Hypomagnesemia is a significant predictor of cardiovascular and non-cardiovascular mortality in patients undergoing hemodialysis

Transcription

1 clinical investigation & 2013 International Society of Nephrology see commentary on page 17 Hypomagnesemia is a significant predictor of cardiovascular and non-cardiovascular mortality in patients undergoing hemodialysis Yusuke Sakaguchi 1, Naohiko Fujii 2, Tatsuya Shoji 3, Terumasa Hayashi 3, Hiromi Rakugi 1 and Yoshitaka Isaka 1 1 Department of Geriatric Medicine and Nephrology, Osaka University Graduate School of Medicine, Osaka, Japan; 2 Department of Internal Medicine, Hyogo Prefectural Nishinomiya Hospital, Hyogo, Japan and 3 Department of Kidney Disease and Hypertension, Osaka General Medical Center, Osaka, Japan Although previous studies in the general population showed that hypomagnesemia is a risk for cardiovascular diseases (CVD), the impact of magnesium on the prognosis of patients on hemodialysis has been poorly investigated. To gain information on this we conducted a nationwide registrybased cohort study of 142,555 hemodialysis patients to determine whether hypomagnesemia is an independent risk for increased mortality in this population. Study outcomes were 1-year all-cause and cause-specific mortality with baseline serum magnesium levels categorized into sextiles. During follow-up, a total of 11,454 deaths occurred, of which 4774 had a CVD cause. In a fully adjusted model, there was a J-shaped association between serum magnesium and the odds ratio of all-cause mortality from the lowest to highest sextile, with significantly higher mortality in sextiles 1 3 and 6. Similar associations were found between magnesium and both CVD and non-cvd mortality. The proportion of patients with a baseline intact parathyroid hormone level under 50 pg/ml was significantly higher in the highest sextile; however, after excluding these patients, the CVD mortality risk in the highest sextile was attenuated. Thus, hypomagnesemia was significantly associated with an increased risk of mortality in hemodialysis patients. Interventional studies are needed to clarify whether magnesium supplementation is beneficial for improving patient prognosis. Kidney International (2013) 85, ; doi: /ki ; published online 28 August 2013 KEYWORDS: cardiovascular events; hemodialysis; mineral metabolism Correspondence: Yoshitaka Isaka, Department of Geriatric Medicine and Nephrology, Osaka University Graduate School of Medicine, 2-2, Yamadaoka, Suita, Osaka , Japan. isaka@kid.med.osaka-u.ac.jp Received 24 March 2013; revised 12 May 2013; accepted 29 May 2013; published online 28 August 2013 Despite increasing knowledge and technical advances in the field of dialysis therapy, the risk of death is still unacceptably high among patients undergoing hemodialysis. The 5-year survival rate is 35% in the United States and 60% in Japan, far below that in the general population. 1,2 Cardiovascular diseases (CVD) account for nearly 40% of all deaths in patients with end-stage renal disease (ESRD). 1 4 Traditional CVD risk factors such as hypertension, dyslipidemia, insulin resistance, and obesity do not fully explain the increased risk observed in hemodialysis patients, while the elevation of several nontraditional risk factors such as inflammation, oxidative stress, malnutrition, anemia, and uremia has been considered to play a more important role. 5 In particular, much attention has been focused on mineral and bone disorder (MBD) in ESRD as a prominent contributor to the development of atherosclerosis and vascular calcification, in which phosphate retention is considered a key component. 6,7 Magnesium (Mg), the fourth most abundant cation in the human body, plays an essential role in numerous biological processes. The importance of this mineral has been particularly recognized due to its antiatherosclerotic effect. 8 In the general population, a lower serum Mg level and/or lower dietary Mg intake is associated with an increased incidence of hypertension, 9 type 2 diabetes mellitus (DM), 10 metabolic syndrome, 11 and CVD, including myocardial infarction, stroke, atrial fibrillation, and sudden cardiac death A significant association between hypomagnesemia and increased risk for fatal and non-fatal CVD events was also reported in patients with pre-dialysis chronic kidney disease. 18 An interventional study in patients with coronary artery disease demonstrated that oral Mg supplementation could improve endothelial cell dysfunction. 19 Several in vitro studies have shown that Mg deficiency causes vascular constriction, platelet aggregation, inflammation, and oxidative stress, resulting in endothelial cell dysfunction and vascular calcification. 20 The direct protective effects of Mg on vascular calcification have also been demonstrated via multiple pathways, including inhibition of hydroxyapatite 174 Kidney International (2014) 85,

