Twenty-four− and 48-week data (postbaseline visits) represent the number of participants included in the primary analysis using a mixed-effects repeated-measures model. iPTH indicates intact parathyroid hormone; eGFR, estimated glomerular filtration rate; RAAS, renin-angiotensin-aldosterone system.
Error bars indicate 95% CIs. The shaded area corresponds to the normal range of values.
Ravi Thadhani, Evan Appelbaum, Yili Pritchett, Yuchiao Chang, Julia Wenger, Hector Tamez, Ishir Bhan, Rajiv Agarwal, Carmine Zoccali, Christoph Wanner, Donald Lloyd-Jones, Jorge Cannata, B. Taylor Thompson, Dennis Andress, Wuyan Zhang, David Packham, Bhupinder Singh, Daniel Zehnder, Amil Shah, Ajay Pachika, Warren J. Manning, Scott D. Solomon. Vitamin D Therapy and Cardiac Structure and Function in Patients With Chronic Kidney DiseaseThe PRIMO Randomized Controlled Trial. JAMA. 2012;307(7):674–684. doi:10.1001/jama.2012.120
Author Affiliations: Divisions of Nephrology (Drs Thadhani, Tamez, and Bhan and Ms Wenger) and General Medicine (Dr Chang), and Pulmonary and Critical Care Unit (Dr Thompson), Department of Medicine, Massachusetts General Hospital, and Perfuse Cardiovascular Magnetic Resonance Core Laboratory and Cardiovascular Division, Department of Medicine (Drs Appelbaum and Manning) and Department of Radiology (Dr Manning), Beth Israel Deaconess Medical Center, and Department of Medicine, Cardiovascular Division, Brigham and Women's Hospital (Drs Shah, Pachika, and Solomon), Boston; Abbott Laboratories, Abbott Park, Illinois (Drs Pritchett, Andress, and Zhang); Department of Medicine, Indiana University School of Medicine, Indianapolis (Dr Agarwal); Nephrology, Dialysis, and Transplantation Unit and Consiglio Nazionale delle Ricerche. Instituto di Biomedicina e Immunologia Molecolare “Alberto Monroy,” Clinical Epidemiology and Pathophysiology of Renal Diseases and Hypertension, Reggio Calabria, Italy (Dr Zoccali); Division of Nephrology, Department of Medicine, University Hospital, University of Wuerzburg, Wuerzburg, Germany (Dr Wanner); Department of Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois (Dr Lloyd-Jones); Servicio de Metabolismo Óseo y Mineral, Instituto Reina Sofia de Investigación, Hospital Universitario Central de Asturias, RedinRen, Instituto Carlos III, Oviedo, Spain (Dr Cannata); Melbourne Renal Research Group, Reservoir Private Hospital, Department of Nephrology, Royal Melbourne Hospital, and Department of Nephrology, Austin Hospital, Melbourne, Australia (Dr Packham); Southwest Kidney Institute, Tempe, Arizona (Dr Singh); and Clinical Sciences Research Institute, Warwick Medical School, University of Warwick, Coventry, England (Dr Zehnder).
Context Vitamin D is associated with decreased cardiovascular-related morbidity and mortality, possibly by modifying cardiac structure and function, yet firm evidence for either remains lacking.
Objective To determine the effects of an active vitamin D compound, paricalcitol, on left ventricular mass over 48 weeks in patients with an estimated glomerular filtration rate of 15 to 60 mL/min/1.73 m2.
Design, Setting, and Participants Multinational, double-blind, randomized placebo-controlled trial among 227 patients with chronic kidney disease, mild to moderate left ventricular hypertrophy, and preserved left ventricular ejection fraction, conducted in 11 countries from July 2008 through September 2010.
Intervention Participants were randomly assigned to receive oral paricalcitol, 2 μg/d (n =115), or matching placebo (n = 112).
Main Outcome Measures Change in left ventricular mass index over 48 weeks by cardiovascular magnetic resonance imaging. Secondary end points included echocardiographic changes in left ventricular diastolic function.
Results Treatment with paricalcitol reduced parathyroid hormone levels within 4 weeks and maintained levels within the normal range throughout the study duration. At 48 weeks, the change in left ventricular mass index did not differ between treatment groups (paricalcitol group, 0.34 g/m2.7 [95% CI, −0.14 to 0.83 g/m2.7] vs placebo group, −0.07 g/m2.7 [95% CI, −0.55 to 0.42 g/m2.7]). Doppler measures of diastolic function including peak early diastolic lateral mitral annular tissue velocity (paricalcitol group, −0.01 cm/s [95% CI, −0.63 to 0.60 cm/s] vs placebo group, −0.30 cm/s [95% CI, −0.93 to 0.34 cm/s]) also did not differ. Episodes of hypercalcemia were more frequent in the paricalcitol group compared with the placebo group.
