Clinical characteristics
A total of 284 patients with CTD were included in the study. Median (IQR) age was 64.0 (48.0–73.0) years and 83.8% were female (Table 1). Underlying CTDs were RA in 108 patients (38.0%), SLE in 59 (20.8%), PM/DM in 20 (7.0%), SSc in 24 (8.5%) MCTD in 10 (3.5%), polymyalgia rheumatica in 10 (3.5%), microscopic polyangiitis in 8 (2.8%), IgG4-related disease in 9 (3.2%), SS in 6 (2.1%), adult Still’s disease in 6 (2.1%), arthritis with palmoplantar pustulosis in 6 (2.1%), eosinophilic granulomatous polyangiitis in 4 (1.4%), psoriatic arthritis in 4 (1.4%), sarcoidosis in 2 (0.7%), Takayasu’s arteritis in 2 (0.7%), granulomatous polyangiitis in 2 (0.8%), Behçet’s disease in 2 (0.8%), diffuse fasciitis in 2 (0.8%), and familial Mediterranean fever in 1 (0.4%). Glucocorticoid was used by 41.5% of total patients and median (IQR) dose was 0.0 (0.0–4.0) mg/day. Among all patients, 141 (49.6%) used PPIs and 68 (23.9%) used TAC. Hospitalization for severe infection was seen in 25 (8.8%) patients. Hypomagnesemia was observed in 63 (22.2%) patients.
Table 1
| All (n = 284) | Normal Mg (n = 221) | Hypomagnesemia (n = 63) | p |
Age, years | 64.0 (48.0–73.0) | 71.0 (51.0–74.0) | 55.0 (38.0–69.0) | 0.029 |
Male:Female | 46:238 | 35:186 | 11:52 | 0.752 |
Underlying disease | | | | |
RA (%) | 108 (38.0) | 94 (42.5) | 14 (22.2) | |
SLE (%) | 59 (20.8) | 36 (16.3) | 23 (36.5) | |
PM/DM (%) | 20 (7.0) | 11 (4.9) | 9 (14.2) | |
SSc (%) | 24 (8.5) | 23 (10.4) | 1 (1.6) | |
MCTD (%) | 10 (3.5) | 9 (4.1) | 1 (1.6) | |
PMR (%) | 10 (3.5) | 7 (3.2) | 3 (4.8) | |
Others* (%) | 53 (18.7) | 41 (18.6) | 12 (19.0) | |
Serum electrolytes | | | | |
Mg, mg/dL | 2.1 (1.9–2.2) | 2.1 (2.0-2.2) | 1.7 (1.7–1.8) | < 0.001 |
Na, mEq/L | 140.9 (139.7-142.3) | 141.2 (139.8-142.4) | 140.5 (139.3-141.9) | 0.163 |
K, mEq/L | 4.2 (3.9–4.4) | 4.2 (4.0-4.4) | 4.1 (3.9–4.3) | 0.087 |
CL, mEq/L | 105.0 (104.0-107.0) | 106.0 (104.0-107.0) | 105.0 (103.0-106.0) | 0.241 |
Ca, mEq/L | 9.2 (8.9–9.4) | 9.1 (8.9–9.4) | 9.2 (8.9–9.4) | 0.608 |
P, mEq/L | 3.5 (3.2–3.9) | 3.5 (3.2–3.9) | 3.5 (3.1–3.9) | 0.489 |
eGFR, ml/min/1.73 m2 | 68.0 (56.0–80.0) | 63.0 (56.0–80.0) | 70.0 (56.0–81.0) | 0.462 |
Urine markers | | | | |
β2-microglobulin, ×102µg/L | 1.5 (0.9–2.9) | 2.1 (0.9–3.2) | 1.2 (0.7–1.9) | 0.386 |
α1-microglobulin, mg/L | 3.3 (1.6–6.5) | 3.8 (1.6–6.9) | 3.3 (1.6–5.8) | 0.636 |
L-FABP, µg/gཥCre | 2.4 (1.6–4.7) | 3.1 (1.7–4.8) | 2.2 (1.4–4.4) | 0.164 |
NAG, IU/L | 5.0 (2.5–8.2) | 5.0 (2.4–8.1) | 5.3 (3.4–9.6) | 0.120 |
NGAL, µg/gཥCre | 21.7 (14.0-38.5) | 21.7 (15.1–37.1) | 21.7 (12.9–46.0) | 0.770 |
Medication | | | | |
GC, (%) | 118 (41.5) | 78 (35.3) | 40 (63.4) | 0.001 |
GC dose, median (IQR) mg/day | 0 (0–4) | 0 (0–3) | 3 (0–5) | 0.001 |