2 Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients clinical investigation formation 21 and suppression of transdifferentiation of vascular smooth muscle cells into osteoblast-like cells. 22,23 Despite the well-accepted associations of CVD and mineral disturbances with hemodialysis, the role of Mg in this population is nearly unexplored. Previous cross-sectional studies in ESRD have reported that hypomagnesemia was significantly associated with an increased prevalence of mitral annular calcification, 24 peripheral arterial calcification, 25 and an increased carotid intima-media thickness. 26 Moreover, 2 months of oral Mg supplementation significantly decreased intima-media thickness. 27 However, very few studies have examined whether the serum Mg level is independently associated with mortality in this population. One previous small retrospective cohort study 28 could not demonstrate a significant association between the serum Mg level and CVD mortality, likely due to insufficient statistical power. Therefore, we conducted this cohort study using a nationwide registry of patients with ESRD in Japan to clarify the relationship between serum Mg levels and mortality. RESULTS The study enrollment process is summarized in Figure 1. Of the total number of 263,849 dialysis patients, those on renal replacement therapy other than facility hemodialysis (n ¼ 26,752) were excluded. Of 237,097 hemodialysis patients, baseline serum Mg level data were available for 142,555 subjects. We compared all baseline characteristics between the subjects with and without serum Mg level data and found no meaningful difference between groups (Supplementary Table 1 online). The mean (standard deviation (s.d.)) serum Mg level of the subjects was 2.61 (0.52) mg/dl. The baseline characteristics according to serum Mg sextiles are summarized in Table 1. A lower serum Mg level was significantly associated with older age, lower albumin, calcium, phosphate, and hemoglobin level, higher C-reactive protein (CRP) and alkaline phosphatase (ALP) level, and increased prevalence of DM, prior history of CVD, and hip fracture. To better understand the relationship between serum Mg and intact parathyroid hormone (ipth) levels, which is strongly influenced by other MBD-related factors, univariate and multivariate restricted cubic spline functions were fitted with four knots (Figure 2). As shown, the ipth level was almost at a plateau in the unadjusted model; however, after adjustment for age, sex, serum calcium, and phosphate level, and prescription of active vitamin D analogues and cinacalcet, an overall negative association was observed. During follow-up, a total of 11,454 deaths occurred, in which 4774 (41.7%) were attributed to CVD and 6680 (58.3%) to non-cvd. Only a small proportion of the patients (n ¼ 367 (0.3%)) received renal transplantation during follow-up. In a crude analysis, there was a significant J-shaped relationship between the Mg sextiles and outcomes (Table 2). Their associations were somewhat attenuated after additional adjustment for the relevant clinical factors of demographics, MBD and malnutrition-inflammation atherosclerosis (MIA) complex related factors, yet they remained 263,849 dialysis patients aged 18 years and older, undergoing any type of renal replacement therapy (nearly all dialysis patients in Japan) Subjects on peritoneal dialysis, hemofiltration, hemodiafiltration, and short daily or home hemodialysis (n = 26,752) 237,097 conventional in-center hemodialysis patients Patients without baseline serum magnesium-level data (n = 94,542) 142,555 patients with baseline serum magnesium-level data Figure 1 Flowchart of the study enrollment process. statistically significant (fully adjusted odds ratio (OR) [95% confidence interval (CI)] of the lowest sextile was 1.28 [1.17, 1.41] (Po0.001) for all-cause mortality, 1.24 [1.08, 1.42] (P ¼ 0.002) for CVD mortality, and 1.32 [1.17, 1.49] (Po0.001) for non-cvd mortality; Table 2). The continuous, fully adjusted association between the serum Mg level and all-cause mortality showed a similar relationship (Figure 3). The results of the subgroup analysis are shown in Table 3; the lowest sextile was significantly associated with a higher risk of all-cause mortality in all prespecified subgroups. As the proportion of subjects with a baseline ipth level o50 pg/ml was significantly higher in the highest sextile (Table 1), we assumed that this low ipth level increased the mortality risk of this group. Therefore, we performed the same multivariate analysis after excluding subjects with an ipth level o50 pg/ml. In this subgroup, CVD mortality risk in the highest sextile was attenuated and lost statistical significance (Table 2). We further explored the associations between Mg levels and specific causes of non-cvd deaths. Of all the non-cvd deaths, infection (n ¼ 2156; 32.3%) and cancer (n ¼ 1084; 16.2%) were the two major causes. In the fully adjusted multivariate model, the ORs [95% CI] for infection-related deaths were 1.49 [1.20, 1.85] (Po0.001), 1.33 [1.06, 1.67] (P ¼ 0.01), 1.36 [1.08, 1.71] (P ¼ 0.008), 1.33 [1.00, 1.76] (P ¼ 0.05), 1.00 (Ref.), and 1.21 [0.93, 1.58] (P ¼ 0.15) from the lowest to highest sextile, respectively. Because the inclusion of patients suffering from severe infections (and, consequently, of hypomagnesemia due to cachexia) at baseline would allow the possibility of reverse causality between hypomagnesemia and death from infection, an additional analysis excluding patients with high baseline CRP levels (X3.0 mg/dl) was conducted. This analysis slightly attenuated the risk; the ORs [95% CI] were 1.44 [1.14, 1.81] Kidney International (2014) 85,