Conclusion Forty-eight week therapy with paricalcitol did not alter left ventricular mass index or improve certain measures of diastolic dysfunction in patients with chronic kidney disease.
Trial Registration clinicaltrials.gov Identifier: NCT00497146
Although primarily recommended for improving bone health, treatment with vitamin D recently has been suggested for other conditions, including cardiovascular disease (CVD).1- 3 Multiple lines of evidence suggest a link between vitamin D and CVD, including experimental studies identifying vitamin D receptors in vascular smooth muscle, endothelial cells, and possibly cardiac tissue1,4 and observational studies, small clinical trials, and meta-analyses suggesting that vitamin D therapy reduces cardiovascular events.2,3 Convincing data demonstrating that vitamin D therapy improves cardiovascular health, however, are lacking.
Patients with chronic kidney disease (CKD) frequently develop deficiency of 1,25-dihydroxyvitamin D3 (calcitriol) because of a lack of its precursor, 25-hydroxyvitamin D3, and impaired activity of the kidney enzyme 1α-hydroxylase, which converts this precursor to the active hormone.5 Altered vitamin D metabolism leads to secondary hyperparathyroidism, the primary indication for calcitriol therapy.6 Observational studies in patients with CKD report associations between vitamin D deficiency and increased risk of cardiovascular events and between therapy with calcitriol or related analogs and reduced events.7- 10 Experimental models suggest that intermediate end points for these observations include a reduction of left ventricular hypertrophy (LVH), improved left ventricular diastolic function, and reduced episodes of heart failure.11- 15 Given both the altered vitamin D metabolism and elevated rates of cardiovascular events found among patients with CKD, vitamin D therapy may reduce their risk of CVD-related morbidity and mortality.
The efficacy of vitamin D therapy to modify intermediate cardiac end points has not been prospectively tested, although large-scale outcome trials have recently been initiated.16 We therefore conducted PRIMO (Paricalcitol Capsule Benefits in Renal Failure–Induced Cardiac Morbidity), an investigator-initiated, industry-sponsored, multinational, double-blind, randomized placebo-controlled trial, to test the hypothesis that 48-week treatment with paricalcitol (19-nor-1,25-[OH]2 vitamin D2) reduces left ventricular mass, improves diastolic function, reduces CVD events, and improves cardiac biomarkers in patients with LVH and CKD.
Details of the PRIMO study design have been previously published.17 In brief, eligibility criteria included 2-dimensional transthoracic echocardiographic evidence of mild to moderate LVH (septal wall thickness of 1.1-1.7 cm in women and 1.2-1.8 cm in men)18 without asymmetric septal hypertrophy or valvular disease; left ventricular ejection fraction greater than 50%; and estimated glomerular filtration rate (eGFR) of 15 to 60 mL/min/1.73 m2. Individuals taking renin-angiotensin-aldosterone system (RAAS) inhibitors followed the same regimen for at least 1 month prior to screening and throughout the study. Eligibility also included a serum intact parathyroid hormone (iPTH) level between 50 and 300 pg/mL; serum calcium level between 8.0 and 10.0 mg/dL; serum phosphorus level of 5.2 mg/dL or lower; and serum albumin level of 3.0 g/L or higher. Major exclusion criteria included receiving any active vitamin D therapy, anticipated dialysis initiation or death within 1 year, clinically significant coronary artery disease, cerebrovascular accident or acute renal failure within 3 months, and systolic blood pressure greater than 180 mm Hg or diastolic blood pressure greater than 110 mm Hg at screening. Nutritional vitamin D (cholecalciferol or ergocalciferol) was limited to 400 IU/d. Demographic and clinical characteristics were collected prospectively by site investigators. Self-reported participants' race/ethnicity was obtained given its potential influence on susceptibility to change in left ventricular mass.19
The study was approved by local and central institutional review and ethics committees, and all participants signed written informed consent statements prior to study initiation.
During a 6-week screening period, left ventricular mass, renal function, and laboratory measures were evaluated. On treatment day 1, participants were randomized 1:1 to receive paricalcitol capsules or placebo. Randomization was stratified with respect to country, sex, and baseline RAAS inhibitor use. Baseline physical examination was performed and medical history was recorded. Blood samples were collected for measurement of cardiac biomarkers.
