Vitamin D, Cardiovascular Disease, and Survival in Dialysis Patients
Dialysis patients have an increased burden of cardiovascular disease, with >70% of patients dying from a cardiovascular event. Several recent retrospective cohort studies have shown an association of treatment with calcitriol or its analogs and improved survival. The similar findings of these studies increase the validity of the observations, but these studies are subject to biases inherent in this type of study design. In particular, the nonrandom choice to treat, lack of information on concomitant medications, and differences in characteristics of patients who are treated compared with those not treated can lead to significant bias. Thus, prospective studies are needed. Two, small, uncontrolled, prospective studies showed improvement in left ventricular hypertrophy in dialysis patients treated with α-calcidiol or calcitriol, and there is abundant animal and in vitro studies showing biological plausibility. Thus, given the available evidence, it seems prudent to advise that most dialysis patients receive some form of vitamin D while we await results of a prospective interventional trial. The unknown is what type and how much to give, and whether this approach is safe, especially in patients with low levels of PTH.
In the setting of chronic kidney disease (CKD), PTH is increased in response to hyperphosphatemia and low 1,25(OH)2D. In patients with CKD stages 3 and 4 (estimated glomerular filtration rate [GFR] of 15–60 ml/min), there is deficiency of 25(OH)D (calcidiol), progressive deficiency of 1,25(OH)2D (calcitriol) with loss of GFR, and progressive development of secondary hyperparathyroidism. In patients on dialysis, the prevalence of secondary hyperparathyroidism is >50%. In addition to increased secretion of PTH, there is altered degradation and resistance to its skeletal effects. As a result, levels of PTH are uniformly increased compared with the general population, and yet levels twice the upper limit for the general population may be associated with decreased bone turnover in patients with CKD. Unfortunately, there are significant intra-assay limitations of the currently available intact assays in patients on dialysis. However, despite these limitations of the assay, serum levels of PTH remain the best studied surrogate marker for bone turnover, and Kidney Disease Outcomes Quality Improvement (K/DOQI) clinical practice guidelines recommend a target intact PTH level of 150–300 pg/ml in patients undergoing dialysis. In the United States, injectable calcitriol or its analogs are the predominant treatment for dialysis patients with secondary hyperparathyroidism.
The use of the vitamin D sterol calcitriol has been key to the management of hyperparathyroidism in CKD for nearly 30 yr. However, a common side effect has always been hypercalcemia. Initially, the higher level of serum calcium was thought to be a therapeutic advantage, providing additional PTH suppressive effects independent of vitamin D. However, studies identified hypercalcemia and excess calcium load in the form of phosphate binders to be associated with mortality, vascular calcification, and adynamic (low turnover) bone disease. As hypercalcemia became a concern, new vitamin D analogs were designed to maximize PTH suppression yet minimize intestinal absorption of calcium and phosphorus. Two “less calcemic” analogs are commercially available in the United States: 19-nor-1,25(OH)2D2 (paricalcitol) and 1α(OH)D2 (doxercalciferol), and others are available outside the United States. Unfortunately, there are no published direct comparative trials of doxercalciferol to paricalcitol nor doxercalciferol to calcitriol. A secondary analysis of a trial comparing paricalcitol and calcitriol has recently been published. This analysis found that, although there was no difference between paricalcitol and calcitriol in the number of subjects who had a single episode of hypercalcemia, paricalcitol led to less sustained hypercalcemia. Unfortunately, showing this in a randomized comparative trial is difficult, given that oversuppression or PTH and the concomitant use of different phosphate binders may alter serum calcium and phosphorus levels independent of the effects of the analogs, requiring complex statistical adjustments. More recently, attention has focused on the potential positive effects of these analogs on cardiovascular disease and survival in dialysis patients, apparently independent of their effects on calcium, phosphorus, and PTH. These studies have used databases of clinical data from large dialysis providers (Table 1).
EPIDEMIOLOGICAL STUDIES EVALUATING THE ASSOCIATION BETWEEN THE ADMINISTRATION OF CALCITRIOL OR ITS ANALOGS AND MORTALITY
Table Table 1.. Studies Evaluating the Association of Calcitriol or Its Analogs on Mortality in Dialysis Patients
The first study that evaluated the association between the administration of calcitriol or its analogs and mortality was published in 2003 by Teng et al. This study evaluated 67,399 hemodialysis patients who were started on therapy with paricalcitol or calcitriol from 1999 to 2001. Patients who received paricalcitol were more likely to be black and have arteriovenous fistulas for access, less likely to be diabetic, and had higher calcium, phosphorus, and PTH baseline values. Patients were followed for up to 3 yr after starting either form of vitamin D, censored for death (n = 3417), switch to another formulation, transplantation, or another facility. The results showed a significant difference in survival at 12 mo that increased with time. After adjusting for confounders in patients who only received one type of vitamin D derivative, paricalcitol-treated patients had a hazard ratio of 0.84 (95% CI: 0.79–0.90, p < 0.001). The survival advantage for paricalcitol was present for all quartiles of calcium, phosphorous, and PTH, suggesting that the benefit was independent of mineral metabolism changes. In addition, in the 14,862 patients who switched from calcitriol to paricalcitol, the 2-yr survival was 73% compared with 64% in the 1621 patients who switched from paricalcitol to calcitriol (marginally significant at p = 0.04). Another study using this Fresenius database showed that paricalcitol-treated patients had fewer and shorter hospitalizations than patients treated with calcitriol.
