Vitamin D supplementation: much ado about nothing

Abstract

The scientific interest and the number of papers dealing with vitamin D supplementation has greatly grown in the last decades. Unfortunately, expert consensus on many clinical aspects of this topic is still lacking. In addition, data coming from recent clinical trials and meta-analyses seem to strongly put into doubt the real benefit of vitamin D supplementation, on both skeletal and extra-skeletal outcomes. This is further confusing since they seem to completely contradict the considerable body of evidence provided from previous epidemiological studies. This paper aims to analyze these new data in order to shed light onto the debated issues.

摘要

在过去的几十年中, 有关补充维生素D的科学兴趣和论文数量大大增加。但不幸的是, 专家对这一主题的许多临床方面仍然缺乏共识。此外, 来自最新临床试验和荟萃分析的数据似乎强烈质疑补充维生素D对骨骼和骨骼外结局的真正益处。由于它们似乎完全与先前的流行病学研究提供的大量证据相矛盾, 因此更加令人困惑。本文旨在对这些新数据进行分析, 以阐明存在争议的问题。

The Chinese abstracts are translated by Prof. Dr. Xiangyan Ruan and her team: Beijing Obstetrics and Gynecology Hospital, Capital Medical University, Beijing 100026, China.

Keywords:

• Vitamin D
• osteoporosis
• meta-analysis
• supplementation

Introduction

In the last 10–15 years, the renewed interest on the role of Vitamin D across many physiological and pathological pathways has lead to a considerable number of publications on the clinical aspects of supplementation [1]. Yet, such abundant scientific production has not translated into major breakthroughs in clinical practice or prevention. In fact, findings reported in literature appear somewhat contrasting and inconclusive [2,3], with many aspects of supplementation (starting from the very definition of serum 25OH-vitamin D threshold values that define vitamin D deficiency) remaining subject to debate [3,4].

In the setting of bone disease, for example, although it is universally recognized that low and very low vitamin D levels (i.e. below 20 and 10 ng/mL respectively) have deleterious effects of on skeletal health [4], experts are still divided on the efficacy of supplementation, with some strongly supporting its effects on skeletal and extra-skeletal health, and others confining it to the treatment of rare bone diseases [5,6]. Likewise, for the effects of Vitamin D in preventing musculoskeletal events (i.e. falls and fractures) [7]. In the broader setting of chronic/severe diseases, many observational studies [8,9] have evidenced the association between low serum Vitamin D levels and major diseases (i.e. cancer [10,11] or diabetes [11]), however large studies on the effectiveness of Vitamin D supplementation failed to provide final evidence to justify supplementation.

Amid such differing positions, it is thus necessary to critically appraise the nuances in the evidence available. Unfortunately, comparability of findings and appreciation of their clinical relevance is hindered by the lack of methodological consistency across supplementation studies. Findings need to be evaluated in light of basic contextual factors, such as baseline characteristics of the enrolled population, vitamin D dosages administered, duration of the follow-up and also the reasonably expectable latency of the protective effect (i.e. when dealing with cancer-related outcomes).

To date, it is estimated that 40% of the European population has 25OHD serum levels below 20 ng/mL [13] that vitamin D deficiency among populations of the southern Mediterranean regions (Italy, Spain, France, Cyprus, Turkey and Israel) is widespread. This remains true across all age groups and even when using a reference threshold of 10 ng/mL [14]. Moreover, in over half of these subjects, 25OHD levels were <10 ng/mL [14].

In this paper, we will attempt to provide some guidance on the appropriate use of Vitamin D in terms of prevention potential and clinical outcomes, by reviewing the most recent studies, basing our discussion on few basic assumptions.