Participants assigned to receive paricalcitol started at 2 μg/d with a protocol-specified dose reduction to 1 μg/d if serum calcium exceeded 11 mg/dL, with identical protocol-specified capsule reduction in placebo. Participants returned for study visits at weeks 4, 8, 12, 18, 24, 30, 36, 42, and 48, during which vital signs and serum iPTH were measured and adverse events and concomitant medications were recorded. Limited nonfasting blood was collected at all visits except at weeks 24 and 48, when a full chemistry panel was performed.
The primary end point was change from baseline left ventricular mass index (LVMI) over 48 weeks by cardiovascular magnetic resonance (CMR) imaging. Prespecified secondary end points included transthoracic echocardiographic measures of left ventricular diastolic function (peak early diastolic lateral mitral annular tissue velocity (E′), isovolumetric relaxation time, ratio of early mitral inflow wave velocity E-wave (E) velocity to E′, and E-wave deceleration time); CMR measures of left ventricular end systolic and diastolic volume indexes and ejection fraction; CVD events leading to hospitalization or death; and change in cardiac biomarkers.
CMR Data. Eligible candidates underwent CMR to establish baseline, week 24, and week 48 LVMI. Cardiac magnetic resonance examinations were performed using an electrocardiograph-gated, breath-hold, 2-dimensional, steady-state free precession cine with contiguous, left ventricular, short-axis stack of images acquired from just above the base to below the apex of the left ventricle with a section thickness of 10 mm (no gap), spatial resolution of 2.0 × 2.0 mm, field of view of 32 cm, and temporal resolution of 50 ms.20 All left ventricular mass measurements were made at a central CMR core laboratory (Perfuse) with investigators blinded to treatment group and temporal sequence. The reproducibility and overall excellent quality of the core CMR laboratory have been previously published,21 and any intercenter deviations in CMR image section thickness were subsequently reread after implementing additional quality control measures. Measurements were made by manual planimetry of the endocardial and epicardial left ventricular borders (to define the left ventricular myocardial area). The area of each section at end diastole was multiplied by section thickness and myocardial density. All section values were summated (QMASS MR, version 6.2.3, Medis Inc).22 Left ventricular papillary muscles and trabeculations were excluded from mass measurements. Left ventricular mass index was obtained by normalizing left ventricular mass to height to the 2.7th power.23
Echocardiographic Data. From 2-dimensional, M-mode, and Doppler (spectral, color, tissue) images, the following continuous variables were obtained: E′ (cm/s), E/E′, isovolumic relaxation time (seconds), and deceleration time (seconds).24 Left ventricular volumes were derived as previously described13 and left ventricle mass estimated from linear dimensions according to published formulas.18 To determine the full extent of structural and functional changes, exploratory echocardiographic measures included left ventricular mass, left ventricular volume in end diastole and end systole, left ventricular posterior wall and septal thickness, left ventricular internal dimension, left atrial volume, and mitral regurgitation jet area.
Prespecified end points included cardiovascular hospitalizations and deaths. An independent adjudication committee blinded to treatment assignment reviewed all hospitalizations (no deaths occurred during the study period) and adjudicated those related to cardiovascular events. Admission and discharge records, interventions during hospitalizations, and hospital course were reviewed in detail. Criteria used to define cardiovascular events were similar to those used previously.25 Committee members independently assigned all the same cases to cardiovascular-related hospitalizations (100% agreement).
Details of all laboratory measurements have been previously published.17,26 In brief, we measured iPTH (Immulite 2000, Siemens; normal range, 12-65 pg/mL) at each visit. Levels of cardiac troponin T (Roche Diagnostics; normal range, <0.01 ng/mL) and B-natriuretic peptide (BNP; Abbott Diagnostics; normal range, <100 pg/mL) were measured at 0 and 48 weeks. Because cardiac troponin T and BNP are altered by changes in renal function,27,28 results were adjusted post hoc for changes in eGFR. Vitamin D compounds alter serum creatinine levels independent of GFR17,26,29- 32; therefore, prespecified analyses included GFR estimated by both a serum creatinine–based equation (eGFR [mL/min/1.73 m2] = 186 × [Cr] − 1.154 × [age] − 0.203 × [0.742 if female] × [1.210 if African American])33 and a cystatin C–based equation (eGFR [mL/min/1.73 m2] = 99.43 × cystatin−1.58).34
Because little was known about the variability of LVMI changes in CKD during the planning stage, we prospectively implemented an information-based adaptive design that allowed sample size reestimation when 50% of the data were collected.17,26 No changes to our initial sample size estimate of 220 participants (110 per group) were necessary to achieve more than 85% power to detect a clinically meaningful difference in LVMI of 2.7 g/m2.7 (absolute left ventricular mass difference of approximately 10 g) between groups with a 2-sided α = .05.