These studies were followed by a study using the same database that compared incident hemodialysis patients who had never received injectable vitamin D (n = 13,864) to patients who received any form of injectable vitamin D (n = 37,173; majority paricalcitol), analyzed as a time-dependent variable. The patients treated with calcitriol or paricalcitol had higher PTH values and were more likely to be black. The results, after adjustment, showed that patients who received either paricalcitol or calcitriol had a reduced hazard ratio for mortality of 0.80 (95% CI, 0.76–0.83) that persisted when stratified for each covariate.
An analysis of the Dialysis Clinics (DCI) database found similar findings. This is the only major dialysis chain with a large enough sampling of patients receiving doxercalciferol to assess the effect of this analog compared with others. Patients who started hemodialysis at DCI from 1999 to 2004 and survived at least 30 days were analyzed (n = 14,967), censoring for death (n = 1326), transplantation, change to peritoneal dialysis, change to a different type of vitamin D (n = 2901), and study end. There was no difference in all-cause mortality between doxercalciferol (n = 2432) and paricalcitol (n = 2087) in any model. There was a significant difference in the survival between calcitriol and paricalcitol or doxercalciferol, but this difference disappeared when corrected for baseline laboratories and individual unit mortality rates. Importantly, the effective dose used (when adjusted for potency differences in lowering PTH) was similar for all three drugs. However, compared with any of these three drugs, patients who received no vitamin D had increased all-cause mortality (HR, 1.20; 95% CI, 1.10–1.32). Similar to the studies by Teng et al., the patients who did not receive calcitriol or an analog were different, with less blacks, older age, higher baseline calcium levels, and lower baseline PTH levels.
Many of these databases used baseline values of calcium, phosphorus, and PTH to adjust for the mortality differences. Using a slightly different approach, Kalantar-Zadeh et al. used time-averaged values, averaging values into 3-mo (quarterly time varying) periods. They compared models that adjusted by baseline value and models adjusted using quarterly time varying values for albumin, protein catabolic rate, iron-binding capacity, creatinine, bicarbonate, ferritin, hemoglobin, white blood cell count, calcium, phosphorus, PTH, and a composite of malnutrition-inflammation-cachexia indices. They were also able to quantify the micrograms per week of calcitriol or paricalcitol administered as a quarterly averaged dose. These authors used the DaVita dialysis database evaluating 58,058 patients, 39,305 who received paricalcitol during the 3 mo between 2001 and 2003. Using several different models, the authors found a reduction risk for all-cause mortality in patients who received paricalcitol compared with those who did not receive any injectable vitamin D derivative, with a hazard ratio of 0.865 (95% CI, 0.082–0.912). There were slight differences depending on the dose.
These studies all represent retrospective analyses of clinical databases and were unable to adjust for baseline comorbidities with the exception of diabetes. Thus, baseline cardiovascular disease was not adjusted for. To overcome this, Melamed et al. used existing data from a prospective cohort, the CHOICE (Choices for Healthy Outcomes in Caring for End-Stage Renal Disease) study. This is a prospectively collected national cohort of incident dialysis patients with repeated measures of laboratory values collected from 1995 to 1998, based in DCI units. This study differs from others in that more detailed comorbidity data were collected at baseline. The disadvantage of this cohort is the relatively small sample size compared with the previously described studies. Similar to other studies, there was no information on oral medication use. However, even with adjustment of these detailed comorbid confounders, including C-reactive protein (CRP) and interleukin-6 levels, there was reduced mortality in patients treated with injectable calcitriol compared with those who did not receive calcitriol or any analog. The hazard ratio was 0.62 (95% CI, 0.44–0.86) using baseline values as predictors (n = 552) versus 0.74 (95% CI, 0.56–1.00) using time-dependent predictors (n = 593).