Recent RCTs on vitamin D

The vitamin D assessment (ViDA) study

Performed between 2011-2015, the RCT ViDA Study [7] recruited 5,110 healthy adults to assess the effect of vitamin D supplementation on the incidence of cardiovascular disease (CVD) (deaths and hospitalizations), and as secondary outcomes the prevention of acute respiratory infection, the incidence of falls and non-vertebral fractures: 2,552 received placebo and 2,558 high dose vitamin D (2,00,000 IU of vitamin D in the first month continuing with 1,00,000 IU per month) and were followed up for a median period of 2.5–4.2 years. The rationale of the study was drawn based on several epidemiological studies that reported the association between vitamin D deficiency and increased risk of numerous types of cancers, autoimmune diseases (including both type 1 and type 2 diabetes, rheumatoid arthritis, Crohn’s disease, and multiple sclerosis), infectious diseases and CVD [15]. Results showed no evidence of risk reduction for CVD [15], falls or fractures [17]; given the large number of individuals involved in the trial and its duration, the findings were considered reliable. However, several aspects deserve some pondering [1]:

• Only 34% of these healthy volunteers were over 70 years of age (typically patients at a higher risk for falls, fractures and CVD).

• Only 6% of enrolled population had a history of falls and therefore at higher risk of new falls.

• Roughly 46% of these healthy volunteers reported a history of fracture but only 2% of the vitamin D-supplemented group and 1% of the placebo group were affected by osteoporosis at baseline.

Given the low prevalence of osteoporosis and in the absence of further details, the data regarding the fracture was perhaps more confusing than informative. Most of the reported fractures were arguably of traumatic nature, unrelated to bone health status and consequently independent of vitamin D.

Later publications by the same Authors flagged some limitations regarding the statistical power potentially affecting their studies. Khaw et al. [17] acknowledged that their study had low statistical power, particularly in participants with vitamin D deficiency. Similarly, Scragg et al. [16] reported that the CV event rate was lower than expected and that the study had too little power to detect benefits for the subgroup with vitamin D deficiency alone. Unfortunately, the subgroup of vitamin D deficient patients group in which to investigate the efficacy of supplementation was crucial. Such considerations eventually pointed to a key flaw in the study design: the individuals’ baseline vitamin D status. A negligible number of subjects (2%, corresponding to 91 subjects over 5,110 participants) had serum levels of 25OHD below 10 ng/mL–a threshold that is generally accepted as deficiency. Only 25-30% of the enrolled individuals had baseline levels <20 ng/mL (the threshold below which, the high dose is considered appropriate by most national and international guidelines [4]).

Upon designing the ViDA trial, the Authors aimed to increase 25OHD serum concentrations above 30 ng/mL, which at the time of the design of the trial was considered the optimal threshold from a general health perspective [17]. Though, one third of the subjects enrolled had already achieved that target value at baseline and would have not needed supplementation . In short, the study analyzed the effects of vitamin D supplementation in a sample of healthy subjects who were non-osteoporotic, at very low risk for falls and already vitamin D-replete for the most part. It would have indeed been surprising to find a significant result! Indeed, a recent sub-analysis of the ViDA trial focused on bone mineral density (BMD) [18] and a reexamination of the data from a similar trial in Aberdeen [19], showed that BMD loss was reduced by vitamin D supplementation, in both the adults and elderly when considering the deficient subjects alone. Moreover, the final conclusion of a pre-established analysis of a subsample of participants in the ViDA focused on the effects of supplementation on central blood pressure parameters [20] was that high-dose monthly vitamin D supplementation did actually lower central blood pressure parameters among adults with vitamin D deficiency but not in the total sample.

As for the 839 subjects with serum 25OHD levels over 30 ng/mL received monthly boluses (1,00,000 IU each) of cholecalciferol for 2-4 years, no relevant adverse events have been reported.

The vitamin D and type 2 diabetes (D2d) study

More recently, Pittas et al. [13] reported the results of a large RCT on the effects of vitamin D supplementation in the prevention of type 2 diabetes (T2DM). The trial involved 2,423 individuals (not selected for vitamin D deficiency) at high risk for developing T2DM and concluded that, in this population, vitamin D3 supplementation, at a dose of 4,000 IU daily, did not result in a significantly lower risk of developing T2DM than placebo. However, in a per-protocol analysis that excluded a small proportion (5%) of participants of the placebo arm (due to out-of-study administration of vitamin D above the pre-specified –and arguably too high— threshold allowed of 1,000 IU daily), a 16% reduction in the risk of developing diabetes was found, with a hazard ratio of 0.84 (95% CI 0.71–1.00).