Safety was evaluated as serious adverse events during the 48 weeks of treatment and 30 days following discontinuation of study drug. The number of participants with hypercalcemia (2 consecutive measurements of serum Ca2+ greater than 10.5 mg/dL [corrected to serum albumin of 4.0 g/dL]) and the number requiring dose reductions (from 2 to 1 μg/d) were compared between treatment groups. An external data monitoring committee (independent of the steering committee and study sponsor) operated under a formalized charter to monitor safety and efficacy and to assess LVMI variability so as to recommend sample size adjustment.
All analyses were performed using SAS software, version 9.2 (SAS Institute Inc). Means and standard deviations were used to summarize distributions and means with 95% confidence intervals were used to summarize group differences unless otherwise specified. Efficacy analyses were conducted in the intention-to-treat (ITT) population, defined as all randomized patients who received at least 1 dose of study drug and with at least 2 primary end point measurements. All analyses were also prospectively performed on a subpopulation with more severe LVH (LVH population), defined as the sex-stratified upper 3 quartiles of baseline LVMI.
The primary efficacy analysis used a maximum-likelihood, mixed-effects repeated-measures model (MMRM) with all longitudinal observations in the ITT population. The model included terms of treatment, visit, and treatment × visit interaction with baseline LVMI, sex, RAAS use, and country as covariates. As a penalty of the interim analysis, the final statistical significance level for the primary outcome was to be adjusted using the γ(−8) function.17,26 However, the information at interim analysis was small and the penalty was negligible. A 2-sided P <.05 was retained for LVMI in the final analysis. A sensitivity analysis was conducted using multiple imputation techniques to account for missing data or loss to follow-up.
Changes from baseline to all postbaseline visits for other continuous variables measured on CMR or echocardiographic imaging were analyzed by the same MMRM. Biomarkers and other laboratory variables were analyzed using an MMRM with terms of treatment, country, visit, and treatment × visit interaction as fixed categorical effects and with baseline as a continuous covariate. Means and 95% confidence intervals from the MMRM are presented throughout. These analyses were performed using PROC MIXED with denominator degrees of freedom estimated by the Satterthwaite approximation. Within-participant errors were estimated using unstructured covariance unless otherwise specified. P values for postbaseline visits represent the significance level between treatment groups at a specific postbaseline visit from the mixed-effects model. Overall P values represent the significance level for the overall treatment group effect with both follow-up times (24 and 48 weeks) combined. Post hoc analyses controlling for change in estimated eGFR by cystatin C were performed for both log-transformed BNP and percentage of participants with cardiac troponin T levels of 0.01 or higher.27,28 Estimated GFR adjustment was also conducted to examine the partial correlation between change in left atrial volume index and change in log-transformed BNP levels.
Safety measures were assessed in participants receiving at least 1 dose of study drug. The frequency of at least 1 adverse event was compared between groups using the Fisher exact test. Hospitalizations were examined using the Fisher exact test for frequency (multiple events per participant are counted only once) and Poisson regression for event rates (the total number of separate events is considered even if recurring within the same participant).
A total of 811 participants from 60 centers in 11 countries were screened from July 2008 through September 2010, leading to enrollment of 227 participants; 115 were randomly assigned to receive paricalcitol and 112 to receive placebo (Figure 1). Demographics were balanced between groups (Table 1). Participants were predominantly male and had hypertension. Other CVD risk factors were frequent in both groups. Blood pressure was well controlled in both groups. Most participants were receiving RAAS inhibitors, and the use of diuretics and erythropoiesis-stimulating agents (9.6% in paricalcitol group vs 9.3% in placebo group) also were balanced. Baseline eGFR was lower and urine albumin-creatinine ratio higher in the group randomized to paricalcitol.
Baseline cardiac imaging (Table 2) showed that left ventricular ejection fraction was well preserved in both groups. Peak early diastolic lateral mitral annular tissue velocity (E′), a measure of diastolic function, was below normal, consistent with impaired diastolic function.35
Serum iPTH levels were approximately 1.5 times the upper limit of normal at baseline. Treatment with paricalcitol reduced iPTH within the first 4 weeks, and levels remained in the normal range throughout the study duration (Figure 2). Overall, 85.7% of the paricalcitol group and 16.5% of the placebo group demonstrated a reduction of iPTH of greater than 30% by week 48 (P < .001). Levels of bone-specific alkaline phosphatase also had a larger decrease in the paricalcitol group (eTable 1). Serum calcium levels increased a mean of 0.32 mg/dL (95% CI, 0.19-0.45 mg/dL) in the paricalcitol group and decreased 0.25 mg/dL (95% CI, −0.37 to −0.12 mg/dL) in the placebo group (between-group difference in baseline-to-48-week change, P < .001). Serum phosphorus levels increased 0.23 mg/dL (95% CI, 0.07-0.39 mg/dL) in the paricalcitol group and increased 0.04 mg/dL (95% CI, −0.12 to 0.20 mg/dL) in the placebo group (between-group difference in baseline-to-48-week change, P = .05). Systolic (P = .82) and diastolic (P = .33) blood pressure and heart rate (P = .44) did not markedly differ throughout the study period between groups (eTable 2).
