Proton Pump Inhibitor and Tacrolimus Uses are Associated With Hypomagnesemia in Connective Tissue Disease: a Potential Link With Renal Dysfunction and Recurrent Infection

Background: Low levels of serum magnesium perturb renal tubular cell function and lymphocytes, resulting in renal deterioration and an imbalance in mononuclear cells. This study investigated the mechanism and influence of hypomagnesemia in patients with connective tissue disease. Methods: We retrospectively evaluated patients with connective tissue disease and available serum magnesium data who visited Keio University Hospital in 2019. Patients were divided into two groups: those with (serum magnesium < 1.8 mg/dl) and those without hypomagnesemia; their rates of hospitalization for severe infection and cumulative renal deterioration were compared. Patients’ fractions of lymphocytes and natural killer and dendritic cell subsets, as measured by fluorescence-activated cell sorting (FACS) analysis, were also compared. Results: Among 284 patients, hypomagnesemia was detected in 63 (22.2%). Multivariate analysis revealed that the use of proton pump inhibitors [odds ratio (OR), 1.48; p = 0.01] and tacrolimus (OR, 6.14; p < 0.01) was independently associated with hypomagnesemia. In addition, the renal deterioration rate was significantly higher in tacrolimus and/or proton pump inhibitor users with hypomagnesemia (p = 0.01). The hospitalization rate for severe infection was also higher in patients with hypomagnesemia (p = 0.04). FACS analysis showed lower CD8+ T cell, CD19+ B cell, natural killer cell, and dendritic cell counts in patients with hypomagnesemia (p = 0.03, p = 0.02, p = 0.02, and p = 0.03, respectively). Conclusion: The use of tacrolimus and proton pump inhibitors may be associated with hypomagnesemia and lead to poor renal outcomes and severe infection in patients with connective tissue disease.


INTRODUCTION
Magnesium (Mg) is an abundant intracellular cation that acts as a co-factor for more than 300 enzymes involved in a number of fundamental functions. Mg deficiency leads to many pathogenic conditions, including cardiovascular mortality, stroke, chronic kidney disease (CKD) progression, osteoporosis, and insulin resistance (Mountokalakis, 1990;Kao et al., 1999;He et al., 2006;Ohira et al., 2009;Castiglioni et al., 2013;Al Alawi et al., 2018). Hypomagnesemia is also associated with the development of recurrent infections due to the role of Mg as a second messenger in T cell activation and a contributor to the cytotoxicity of natural killer (NK) cells and CD8+ T cells (Mazur et al., 2007;Li et al., 2011;Chaigne-Delalande et al., 2013). Although the association of hypomagnesemia with wide spectrum disorders has been shown, their causal relationship remains to be proven in most cases.
The causes of hypomagnesemia are categorized into three groups: decreased dietary intake, impaired gastrointestinal absorption, and increased renal loss (Blaine et al., 2015). Several medications are known to influence serum Mg levels through these mechanisms. Although the association between proton pump inhibitors (PPIs) use and development of hypomagnesemia has not been confirmed, PPIs may potentially inhibit pH-dependent active Mg absorption (Perazella, 2013). Calcineurin inhibitors (CNIs) are also associated with low serum Mg concentrations (Barton et al., 1987;Navaneethan et al., 2006), Although both PPIs and CNIs, including tacrolimus (TAC), are frequently used in patients with connective tissue disease (CTD) (Takahashi et al., 2011;Takeuchi et al., 2014;Mok, 2016;Mok et al., 2016;Ueno et al., 2016;Hanaoka et al., 2019;Kaneko et al., 2020;Takada et al., 2020), little is known about their influence on serum Mg levels in the management of CTD.
The aim of this study was to investigate the prevalence of hypomagnesemia and its clinical impact on patients with CTD in relation to PPI and TAC use.