TAC (%) | 68 (23.9) | 34 (15.3) | 34 (53.9) | 0.001 |
MMF (%) | 13 (4.6) | 6 (2.7) | 7 (11.1) | 0.006 |
MTX (%) | 81 (28.5) | 71 (32.1) | 10 (15.9) | 0.001 |
AZA (%) | 19 (6.7) | 17 (7.6) | 2 (3.2) | 0.213 |
HCQ (%) | 27 (9.5) | 13 (5.9) | 14 (22.2) | 0.001 |
PPI (%) | 141 (49.6) | 93 (42.1) | 48 (76.1) | 0.001 |
Hospitalization due to infection | 25 (8.8) | 15 (6.7) | 10 (15.8) | 0.042 |
Results show median (interquartile range) unless otherwise indicated. |
*Others include microscopic polyangiitis, IgG4-related disease, Sjogren syndrome, adult Still’s disease, arthritis with palmoplantar pustulosis, eosinophilic granulomatous polyangiitis, psoriatic arthritis, sarcoidosis, Takayasu’s arteritis, granulomatous polyangiitis, Behçet’s disease, diffuse fasciitis, and familial Mediterranean fever. |
Mg, magnesium; RA, rheumatoid arthritis; SLE, systemic lupus erythematosus; SSc, systemic sclerosis; MCTD, mixed connective tissue disease; PMR, polymyalgia rheumatica; eGFR, estimated glomerular filtration rate; L-FABP, liver-type fatty acid binding protein; NAG, N-acetyl-β-D-glucosaminidase; NGAL, neutrophil gelatinase-associated lipocalin; GC, glucocorticoid; PPIs, proton-pump inhibitors; TAC, tacrolimus; MMF, mycophenolate mofetil; MTX, methotrexate; AZA, azathioprine; HCQ, hydroxychloroquine. |
Factors associated with hypomagnesemia
The patients were divided into two groups according to the presence of hypomagnesemia. Clinical characteristics were compared between the normal Mg (n = 221) and hypomagnesemia groups (n = 63) (Table 1). Median age was significantly higher in the normal Mg group than in the hypomagnesemia group (71.0 years vs 55.0 years, p = 0.029, respectively). Renal function, electrolyte concentrations except for Mg, and urine markers were not different between the two groups. Rates of use of glucocorticoid, PPI, TAC, mycophenolate mofetil (MMF), and hydroxychloroquine (HCQ) were significantly lower in the normal Mg group than in the hypomagnesemia group (glucocorticoid, 35.3% vs 63.4%, p = 0.001; PPI, 42.1% vs 76.1%, p = 0.001; TAC, 15.3% vs 53.9%, p = 0.001; MMF, 2.7% vs 11.1%, p = 0.006; HCQ, 5.9% vs 22.2%, p = 0.001, respectively). Use of MTX was significantly higher in the normal Mg group (32.1% vs 15.9%, p = 0.001).
Hospitalization due to severe infection from the diagnosis of CTDs to December 2019 occurred significantly less frequently in the normal Mg group than in the hypomagnesemia group (6.7% vs 15.8%, p = 0.042).
Multiple logistic regression analysis identified the use of PPI (odds ratio 1.45, confidence interval 1.0-3.29, p = 0.009) and TAC (odds ratio 5.99, confidence interval 2.93–12.24, p < 0.001) as independent factors associated with hypomagnesemia (Table 2).
