Hypomagnesemia is associated with new-onset diabetes mellitus following heart transplantation

Heart transplantation (HT) is currently the “gold standard” therapy for selected patients with end-stage heart failure. Still, despite significant advances in the field, this treatment is associated with high rates of morbidity and mortality. A major cause of morbidity and mortality following HT is diabetes mellitus: it is known that at 1 and 5 years following HT, 21% and 35% of survivors, respectively, suffer from diabetes [1]. In addition to pre-existing diabetes, new-onset diabetes after transplantation (NODAT) may also develop as a complication that has a detrimental impact on patient survival or on other transplant-related adverse events [2].

Another unfavorable post-HT occurrence may be the development of a magnesium deficiency, to which a possible contributory factor is the administration of calcineurin inhibitors (CNIs), which are known to induce magnesium urinary wasting [3, 4]. Moreover, MiR-133a-regulated calcineurin-nuclear factor in activated T cells c4 (NFATc4) signalling and DNA methyltransferases-1 (DNMTs-1)-3a is changed in diabetic hearts and has been shown to be associated with a hypertrophic response and cardiac remodeling [5]. Indeed, it has been reported that hypomagnesemia frequently develops within the first few weeks following kidney transplantation, with a nadir in the serum magnesium (s-Mg) level in the second month post-transplantation and persistent hypomagnesemia is invariably accompanied by myocardial magnesium depletion in the transplanted heart [6]. Several studies have indicated that magnesium deficiency is nonetheless a potentially modifiable risk factor for diabetes in both non-transplant patients and in kidney transplant recipients [7‐9]; it is also known that magnesium in the high-normality range is associated with a lower cardiovascular risk [10] and its levels are independently and inversely associated with prediabetes and overt diabetes [11]. The molecular basis for the involvement of magnesium in the pathogenesis of diabetes may lie in its role as a co-factor in several pathways, including glucose transport and insulin sensitivity and secretion [8, 12]. In healthy individuals, binding of insulin to insulin receptor in vitro leads to translocation of magnesium to platelets, leading to reduced platelet aggregation and decreased release of pro-aggregatory agents like thromboxane; this protective effect is lost in diabetics [13].

Given the high prevalence of both hypomagnesemia and diabetes in HT patients, and the proposed association of low s-Mg with an increased risk for diabetes in non-transplant patients and for NODAT in kidney transplant recipients, we designed a study to determine the association between hypomagnesemia and the incidence of NODAT in HT patients.

Diabetes is a major confounder of mortality and morbidity following HT, and therefore every effort should be made to reduce the diabetes burden in HT recipients. To the best of our knowledge, our current study is the first to demonstrate that low s-Mg after HT is independently associated with a significant > 8-fold increased risk for NODAT. Additional important findings were that the incidence of stroke was significantly higher in patients with low s-Mg levels compared to those with high s-Mg (14% vs 0, p = 0.031) as well as higher mortality rate.

Diabetes is common in HT recipients and is associated with high incidence of infection, cardiac allograft vasculopathy (CAV), graft loss, and reduced survival [20, 21]. Although risk factors for NODAT in transplant patients are similar to those in the general population (including older age, higher BMI, and male gender), immunosuppressive medications significantly further contribute to NODAT [22]. It has been shown that tacrolimus reduces insulin secretion more potently than cyclosporine both in vitro and in vivo [23, 24]. In the DIRECT trial, the incidence of NODAT was significantly lower for patients treated with cyclosporine than for those receiving tacrolimus [24]. In our study cohort, tacrolimus and cyclosporine therapies were equally distributed among the two groups, with similar mean trough levels for the two groups. This distribution, taken together with the Cox proportional hazards model, suggest that hypomagnesemia is an independent risk factor for NODAT.

Diabetes per se may also induce urinary magnesium loss, caused by hyperglycemia, hyperfiltration or a direct effect of insulin on the kidneys’ magnesium Mg channels [25]. In contrast, in the non-diabetes or pre-diabetes general population, serum glucose levels are below the threshold for urinary Mg wasting and are hence unlikely to influence s-Mg levels [8]. One of the strengths of our study is thus that by excluding patients with diabetes before the HT or those developing diabetes within the first year after HT, and following s-Mg levels with repeated measurements for 12 months post-HT, while evaluating outcomes beyond 12 months, we precluded reverse causation (as diabetes per se may induce urinary magnesium loss [25]).

