This was a single center, parallel group, randomized, controlled, double–blinded prospective trial. Enrollment and all assessment were conducted at a specialty heart failure clinic by cardiologists, in Western Montana between fall of 2012 and spring of 2015. Clinic patients with New York Heart Association (NYHA) Class II or III heart failure were eligible for inclusion into the study if they were 18 years or older, had been on stable and guideline directed medical therapy (GDMT) for greater than 3 months, and had a 25(OH)D level of ≤32 ng/ml. Patients who were pregnant, breastfeeding, or with any clinically unstable medical disorder were excluded. All data, including QOL, laboratory values, and cardiopulmonary function were collected from the clinic by research coordinators and nursing staff. Patients currently taking ≤1000 IU vitamin D daily could enroll in the trial. Randomization was computerized in a permuted block size of 4 and occurred off-site at a university pharmacy department. Vitamin D3 and placebo bottles were given sequentially numbered identical containers and administered serially. The trialists, patients, and caregivers were blinded to the randomization until all data were collected and collated. Patients were excluded from the trial if they presented with any of the following conditions: hypercalcemia, nephrolithiasis, sarcoidosis, acute decompensation of heart failure, and/or any clinically unstable systemic medical disorder, including severe pulmonary disease. Vitamin D3 has been shown to be safe long-term at doses of 10,000 IU per day [22]. Vitamin D dose at a given amount seems to result in a plateau in 25 (OH)D at or around 1 month, so we estimated that 6 months of dosing would be sufficient to have an effect on hormonal and cardiopulmonary status [22]. All patients provided written, informed consent and the trial was approved by the Providence Saint Patrick Hospital Institutional Review Board. Patient variables were divided into gender differences for men and women. Vitamin D3 at 10,000 IU per day (Biotech Pharmacal, Fayetteville, AR) or identical placebo (dextrose) was given in parallel design. Patients remained on all current medications as prescribed, and changes in any baseline medication dosages or forms were recorded. All patients enrolled in the study were receiving best tolerated and GDMT, which included Angiotensin Converting Enzyme Inhibitor (ACE), Angiotensin Receptor Blocker (ARB), beta blockers (Table 1), and aldosterone antagonists. They were also receiving statins and loop and/or non-loop diuretics as indicated. All medications had been stable for at least 3 months before enrollment.

Patients were evaluated at baseline and at 6 months for all clinical and laboratory data, and they had complete medical assessments, QOL, and CPX tests at both visits. At 3 months, both 25(OH)D and serum calcium levels were drawn. Due to the high number of patients screened taking vitamin D at 1000 IU daily that remained deficient in 25(OH)D, we modified the protocol to include patients who were taking up to 2000 IU per day of vitamin D prior to enrollment that had baseline levels of 25(OH)D ≤ 32 ng/ml. If 25-hydroxyvitamin D levels exceeded 150 ng/ml or calcium level exceeded 10.5 mg/dl at 3 months, vitamin D3 dose was decreased to 5000 IU per day. If any adverse effects were observed by the patient or cardiologist, the patient was withdrawn from the trial.

BNP, PTH, hsCRP, and 25(OH)D, hsCRP, complete metabolic panels and complete blood counts were all measured using commercially available kits.

The Kansas City Cardiomyopathy (KCCQ), a validated tool that is both sensitive and specific for assessment of change in health-related QOL, was interviewer administered at baseline and at 6 months of treatment [23]. Briefly, the KCCQ has domains of physical function, symptom frequency and severity, social function and QOL; composite scores are derived from these domains. Scores are computed for a range of 0–100 for each domain, in which higher scores reflect better health status.

All subjects, following 2 min of rest, completed an initial 1-min, constant-rate stair step. After the initial workload, speed and grade were incrementally increased at 1-min intervals to peak attained exertion. Ventilatory expired gas analysis was collected at rest and throughout exercise using a miniaturized metabolic cart (Shape HF cardiopulmonary testing system, St. Paul, MN). The Shape analyzer was self-calibrated daily using an auto-cal performance test. Heart rhythm was assessed on a 4-lead electrocardiogram, and arterial O₂ saturation was monitored with finger pulse-oximetry (Envitec D-23966, Wismar, Germany).

