Relative Efficacy of Vitamin D2 and Vitamin D3 in Improving Vitamin D Status: Systematic Review and Meta-Analysis

Background: Widespread prevalence of vitamin D deficiency has been documented globally. Commonly used interventions to address this deficiency include supplementation and/or fortification with either ergocalciferol (vitamin D2) or cholecalciferol (vitamin D3), but the relative efficacy of these two vitamers is unclear. The current study aimed to evaluate the relative efficacy of ergocalciferol (vitamin D2) and cholecalciferol (vitamin D3) for raising the serum levels of vitamin D metabolites and functional indicators including serum parathyroid (PTH) levels, isometric muscle strength, hand grip strength and bone mineral density. Methods: Randomized and non-randomized controlled studies evaluating relative efficacy of ergocalciferol and cholecalciferol were systematically reviewed to synthesize quantitative and qualitative evidence as per the recommendations of according to “Preferred Reporting Items for Systematic reviews and Meta-analysis” guidelines. Search terms were constructed on the basis of the “participants”, “intervention”, “control”, “outcome” and “study type” (PICOS) strategy to systematically search the popular electronic databases. Relevant data from studies meeting inclusion and exclusion criteria were extracted and analyzed. Meta-regression, subgroup and sensitivity analyses were performed to investigate the influence of study-level characteristics including intervention dosage, frequency of dosing, interval between the last dose and test for outcome assessment, participant characteristics and analytical methods. Results: Apparently healthy human participants (n = 1277) from 24 studies were included for meta-analysis. The quantitative analysis suggested higher efficacy of cholecalciferol than ergocalciferol in improving total 25(OH)D (mean difference: 15.69, 95%CI: 9.46 to 21.93 nmol/L) and reducing PTH levels, consistently across variable participant demographics, dosage and vehicle of supplementation. Meta-regression suggested smaller differences in the efficacy of cholecalciferol and ergocalciferol at lower doses. Average daily dose was the single significant predictor of effect size, as revealed by multivariate meta-regression analysis. Conclusions: Compared to ergocalciferol, cholecalciferol intervention was more efficacious in improving vitamin D status (serum levels of total 25(OH)D and 25(OH)D3) and regulating PTH levels, irrespective of the participant demographics, dosage and vehicle of supplementation.