Patients and the Evaluation of Clinical Data
We reviewed the data of consecutive patients who visited Keio University Hospital from January 2019-December 2019, and were diagnosed with rheumatoid arthritis (RA), systemic lupus erythematosus (SLE), polymyositis/dermatomyositis (PM/DM), Sjögren syndrome (SS), systemic sclerosis (SSc), mixed connective tissue disease (MCTD), anti-neutrophil cytoplasmic antibody-related vasculitis, or other rheumatic diseases according to their respective classification criteria (Bohan and Peter, 1975; Subcommittee for scleroderma criteria of the American Rheumatism Association Diagnostic and Therapeutic Criteria Committee, 1980;Benhamou et al., 1988;Yamaguchi et al., 1992;Hochberg, 1997;Vitali et al., 2002;Taylor et al., 2006;Watts et al., 2007;Aletaha et al., 2010;Dasgupta et al., 2012;Tani et al., 2014;Bashardoust, 2015;Jinnin et al., 2018;Wallace et al., 2020;Heinle and Chang, 2014;Behcet's Disease Research Committee of Japan.1974). Patients with available serum Mg data were included in the study. Patients who had been treated with oxidized magnesium were excluded. We collected clinical characteristics, treatments administered for more than 3 years, laboratory data, estimated glomerular filtration rate (eGFR) (Levey A.S et al., 2009), and hospitalization due to infection from diagnosis to December 2019. For patients with SLE, we evaluated disease activity using systemic lupus erythematosus activity index (SLEDAI) (Bombardier et al., 1992).
This study was approved by the Ethics Committee of Keio University School of Medicine. Written informed consent was obtained from all subjects prior to blood sample collection as approved by the Institutional Review Board and in accordance with the tenets of the Declaration of Helsinki.

Definitions
Hypomagnesemia was defined as a serum Mg concentration <1.8 mg/dl (Elgend et al., 2012), and renal deterioration was defined as a >30% elevation in serum creatinine levels from baseline (Badve et al., 2016). Fractional excretion of Mg (FEMg, %) was calculated using the following formula: FE Mg (U Mg × P cr )/(0.7 × P Mg × U Cr ) × 100 (Agus, 1999), where U and P refer to the urine and plasma concentrations of Mg and creatinine (Cr), respectively. Serum Mg concentration was multiplied by 0.7 as only approximately 70% of circulating Mg is free and filters across the glomerulus. The normal limit of FEMg was defined as <2% (Agus, 1999). Severe infection is defined according to previous report (Díaz-Lagares et al., 2011) as those that required intravenous treatment or that led to hospitalization.

TAC Measurement
TAC concentration was measured using fresh whole blood samples collected 12 h after the last TAC administration by the TACR Flex Dimension immunoassay method using a Dimension EXL analyzer (Siemens Healthcare Diagnostics, Tokyo, Japan) (Takahashi et al., 2011).

Flow Cytometry
Cell bank blood samples collected at the time of serum Mg measurement from RA patients treated with methotrexate (MTX) monotherapy were analyzed by fluorescence-activated cell sorting (FACS) analysis. Samples were stained with antibodies (BD Biosciences and BioLegend; Supplementary Table S1) and fixed with Phosflow Lyse/Fix Buffer (BD Bioscience). Flow cytometric analysis was conducted on an LSRFortessa TM X-20 (Becton Dickinson) and analyzed using FlowJo ver. 10 (FlowJo, LLC). The phenotypes of immune cell subsets were defined based on the Human Immunology Project protocol (Supplementary Table S2) (Maecker et al., 2012). The mean numbers of immune cell phenotypes were compared.

Statistical Analyses
Continuous values are shown as the median and interquartile range (IQR). Comparisons between two groups were performed with the Mann-Whitney U-test for continuous variables and the chisquared test or Fisher's exact test for categorical variables. The four groups were compared by analysis of variance. Cumulative renal deterioration rates were analyzed using the Kaplan-Meier method with the log-rank test. Correlations between two continuous variables were analyzed using Spearman's rank correlation coefficient. To identify independent parameters, binary logistic regression analysis was used with variables having a p-value < 0.005 in a previous univariate analysis as covariates. A p-value < 0.05 was considered statistically significant.

Factors Associated With Hypomagnesemia
Patients were divided into two groups-the normal Mg group (n 221) and the hypomagnesemia group (n 63)-according to the presence of hypomagnesemia. The clinical characteristics of the two groups were compared (

Association of Drugs and Mg Levels
To further investigate the effects of TAC and PPIs on serum Mg levels, we divided all patients into four groups according to their use of TAC and PPIs ( Figure 1A) and compared their Mg levels. Median levels of serum Mg were 2.1 (2.0-2.2) mg/dl in patients that did not use TAC or PPIs, 2.1 (1.9-2.2) mg/dl in those that only used PPIs, 1.9 (1.8-1.9) mg/dl in those that only used TAC, and 1.8 (1.8-2.9) mg/dl in those that used both TAC and PPIs (p < 0.0001). When we compared Mg levels in SLE patients, the difference remained (Supplementary Figure S2).
FEMg was then calculated in patients with hypomagnesemia (n 57) ( Figure 1B). It was 1.7% (1.5-2.7%) in those that did not use TAC or PPIs, 2.2% (1.6-3.2%) in those that only used PPIs, 3.9% (2.6-4.9%) in those that only used TAC, and 2.7% (1.3-4.7%) in those that used both TAC and PPIs (p 0.04). These findings reflect the different causal mechanisms of hypomagnesemia, namely, that TAC inhibits reabsorption of Mg in the kidneys with a consequent increase in Mg excretion in the urine, while PPIs cause Mg to be wasted in the intestine.