The primary end point was change in LVMI at 48 weeks, which did not significantly differ between groups in the ITT analysis (Table 3). Results were similar when changes in left ventricular mass were not indexed to height (paracalcitol, 1.29 g [95% CI, −0.72 to 3.29 g] vs placebo, −0.20 g [95% CI, −2.19 to 1.80 g]; P = .12). We conducted a sensitivity analysis using multiple imputation techniques to account for participants with missing data or lost to follow-up and the conclusion remained the same for the primary end point (P = .32). In the LVH population, LVMI increased slightly at week 48 in the paricalcitol group (paricalcitol, 0.46 g/m2.7 [95% CI, −0.15 to 1.08 g/m2.7] vs placebo, −0.23 g/m2.7 [95% CI, −0.87 to 0.41 g/m2.7] ; P = .05) (eTable 3). In both the ITT and LVH analyses, CMR measures of left ventricular end-diastolic volume index and ejection fraction tended to increase in the paricalcitol group and decrease in the placebo group, but the comparisons did not reach statistical significance. The change in prespecified echocardiographic measures of diastolic function did not significantly differ between the paricalcitol and placebo groups (Table 4 [ITT population] and eTable 4 [LVH population]).
The number of hospitalizations from any cause (paricalcitol, 15.7% vs placebo, 17.0%; P = .86) and from noncardiovascular causes (paricalcitol, 15.7% vs placebo, 11.6%; P = .44) did not differ between groups. In contrast, there were fewer hospitalizations for CVD events in the paricalcitol group (Table 5). Most CVD-related hospitalizations occurred near the study mid point (163 days; interquartile range, 104-241 days), and all events occurred in the LVH population. The most common cardiovascular event was congestive heart failure (paricalcitol, n = 0; placebo, n = 5).
Plasma levels of BNP increased in both groups; however, the increase was attenuated in the paricalcitol group compared with the placebo group in the ITT analysis (+21% vs +41%, respectively; P = .14) and in the LVH subgroup (+16% vs +50%, respectively; P = .04). After adjusting for changes in eGFR, changes in plasma levels of BNP remained similar (ITT, +23% vs +46%; P = .11; LVH subgroup, +19% vs +64%; P = .02). At baseline, 25.3% in the paricalcitol group and 26.1% in the placebo group had a serum cardiac troponin T level of at least 0.01 ng/mL. In the ITT analysis, participants with cardiac troponin T levels of at least 0.01 increased to 39.2% in the paricalcitol group and to 27.3% in the placebo group (P = .01). The difference between groups was reduced after controlling for changes in eGFR (P = .05). The results were similar in the LVH subgroup.
As expected, LVMI as measured by echocardiography differed from that obtained by CMR,36,37 but within each mode the baseline measures were similar (Table 2 and eTable 5). There was no evidence of change in LVMI by echocardiography (eTable 6 [ITT population] and eTable 7 [LVH population]). We then examined additional cardiac structural and functional changes potentially linked with reduced hospitalizations for congestive heart failure and noted a monotonic decline in left atrial volume index in the paricalcitol group but not in the placebo group (eTable 6 [ITT population], eTable 7 [LVH population], and the eFigure). Combining both treatment groups, changes in BNP levels significantly correlated with changes in left atrial volume index (ITT population, r = 0.24; P = .01 and LVH population, r = 0.32; P = .005).
A similar number of participants in each group reached the final visit at 48 weeks (paricalcitol, 76.5% vs placebo, 81.3%; P = .42). There was no difference in the overall incidence of adverse events between groups (paricalcitol, 80.0% vs placebo, 77.7%; P = .75). More adverse events were judged to be probably or possibly drug related in the paricalcitol group (20.9% vs 5.4%; P < .001) (eTable 8). These events were primarily due to hypercalcemia (paricalcitol, 22.6% vs placebo, 0.9%; P < .001). There was a slightly higher number of participants withdrawing from the study because of adverse events in the paricalcitol group (9.6% vs 4.5%; P = .19), again primarily because of hypercalcemia. A similar number of participants reported serious adverse events in each group (paricalcitol, 17.4% vs placebo, 17.9%; P = .93) (eTable 8), and none were judged to be related to the study drug.