As in the case of ViDA the authors had based their rational on the observational studies available at the time they laid out their study design and which suggested a biologically plausible association between low serum 25OHD levels and the risk of T2DM. Unfortunately, 78.4% of subjects included in the study by Pittas et al. had normal or even ideal vitamin D serum levels (≥20 ng/mL), which thus could have actually diluted the evidence for any benefit of vitamin D supplementation. Interestingly, in the post-hoc analysis (not pre-specified) limited to the small proportion of patients with true vitamin D deficiency at baseline (25OHD less than 12 ng/mL; 103 participants) the risk was considerably reduced by 62%. Therefore, the D2d study presents several flaws in terms of both study design (i.e. including patients with normal vitamin D serum levels) and study conduction (i.e. allowing out-of-study administration of vitamin D) which need to be taken into due consideration.

In our opinion, we should start to question cost to benefit ratio of these large trials on cholecalciferol administration involving individuals with normal 25OHD serum levels. These studies are expensive, and their conclusions cannot be generalized and ultimately applied to the populations of clinical interest, such as those at high risk of vitamin D deficiency.

VITamin D and OmegA-3 TriaL (VITAL)

Vitamin D deficiency has been shown to be associated to increased cancer incidence and mortality, in numerous studies (both observational and on animal models) [21,22], yet randomized clinical trials (RCTs) on the effectiveness of vitamin D supplementation on cancer prevention have not provided any encouraging findings to support these findings. Among these studies is the recent VITAL (VITamin D and OmegA-3 TriaL) [23], sponsored by National Institute of Health (NIH). The trial enrolled a total of 25,871 healthy subjects (males over 50 and females over 55 year of age) and randomly assigned to receive vitamin D 2,000 UI/day or placebo. After 5.3 years of follow up, vitamin D supplementation failed the primary end point (decreased incidence of invasive cancer). The study, which is the largest so far, had an adequate follow up, tested a high dose of vitamin D with an 82%, adherence to the trial regimen In summary, the trial had adequate power to detect a potential vitamin D effect. However, the authors did not take in account that baseline serum 25OHD (among 15,757 participants who provided samples) was 30.8 ± 10.0 ng/mL. 25OHD was < 20 ng/mL only in 2,001 subjects and ≥20 ng/mL in 13,786 out of 15,787. In other words, about 87% of the treated subjects had normal 25OHD levels. No data were disclosed on the number of subjects with severe 25OHD deficiency (i.e. lower than 10 ng/mL). The Authors reported that most of the supplemented subjects attained serum 25OHD levels of 40 ng/mL or above after the first year of follow up. In our opinion, these results simply show that, in healthy subjects, values of 25OHD above 20 ng/mL do not affect health in a significant way. Lastly, the aim of the Authors was to obtain new information through the use of different vitamin D regimens [10] and not by selection of the patients that were likely to benefit from supplementation. Nevertheless, post-hoc analyses showed interesting results. Excluding 1 and 2 years of follow-up, the rate of death from cancer was significantly lower with vitamin D than with placebo (hazard ratio, 0.79 [95% CI, 0.63 to 0.99], and hazard ratio, 0.75 [95% CI, 0.59 to 0.96], respectively). Furthermore, another post-hoc analysis suggested an influence of BMI on the effect of vitamin D on cancer, with a protective effect limited to the <25 BMI subgroup hazard ratio of 0.76 [95% CI 0.63–0.90] [23].

Meta-analyses on vitamin d

Meta-analyses on vitamin D and musculoskeletal outcomes

Bolland and colleagues recently published a systematic review and meta-analysis on RCT testing vitamin D supplementation on musculoskeletal outcomes [6], the Authors showed that vitamin D supplementation does not prevent fractures or falls, nor have clinically meaningful effects on bone mineral density. Accordingly, they concluded that there is little justification in using vitamin D supplements to maintain or improve musculoskeletal health but can may be useful in cases of rare bone conditions, such as rickets and osteomalacia, associated with prolonged states of 25OHD deficiency (<10 ng/mL) [6].