Introduction
Vitamin D is a group of fat-soluble vitamins traditionally recognized for its role in maintaining the homeostasis of calcium and phosphorous. Vitamin D commonly occurs in two forms: vitamin D 2 and vitamin D 3 . Vitamin D 3 , also known as cholecalciferol, is synthesized de novo in the skin on exposure to ultraviolet-B radiation, and it is also available from animal source foods [1]. Vitamin D 2 (ergocalciferol) is obtained from plants, particularly mushrooms and yeast. Structurally, vitamin D 2 differs from vitamin D 3 in having a double bond between C 22 and C 23 and a methyl group at C 24 [2]. Vitamin D 2 and vitamin D 3 undergo two sequential enzymatic hydroxylation reactions to be biologically active. The first hydroxylation occurs in the liver, which results in conversion of vitamin D 2 and vitaminD 3 to 25(OH)D 2 and 25(OH)D 3 , respectively. The second reaction occurs in the kidneys, wherein 25(OH)D 2 and 25(OH)D 3 are converted to their respective biologically active forms 1,25 dihydroxy vitamin D 2 and 1,25 dihydroxy vitamin D 3 [1]. As the circulating levels of total 1,25(OH) 2 D are homeostatically regulated, serum total 25(OH)D is considered to reflect the vitamin D status [3]. Both ergocalciferol and cholecalciferol are reported to exhibit similar potency in terms of their ability to cure vitamin D deficiency rickets [1].
Vitamin D deficiency is currently a global health problem. It is estimated that about 30% of adults have vitamin D deficiency (serum 25(OH)D < 50 nmol/L) and about 60% have insufficiency (serum 25(OH)D 50-75 nmol/L) [4]. The underlying reasons are probably multi-factorial including socio-cultural practices of avoiding sun exposure, dietary restrictions, environmental pollution, increased prevalence of obesity and genetic causes [5,6]. Tropical countries (such as India) with abundant sunlight are no exception, as high prevalence of vitamin D deficiency (30-80%) has been reported among adults [4] as well as among children and adolescents [7]. In addition to its classic functions, recent research also suggests the potential benefits of vitamin D in diabetes mellitus, metabolic syndrome, malignancy, hypertension, cardiovascular illness and neuropsychiatric disorders [8][9][10]. Alleviating vitamin D deficiency is, therefore, of public health significance.
Therapeutic supplementation and food fortification are the commonly used strategies for improving vitamin D status. Multiple intervention studies have demonstrated the efficacy of vitamin D (vitamin D 2 or vitamin D 3 ) supplementation, either as a single large bolus or given in divided doses by oral and parenteral routes, in raising the serum levels of the respective forms of 25(OH)D to varying levels [11,12]. Though both the vitamers increase the serum or plasma total 25(OH)D levels, their relative efficacy remains unclear. The national guidelines on food fortification in many countries including India do not specify the choice of the vitamin D fortificant and recommend a similar dose of vitamin D 2 and D 3. This is based on the assumption that the two vitamers have similar biological activities and are equally potent [13]. However, the equivalent potency of the two forms of vitamin D is based on studies on prevention and cure of rickets with either of the two vitamers in experimental animals and humans [14]. However, in order to enhance the effectiveness of the food fortification program, there is a need to evaluate the relative efficacy of these two vitamers in improving the serum vitamin D levels and influencing parathyroid hormone (PTH), a biomarker of bone mineral metabolism.
Previous systematic review concluded that cholecalciferol is more efficacious than ergocalciferol in raising the serum levels of total 25(OH)D [15]. As the metabolites of vitamin D 2 and vitamin D 3 are structurally different, studies comparing the efficacy of these two forms should ideally also estimate the individual metabolites: 25(OH)D 2 and 25(OH)D 3 . However, some of the studies included in the above meta-analysis did not report this crucial information [16][17][18][19]. Further, the relative effect of these two vitamers on serum PTH levels was also not evaluated. There is thus a need for a systematic review to evaluate the relative efficacy of vitamin D 2 and D 3 in raising the serum levels of different metabolites of vitamin D (total 25(OH)D, 25(OH)D 2 and 25(OH)D 3 ) and in modulating calcium homeostasis, as measured by serum PTH levels. Further, it is imperative to examine the relative efficacy of these two vitamers in relation to the baseline vitamin D status for better targeting of the intervention and in relation to the intervention dosage, frequency of dosing and duration of supplementation in order to understand their relative efficacy at different dosage regimes and during short-term and long-term use. Information on these aspects would be helpful for public health policy and practice.
We, therefore, conducted this systematic review to evaluate the relative efficacy of ergocalciferol and cholecalciferol supplementation in raising the serum levels of vitamin D metabolites (total 25(OH)D, 25(OH)D 2 and 25(OH)D 3 ) and functional indicators such as serum PTH, isometric muscle strength, hand grip strength and bone mineral density. Additionally, we explored the influence of various study-level characteristics including the dose of the intervention, dosing frequency, interval between the last dose and time of sample collection for the outcome assessment and average age of the participants on the outcome parameters using meta-regression analyses.

Methods
The study protocol was registered at PROSPERO (ID = CRD42018108202) [20] and executed as per the recommendations of "Preferred Reporting Items for Systematic reviews and Meta-analysis (PRISMA)" [21].

Criteria for Considering the Studies
Randomized and non-randomized controlled studies directly investigating the relative efficacy of ergocalciferol and cholecalciferol intervention (by either conventional supplementation/food fortification) in apparently healthy human participants were considered for the review. Studies explicitly intervening in patients with either acute or chronic conditions such as cardiovascular, liver, kidney, neuropsychiatric disorders, Human Immunodeficiency Virus (HIV) infection, cystic fibrosis and cancer were excluded.  [20]. The search terms were constructed on the basis of the PICOS (i.e., participants, intervention, control, outcome and study type) strategy endorsed by Cochrane collaboration [21]. The details of electronic search terms and inclusion/exclusion criteria are provided at the PROSPERO registration [20] and in the Appendix. The electronic search was initially performed from the date of inception to 31 September 2019 and updated on 19 June 2021. We employed "sensitivity and precision maximizing version" strategy to identify the relevant studies [21].