3 clinical investigation Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients Table 1 Baseline characteristics of 142,555 hemodialysis patients according to serum magnesium level sextiles Characteristics No. of missing data (%) Total n ¼ 142,555 Sextiles of serum magnesium level (serum magnesium level range (mg/dl)) 1(o2.3) n ¼ 28,764 2(X2.3, o2.5) n ¼ 25,798 3(X2.5,o2.7) n ¼ 28,731 4(X2.7, o2.8) 5(X2.8, o3.1) n ¼ 13,013 n ¼ 27,178 6(X3.1) n ¼ 19,071 P for trend Age, years 0 (0) 66.0± ± ± ± ± ± ±12.8 o0.001 Gender, male 0 (0) 88,290 (61.9) 18,034 (62.7) 16,264 (63.0) 18,158 (63.2) 7973 (61.3) 16,596 (61.1) 11,265 (59.1) o0.001 HD vintage, years 0 (0) 7 [4, 12] 6 [3, 10] 6 [4, 11] 7 [4, 12] 7 [4, 12] 8 [5, 13] 8 [5, 13] o0.001 HD, hours/week 1294 (0.9) 11.6± ± ± ± ± ± ±1.6 o0.001 BMI, kg/m 2 20,113 (14.1) 22.2± ± ± ± ± ± ±4.5 o0.001 DM, % 0 (0) 51,184 (35.9) 11,041 (38.4) 9412 (36.5) 10,147 (35.3) 4445 (34.2) 9176 (33.8) 6963 (36.5) o0.001 BUN, mg/dl 214 (0.2) 64.5± ± ± ± ± ± ±16.3 o0.001 Adj. Ca, mg/dl 2337 (1.6) 9.3± ± ± ± ± ± ±0.9 o0.001 P, mg/dl 486 (0.3) 5.1± ± ± ± ± ± ±1.6 o0.001 Alb, g/dl 2244 (1.6) 3.7± ± ± ± ± ± ±0.4 o0.001 CRP, mg/dl 22,626 (15.9) 0.53± ± ± ± ± ± ±1.49 o0.001 Hb, g/dl 699 (0.5) 10.6± ± ± ± ± ± ±1.3 o0.001 Hbo10 g/dl, % 38,652 (27.1) 10,089 (35.1) 7662 (29.7) 7560 (26.3) 3221 (24.8) 6005 (22.1) 4115 (21.6) o0.001 ALP, IU/l 3351 (2.4) 263± ± ± ± ± ± ±127 o0.001 ipth, pg/ml 7590 (5.3) 118 [59, 201] 120 [61, 202] 119 [61, 201] 120 [61, 202] 122 [62, 205] 120 [60, 202] 106 [50, 190] o0.001 iptho50 pg/ml, % 27,803 (19.5) 5350 (18.6) 4854 (18.8) 5409 (18.8) 2402 (18.5) 5312 (19.5) 4476 (23.5) o0.001 Prescription CaCO 3, % 7416 (5.2) 82,279 (60.9) 13,501 (49.6) 14,488 (59.4) 17,389 (63.6) 8144 (66.0) 17,381 (67.4) 11,376 (62.9) o0.001 Sevelamer, % 8080 (5.7) 38,105 (28.3) 4209 (15.5) 5379 (22.2) 7512 (27.7) 3965 (32.2) 9609 (37.4) 7431 (41.2) o0.001 Lanthanum, % 8326 (5.8) 17,334 (12.9) 2307 (8.5) 2699 (11.2) 3562 (13.1) 1823 (14.9) 4051 (15.8) 2892 (16.1) o0.001 Cinacalcet, % 8744 (6.1) 15,179 (11.3) 2116 (7.8) 2389 (9.9) 3059 (11.3) 1624 (13.3) 3562 (14.0) 2429 (13.6) o0.001 Vit. D (p.o.), % 7936 (5.6) 51,964 (38.6) 11,051 (40.7) 9950 (40.9) 10,841 (39.9) 4642 (37.7) 9523 (37.1) 5957 (33.1) o0.001 Vit. D (i.v.), % 8887 (6.2) 35,547 (26.6) 6219 (23.1) 6255 (25.9) 7360 (27.3) 3557 (29.1) 7361 (28.9) 4795 (26.8) o0.001 Past history PTx, % 13,525 (9.5) 6886 (5.3) 1204 (4.6) 1190 (5.1) 1427 (5.5) 732 (6.2) 1500 (6.1) 833 (4.8) o0.001 MI, % 15,419 (10.8) 9270 (7.3) 2431 (9.4) 1878 (8.2) 1835 (7.2) 728 (6.3) 1460 (6.0) 938 (5.5) o0.001 CI, % 15,215 (10.7) 19,107 (15.0) 4797 (18.6) 3786 (16.4) 3788 (14.7) 1531 (13.2) 3028 (12.5) 2177 (12.8) o0.001 CH, % 15,362 (10.8) 6162 (4.8) 1338 (5.2) 1152 (5.0) 1157 (4.5) 519 (4.5) 1132 (4.7) 864 (5.1) 0.12 Amputation, % 14,736 (10.3) 3715 (2.9) 988 (3.8) 659 (2.9) 693 (2.7) 287 (2.5) 596 (2.5) 492 (2.9) o0.001 Hip fracture, % 15,586 (10.9) 3660 (2.9) 1114 (4.3) 743 (3.2) 654 (2.6) 286 (2.5) 505 (2.1) 358 (2.1) o0.001 Abbreviations: Adj. Ca, albumin-adjusted calcium; Alb, albumin; ALP, alkaline phosphatase; BMI, body mass index; BUN, blood urea nitrogen; CH, cerebral hemorrhage; CI, cerebral infarction; CRP, C-reactive protein; DM, diabetes mellitus; Hb, hemoglobin; HD, hemodialysis; ipth, intact parathyroid hormone; MI, myocardial infarction; P, phosphate; PTx, parathyroidectomy; Vit. D, active vitamin D analog. Data are presented as mean±standard deviation, median [interquartile range], or number (percent). ipth (pg/ml) (P ¼ 0.002), 1.24 [0.98, 1.58] (P ¼ 0.08), 1.26 [0.99, 1.60] (P ¼ 0.07), 1.19 [0.88, 1.62] (P ¼ 0.26), 1.00 (Ref.), and 1.14 [0.86, 1.52] (P ¼ 0.37) from the lowest to highest sextile, respectively. On the other hand, there was no significant association between the serum Mg level and death from ipth (pg/ml) Serum magnesium (mg/dl) Serum magnesium (mg/dl) Figure 2 Restricted cubic spline showing the associations between the serum magnesium and intact parathyorid hormone (ipth) levels. (a) Unadjusted and (b) adjusted associations between the serum magnesium and ipth levels. The dashed line represents the 95% confidence interval. cancer (OR [95% CI]: 1.22 [0.92, 1.62] (P ¼ 0.17), 1.19 [0.89, 1.60] (P ¼ 0.25), 1.01 [0.74, 1.37] (P ¼ 0.94), 0.97 [0.65, 1.45] (P ¼ 0.89), 1.00 (Ref.), and 0.79 [0.54, 1.17] (P ¼ 0.25)), from the lowest to highest sextile, respectively. DISCUSSION The major finding of our study is that a lower serum Mg level was a significant and independent predictor of CVD mortality among chronic hemodialysis patients. To the best of our knowledge, this is the first published study to demonstrate the adverse effect of hypomagnesemia on CVD mortality in the hemodialysis population. We also found a significant association between hypomagnesemia and non- CVD mortality, especially deaths from infection. The large sample size, dose response relationship, and extensive subgroup analysis increased the robustness of our findings. Several mechanisms may explain the increased CVD risk of patients with hypomagnesemia. 29 First, Mg possesses an antiatherosclerotic effect, which is mediated partly via its anti-inflammatory and antioxidant properties; conversely, 176 Kidney International (2014) 85,

