, lansoprazole, by approximately 6 years (released table 3. Factors Associated with Presence of Hypomagnesemiaa in 1995), thus providing a greater opportunity hypomagnesemia, for use.1 Esomeprazole’s reported superior effivariable n/n (%) odds ratio 95% ci p value cacy of gastric acid control compared to the Proton pump inhibitor other PPIs has seen its widespread use postesomeprazole 93/22,969 (0.4) 1 (reference) marketing, and this might explain the large lansoprazole 72/8,887 (0.8) 1.7 1.2-2.3 0.001 number of adverse reaction reports with this omeprazole 368/22,269 (1.7) 3.8 3.0-4.8