Minute ventilation, the end-tidal partial pressures of O₂ (PETO₂) and CO₂ (PETCO₂), oxygen consumption (VO₂), and carbon dioxide production (VCO₂) was acquired breath by breath and averaged over 30-s intervals every minute until exhaustion (progressive exercise to peak oxygen consumption, VO₂peak). The data points between the averaged value during the first 30 s of the initial work load and at peak exercise was used to calculate the ventilator efficiency (V_E) to VCO₂ slope and changes in partial pressure end-tidal CO₂. In addition, the chronotropic response index (CRI) and O₂ uptake efficiency slopes were determined during exercise as previously described [24]. Greater values of oxygen uptake efficiency slope indicate a higher ventilatory efficiency, and a larger CRI slope is an indicator of improved heart rate reserve.

We set a threshold of 90% power to detect a between-group difference in BNP (based on 74 ± 25 pg/ml between-group differences seen in the literature) to determine a need for at least 6 patients in each [25]. The differences between the means of the two data sets were assessed by 2-sided independent samples t test. For data not normally distributed, the analysis was performed by means of Wilcoxon signed-rank test. A p-value of 0.05 or less was considered statistically significant. For the study to have >90% power to detect a between-group difference for our primary outcome measure of BNP of 74 pg/ml with SD value of 25 mg/ml found in the literature, the number of patients in the participating group must be 6 or more [25]. A two-tailed z-value related to α = 0.05and a two-tailed z-value related to β = 0.1 were used for the estimate together with the expectation that vitamin D₃ supplementation would lead to a reduction in BNP.

We increased our goal recruitment size to 20 per group to allow for gender differences to be assessed. Data presented was analyzed using SPSS version 23 (IBM® Corporation, Armonk, New York), and expressed as the mean ± standard deviation (SD). Normality was assessed for each variable by the Shapiro Wilk test. Values between the two treatment groups were assessed using Student’s t-test or Mann-Whitney Test for normally distributed and non-normally distributed data, respectively. Univariate analysis of covariance was conducted to adjust for variance in baseline vitamin D status. Significance was set a priori at α = 0.05. Other continuous variables (expressed as means, SD) and categorical variables (expressed as frequencies) were collected during the study to characterize the group. Subgroup analysis was conducted for gender differences for all outcome variables. Differences in race could not be conducted due to the patient demographic of the region was primarily white. Data were analyzed per protocol analysis. All data is available in the attached data repository (Additional file 1). The CONSORT checklist is attached in Additional file 2.

Our findings demonstrate that therapeutic vitamin D3 treatment is significantly better than placebo for BNP measures when adjusted for baseline vitamin D in patients with reduced ejection fraction heart failure. These findings are noteworthy because BNP has been shown to be strongly associated with recurrent hospitalization and mortality in this patient population. BNP was the most powerful independent marker of outcome in heart failure patients in the Valsartan Heart Failure study [26]. A limitation to our current finding is that BNP variability between patients was large, and our enrollment criteria did not specify a range for BNP.

Prior research is conflicting related to vitamin D and its influences on BNP. A study by Jiang et al. failed to show a change in BNP after vitamin D supplementation, but a more recent supplement trial found that vitamin D dramatically improved BNP levels using 200,000 IU vitamin D per week [27, 28]. However, a limitation of this trial was a lack of randomization [28]. In the meta-analysis failing to find a change in BNP, biological flaws were present, as the average 25(OH)D level was not taken into consideration. Our current study is the highest dose per day of any study to our knowledge. Mechanisms of vitamin D3 actions related to BNP have not been fully described, but the vitamin D receptor likely plays a role in regulating the BNP gene promoter [29].

In our study, QOL scores improved significantly in most domains, and this remained robust after adjustment. Our study is the first to provide vitamin D3 at a dose shown to replenish vitamin D stores, and also measure QOL. Prior research demonstrates that improvement in QOL predicts event-free survival in heart failure patients [30]. Comparatively, current treatments don’t approach the effect size for QOL that our study has shown. For example, the Angiotensin-Neprilysin Inhibition (LCZ696) versus Enalapril trial [31] demonstrated a 2.99-point worsening in KCCQ clinical summary with the use of LCZ696, which was statistically better than conventional therapy with Enalapril (worsened KCCQ by 4.63 points). In contrast, our KCCQ clinical summary had an 8 point average improvement. The implications of this could be large if replicated.

Heart failure results in diminished QOL, in part due to frailty; vitamin D3 may help by preserving and improving lean body mass and strength [32‐34]. A recent trial in heart failure patients failed to show improvements in physical function and QOL in heart failure patients. However, serum 25(OH)D never surpassed 32 ng/ml on average [34]. Two pulse doses of vitamin D2 is likely insufficient to overcome a deficiency in heart failure [22]. In contrast, when 200,000 IU vitamin D3 per week was given to heart failure patients, the 6-min walk test was significantly improved from baseline [28].