All of these studies were from patients undergoing dialysis in the United States; however, this effect was also observed in the Japanese national database of 77,486 hemodialysis patients in 1998 where the relative risk of mortality was 0.76 (p < 0.0001) for patients using any form of calcitriol or its analogs compared with nonusers. In a smaller prospective cohort of patients in Japan, those treated with α-calcidiol (1-α-hydroxyvitamin D3) over 24 mo (n = 162) had decreased all-cause mortality and specifically decreased cardiovascular mortality compared with 80 patients who did not use α-calcidiol (HR, 0.377; 95% CI, 0.246–0.578; p = 0.022). Of note, because this was a prospective cohort study, they were able to document that the α-calcidiol users received drug 85% of the 24 mo, with 63% of the subjects using the drug for the entire 24-mo period. Conversely, 56% of the non-vitamin D users never received α-calcidiol, with 88% of the patient months documented to be true nonadministration.
In summary, these cohort studies consistently showed improved survival in dialysis patients receiving calcitriol or its analogs compared with patients who did not (Table 1). Although one study suggests that paricalcitol is superior to calcitriol, another study failed to confirm these observations. Thus, although these data support that patients on dialysis should probably be treated with some form of calcitriol or its analogs, which form is not yet clear. In addition, although these data have been duplicated in several large databases, none of the studies can completely control for all potential biases, especially the nonrandom assignment of therapy by the patients' physicians and concomitant medication use. Furthermore, it is important to point out studies such as these may find statistical significance because of the large sample size, but this does not necessarily mean clinical significance. In addition, patients treated with any form of calcitriol or its analogs are less likely to be diabetic and more likely to be black, both traits that convey a survival advantage. Thus, the results clearly need to be confirmed in a prospective interventional trial. In addition, we must also seek biological plausibility for these observations.
CARDIOVASCULAR EFFECTS OF CALCITRIOL OR ITS ANALOGS
Up to 70% of patients on dialysis die of cardiovascular disease; thus, it is most likely that calcitriol or its analogs must impact cardiovascular disease to be associated with a survival benefit. Although there is a high prevalence of traditional Framingham risk factors (diabetes, hypertension, etc.) in patients on dialysis, studies have shown that these risk factors cannot completely account for the increased cardiovascular risk compared with the general population. Thus, there is increasing awareness of so-called nontraditional risk factors that may be unique, or accelerated, in the setting of CKD, including derangements in vitamin D. There is a deficiency of both calcidiol and calcitriol in patients with CKD at all stages, which complicates differentiating endocrine (lowering PTH) from autocrine (immunologic, cardiovascular) effects of vitamin D.
The vitamin D receptor (VDR) is found in cardiac myocytes and vascular smooth muscle cells, and thus it is possible that these drugs have a direct effect on the cardiovascular system. In the general population, low calcidiol levels are associated with congestive heart failure, high blood pressure, and low matrix metalloproteinase-9 (MMP-9) and high CRP levels. In elderly patients, treatment with cholecalciferol lowered systolic blood pressure and PTH levels.
In 44 dialysis patients, low 25(OH)D levels were inversely correlated with left ventricular hypertrophy (LVH); treatment for 6 wk with intravenous 1-α-cholecalciferol led to improvement in LVH. In another study, treatment of 15 hyperparathyroid dialysis patients with intravenous calcitriol for 15 wk led to a reduction in LVH, suppression of renin and angiotensin levels, and lowering of atrial natriuretic peptide (ANP) levels. Unfortunately, these human studies in dialysis patients do not allow the separation of a direct effect from an indirect effect of lowering PTH. In addition, in dialysis patients, there is concern that the administration of calcitriol or its analogs may be associated with vascular calcification, although it is nearly impossible to separate the direct actions from the indirect effect of raising the calcium × phosphorous product on vascular calcification in humans. However, the potential adverse effect of calcitriol or its analogs raising blood calcium and phosphorous levels, especially in patients with low PTH, suggests cautious use in dialysis patients. Despite these limitations in clinical studies, there is increasing data in in vitro and in animal studies that vitamin D deficiency may have an anti-inflammatory effect that protects against blood vessel injury, inhibits cardiac hypertrophy, and regulates the renin-angiotensin system. Thus, there is mounting evidence for the biological plausibility of the findings from these observational studies.
Dialysis patients have an increased burden of cardiovascular disease, with >70% of patients dying from a cardiovascular event. Several recent large cohort studies showed an association of treatment with calcitriol or its analogs and improved survival. The similar findings of these studies increase the validity of the observations, but these studies are subject to biases inherent in this type of study design. In particular, the nonrandom choice to treat, concomitant medication use, and differences in characteristics of patients who are treated compared with those not treated can lead to significant bias. Thus, prospective studies are needed. Two, small, uncontrolled, prospective studies showed improvement in left ventricular hypertrophy in dialysis patients treated with calcitriol, and there is abundant animal and in vitro studies showing biological plausibility. Thus, given the available evidence, it seems prudent to advise that most dialysis patients receive some form of vitamin D while we await results of a prospective interventional trial. The unknown is what type and how much to give, and whether this approach is safe, especially in patients with low levels of PTH. Thus, further research is needed.
SMM is supported by an NIH K24 Award and a VA Merit Award.