Data Collection and Analysis
All citations resulting from the electronic search were compiled using Endnote (Version 9), and duplicates were removed. Authors (R.B., R.P.) independently screened titles and abstracts of all the articles for their inclusion.
Full texts of articles identified during screening were further scrutinized for their inclusion. Information on the estimates of vitamin D metabolites such as serum levels of total 25(OH)D, 25(OH)D 2 , 25(OH)D 3 , functional indicators such as serum PTH, isometric muscle strength, hand grip strength and bone mineral density were extracted from each of the included studies, wherever available.

Data Extraction and Management
A structured data sheet was used to extract details from the included studies such as the year of publication, country/place of study, details of the intervention (duration, dosage, route of administration, vehicle used for supplementation and season, interval between the last dose and the test for outcome assessment), sample size, male-female ratio, mean and standard deviation of outcome parameters (vitamin D metabolites and its functional markers described above) and techniques employed to measure the outcome parameters. Duplication of data (publication) was investigated in the included studies as recommended by Cochrane (Section 5) [21].

Assessment of Risk of Bias
The risk of bias was independently evaluated by the authors using a structured spread sheet. The domains-random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective reporting and other bias were rated according to the 'Risk of bias' assessment tool described in Cochrane Handbook for Systematic Reviews of Interventions [21]. Disagreement was resolved by discussion among authors (R.B., R.P. and B.K.). Additionally, funnel plot symmetry was visually inspected to assess publication bias as a source of heterogeneity.

Measures of Treatment Effect
Mean and standard deviation (SD) (or equivalent) of the outcome variables were pooled from all the included studies to execute the meta-analysis. Studies reporting post-intervention changes (∆) from baseline were directly recorded for quantitative analysis [18,[22][23][24][25][26][27][28]. In case of studies reporting baseline and final (post-intervention) values [16,19,[29][30][31][32][33][34][35][36][37][38][39][40], the mean and SD of post-intervention changes were calculated using Monte Carlo simulation (Microsoft excel function). Lastly, for those studies reporting the results with box-and-whisker plots, a web-plot application was used to manually extract mean and confidence intervals [41]. The SD was derived from confidence interval using (where, SD, CI, Tα and df indicate standard deviation, confidence interval, t value distribution and degree of freedom respectively) [21]. All included studies analyzed and reported the results as per the principles of intention-to-treat.
Meta-analysis was conducted using RevMan 5.3 [42] utilizing the generic inverse variance method. A random-effect model was used in anticipation of contextual heterogeneity among the studies. Two-sided p < 0.05 was considered statistically significant.

Assessment of Heterogeneity
The influence of heterogeneity was evaluated by (1) visual inspection (inconsistency) of forest plots, (2) standard Chi 2 test (p < 0.1) and (3) I 2 statistic (>75%). Further, the source of heterogeneity was investigated by manually inspecting variables (sensitivity analysis) such as study participants, study setting, dose and duration of the intervention and co-interventions as well as methodological factors including study duration, season, method of sequence generation, allocation concealment, blinding of outcome assessment and losses to follow-up. Additionally, the heterogeneity due to study-level characteristics was explored using sub-group analyses and meta-regression with the random effect model.

Meta-Regression
Meta-regression was conducted to explore the contribution of various study characteristics on heterogeneity (for continuous variables). Bubble plots were constructed for those variables identified to be significant (p < 0.05). R-studio (Metafor package) was used for conducting meta-regression analysis.