While most CPX remained unchanged in our study, we did find that vitamin D seemed to improve CRI. However, adjustment for baseline vitamin D attenuated this. Our study may not have been long enough in duration to find differences in cardiopulmonary exercise testing as a whole. Our finding that chronotropic response may be improved by vitamin D deserves further attention since this index is inversely associated with mortality in heart failure patients [35].

Treatment of heart failure with Angiotensin Converting Enzyme Inhibitors and Angiotensin Receptor Blockers has been documented to improve cardiopulmonary function and exercise capacity [36]. The mechanism of action of these medications is inhibition of the Renin-Angiotensin-Aldosterone System (RAAS), in particular, angiotensin II. Vitamin D analogs block renin upregulation and ANGII accumulation in animals [37]. Vitamin D3 thus may also improve CPX with synergistic effects of established treatments. Further research with vitamin D3 and vitamin D cofactors with longer duration may further elucidate the effects of vitamin D3 on CPX.

PTH was reduced by 20 pg/ml in our vitamin D3 group; other studies have had similar findings in heart failure to a lesser magnitude [38]. Adjustment for baseline 25(OHD status in our study did attenuate this significance; however, PTH and 25(OH)D are quite inter-related, so this does not appear to minimize our treatment effect. However, our baseline PTH levels were higher than most prior trials [39]. Vitamin D3, therefore, may be more beneficial in patients with markedly elevated PTH. Our groups were not matched at baseline in regard to PTH. This is possibly true because our enrollment criteria did not include PTH parameters for exclusion or inclusion. Serum calcium also didn’t vary between the groups, but we did not allow patients with hypercalcemia into the study for safety concerns. This may explain why the baseline calcium levels looked similar between groups. However, with no change for PTH in the placebo group after 6 months, our reduction of PTH in the vitamin D group does appear to be related to vitamin D repletion. National Health and Nutrition Examination Survey (NHANES) data have demonstrated that PTH is associated with CRP [40]. We found that PTH also was associated with hsCRP, and selecting for patients with hsCRP greater than 1 further strengthened the effect of vitamin D3 on PTH.

Vitamin D directly affects the parathyroid gland, and deficiency of vitamin D results in secondary hyperparathyroidism. Hyperparathyroidism is strongly associated with HF outcomes, with elevated PTH directly related to both hospitalizations and mortality [41, 42]. Additionally, PTH levels in heart failure patients are elevated compared to healthy adults [43], suggesting that PTH dysregulation is a component in the complex pathology of heart failure. Optimization of PTH does not occur until 25(OH)D surpasses 40 ng/ ml [44].

Our results indicate that a vitamin D3 supplement of 10,000 IU per day effectively corrects vitamin D deficiency in heart failure without evidence of adverse effects. Our dose-response is like other studies that have used 10,000 IU of vitamin D3 daily, and a steady state of serum 25(OH)D was reached by month 3 [22]. Further research is needed to evaluate if vitamin D dosing guided by 25(OH)D levels would be equally effective or superior, or if additional nutrient cofactors would further improve patients’ response [15, 16].

Vitamin D deficiency increases the risk of all-cause mortality, incident cardiovascular disease, and sudden cardiac death in the general population [45]. Patients with heart failure and vitamin D deficiency have an almost 3-fold risk of death and 5-fold risk of sudden cardiac death compared with those who are vitamin D replete [6]. Vitamin D3 affects genetic expression related to all cardiovascular cell types; it thereby helps regulate cellular differentiation, apoptosis, hormone production, and oxidative stress [46, 47]. The efficacy of vitamin D3 is related to its ability to provide enough substrate, so the translation of vitamin D studies should involve serial measurements of vitamin D status. Vitamin D repletion is an appealing target therapy due to its numerous physiological benefits for humans [15, 16].

Vitamin D activated to 1,25(OH)2D calcitriol is considered a pleiotropic hormone, and vitamin D receptors occupy 2776 genome positions. Vitamin D directly affects the expression of many genes, and a recent study showed that improvement in vitamin D status after supplementation results in correction of genetic expression [47]. In this study, there was a 1.5 fold change in the expression of 291 genes with vitamin D supplementation in vitamin D deficient healthy adults; 66 of those genes had dramatic changes in expression [47]. Tripartite Motive Containing 27 gene (TRIM27) and Early Growth Response (EGR-1) gene among many others were down-regulated; TRIM27 gene is associated with oxidative stress, immunoglobin E mediated immunity among others, and EGR-1 is linked to vascular health.