Measures of renal damage (reduced eGFR and increased proteinuria) suggested worse renal disease at baseline in the paricalcitol group vs the placebo group (Table 1). Estimated GFR by creatinine-based methods showed a greater decline in the paricalcitol group (mean, −4.1 [SD, 0.9] mL/min/1.73 m2 with paricalcitol vs −0.1 [SD, 0.7] mL/min/1.73 m2 with placebo; P < .001). Estimated GFR by cystatin C–based methods also demonstrated a greater decline with paricalcitol (mean, −9.5 [SD, 2.7] mL/min/1.73 m2 with paricalcitol vs −3.8 [SD, 2.7] mL/min/1.73 m2 with placebo), but this difference was not statistically significant (P = .06). More participants in the paricalcitol group vs the placebo group initiated long-term dialysis (6 vs 1, respectively; P = .12), which tended to occur near the end of the study (median, 281 days; interquartile range, 151-301 days). The mean creatinine-based eGFR at study start among those initiating long-term dialysis was 23 (SD, 4) mL/min/1.73 m2. Blinded field investigators did not attribute long-term dialysis initiation to study drug in any participant, similar to a blinded adjudication of medical records at the study end. There were no deaths during the study period; however, 1 death occurred in each group more than 30 days after study completion.
The links among vitamin D deficiency, vitamin D therapy, and chronic disease are of considerable clinical and research interest. Although alterations in vitamin D metabolism have been associated with CVD and experimental data suggest that the vitamin D pathway is involved in modifying cardiac structure and function, corresponding clinical trial evidence is limited. Our goal was to rigorously examine whether an active vitamin D compound improves intermediate cardiac end points and thus inform larger outcome studies with vitamin D therapy. We found that paricalcitol at doses sufficient to suppress blood levels of iPTH did not reduce LVMI as measured by CMR over a 48-week period in participants with mild to moderate LVH. Additionally, paricalcitol did not modify certain echocardiographic measures of diastolic function. However, paricalcitol did reduce CVD hospitalizations and attenuate the increase in blood levels of BNP, particularly in those with prominent LVH at baseline.
We studied an active vitamin D compound rather than a nutritional vitamin D supplement because active vitamin D compounds (eg, calcitriol) have a greater than 100-fold affinity for the vitamin D receptor compared with the precursor 25-hydroxyvitamin D3.38 Furthermore, rats null for 1α-hydroxylase develop LVH in the setting of elevated levels of 25-hydroxyvitamin D3 and almost absent levels of calcitriol.39 In this rat model, LVH is attenuated following calcitriol administration, suggesting that the active hormone has the greatest effect on modification of this end point. Given our results with an active vitamin D analog, it is unlikely that nutritional vitamin D supplementation (ergocalciferol or cholecalciferol) of similar duration modifies LVH.
Our results differ from reports of vitamin D therapy in animal models of LVH.12,14,39,40 One possibility is that our sample size was too small. We used an adaptive design involving an interim analysis to ensure that our sample size provided more than 85% power to detect differences between groups.26 Confidence intervals for the primary end point were narrow and exclude a clinically meaningful change, suggesting that a larger sample size would have yielded similar results. In fact, although not clinically meaningful, the results in the LVH population were contrary to our original hypothesis, suggesting a longer duration would also not have yielded a similar result. Notably, our primary end point was measured with CMR, a highly sensitive technique requiring a sample size markedly lower than that required in echocardiography studies.41,42 We do not believe that either inadequate dose or poor adherence led to a neutral effect as the intervention resulted in a strong physiological response (marked reduction in blood levels of iPTH) in all treated participants. A significant determinant of LVH is blood pressure.43 Stringent guidelines to manage blood pressure were not implemented; however, blood pressure was well controlled throughout the study period, which may have attenuated a treatment effect. Overall, 48 weeks of paricalcitol treatment does not influence LVH in humans.
Vitamin D deficiency is associated with congestive heart failure,44 including in infants with vitamin D–deficient rickets.45 In animal studies, calcitriol or related analogs such as paricalcitol augment diastolic relaxation and reduce end-diastolic pressures, reduce cardiac mRNA expression and blood levels of natriuretic peptides, and reduce episodes of congestive heart failure.12,13,15,46 In prespecified analyses, paricalcitol attenuated the rise in BNP levels and was associated with fewer cardiovascular hospitalizations, primarily for congestive heart failure. This is consistent with previous observational studies suggesting that therapy with active vitamin D is associated with fewer CVD outcomes.8- 10,47,48 In post hoc analysis, paricalcitol also reduced left atrial volume, a measure linked to adverse cardiovascular events, particularly congestive heart failure.49- 52 Left atrial enlargement reflects chronically impaired diastolic relaxation and elevated end-diastolic pressures and is less prone to preload changes, which limit Doppler measures.53 Left atrial stretch and ventricular stiffness result in release of natriuretic peptides, and we observed a correlation between reduced left atrial size and reduced BNP levels. Given the high risk of cardiovascular events and heart failure in patients with CKD51,52,54 and the paucity of available lusitropic agents, future outcome studies with paricalcitol should target patients with CKD and a history of congestive heart failure with preserved systolic function.