4 Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients clinical investigation Table 2 Univariate and adjusted odds ratios for (1) all-cause, (2) cardiovascular, and (3) non-cardiovascular mortality Adjusted Unadjusted Model 1 Model 2 Model 3 Model 3 (excluding subjects with ipth levelo50 pg/ml) Magnesium category OR 95% CI P OR 95% CI P OR 95% CI P OR 95% CI P OR 95% CI P (1) All-cause , 3.47 o , 2.14 o , 1.94 o , 1.41 o , 1.46 o , 2.18 o , 1.55 o , 1.50 o , , , 1.64 o , 1.33 o , 1.34 o , , , 1.39 o , , , , Ref. Ref. Ref. Ref. Ref , 1.44 o , 1.48 o , , , (2) CVD , 2.99 o , 1.95 o , 1.84 o , , , 2.03 o , 1.50 o , 1.49 o , , , 1.62 o , , , , , , , , , Ref. Ref. Ref. Ref. Ref , 1.55 o , 1.56 o , 1.43 o , , (3) Non-CVD , 3.70 o , 2.20 o , 1.97 o , 1.49 o , 1.55 o , 2.32 o , 1.62 o , 1.55 o , , , 1.70 o , 1.37 o , , , , 1.44 o , , , , Ref. Ref. Ref. Ref. Ref , 1.42 o , 1.45 o , , , Abbreviations: 95% CI, 95% confidence interval; CVD, cardiovascular disease; ipth, intact parathyroid hormone; OR, odds ratio. Odds ratios of the lowest against fifth sextile of serum magnesium level. Model 1 was adjusted for age, gender, hemodialysis vintage, duration of hemodialysis treatment, blood urea nitrogen level, and diabetes mellitus. Model 2 was adjusted for Model 1 þ mineral and bone disorder-related factors (levels of serum calcium, phosphate, alkaline phosphatase, intact parathyroid hormone, prescription of phosphate binders, cinacalcet hydrochloride, active vitamin D analogue, past history of parathyroidectomy, and hip fracture). Model 3 was adjusted for Model 2 þ malnutrition-inflammation and atherosclerosis complex-related factors (body mass index, serum albumin, C-reactive protein, hemoglobin level, and past history of CVD (myocardial infarction, cerebral infarction, cerebral hemorrhage, and amputation). Full adjusted odds ratio for all-cause mortality Serum magnesium (mg/dl) Figure 3 Restricted cubic spline showing the fully adjusted associations between the serum magnesium level and all-cause mortality. The dashed line represents the 95% confidence interval. Mg deficiency can cause endothelial cell dysfunction. 8,20,30,31 Associations between hypomagnesemia or low Mg intake and the incidence of CVD events, including sudden cardiac death, have been reported in the general population as well as in patients with pre-dialysis chronic kidney disease We found that while the mean CRP level in our total cohort (0.53 mg/dl) was above the reported cutoff level (0.3 mg/dl) for high risk for CVD events in the general population, 32 the level further increased as serum Mg level decreased and reached as high as 0.82 mg/dl in the lowest Mg sextile. Inflammation is a cause of anemia, hypoalbuminemia, and the high prevalence of CVD comorbidities, all of which were observed in the low-mg groups. In this context, we believe that MIA complex-related factors may be in part intervening factors rather than confounders of the association between Mg deficiency and high mortality; as a result, a substantial attenuation of mortality risk was observed in the model adjusted for MIA-related factors compared with the model adjusted for only demographics and MBD-related factors. Second, Mg has an anticalcification effect. Mg deficiency promotes hydroxyapatite formation and calcification of vascular smooth muscle cells Notably, we found that a lower Mg level was associated with a higher ALP level. ALP promotes vascular calcification by hydrolyzing pyrophosphate, a potent inhibitor of hydroxyapatite formation. Several previous studies showed significant associations between high ALP levels and arterial calcification and Kidney International (2014) 85,

5 clinical investigation Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients Table 3 Fully adjusted odds ratios for all-cause mortality in each subgroup Subgroup n OR a 95% CI P Age, years o67 43, , 1.83 o0.001 X67 45, , Gender Male 55, , Female 33, , 1.69 o0.001 Hemodialysis vintage, years o7 41, , 1.49 o0.001 X7 47, , 1.46 o0.001 Body mass index, kg/m 2 o , , 1.44 o0.001 X , , Diabetes mellitus Yes 32, , No 56, , 1.52 o0.001 Albumin, g/dl o3.7 36, , 1.52 o0.001 X3.7 52, , Calcium, mg/dl o9.2 44, , X9.2 44, , 1.50 o0.001 Phosphate, mg/dl o5.0 44, , X5.0 44, , 1.52 o0.001 CRP, mg/dl p0.1 53, , 1.73 o , , Hemoglobin, g/dl o , , 1.44 o0.001 X , , 1.52 o0.001 ALP, IU/l o234 44, , 1.56 o0.001 X234 44, , 1.41 o0.001 ipth, pg/ml o118 44, , 1.52 o0.001 X118 44, , Vitamin D use Yes 57, , 1.52 o0.001 No 31, , Past history of CVD Yes 22, , No 66, , 1.48 o0.001 Abbreviations: ALP, alkaline phosphatase; 95% CI, 95% confidence interval; CRP, C-reactive protein; CVD, cardiovascular disease; ipth, intact parathyroid hormone; OR, odds ratio. a Fully adjusted odds ratios of the lowest versus the fifth serum magnesium level sextile. mortality in hemodialysis patients One explanation for the high ALP level of individuals with a low Mg level would be the high bone turnover due to an enhanced ipth level; in fact, Mg, like calcium, suppresses PTH secretion via calcium sensing receptor binding. Finally, Mg deficiency is closely related to insulin resistance and metabolic syndrome. 