Results
The details of the electronic search and studies excluded at intermediate steps are described in the flowchart (Figure 1). The systematic review identified 24 studies; however, data from two studies were not included in the quantitative analysis (meta-analysis) because precise estimates of central tendency and data dispersion were not available due to graphical reporting and lack of response to our request from the primary authors [43,44]. Therefore, 22 studies were finally included in the quantitative data analysis (i.e., metaanalysis), whereas all 24 studies were included in the qualitative analysis (systematic review). All the studies meeting the inclusion and exclusion criteria involved random allocation of participants to receive either ergocalciferol or cholecalciferol (with occasionally an additional placebo group) for evaluating the efficacy of the two vitamers. A total of 1277 participants were included in the meta-analysis, of which 644 received cholecalciferol and 633 received ergocalciferol intervention. Details regarding the study design, objectives, participants, interventions (as well as its adherence and dose received in the month before the outcome assessment), intake of additional calcium supplements (with equal doses in the two arms), exposure to UV-B radiation (e.g., sunlight), analytical methods and outcome variables evaluated are described in Table 1. Further, the risk of bias for each domain for all the included studies is described in Table 2. Ergocalciferol had marginally higher efficacy than cholecalciferol in raising total 25(OH)D during the initial 14 days of daily supplementation. However, the latter was more efficacious with subsequent daily supplementation. Cholecalciferol was more efficacious in fortnightly and monthly supplementation. Daily doses of both forms were equipotent in raising total 25(OH)D levels from their baseline value.  Cholecalciferol was more potent than ergocalciferol in raising total 25(OH)D levels. Details on the outcome parameters evaluated are listed in Table 1. Serum total 25(OH)D was evaluated in all but one study. Ten studies measured 25(OH)D 3 and 25(OH)D 2 levels individually [16,23,25,27,[32][33][34][35]39,45], whereas the remaining studies reported only total 25(OH)D values [17][18][19]22,24,26,28,29,31,[36][37][38]. Serum PTH levels were reported by seven studies [16][17][18]25,27,29,36]. All the studies involved healthy individuals including elderly [18], postmenopausal women [18,32] and pre-pubertal children [28,44].
Except for studies Hammami et al. (2017) [43], Nimitphong et al. (2013) [27] and Thacher et al. (2010) [44], which were conducted at Saudi Arabia, Thailand and Nigeria, respectively, the rest of the studies were conducted in North America, Europe and Australian continents. None of the studies investigated functional outcomes such as muscle strength or bone density. Sheih et al. (2016) reported evaluating bone mineral density and muscle strength in their clinical trial registration (ClinicalTrials.gov Identifier: NCT01848236) [38]. However, the results on these outcomes are not available.
Risk of Bias: Risk of bias for the domains "random sequence generation", "allocation concealment", "blinding of participants and personnel", "blinding of outcome assessment", "incomplete outcome data", "selective reporting" and other bias were rated as "low", "unclear bias" and "high" risk of bias as described by Cochrane Handbook for Systematic Reviews of Interventions. None of the studies were biased by incomplete/selective reporting of outcome, while majority of studies had low risk of bias in terms of random selection of participants (random sequence generation) and blinding the participants and personnel (96% and 60%, respectively). However, majority of the studies did not provide clear description of the allocation concealment and blinding of outcome assessment (65.38% and 76.92%, respectively) ( Table 2).
The multivariate meta-regression analyses revealed that "average dose per day" was a significant predictor of effect size even after controlling for other study-level characteristics "mean age of the participants", "total dose" and "dose-test interval" (Table 3 and Supplementary Figure S6 (Figure 6). Subgroup analy-ses were not possible as fewer studies reported serum 25(OH)D 2 and 25(OH)D 3 levels. In multivariate meta-regression analysis, "total dose", "average dose per day", "participant age" and "dose-test interval" were not significant predictors of effect size (Table 3).
Parathyroid hormone: Although both ergocalciferol and cholecalciferol interventions promoted a fall in serum PTH levels, most studies documented larger reduction in the cholecalciferol group as compared to the ergocalciferol group. Meta-analysis suggested higher efficacy of cholecalciferol in reducing PTH levels than ergocalciferol (MD: −0.56 pmol/L; 95% CI: −0.93 to −0.18, p = 0.005) (Figure 7). There was moderate heterogeneity (I 2 = 41%) within the included studies. Subgroup analysis in seven studies with daily intervention reduced heterogeneity (I 2 = 0), as well as the magnitude of effect (MD = −0.15 pmol/L, 95% CI: −0.01 to −0.3, p = 0.04) (Figure 7). The meta-regression analyses showed that study-level characteristics "total dose", "dose-test interval" "average dose per day" and "participant's age" were not significant predictors of the effect size (Table 3).     Table 3: Table presents the results of random-effect model (multivariate) meta-regression analyses investigating the relationship of "Mean age of the participants" (grand mean of both groups), "total dose (per 100 IU)" (sum of all doses received between baseline to final assessment), "average dose/day (per 100 IU)"(computed as the ratio of total dose and total study duration) and "Dose-Test Interval" (duration in days between the last intervention received and sample collection) with the outcome parameters (viz. the mean differences in 25(OH)D, PTH, 25(OH)D 3 and 25(OH)D 2 ) between the cholecalciferol and ergocalciferol groups.