Treatment with paricalcitol was well tolerated. The most common adverse effect was hypercalcemia, which is known to occur with active vitamin D treatment.6 More participants treated with paricalcitol initiated long-term hemodialysis; however, those randomized to the paricalcitol group had worse renal disease at baseline, an imbalance likely due to chance. Paricalcitol also increased serum creatinine and subsequently decreased creatinine-based measures of eGFR. Because paricalcitol and related agents inhibit renin expression,12,55,56 elevated serum creatinine may have represented a true reduction in GFR, as reported with angiotensin-converting enzyme inhibitor use.29 Alternatively, when GFR is examined by more direct measures including iothalamate and inulin clearance, active vitamin D compounds (including, most recently, paricalcitol) increase serum creatinine without altering renal function.17,26,29- 32 We therefore prospectively examined cystatin C–based measures of eGFR, which were more similar between groups. Although the precise mechanism is unclear, vitamin D receptor activation likely modifies protein (and thus creatinine) metabolism.17,26,30 We could not exclude the possibility that paricalcitol may have adversely affected renal function, although this has not been previously observed.57- 59 Nevertheless, clinicians should be aware that active vitamin D compounds may alter serum creatinine without significantly changing renal function.
Our study has certain limitations. Although previous cross-sectional studies have reported that LVMI increases as GFR declines,18,60,61 we did not find a significant increase in left ventricular mass in either group over 48 weeks despite a decline in kidney function. When patients with CKD and predominantly normal left ventricular mass were followed up for 2 years in randomized trials, little progression was observed.4,62 We targeted patients with CKD and preexisting LVH, expecting more rapid progression, but this was not reported. In experimental models, vitamin D therapy consistently attenuates progression of LVH,12,14,40 a finding we were unable to test. Importantly, despite the finding of no improvement in LVH (even with highly sensitive CMR), cardiovascular events and related biomarkers did differ.
In conclusion, in this 48-week study of patients with CKD and mild to moderate LVH, the active vitamin D compound paricalcitol did not regress left ventricular mass or improve certain Doppler measures of diastolic function. Paricalcitol appeared to be associated with fewer cardiovascular-related hospitalizations, an attenuated increase in blood levels of BNP, but a greater incidence of hypercalcemia; however, these results warrant further confirmation.
Corresponding Author: Ravi Thadhani, MD, MPH, Massachusetts General Hospital, 55 Fruit St, Bulfinch 127, Boston, MA 02114 (firstname.lastname@example.org).
Author Contributions: Dr Thadhani, Ms Wenger, and Dr Chang had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Thadhani, Applebaum, Pritchett, Chang, Tamez, Manning, Solomon.
Acquisition of data: Thadhani, Applebaum, Chang, Wenger, Manning, Solomon.
Analysis and interpretation of data: Thadhani, Applebaum, Pritchett, Chang, Wenger, Tamez, Bhan, Agarwal, Zoccali, Wanner, Lloyd-Jones, Cannata, Thompson, Andress, Zhang, Packham, Singh, Zehnder, Shah, Pachika, Manning, Solomon.
Drafting of the manuscript: Thadhani, Chang, Wenger, Tamez, Solomon.
Critical revision of the manuscript for important intellectual content: Thadhani, Applebaum, Pritchett, Chang, Wenger, Tamez, Bhan, Agarwal, Zoccali, Wanner, Lloyd-Jones, Cannata, Thompson, Andress, Zhang, Packham, Singh, Zehnder, Shah, Pachika, Manning, Solomon.
Statistical analysis: Thadhani, Pritchett, Chang, Wenger, Zhang.
Obtained funding: Thadhani.
Administrative, technical, or material support: Thadhani, Chang, Wenger.
Study supervision: Thadhani, Applebaum, Pritchett, Chang, Tamez, Bhan, Agarwal, Zoccali, Wanner, Lloyd-Jones, Cannata, Thompson, Andress, Zhang, Packham, Singh, Zehnder, Shah, Pachika, Manning, Solomon.