36 Mg is an essential cofactor for multiple enzymes involved in glucose metabolism. It was reported that increased Mg intake was significantly associated with a lower incidence of type 2 DM. 10 Mg supplementation improves the insulin resistance index and beta-cell function, and decreases hemoglobin A1c levels in type 2 DM patients. 37,38 We also observed a slightly higher prevalence of DM in the lower Mg groups. On the other hand, Mg deficiency is also implicated in hypertension and dyslipidemia in the general population. 9,11 Unfortunately, blood pressure and lipid profile data were not available in our database; thus, their relationship with Mg levels and influence on mortality cannot be determined. Another important finding of this study is that the patients in the highest serum Mg sextile also had a significantly higher mortality risk. The reason for this is uncertain, but might be partly explained by the oversuppression of PTH under high Mg levels, as the proportion of subjects with a baseline ipth level o50 pg/ml was significantly higher in this group. It was reported that those with an ipth level o50 pg/ml had a 3.1- fold higher hazard ratio for CVD mortality than those with an ipth level between 150 and 300 pg/ml. 39 This speculation is further supported by our additional analysis that excluded subjects with an ipth level o50 pg/ml, which resulted in a decrease in the CVD mortality risk of the highest sextile. Therefore, it should be considered that the association between serum Mg levels and mortality is potentially J-shaped. This is particularly important when Mg supplementation is conducted, and the PTH level should be monitored to avoid its oversuppression. We unexpectedly found a significant association between serum Mg levels and infection-related mortality. This may not be surprising, as Mg plays a key role in immunity, e.g., as a cofactor for immunoglobulin synthesis and immune cell adhesion. 40,41 In particular, the functional importance of Mg in CD4-positive T-cell activation has recently been demonstrated. 42 Therefore, our finding is likely reasonable; however, it should be noted that our analysis permits the possibility of reverse causality, as any patients with severe infections and consequent hypomagnesemia due to low Mg intake at baseline may also have been more likely to die from infection. In fact, excluding patients with high CRP levels at baseline from the multivariate analysis attenuated the association. Although some studies have suggested a possible link between Mg levels and malignancy, the results have been inconsistent. 43,44 In vitro and animal studies have also shown contradictory data regarding Mg deficiency and the occurrence or metastasis of neoplasm. 45 We did not find a significant association between serum Mg levels and deaths from malignancy; however, the short follow-up period of our study was insufficient for the estimation of a time-course association of hypomagnesemia followed by the occurrence or progression of malignancy. Long-term studies are needed 178 Kidney International (2014) 85,

6 Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients clinical investigation to elucidate whether a low Mg level is a significant predictor of cancer mortality in patients with ESRD. What makes Mg particularly attractive for future research is not only its prognostic value but also its potential for intervention. Patients undergoing hemodialysis are at an extremely high risk of CVD, and the current therapeutic strategy is obviously suboptimal. The results of this study shed new light on a possible beneficial effect of Mg supplementation on CVD and non-cvd mortality in the ESRD population. To date, there is only a single interventional study in hemodialysis patients that reported a beneficial effect of oral Mg administration on decreasing carotid intima-media thickness, 27 although this study was limited by its small sample size. The effect of Mg administration on mortality has never been investigated. As mentioned, the relationship between serum Mg levels and mortality is J-shaped, and oversuppression of PTH under high Mg levels should be avoided. As many ESRD patients already have advanced atherosclerotic lesions at the induction of hemodialysis, it should also be investigated whether Mg supplementation in the predialysis CKD population could improve their prognosis after starting hemodialysis. The limitations of this study include its observational nature, which precludes conclusions about causality. We adjusted for numerous traditional and nontraditional CVD risk factors, performed a detailed subgroup analysis, and showed the dose response relationship between Mg and mortality; however, residual confounding variables may exist. Specifically, no information was available on blood pressure level, lipid profile, and smoking status in this study. The serum potassium level is another potential confounding factor because (1) it correlates positively with the serum Mg level 46 and (2) hypokalemia is reported to be associated with increased mortality in patients undergoing hemodialysis. 