We may however comment on several aspects of this meta-analysis such as:

• The choice of the included and excluded trials for the analysis.

• The inclusion of studies adopting different forms of vitamin D.

• The inclusion of different administration schedules (frequent doses, or boluses), with doses ranging from insufficient to extreme.

• Unsuitable observation periods (too limited to detect a recognizable effect on long-term events).

• The involvement of subjects at low fracture risk.

• The lack of consideration of drug adherence [1].

The most relevant limitation is that the vast majority of the studies considered had enrolled individuals who had baseline vitamin D serum levels >20 ng/mL. Moreover, the small proportion of patients with vitamin D deficiency at baseline had not even attained sufficient levels throughout the studies (>30 ng/mL) thus being unlikely to experience any benefit from the supplementation [1]. Interestingly, few days after his meta-analysis, Bolland himself published two papers in which he underlined that most of the available studies on the efficacy of vitamin D supplementation include a large proportion of vitamin D-replete participants [2,3], contributing to the underestimation of benefits of vitamin D supplementation in deficient subjects. We agree with Bolland in defining such studies ‘research waste’ and share the concern that meta-analyses stemming from these studies may be tainted to a similar extent. There can be hardly any benefit in treating healthy subjects who are not vitamin D deficient. Unfortunately, the strong conclusions of Bolland’s meta-analysis might induce drop on the prescription vitamin D supplementation indiscriminately to all patients. This concern is especially strong for the elderly frail subjects and for the patients with osteoporosis who are receiving treatment with bone acting agents, in whom vitamin D deficiency has been shown to be a strong determinant in their anti-fracture efficacy [24–26].

Meta-analyses on Vitamin D and cancer

In 2017 Rejnmark et al. [27] published a systematic review of meta-analyses evaluating the harmful effects of low vitamin D on non-skeletal outcomes. Of the 54 meta-analyses found, 5 were specifically focused on cancer and had considered a total of 19 RTCs, none of which had the effect on cancer as a primary outcome, and only one of which reported a benefit for the primary outcome (which was not cancer). Among the 5 meta-analyses found all reported null findings. In reference to the overall RCTs analyzed in these meta-analyses, these apparently failed to prove the clinical benefit of vitamin D supplementation because of several limitations, such as: insufficient statistical power, short duration, low doses of vitamin D, small sample sizes and different dosing schedules. Again, the main weak points emerging are the lack of baseline and/or end of the study levels of 25OHD, or the selection of patients with 25OHD levels >10 ng/mL. In fact, none (except one) of the 19 RCTs which addressed the relationship between supplementation and cancer had set 25OHD levels below 20 ng/mL as inclusion criterion. Mention to baseline 25OHD values found only in 11 out of 19 RCTs reported values above 20 ng/mL and the intervention with vitamin D supplementation increased plasma 25OHD levels by more than 50% in only one-third of the studies [27].

In this same study one recent meta-analysis that considered 10 RCTs with a 3-10 years of follow up (and including the data from the ViDA and VITAL studies that have been published after the meta-analyses by Bolland and Rejnmark) concluded that supplementation did not significantly reduce total cancer incidence [28]. However, vitamin D supplementation did show significant results on mortality outcomes, with an RR of 0.87 [95% CI 0.79–0.96, p = .02 I² = 0%] – despite only one out of the10 RCTs having mean baseline 25OHD <20 ng/mL (i.e. 15.2 ng/mL) [28]. Interestingly, this result was largely attributable to interventions with daily dosing and not to infrequent bolus dosing.

Final remarks

Taken together, the considerations above evidence the scarcity of subjects with vitamin D deficiency unlike the general population in which a large portion is Vitamin D deficient. Therefore results from those studies (originating from a prevalently vitamin D-replete population) cannot be easily generalizable [13].

On the other, there appears to be sufficient evidence to confirm beyond any reasonable doubt the benefits of calcium and vitamin D supplementation in the pharmacological treatment of osteoporosis. In all RCTs aimed at establishing the protective effects of osteoporotic medications, [29–33] calcium and vitamin D were administered in addition to the study drugs in both treatment and placebo. In addition several studies have established beyond any doubt the link between vitamin D-and increased risk of fracture [9,34] and reduced efficacy of several drugs [26,35]. Such evidence is in agreement with findings on data from administrative databases of five Italian Local Health Units (representing a population of 3.3 million health-assisted inhabitants [36]), which confirmed a reduction in risk of both re-fracture and all-cause mortality particularly when treatment for osteoporosis was administered concurrently to calcium and vitamin D [36].

