Impact of Vitamin D3 Dietary Supplement Matrix onClinical Response

Michael L. Traub, John S. Finnell, Anup Bhandiwad, Erica Oberg, Lena Suhaila,Ryan Bradley

Lokahi Health Center, Kailua Kona, HI, USA; AOMA Graduate School of Integrative Medicine; Austin,TX, USA; Wayne State University, Detroit, MI, USA; Pacific Pearl Center for Health and Healing; La Jolla,CA, USA; Cancer Treatment Centers of America Western Regional Medical Center; Phoenix, AZ, USA;Bastyr University Research Institute; San Diego, CA, USA

Context: As a result of research suggesting increased health risk with low serum 25-hydroxy-cholecalciferol (25(OH)D), healthcare providers are measuring it frequently. Providers and pa-tients are faced with treatment choices when low status is identified.

Objective: To compare the safety and efficacy of three vitamin D3 dietary supplements withdifferent delivery matrices.

Design & Setting: 12-week, parallel group, single-masked, clinical trial conducted in Seattle, WAand Kailua Kona, HI.

Participants & Intervention: 66 healthy adults with (25(OH)D) < 33 ng/ml were randomly assignedto take one of three D3 supplements, i.e., a chewable tablet (TAB), an oil-based drop (DROP) oran encapsulated powder (CAP), at a label-claimed dose of 10,000 IU/day. Actual D3 content wasassessed by a third-party, and the results adjusted based on the actual D3 content administered.

Main Outcome Measures: mean change in 25(OH)D/mcg D3 administered; difference in theproportion of D3 insufficient participants (i.e., 25(OH)D <30ng/ml) reachingsufficiency (i.e.,25(OH)D >30ng/ml); and mean change in serum 1, 25-dihydroxycholecalciferol.

Results: In two of the three products tested, the measured vitamin D3 content varied considerablyfrom the label-claimed dose. Differences in 25(OH)D/mcg D3 administered were significantlydifferent between groups (p!0.04; n!55). Pairwise comparisons demonstrated DROP resultedin a greater increase than TAB (p<0.05) but not than CAP. TAB was not different from CAP. Theproportions reaching sufficiency were: 100% (TAB and CAP) and 80% (DROP) (p!0.03 betweengroups, n!55). 1, 25-dihydroxycholecalciferol did not change significantly in any group.

Conclusions: Oil-emulsified vitamin D3 supplements resulted in a greater mean change in serum25(OH)D concentration, but fewer patients reaching vitamin D sufficiency, than chewable orencapsulated supplements.

The importance of cholecalciferol, ie, vitamin D3 (D3),in human health conditions has gained increased re-

search attention in recent years. As a result of observa-tional findings suggesting increased health risk with lowserum 25-hydroxycholecalciferol (25(OH)D) (1–11),healthcare providers of all categories are measuring serum

25(OH)D more frequently, and providers and patients arefaced with choices in treatment to achieve repletion whenlow status is identified, including prescription and over-the-counter dietary supplement forms of D3.

The 2011 report on dietary reference intakes for cal-cium and vitamin D from the Institute of Medicine sug-

ISSN Print 0021-972X ISSN Online 1945-7197Printed in U.S.A.Copyright © 2014 by the Endocrine SocietyReceived August 15, 2013. Accepted March 19, 2014.

Abbreviations:

doi: 10.1210/jc.2013-3162 J Clin Endocrinol Metab jcem.endojournals.org 1

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

gested substantial gaps in D3 research requiring attention,including the effectiveness of high-dose supplementationstrategies with structured analysis of adverse events (12–14). Recent evidence suggests high doses of D3 are safe andmay be necessary to achieve cholecalciferol “sufficiency.”(15–17) Vitamin D3 dosages of up to 10,000 IU/d havebeen determined to be safe, whereas dosages over 40,000IU/d have been associated with vitamin D toxicity (18–21). Clinical trials evaluating the effectiveness of dietarysupplementation to correct D3 insufficiency have demon-strated relative safety without hypercalcemia or othermanifestations of toxicity (21–25). However, the effec-tiveness of different repletion strategies to reach “opti-mal” 25(OH)D concentration was highly variable andranged between 33%–86% in participants receiving oralD2 (23, 24) and 5%–89% in the participants receivingoral D3 (22, 25).

It is important to consider sources of potential variabil-ity when interpreting the results of these and other clinicaltrials of vitamin D, especially when dietary supplementsare being used. One potential source of variability in theclinical effectiveness of dietary supplements, which mayalso influence patient preference and adherence, is the de-livery matrix used in the supplement; numerous matricesexist in the marketplace, however the most common ma-trices include dry powder in capsules, chewable tabletsand oil-emulsified drops. Thus, if absorption and overalleffectiveness differs by delivery matrix, it is important toestablish dosing equivalents of these products to ensurepatient safety and to assist providers.