47 The short follow-up period might increase the probability of reverse causality contributing to the findings, and a longterm study is needed to overcome this drawback. On the other hand, the baseline serum Mg level represented only a single measurement, whereas the level may change over time. The potential association between changes in the serum Mg level over time and mortality should be explored in future research. Owing to the nature of questionnaire-based surveillance, the methodology of data collection, including the ascertainment of the causes of death, was not standardized. Although many baseline covariates in our data set contained missing values, most accounted for 10% or less, and the large sample size would minimize such bias. It should be noted that, as there are considerable international differences in the clinical pattern and prognosis of patients on hemodialysis, the generalizability of our findings to other ethnicities is uncertain. In conclusion, we demonstrated that a low serum Mg level was a significant predictor of increased risks for CVD and non-cvd mortality in a large cohort of hemodialysis patients in Japan. This relationship remained significant even after extensive adjustment for relevant clinical factors and subgroup analyses. The increased risk for CVD mortality observed in association with high serum Mg levels might be attributable to some extent to PTH oversuppression. Randomized trials should elucidate whether Mg supplementation could have any potential benefit in improving mortality risk in hemodialysis patients. MATERIALS AND METHODS Registry of ESRD patients All data used in this study were collected from the database of the Japanese Society for Dialysis Therapy-Renal Data Registry (JRDR). The design and detailed methods of this survey have been described elsewhere. 2 Briefly, the Japanese Society for Dialysis Therapy (JSDT) started conducting annual questionnaire surveys of dialysis facilities throughout Japan in Since 1983, the JSDT began collecting patients information prospectively and has been compiling a computer-based registry to investigate national trends in dialysis care. The rate of response to the questionnaire exceeded 98% every year, meaning that the database includes nearly all hemodialysis patients in Japan. Details on the inception, limitations, validity, variables, and questionnaires used are available online at the JSDT website ( In Japan, a dialysate Mg concentration of 1.0 meq/l is generally prescribed. The study protocol was approved by the Medicine Ethics Committee of JSDT. Study sample For the current study, a data set (JRDR 11002) was created from the database in 2009, when serum Mg levels were examined for the first time in the survey. The original data set initially contained a total of 263,849 dialysis patients aged 18 years or older, representing 99% of the dialysis facilities in Japan. We identified all patients on conventional hemodialysis and whose serum Mg level was available. Patients on peritoneal dialysis, hemofiltration, hemodiafiltration, and short daily or home hemodialysis were excluded. Baseline covariates Baseline demographic and clinical characteristics of these patients were obtained from the database, including age, gender, body mass index, the primary kidney disease (DM or non-dm), hemodialysis vintage (year), duration of hemodialysis treatment (hours per week); laboratory measurements (predialysis serum levels of albumin, urea nitrogen, calcium, phosphate, CRP, hemoglobin, ALP, ipth); prescription of phosphate binders (calcium carbonate, sevelamer hydrochloride, and lanthanum carbonate), cinacalcet hydrochloride, and oral and/or intravenous active vitamin D analogs; and past history of parathyroidectomy, CVD (myocardial infarction, cerebral infarction, cerebral hemorrhage, and amputation of the extremities), and hip fracture. When the serum albumin level was o4.0 g/dl, the serum calcium level was adjusted as follows: corrected serum calcium level (mg/dl) ¼ measured serum calcium level (mg/dl) þ (4.0 serum albumin level (g/dl)). Exposure The baseline serum Mg level for each subject was the latest measurement of serum Mg made in Given a possible nonlinear relationship with mortality rates, the serum Mg level was treated as a categorical variable and divided into sextiles (see Table 1 for the cutoff points). We also treated the serum Mg level as a continuous variable and modeled a nonlinear effect by using a restricted cubic spline function. Kidney International (2014) 85,