Conflict of Interest Disclosures: All authors have completed and submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Dr Thadhani reported receiving a coordinating grant from Abbott Laboratories to the Massachusetts General Hospital and speaker's fees and travel support from Abbott Laboratories. Drs Pritchett, Andress, and Zhang are employees of Abbott Laboratories and may own Abbott stock or options. Dr Agarwal reported receiving honoraria for serving on the speaker's bureau and steering committees of Abbott Laboratories. Dr Zoccali reported receiving consulting/honorarium fees and travel support from Abbott Laboratories, honoraria for lectures from Genzyme, Roche, Amgen, Abbott Laboratories, and Malesci and research grants from Pfizer, Roche, and Abbott Laboratories. Drs Wanner and Packham serve on an advisory board and have received travel support from Abbott Laboratories. Dr Wanner has received research support and speakers fees from Abbott Laboratories. Dr Zehnder reported receiving honoraria for serving on a steering committee and for lectures from Abbott Laboratories and receiving grants from Abbott Laboratories. Dr Manning received travel support for a CMR meeting from Abbott Laboratories. Dr Solomon is supported by a research grant from Abbott Laboratories to the Brigham and Women's Hospital. No other disclosures were reported.
PRIMO Investigators:Australia: Stephen McDonald, Grahame Elder, Michael Suranyi, Rosemary Masterson, Ian Fraser, David Packham; Czech Republic: Vladimir Tesar, Ondrej Viklicky; Germany: Frank Dellanna, Stefan Degenhardt, Kai Hahn, Martin Nitschke, Christoph Wanner; Italy: Francesco Pugliese and Vincenzo Panichi; Poland: Michal Nowicki; Puerto Rico: Rafael Burgos-Calderon, Jose Cangiano, Angel Comulada-Rivera, Maria Rios, Jose Vazquez-Tanus; Romania: Mihai Voiculescu; Russia: Valentin Ermolenko, Natalia Tamilina, Galina Volgina; Spain: Angel Luis Martin de Francisco, Luis Miguel Ruilope, Maria Jose Soler; Taiwan: Yung-Ming Chen, Hung-Hsiang Liou, Shin-Hung Tsai, Chun-Chen Yu; United Kingdom: Daniel Zehnder; United States: Lionel Abbott, Hanna Abboud, Ali Assefi, Mario Belledonne, Marializa Bernardo, Geoffrey Block, David Burstein, Gabriel Contreras, George Fadda, Seth Goldberg, Mandeep Grewal, Terry Hammond, Nelson Kopyt, Richard Lund, Peter McCullough, William Moore, Joseph Muhlestein, Kaldun Nossuli, Anita Patel, German Ramirez, Amit Sharma, Bhupinder Singh, Nicole Stankus, Samir Tuma, Thomas Wiegmann.
Funding/Support: This study was funded by an investigator-initiated grant from Abbott Laboratories.
Role of the Sponsor: This was an investigator-initiated study that was funded by industry (Abbott Laboratories) and led by a steering committee composed of academic members, 2 sponsor members (nonvoting), and statisticians. The steering committee oversaw the design, conduct, and all analyses. Data were collected by the sponsor and shared with the principal investigator at the study mid point and termination. Final assignment code was sent to the principal investigator after the database lock. The principal investigator, with statisticians at Massachusetts General Hospital, independently performed all prespecified and exploratory analyses and resolved any discrepancies with industry statisticians. The sponsor provided the active medication and matching placebo. The principal investigator wrote the first draft of the manuscript and the steering committee was responsible for data interpretation and manuscript completion. The sponsor reviewed the manuscript, but decisions about the final manuscript were made by the principal investigator and academic members of the steering committee. All authors vouch for the integrity of the data.
Independent Statistical Analysis: The principal investigator, Dr Thadhani, with Ms Wenger and Dr Chang, statisticians at Massachusetts General Hospital, independently performed all prespecified and exploratory analyses and resolved any discrepancies with industry statisticians. The results of the independent analyses are the results that are reported in the article.
Additional Contributions: We thank Anne Kavanaugh, Stacey Orzel, MBA, MT(ASCP), and Keith Boardway, Abbott Laboratories, for their organizational support of the PRIMO trial, and we thank all participants. We also thank members of the data and safety monitoring committee: Theodore Steinman, MD (chair), Beth Israel Deaconess Medical Center; Mark Sarnak, MD, MS, Tufts Medical Center; and Sheryl F. Kelsey, PhD, University of Pittsburgh. All members of the data and safety monitoring committee received compensation for their role in the study. Finally, we thank Dalane Kitzman, MD, Wake Forest Baptist Medical Center, for his critical review of the manuscript.