Relationship Between Hypomagnesemia and Renal Deterioration
As the use of TAC and PPIs were the major causes of hypomagnesemia in this study, we investigated the sequential  Figure 4). Furthermore, we selected SLE patients and additionally compared renal deterioration-free rates between those with hypomagnesemia and those without (Supplementary Figure S3). The renal deterioration-free rate in SLE patients with normal Mg levels was higher than in those with normal hypomagnesemia (p 0.019). Of note, renal deterioration was not related to TAC use because renal deterioration-free rates at last observation was 84.5% in the TAC users and 90.1% in the non-TAC users (p 0.34).

Effect of Hypomagnesemia on Immune Cells
Among the 283 patients enrolled in this study, 17 patients were concurrently registered in another cohort study at our university. These patients had peripheral blood mononuclear cells available for FACS analysis when their Mg levels were measured. All 17 of these patients had RA and were only treated with MTX. Six of the patients had hypomagnesemia and 11 had normal Mg levels. These patients did not differ with regard to sex, disease duration,   Table S4). The number of CD8+ T cells, CD19+ B cells, NK cells, and dendritic cells (DCs) were significantly lower in patients with hypomagnesemia than in patients with normal Mg levels (p 0.03, p 0.02, p 0.02, and p 0.03, respectively, Figure 5). Hospitalization due to infection was observed in one patient with hypomagnesemia (16.6%) and 1 with normal Mg levels (9.1%) (p 0.64).