7 clinical investigation Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients Outcomes The study outcomes were 1-year all-cause and cause-specific mortality, which were obtained from the JRDR database in The classification codes for the underlying cause of each death have been reported elsewhere. 48 CVD mortality included deaths coded as heart failure, pulmonary edema, ischemic heart disease, arrhythmia, and cerebrovascular disease. All other causes of death were categorized as non-cvd death. These data were ascertained essentially by a review of the subjects medical records by the questionnaire respondents. Statistical analysis Data were presented as the number (percent) for categorical variables and as the mean (s.d.) for continuous variables with a normal distribution or median (interquartile range) for those with a skewed distribution. The distributions of the baseline characteristics stratified by serum Mg sextiles were compared using trend analysis. Logistic regression models were constructed to calculate the OR and 95% CI for all-cause, CVD, and non-cvd deaths, in which the fifth sextile was taken as a reference group. We constructed three multivariate models: model 1 adjusted for age, gender, hemodialysis vintage, duration of hemodialysis treatment, serum urea nitrogen level, and DM; model 2 adjusted for variables in model 1 plus MBDrelated factors (serum calcium, phosphate, ALP, and ipth levels, prescription of phosphate binders, cinacalcet hydrochloride and active vitamin D analogs, and past history of parathyroidectomy and hip fracture); and model 3 adjusted for variables in model 2 plus MIA complex-related factors (body mass index, serum albumin, CRP, hemoglobin level, and past history of CVD). The variables for which the data were not normally distributed (i.e., hemodialysis vintage, CRP, and ipth) were logarithmically transformed before incorporation into the models. We additionally explored the continuous, potentially nonlinear relationship between the serum Mg level and mortality by using fully adjusted, restricted cubic spline models with four knots. To test the robustness of our findings, we performed subgroup analyses based on a priori defined variables, i.e., age, gender, dialysis vintage, body mass index, DM, serum levels of albumin, calcium, phosphate, CRP, hemoglobin, ALP and ipth, prescription of vitamin D analogs, and past history of CVD. Subgroups of the continuous variables were created by dichotomization of these variables at the median value. All reported P values were two-sided and values of Po0.05 were considered statistically significant. All statistical analyses were performed using StataIC 12 Statistical Software (StataCorp LP, College Station, TX). DISCLOSURE All the authors declared no competing interests. SUPPLEMENTARY MATERIAL Table S1. Comparison of baseline characteristics between subjects with and without serum Mg level data. Supplementary material is linked to the online version of the paper at REFERENCES 1. US Renal Data System (USRDS 2009). Annual Data Report: Atlas of Chronic Kidney Disease and End-Stage Renal Disease in the United States. National Institute of Diabetes and Digestive and Kidney Diseases: Bethesda, MD, Nakai S, Iseki K, Itami N et al. Overview of regular dialysis treatment in Japan (as of 31 December 2009). Ther Apher Dial 2012; 16: de Jager DJ, Grootendorst DC, Jager KJ et al. Cardiovascular and noncardiovascular mortality among patients starting dialysis. JAMA 2009; 302: Foley RN, Parfrey PS, Sarnak MJ. Clinical epidemiology of cardiovascular disease in chronic renal disease. Am J Kidney Dis 1998; 32: S112 S Stenvinkel P, Carrero JJ, Axelsson J et al. Emerging biomarkers for evaluating cardiovascular risk in the chronic kidney disease patient: how do new pieces fit into the uremic puzzle? Clin J Am Soc Nephrol 2008; 3: Block GA, Klassen PS, Lazarus JM et al. Mineral metabolism, mortality, and morbidity in maintenance hemodialysis. J Am Soc Nephrol 2004; 15: Tentori F, Blayney MJ, Albert JM et al. Mortality risk for dialysis patients with different levels of serum calcium, phosphorus, and PTH: the Dialysis Outcomes and Practice Patterns Study (DOPPS). Am J Kidney Dis 2008; 52: Van Laecke S, Van Biesen W, Vanholder R. Hypomagnesaemia, the kidney and the vessels. Nephrol Dial Transplant 2012; 27: Peacock JM, Folsom AR, Arnett DK et al. Relationship of serum and dietary magnesium to incident hypertension: the Atherosclerosis Risk in Communities (ARIC) Study. Ann Epidemiol 1999; 9: Schulze MB, Schulz M, Heidemann C et al. Fiber and magnesium intake and incidence of type 2 diabetes: a prospective study and meta-analysis. Arch Intern Med 2007; 167: He K, Liu K, Daviglus ML et al. Magnesium intake and incidence of metabolic syndrome among young adults. Circulation 2006; 113: Liao F, Folsom AR, Brancati FL. Is low magnesium concentration a risk factor for coronary heart disease? The Atherosclerosis Risk in Communities (ARIC) Study. Am Heart J 1998; 136: Zhang W, Iso H, Ohira T et al. Associations of dietary magnesium intake with mortality from cardiovascular disease: the JACC study. Atherosclerosis 2012; 221: Larsson SC, Orsini N, Wolk A. Dietary magnesium intake and risk of stroke: a meta-analysis of prospective studies. Am J Clin Nutr 2012; 95: Ohira T, Peacock JM, Iso H et al. Serum and dietary magnesium and risk of ischemic stroke: the Atherosclerosis Risk in Communities Study. Am J Epidemiol 2009; 169: Khan AM, Lubitz SA, Sullivan LM et al. Low serum magnesium and the development of atrial fibrillation in the community: the Framingham Heart Study. Circulation 2013; 127: Chiuve SE, Korngold EC, Januzzi JL Jr et al. Plasma and dietary magnesium and risk of sudden cardiac death in women. Am J Clin Nutr 2011; 93: Kanbay M, Yilmaz MI, Apetrii M et al. Relationship