Table 2
Multivariate analysis for factors associated with hypomagnesemia
Factor | Odds ratio (95%CI) | p |
Age | 0.96 (0.95–1.05) | 0.152 |
SLE | 1.47 (0.54–3.97) | 0.445 |
RA | 0.84 (0.32–2.16) | 0.727 |
GC use | 1.14 (0.49–2.71) | 0.753 |
PPI use | 1.45 (1.01–3.29) | 0.009 |
TAC use | 5.99 (2.93–12.24) | < 0.001 |
MTX use | 0.72 (0.27–1.95) | 0.523 |
HCQ use | 1.71 (0.53–5.52) | 0.371 |
SLE, systemic lupus erythematosus; RA, rheumatoid arthritis; GC, glucocorticoid; PPIs, proton-pump inhibitors; TAC, tacrolimus; MTX, methotrexate; HCQ, hydroxychloroquine. |
Association of drugs and Mg levels
To investigate the effect of TAC and PPI on serum Mg levels further, we divided all patients into 4 groups according to the use of TAC and PPI (Fig. 1A) and compared magnesium levels. Median (IQR) levels of serum Mg were 2.1 (2.0-2.2) mg/dL in patients without TAC and PPI, 2.1 (1.9–2.2) mg/dL in those with only PPI, 1.9 (1.8–1.9) mg/dL in those with only TAC, and 1.8 (1.8–2.9) mg/dL in those with both TAC and PPI (p < 0.0001).
We then calculated FEMg in patients with hypomagnesemia (n = 57) (Fig. 1B). The FEMg was 1.7 (1.5–2.7)% in those without TAC and PPI, 2.2 (1.6–3.2)% in those with only PPI, 3.9 (2.6–4.9)% in those with only TAC, and 2.7 (1.3–4.7)% in those with both TAC and PPI (p = 0.04), reflecting the different mechanisms of causation of hypomagnesemia, namely that TAC inhibits reabsorption of Mg in the kidneys with a consequent increase in excretion of Mg in urine, and that PPI causes wasting of Mg in the intestine.
The relationship of serum TAC concentrations with serum Mg levels and FEMg corroborated the effect of the drugs. In patients who did not use PPI, TAC concentrations were negatively correlated with Mg levels (r = -0.61, p < 0.01, Fig. 2A) and positively with FEMg (r = 0.38, p = 0.05, Fig. 2B). In patients who used PPI, these correlations disappeared (Mg levels, r = -0.25, p = 0.19, Fig. 2C; FEMg, r = -0.07, p = 0.73, Fig. 2D). We further investigated 28 patients who had discontinued PPI at the attending physician’s discretion (without tacrolimus, n = 22; with tacrolimus, n = 6): in patients who did not use TAC, serum Mg levels were significantly increased after PPI discontinuation from 2.0 (2.0-2.2) mg/dL to 2.2 (2.0-2.4) mg/dL, p = 0.04, Fig. 3A), while FEMg did not change (from 1.9 [1.4–2.7] % to 2.2 [1.9–2.7]%, p = 0.16, Fig. 3B); in patients treated with TAC, in contrast, serum Mg concentrations and FEMg were not changed (Mg, 1.8 [1.6-3.0] mg/dL to 2.0 [1.6–4.6] mg/dL, p = 0.82, Fig. 3C; FEMg, 2.3 [1.3-3.0]% to 2.1 [1.6-.8]%, p = 0.56, Fig. 3D).
Relationship between hypomagnesemia and renal deterioration
As the use of TAC and PPI was the major cause of hypomagnesemia, we investigated the sequential renal function of patients treated with TAC and/or PPI (n = 173) from the initiation of these drugs until the last observation. When we divided the patients according to the presence of hypomagnesemia, the cumulative renal deterioration-free rates were significantly higher in patients with normal Mg (n = 124; 80.7%; observation period, 5.0 ± 2.9 years) than in those with hypomagnesemia (n = 49, 65.7%, observation period, 5.3 ± 3.4 years) (p = 0.007, Fig. 4). Of note, renal deterioration was not related with TAC use (renal deterioration-free at last observation, TAC users 84.5% and non-TAC users 90.1%, p = 0.34).
Effect of hypomagnesemia on immune cells
Among the 283 patients enrolled in this study, 17 patients who were also registered in another cohort of our university had information on peripheral blood mononuclear cells analyzed with FACS at the time Mg concentrations were measured. All 17 of these patients had RA and were treated with methotrexate alone. Six of these 17 patients had hypomagnesemia while the other 11 had normal Mg levels. These patients did not differ with regard to sex, disease duration, disease activity, or MTX dose (Table S3). In contrast, numbers of CD8 + T cells, CD19 + B cells, NK cells, and DC were significantly lower in the patients with hypomagnesemia than in the patients with normal Mg (p = 0.03, p = 0.02, p = 0.02, and p = 0.03, respectively, Fig. 5). Hospitalization due to infection was observed in 1 patient with hypomagnesemia (16.6%) and 1 with normal Mg (9.1%) (p = 0.64).