Discussion
We analyzed the relative efficacy of ergocalciferol and cholecalciferol through a systematic review and meta-analysis, particularly focusing on different vitamin D metabolites (total 25(OH)D, 25(OH)D 2 and 25(OH)D 3 ) and a functional marker of calcium metabolism, PTH levels). Cholecalciferol supplementation was more efficacious than ergocalciferol in increasing total 25(OH)D levels and reducing PTH levels.
The qualitative analysis showed that, irrespective of the dosing frequency (single bolus/weekly/monthly/daily doses) or the mode or vehicle of administration (such as intramuscular injections, capsules, tablets, fortified orange juice, malt drink, biscuits or bread), cholecalciferol was more efficacious in raising serum total 25(OH)D levels. These results are in conformity with the earlier systematic review [15]. Our meta-analysis included fourteen randomized controlled trials in addition to the seven studies included in the previous meta-analysis. The mean difference in the ∆ total 25(OH)D (15.69 nmol/L, 95%CI: 9.46 to 21.93) observed in our study was similar to the earlier review [15], suggesting this to be a stable estimate. None of the studies included in our review investigated functional outcomes. However, a previous systematic review has reported lower relative risk of mortality among those supplemented with cholecalciferol than those with ergocalciferol [46].
The studies included in the present meta-analysis were heterogeneous. The sub-group and meta-regression analyses conducted to explore the source of heterogeneity provide interesting insights. The sub-group analysis of studies which measured the outcome more than two weeks after the last dose of the intervention showed greater difference in ∆ total 25(OH)D levels in the two groups (Supplementary Figure S2). Additionally, there was a greater difference in ∆ total 25(OH)D at higher intervention doses and when the intervention was delivered as a bolus dose as against the daily doses (Figures 2 and 3). These findings differ from the previous meta-analysis, which did not find significant difference in the impact of ergocalciferol and cholecalciferol in ∆ total 25(OH)D when the data from 6 RCTs (n = 248 participants) implementing daily dosing were pooled. However, our analysis included 14 RCTs with daily dosing (n = 965 participants) and had higher statistical power. Greater difference in ∆ total 25(OH)D levels in the two groups was also associated positively with average dose per day in the meta-regression analysis (Table 3 and Supplementary Figure S6).
Together, these findings suggest that the greater efficacy of cholecalciferol in raising serum levels of total 25(OH)D is likely with higher intervention doses, especially as bolus, and when the measurement is made more than two weeks after the last dose. The relatively lower potency of ergocalciferol in raising and maintaining total 25(OH)D could be attributed to the differences in structure (presence of additional methyl group at the 22nd carbon), its poor affinity to vitamin D binding protein leading to early degradation with shorter plasma half-life (13.9 days versus 15.1 days) [47].
Additionally, both ergocalciferol and cholecalciferol were relatively more beneficial in raising their respective forms of 25(OH)D compared to the other vitamer. As for ∆ total 25(OH)D, the differences in the two groups were greater at higher intervention doses in case of ∆ 25(OH)D 2 and ∆ 25(OH)D 3 . Interestingly, a few studies, but not all, have demonstrated decline in 25(OH)D 3 after ergocalciferol supplementation indicating its degradation [32,35]. This could be linked with competition between ergocalciferol and cholecalciferol in binding 25-hydroxylase or vitamin D binding protein [48]. The studies included in the present meta-analysis, however, did not show a negative impact of cholecalciferol supplementation on 25(OH)D 2 .
Parathyroid hormone tightly regulates calcium homeostasis, at the expense of bone resorption; vitamin D induced regulation of PTH is therefore essential for bone health and integrity. The PTH suppression following vitamin D supplementation is due to the paracrine 1-hydroxylase in the parathyroid gland and other tissues [49]. There was a greater reduction in PTH levels with cholecalciferol, and meta-regression suggested lower difference at lower intervention doses. It is, however, noteworthy that majority of the included studies [17,25,27,29,31,35] did not report a significant difference in PTH (p > 0.05).
Our study has several strengths including comprehensive assessment of vitamin D metabolites as well as PTH, a larger sample size compared to previous meta-analysis, important sub-group analyses in relation to baseline vitamin D levels, intervention dose and frequency of dosing, analytical methods and dose-test interval, which provide important insights. Further, meta-regression analyses provide valuable information on predictors of the magnitude of difference between the impact of the two vitamers. However, the study has limitations that need to be acknowledged. The studies included in our systematic review were heterogeneous as some involved only women [18,32] or elderly [16,18,30,31]; participants with low vitamin D at baseline (<50 nmol/L); variable dosages and differing frequency of dosages such as single bolus or daily, weekly [26] or monthly doses [30]; and different methods (RIA, HPLC or LCMS) were used to estimate vitamin D metabolites. Bias assessment revealed that only two studies [25,39] provided clear description of methods and were deemed high quality, while the remaining studies had incomplete description of methods and were regarded as moderate quality. Moreover, the bulk of evidence in our meta-analysis is based on studies from the North America, Europe and Australia, with low representation of studies from the lower-and middle-income countries in Africa and South Asia, where dietary patterns and sun exposure are different and the results may not be generalizable. Studies in children and infants are also underrepresented in the current analysis.