between serum magnesium levels and cardiovascular events in chronic kidney disease patients. Am J Nephrol 2012; 36: Shechter M, Sharir M, Labrador MJ et al. Oral magnesium therapy improves endothelial function in patients with coronary artery disease. Circulation 2000; 102: Maier JA. Endothelial cells and magnesium: implications in atherosclerosis. Clin Sci (Lond) 2012; 122: Rüfenacht HS, Fleisch H. Measurement of inhibitors of calcium phosphate precipitation in plasma ultrafiltrate. Am J Physiol 1984; 246: F648 F Montezano AC, Zimmerman D, Yusuf H et al. Vascular smooth muscle cell differentiation to an osteogenic phenotype involves TRPM7 modulation by magnesium. Hypertension 2010; 56: Kircelli F, Peter ME, Sevinc Ok E et al. Magnesium reduces calcification in bovine vascular smooth muscle cells in a dose-dependent manner. Nephrol Dial Transplant 2012; 27: Tzanakis I, Pras A, Kounali D et al. Mitral annular calcifications in haemodialysis patients: a possible protective role of magnesium. Nephrol Dial Transplant 1997; 12: Ishimura E, Okuno S, Kitatani K et al. Significant association between the presence of peripheral vascular calcification and lower serum magnesium in hemodialysis patients. Clin Nephrol 2007; 68: Tzanakis I, Virvidakis K, Tsomi A et al. Intra- and extracellular magnesium levels and atheromatosis in haemodialysis patients. Magnes Res 2004; 17: Turgut F, Kanbay M, Metin MR et al. Magnesium supplementation helps to improve carotid intima media thickness in patients on hemodialysis. Int Urol Nephrol 2008; 40: Kidney International (2014) 85,

8 Y Sakaguchi et al.: Mg and mortality risk in hemodialysis patients clinical investigation 28. Ishimura E, Okuno S, Yamakawa T et al. Serum magnesium concentration is a significant predictor of mortality in maintenance hemodialysis patients. Magnes Res 2007; 20: Massy ZA, Drueke TB. Magnesium and outcomes in patients with chronic kidney disease: focus on vascular calcification, atherosclerosis and survival. Clin Kidney J 2012; 5(Suppl 1): i52 i Maier JA, Malpuech-Brugère C, Zimowska W et al. Low magnesium promotes endothelial cell dysfunction: implications for atherosclerosis, inflammation and thrombosis. Bochim Biophys Acta 2004; 1689: Dickens BF, Weglicki WB, Li YS et al. Magnesium deficiency in vitro enhances free radical-induced intracellular oxidation and cytotoxicity in endothelial cells. FEBS Lett 1992; 311: Pearson TA, Mensah GA, Alexander RW et al. Markers of inflammation and cardiovascular disease: application to clinical and public health practice: a statement for healthcare professionals from the Centers for Disease Control and Prevention and the American Heart Association. Circulation 2003; 107: Sigrist MK, Taal MW, Bungay P et al. Progressive vascular calcification over 2 years is associated with arterial stiffening and increased mortality in patients with stages 4 and 5 chronic kidney disease. Clin J Am Soc Nephrol 2007; 2: Regidor DL, Kovesdy CP, Mehrotra R et al. Serum alkaline phosphatase predicts mortality among maintenance hemodialysis patients. J Am Soc Nephrol 2008; 19: Beddhu S, Baird B, Ma X et al. Serum alkaline phosphatase and mortality in hemodialysis patients. Clin Nephrol 2010; 74: Belin RJ, He K. Magnesium physiology and pathogenic mechanisms that contribute to the development of the metabolic syndrome. Magnes Res 2007; 20: Rodríguez-Morán M, Guerrero-Romero F. Oral magnesium supplementation improves insulin sensitivity and metabolic control in type 2 diabetic subjects: a randomized double-blind controlled trial. Diabetes Care 2003; 26: Guerrero-Romero F, Rodríguez-Morán M. Magnesium improves the beta-cell function to compensate variation of insulin sensitivity: doubleblind, randomized clinical trial. Eur J Clin Invest 2011; 41: Naves-Díaz M, Passlick-Deetjen J, Guinsburg A et al. Calcium, phosphorus, PTH and death rates in a large sample of dialysis patients from Latin America. The CORES Study. Nephrol Dial Transplant 2011; 26: Tam M, Gómez S, González-Gross M et al. Possible roles of magnesium on the immune system. Eur J Clin Nutr 2003; 57: Galland L. Magnesium and immune function: an overview. Magnesium 1988; 7: Li FY, Chaigne-Delalande B, Kanellopoulou C et al. Second messenger role for Mg2 þ revealed by human T-cell immunodeficiency. Nature 2011; 475: Zhang X, Giovannucci EL, Wu K et al. Magnesium intake, plasma C-peptide, and colorectal cancer incidence in US women: a 28-year follow-up study. Br J Cancer 2012; 106: Chen GC, Pang Z, Liu QF. Magnesium intake and risk of colorectal cancer: a meta-analysis of prospective studies. Eur J Clin Nutr 2012; 66: Wolf FI, Trapani V, Cittadini A et al. Hypomagnesaemia in oncologic patients: to treat or not to treat? Magnes Res 2009; 22: Wyskida K, Witkowicz J, Chudek J et al. Daily magnesium intake and hypermagnesemia in hemodialysis patients with chronic kidney disease. J Ren Nutr 2012; 22: Hwang JC, Wang CT, Chen CA et al. Hypokalemia is associated with increased mortality rate in chronic hemodialysis patients. Blood Purif 2011; 32: Nakai S, Iseki K, Itami N et al. An overview of regular dialysis treatment in Japan (as of 31 December 2010). Ther Apher Dial 2012; 16: Kidney International (2014) 85,