Hypomagnesemia after transplantation has been attributed to a number of factors: the use of CNIs, which induce urinary loss of magnesium. In this context, a sub-population of L-type Ca2+ channels (LTCCs) has also been identified in caveolae domains that appears critical in regulating β-adrenoceptor and hypertrophic calcineurin/nuclear factor of activated T-cells (NFAT) heading to a reduced ischemic tolerance and lesser cardioprotection [26]. Additional factors are loop diuretic agents, which are frequently administered in these patients [27], and impaired gastrointestinal absorption of magnesium due to the diarrhea commonly occurring post-HT. It has been shown in kidney transplant patients and non-transplanted patients that lower s-Mg levels are an independent risk factor for new onset diabetes [7‐9, 28‐30]. We note that most of these studies showing an association between low s-Mg and diabetes mellitus, were based on a single measurement [8], a median or 30-day moving average assessment [9, 29], or s-Mg concentrations measured 1 year apart [8, 31]. A retrospective study of kidney transplant recipients evaluating time-dependent exposure (i.e., using 3 monthly time-varying and rolling-average s-Mg levels) indicated that low s-Mg are an independent risk factor for NODAT [7]. Another strength of our study is thus that it is based on a comprehensive assessment and repeated consistent measures of s-Mg, along with other detailed clinical parameters, allowing a detailed analysis of the study question. Our analysis thus suggests that it is the consistency of low s-Mg over time that influences the diabetes risk.

The pathophysiological mechanisms underlying the association between hypomagnesemia and NODAT are complex and have not yet been fully elucidated. It has been suggested that a number of mechanisms may provide the molecular and functional basis for the involvement of magnesium in the pathogenesis of diabetes mellitus, as follows: (1) Magnesium is a co-factor necessary for glucose metabolism in several pathways, including transport between membranes, glucose oxidation, and insulin-mediated tyrosine kinase pathways [12, 32, 33], and it may therefore be involved in insulin secretion or insulin signaling [34]. Magnesium deficiency has been shown to promote insulin resistance [32, 33, 35], and magnesium supplementation has been reported to improve both glucose tolerance and insulin sensitivity [36] in animal and clinical studies in non-transplant patients with diabetes mellitus [37, 38]. In addition, it should be pinpointed that after administration of SGLT2 inhibitors, a raise in serum magnesium concentrations (not coupled with a reduced urinary excretion) is associated with a significant improvement in endothelial function, arterial stiffness and renal resistive index [39]. (2) Common genetic variations in the magnesium-regulating genes TRPM6, SLC41A2, CLDN19, CNNM2, and FXYD2 have been shown to significantly modify the risk of diabetes through s-Mg levels [8, 40, 41]. (3) It is also possible that mitochondrial dysfunction could underlie the association between low s-Mg and NODAT, as both hypomagnesemia and diabetes can result from mitochondrial dysfunction, as found in large pedigree with hypomagnesemia and metabolic syndrome [29, 42]. This association is further enforced by the findings in kidney transplant recipients that tacrolimus can cause secondary mitochondrial respiratory chain dysfunction [43].

Before concluding the Discussion, we touch on two findings that are relevant not only to HT patients but also to the general population at large—stroke and calcium metabolism. Prospective studies in large populations have indicated a dose-dependent inverse association between dietary magnesium and stroke incidence [44‐46], with a recent meta-analysis reporting a 22% lower risk of stroke in people in the highest vs the lowest categories of dietary magnesium intake [47]. Consistent with these findings, we report here a significantly higher rate of cerebrovascular events in patients with low s-Mg. Magnesium depletion has also been associated with multiple biochemical abnormalities, among them bone and calcium metabolism. There is evidence for a suppressive effect of hypomagnesemia on parathyroid hormone secretion and resistance [48], further supporting the importance and potential clinical implications of managing low-Mg state in HT patients known to suffer frequently from osteoporosis, partially due to prolonged steroid treatment.

Our findings may have potentially important implications for the management of patients following HT. As hypomagnesemia is a potentially modifiable risk factor for diabetes and NODAT, a number of studies were conducted aiming to verify its clinical implications. In non-transplant populations higher consumption of magnesium was associated with a lower risk of diabetes [37, 38, 49, 50], and in a randomized trial of kidney transplant patients, magnesium supplementation 3 months post-transplant, was shown to improve fasting blood glucose levels [51]. The increased s-Mg level observed following SGLT2 inhibition, combined with a favourable cardiovascular profile of these drugs, are suggestive of a therapeutic potential for diabetic and pre-diabetic HT recipients, that should be further studied.

There are several limitations to our study. First, there is the limitation inherent in observational trials that uncovers associations but precludes the determination of cause-and-effect relationships. Second, this study was limited by being based on a single-center experience. Third, measuring s-Mg and not intracellular magnesium levels may influence the assessment of the patients’ magnesium status. Finally, magnesium intake was not controlled or assessed. Nevertheless, the intake of magnesium should be reflected in the measured s-Mg. Thus, any conclusions drawn from the data must be replicated with a larger sample size and a prospective study design.

In conclusion, our study demonstrates that low post-HT s-Mg level is independently associated with an increased risk for NODAT in HT patients. Nevertheless, the implications of interventions, focusing on preventing or correcting low s-Mg, on the risk for NODAT and on clinical outcomes should be further evaluated.