A recent meta-analysis demonstrated that supplemental vitamin D3 increases circulating 1,25(OH)2D levels significantly, and its actions are mediated through vitamin D receptors [48]. This is an important finding because circulating 1,25(OH)2D levels are associated with better outcomes in end stage HF [49]. Cardiomyocytes, immune cells, endocrine, and vascular smooth muscle cells have vitamin D receptors and outcomes in HF are associated with circulating calcitriol levels [50‐52]. These receptors throughout the body also allow for onsite conversion of1,25(OH)2D calcitriol and calcium-dependent cellular processes, and high concentrations of calcitriol in these cell types are likely necessary to trigger gene expression [51]. These key effects of vitamin D on post-translation and non-genomic actions, such as transmembrane calcium transport, provide insight into the beneficial mechanistic action of vitamin D on the cardiovascular system [53].

Vitamin D deficiency, as defined by the Institute of Medicine is <20 ng/ml for 25(OH)D levels, while insufficiency is generally considered to be 20–30 ng/ml [54]. New research suggests, however, that a circulating level of 20 ng/ml presents a 2-fold risk of secondary hyperparathyroidism over levels greater than 30 ng/ml [55]. Elevated PTH is associated with poor outcome in HF patients [56]; our findings demonstrate that vitamin D3 supplementation can markedly improve PTH concentrations at higher circulating levels of 25(OH)D, further substantiating the existing data showing that higher circulating 25(OH)D concentrations help mitigate hyperparathyroidism. Vitamin D status is closely related to PTH status and 25(OH)D appears to suppress PTH across at all levels of 25(OH)D status [57]. Additionally, it has become clear that the extraskeletal benefits of vitamin D occur at higher circulating blood levels of 25(OH)D; for example, respiratory infection risks are reduced for those individuals with 25(OH)D in the range of 50 ng/ml or more [54]. For these reasons, we chose a higher cutoff criterion for enrollment than Institute of Medicine guidelines.

A question remains as to whether heart failure patients require vitamin D3 treatment only in the cases of 25(OH)D < 20–30 ng/ml, or if all patients with heart failure would benefit from supplementation despite normal circulating 25(OH)D levels. As of yet, evaluation of vitamin D status is not routinely performed in heart failure. However, patients with chronic kidney disease patients do have routine monitoring in their guidelines (56?). The case is made for class 1A indication for beta-blocker use in patients with heart failure with reduced ejection fraction even when the heart rate is not elevated (26–27). In that situation, heart rate is a surrogate and possibly imperfect marker for upregulation of sympathetic nervous system in heart failure. In our small study, we found that monitoring 25(OH)D and subsequent repletion may have clinical merit with regard to improved QOL and neurohormonal indices. Determining if supplementation of vitamin D in all CHF patients without regard to baseline status will require further research.

Vitamin D3 in our study did not appear to have a strong impact on hsCRP as a whole, although men had an almost 2 mg/L reduction at 6 months. A prior study showed that vitamin D3 given at 2000 IU per day for nine months in heart failure improved cytokine profiles [38], and both vitamin D and inflammation are related in heart failure [55]. Vitamin D impacts several inflammatory pathways and acts to inhibit the activity of nuclear factor B activity, and by reducing interleukin 6, interleukin 12, interferon C, and tumor necrosis factor alpha production [38]. A recent review of vitamin D and inflammation suggests that vitamin D possibly modestly improves inflammation overall and that it may have more anti-inflammatory effect with greater levels of inflammation in diseases such as end stage renal disease, lupus, and tuberculosis, but also has effects on inflammation in acute coronary syndrome [58]. Additionally, inflammation related to vitamin D defiency is associated with poor HF outcomes [59]. Both of our treatment arms were equally matched in regard to statin usage at baseline and had been on stable doses for at least 3 months, so this likely did not impact hsCRP in this study. A recent review of inflammation and vitamin D found that 6 out of 7 randomized clinical trials supplementing vitamin D found reductions in inflammatory markers [60]. Our research was exploratory in nature and was not powered to detect differences in hsCRP. Future studies with marked elevation in inflammation levels at baseline may be useful to further understand the role of vitamin D3 for inflammation in HF.