As to the effects of vitamin D supplementation on the extra-skeletal effects the trials (D2d study and ViDA) [12], benefits were found for those subjects with baseline vitamin D deficiency. Being vitamin D a nutrient and not a medication, the response to an appropriate nutritional intervention is obviously dependent on the levels prior to its supplementation [37].

A recent and witty paper reported the negative results of the first placebo-controlled trial evaluating the efficacy of parachute use to reduce the risk of injury of gravitational challenge [38]. Participants jumped from a small stationary aircraft on the ground and thus from a height without any risk of injury! Clearly, the involvement of individuals exposed to a condition free from any risk for ‘adverse events’ influenced the final results. Likewise, all the trials on vitamin D administration that involved individuals with normal 25OHD serum levels fell in the same paradox and therefore it is not surprising that the findings were negative.

In two recent papers of ours on patients with anorexia nervosa undergoing an intensive weight-restoration program [39,40], a strong relationship was found between vitamin D status and bone mineral density [39], as well as between BMD and the end-of-study vitamin D serum levels [40]. The gains in terms of BMD were significant only in the patients who had achieved optimal vitamin D status [40]. This suggests the need to include also to the end-of-study 25OHD serum levels. among the study parameters and not analyze outcomes in terms of randomization to treatment alone (vitamin D vs placebo). We strongly suspect that, in the end-of-study vitamin D deficient subgroup, the observed rate of new-onset events will be higher.

In conclusion, vitamin D deficiency is the true problem, and there are few doubts of its negative effects on health. Despite the large agreement on 20 ng/mL being an appropriate threshold for deficiency, consensus on the threshold for the diagnosis of deficiency needs to be quickly addressed.

Currently, misinterpretation of findings has been feeding a perception of all-round uselessness of vitamin D supplementation, Our sincere hope is that nobody will discontinue vitamin D supplementation in those cases in which this intervention is necessary, solely on the basis of faulty conclusions drawn from flawed studies.

In our opinion, experts should refuse to be labeled as ‘supporters’ or ‘opponents’ of vitamin D supplementation. On the contrary, we need to stress the importance of adequate supplementation in individuals with established deficiency, especially in the frail subjects and the elderly.

In closing, it is worth reminding, however, that Vitamin D status is not only a medical issue, but also a healthcare policy issue as well, with a relevant cost burden on the national healthcare system. With specific reference to Italy, treatment of Vitamin D deficiency in 2017 has accounted for 10.5 Defined Daily Doses of cholecalciferol consumed for every 1,000 subjects, with a total cost of 234 million euros and a per-capita cost of 3.89 euros [1]. Certainly, the supplementation in subjects with normal or even ideal vitamin D serum levels who cannot obtain further health benefits is certainly a waste of economical resources, but lack of supplementation to Vitamin D-deficient individuals to reduce risk of chronic or debilitating disease is a waste of opportunity.

Disclosure statement

Davide Gatti reports personal fees from Abiogen, Amgen, Janssen-Cilag, Mundipharma and Pfizer, outside the submitted work.

Francesco Bertoldo reports personal fees from: Abiogen, Amgen, Chiesi, Janssen, Grunenthal, Astellas, outside the submitted work.

Giovanni Adami has no conflict of interest to declare.

Ombretta Viapiana reports personal fees from: Abbvie, Novartis, MSD, Biogen, Fresenius Kabi, UCB, Eli-Lilly.

Stefano Lello has no conflict of interest to declare.

Maurizio Rossini reports personal fees from AbbVie, Abiogen, Amgen, BMS, Eli-Lilly, Novartis, Pfizer, Sanofi, Sandoz and UCB, outside the submitted work.

Angelo Fassio reports personal fees from: Abiogen and Novartis, outside the submitted work.