Conclusions
The results suggest cholecalciferol to be more efficacious than ergocalciferol for increasing 25(OH)D levels and reducing serum PTH levels. However, both ergocalciferol and cholecalciferol interventions had higher efficacy in raising the serum levels of their respective forms of 25(OH)D (i.e., 25(OH)D 2 and 25(OH)D 3 ) when compared to the other vitamer. Cholecalciferol was more efficacious than ergocalciferol with bolus/intermittent doses, but frequent (daily) dosing was associated with lower differences for serum 25(OH)D and PTH levels. Thus, with lower doses typically used in fortified foods, cholecalciferol may be only marginally better than ergocalciferol for improving vitamin D status. Future studies evaluating the relative efficacy of ergocalciferol and cholecalciferol should also evaluate functional markers such as bone mineral density and muscle strength, and they should include longitudinal assessment at multiple time points to provide deeper insights on kinetics and dynamics of vitamin D. Lastly, studies from tropical areas, low-and middleincome country settings and younger populations (children and adolescents) are needed to understand the roles of nutrition and sun exposure in influencing the relative efficacy of the two vitamers.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/nu13103328/s1, Figure S1: Forest Plot analysis of total 25(OH)D: subgroup analyses in relation to the method of estimation of 25(OH)D. Figure S2: Forrest Plot analysis of total 25(OH)D: dose test interval wise sub-group analysis. Figure S3: Forrest Plot analysis of 25(OH)D: total dose wise sub-group analysis. Figure S4: Forrest Plot analysis of 25(OH)D: age group wise sub-group analysis. Figure S5; Forrest Plot analysis of 25(OH)D: sub-group analysis based on baseline vitamin D status. Figure S6: Bubble plot demonstrating the association between "average dose per day" as and mean difference between the 2 interventions Author Contributions: R.B.: conceptualization, data curation, formal analysis, investigation, methodology, statistical analysis, result interpretation, original draft preparation, editing revision and submission. R.P.: conceptualization, data curation, methodology, result interpretation, manuscript draft edit and revision. B.K.: conceptualization, data curation, investigation, methodology, result interpretation, manuscript draft edit and revision H.S.S.: conceptualization, result interpretation, manuscript draft edit and revision. All authors have read and agreed to the published version of the manuscript.

Funding:
The research did not receive any specific funding.

Acknowledgments:
The authors acknowledge Soundarya Soundararajan for providing codes and aiding with R-studio analysis.

Conflicts of Interest:
The authors declare no conflict of interest.