DISCUSSION
In this study, hypomagnesemia was observed in approximately 20% of the patients with CTD and was associated with renal deterioration and hospitalization due to severe infection. The development of hypomagnesemia might have been caused by the use of TAC and PPIs.
In addition, a high renal deterioration rate was seen in patients with hypomagnesemia. This finding is consistent with previous studies that have reported the association of hypomagnesemia with incident CKD (Tin et al., 2015), a decline in eGFR in CKD patients (Van Laecke et al., 2013), and the progression to endstage renal disease in diabetic nephropathy (Sakaguchi et al., 2012). Laecke et al. investigated 1,650 patients with CKD with a median follow-up of 5.1 years and reported that a 1 mg/dl decrease in baseline serum Mg was associated with a 5.1% annual decrease in eGFR. In a study of Japanese patients with diabetic nephropathy (n 144), patients with hypomagnesemia were twice as likely to progress to end-stage renal disease as were those within the normal range. Although the pathogenic mechanism of hypomagnesemia in relation to renal deterioration is not fully understood, hypomagnesemia is considered to damage renal tubules. In one study, incubation of tubular epithelial cells in low-Mg medium increased the rate of apoptosis, whereas this effect was significantly suppressed when Mg concentrations were increased (Sakaguchi et al., 2015).
In the current study, 22.2% of patients with CTDs showed serum Mg levels <1.8 mg/dl (hypomagnesemic). Furthermore, low Mg levels were associated with a high hospitalization rate due to severe infection. The association between hypomagnesemia and recurrent infection was reported in a previous study in which patients with X-linked XMEN, a hereditary immune deficiency syndrome in which dysfunction of the Mg channel, MAGT1, in T lymphocytes leads to a low intra-lymphocytic free Mg concentration, suffered from recurrent infection (Ravell et al., 2014). In another study, hypomagnesemic rats were shown to die earlier than control rats when injected with intravenous Escherichia coli endotoxin; however, Mg supplementation improved survival (Salem et al., 1995). In a clinical report on kidney transplantation, low serum Mg was associated with an increased risk of infection, and every 0.1 mg/dl reduction in serum Mg below 2.0 mg/dl increased the hazard ratio by 15% FIGURE 3 | Serial change in magnesium level after the discontinuation of proton pump inhibitors. Mg levels significantly increased after discontinuing PPI use in patients not using TAC (p 0.04) (A); however, no significant difference was seen in FEMg (B). In patients using TAC, no change in Mg level or FEMg was observed after PPI discontinuation (C,D). Mg, magnesium; PPI, proton pump inhibitor; TAC, tacrolimus; FEMg, fractional excretion of magnesium.
FIGURE 4 | Cumulative renal deterioration-free rate. A significantly lower renal deterioration-free rate was observed in patients with hypomagnesemia than in patients with normal Mg levels (p 0.007) Mg, magnesium.
Frontiers in Pharmacology | www.frontiersin.org May 2021 | Volume 12 | Article 616719 (Van Laecke et al., 2016). Hypomagnesemia decreases T cell count and causes activation and cytotoxicity in CD8+ T cells and NK cells (Gröber, 2019). Although this study did not examine lymphocyte function, we did identify a decrease in the number of CD8+ T cells, CD19+ B cells, NK cells, and DCs in RA patients with hypomagnesemia. Taken together, the decreased function and number of mononuclear cells caused by hypomagnesemia may be associated with impaired immune function in hypomagnesemic patients. In this study, hypomagnesemia in patients with CTD was significantly associated with the use of TAC and PPIs. These findings are consistent with the action of these drugs, specifically, that TAC interferes with Mg-reabsorption from urine, and PPIs interfere with Mg absorption from the intestines (Barton et al., 1987;Navaneethan et al., 2006;Perazella, 2013). In addition, the interference of TAC with systemic Mg transportation was much greater than that of PPI, as the kidneys can withstand a reabsorption of 20 times more dietary Mg than the intestines (Al Alawi et al., 2018). In fact, our study showed that patients with TAC had lower Mg concentrations than those with PPIs, and the combination of PPIs and TAC did not show additional lowering effect on Mg concentrations than TAC alone. Furthermore, we found that the discontinuation of PPIs increased Mg levels in patients who were not using TAC, suggesting that hypomagnesemia caused by PPIs is reversible. Therefore, we recommend monitoring the serum Mg levels of patients treated with PPIs and considering discontinuation in cases of hypomagnesemia. Mg levels did not change when PPI use was discontinued in patients using both TAC and PPIs. We speculate that this was due to the stronger effect of TAC on lowering serum Mg levels. While we cannot conclude that TAC discontinuation would cause an increase in serum Mg concentrations because no patient discontinued TAC use in our study, based on the findings that TAC concentrations were negatively correlated with serum Mg concentrations, we recommend that TAC dosage be monitored and reduced as much as possible to prevent hypomagnesemia.
Our study has several limitations. First, it was a retrospective, single-center cohort study with a small sample size. This could have caused a degree of selection bias. Second, serum Mg levels were measured cross-sectionally. Therefore, changes in Mg levels during the renal function observation period were unclear. This weakened the discussion regarding the relationship between hypomagnesemia and renal deterioration. Third, PPI use was discontinued at the discretion of attending physicians. This may FIGURE 5 | Flow cytometric analysis in patients with and without hypomagnesemia. Lower cells counts were observed for CD8 + T cells, CD19 + B cells, NK cells, and dendritic cells in patients with hypomagnesemia than in patients with normal Mg levels (p 0.03, p 0.02, p 0.02, and p 0.03, respectively). NK, natural killer; Mg, magnesium.
Frontiers in Pharmacology | www.frontiersin.org May 2021 | Volume 12 | Article 616719 have also resulted in a degree of selection bias. Fourth, although cyclosporin, one of the calcineurin inhibitors, was used widely in the world, we focused on TAC in this study. The influence of TAC on magnesium level was reported to be greater than that of cyclosporin (Racca et al., 2020), and further investigations may be needed to clarify the difference between the two drugs. Confirmation of our findings will require a multi-center prospective study.

CONCLUSION
The use of TAC and PPIs was associated with hypomagnesemia and led to poor renal outcomes and severe infections in patients with CTD. The lowest possible dose of TAC should be prescribed in the management of CTD, and the need for PPIs should be periodically reassessed.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Ethics Committee of Keio University School of Medicine. The patients/participants provided their written informed consent to participate in this study.

AUTHOR CONTRIBUTIONS
HH performed the majority of research, and analyzed and interpreted the data; NS, HT, and KC performed the FACS analysis; JK and YK collected the patients' data and samples; TT designed the research, interpreted the data, and supervised and organized the study; all authors wrote the manuscript. All authors have read and approved the final manuscript.