Another source of potential variability in results fromvitamin D clinical research is dosing disparities from thelabel-claimed dose and the actual vitamin D content. In-dustry guidance for the manufacturer of vitamin D sup-plements comes from the U.S. Pharmacopoeial Conven-tion National Formulary (USP-NF) and the U.S. Food andDrug Administration (FDA) (26–28). The USP-NF guid-ance to dietary supplement manufacturers allows D3 di-etary supplements to contain no less than 90% and nomore than 165% of the product label claim for cholcal-ciferol (D3). The FDA (21CFR101.9) provides the follow-ing regulatory guidance on D3 supplements: 1) D3 contentin the composite that is at least equal to the D3 content ofthe label claim; 2) reasonable excesses are acceptablewithin current good manufacturing practices; and 3) D3

content does not exceed 120% (%) of the label claim orfalls within the variability generally recognized for theanalytical method for analysis (28). Therefore, label-claimed doses are not always accurate for vitamin D sup-plements, and third party measurement and quantifica-tion of vitamin D content of dietary supplements used inclinical trials should be considered a “best practice”.

To better assess the degree of inconsistency betweenproduct label claims and actual vitamin D content, and todirectly compare different delivery matrices on clinical re-sponse, we performed a randomized, pragmatic clinicaltrial (PCT) of three common supplement matrices, andadjusted our results for the actual dose administered basedon third party analysis of each product.

Materials and Methods

The clinical trial protocol was approved by the Bastyr UniversityInstitutional Review Board (#09A-1241) and was registered atclinicaltrials.gov as NCT01524874.

Settings and ParticipantsThe trial protocol was standardized and implemented in three

clinical settings: Bastyr Center for Natural Health (Seattle,Washington); Bastyr University Clinical Research Center (Ken-more, Washington); and Lokahi Health Center (Kailua Kona,Hawaii). Participants were recruited from active patients at eachclinical site, as well as through flyers posted throughout thegreater Seattle metropolitan area. Candidate participants wereinformed about the study by their physicians and/or a studycoordinator and invited for screening at the discretion of theirphysician.

Candidate participants provided informed consent and com-pleted a baseline health assessment and laboratory screening in-cluding measurement of serum 25(OH)D and 1,25(OH)2D, acomprehensive metabolic panel and complete blood count(CBC). We then enrolled participants who were generallyhealthy adults, aged 18–65 years of age, with serum 25(OH)D !33ng/ml and willing to be randomized to one of three D3 treat-ments for 12-weeks. Candidates were excluded if they met any ofthe following criteria: unwilling to provide consent; extradietaryvitamin D intake " 1,000 IU per day; unwilling to use sunscreen;allergy to sunscreen or sesame oil; serum 25(OH)D ! 33 ng/ml(82.5 nmol/ml); aspartate transaminase (AST) "60 U/L; alaninetransaminase (ALT) "65 U/L; alkaline phosphatase " 120 U/L;total bilirubin"1.5 mg/dL; serum creatinine"1.4 mg/dL; bloodurea nitrogen (BUN) "25 mg/dL; pregnancy or unwillingness toavoid pregnancy by using contraceptives; osteoporosis; parathy-roid disorder; difficulty swallowing pills; psychological condi-tions or substance abuse that may challenge adherence to theprotocol; cardiac arrhythmia; other severe illness limiting activ-ities of daily living (ADL); and/or currently taking medicationsthat interfere with the metabolism of vitamin D (eg, anticonvul-sants, anticoagulants, oral corticosteroids, or barbiturates) (29,30).

Randomization and InterventionsRandomization was conducted centrally in blocks of three.

Upon confirmation of participation criteria each participant wasrandomized to one of three dietary supplements donated by threedifferent private manufacturers: an oil emulsified drop (DROP);a dry, encapsulated powder (CAP); or a chewable tablet (TAB).Participants were each instructed to take five dosage units perday, equaling a label-claimed daily dose of 10,000 IU. The TABand CAP were from a single manufacturing batch; the DROP wasfrom two manufacturing batches due to the short expiry life of

2 Response to Vitamin D3 Supplement Matrix J Clin Endocrinol Metab

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

this product. The supplements were provided free of charge inadequate quantity for the entire 12-week study period. Studyparticipants were not masked to their product assignment due tothe inherent challenges in masking the administration of thechewable tablets and the oil-emulsified drops; the trial was in-tended as pragmatic.

Assessment of D3 Content of Each SupplementEach of the supplements was sent to an independent analyt-

ical lab (Flora Research Laboratories; Grants Pass, OR) to mea-sure the D3 content of each product before the trial began, andat the end of the trial to evaluate for product degradation. Asample from each product was prepared in triplicate, with anadditional sample for spike recovery, and the samples were an-alyzed using ion-trap liquid chromatography tandem mass spec-trometry (LC-MS/MS).

Assessment and Control of Potential ResidualConfounders

Participants completed a standardized food frequency ques-tionnaire (FFQ) to quantify their dietary intake of vitamin D(31–33) and a standardized sun exposure questionnaire (SEQ) toquantify their use of sun protection, eg, long-sleeved clothing,and exposure, ie, hours per day (34). To further reduce potentialsources of variability in our results, participants were providedwith an ample supply of SPF 30 sunscreen (SolRx, Buford, GA orMyChelle Dermaceuticals, LLC., Louisville, CO) and instructedto apply it daily to sun-exposed skin. Special attention was paidto the measure of human PTH using the HBN1B-51K MilliplexMap Kit (Millipore Corp., Billerica, MA); PTH kits were read ona Luminex 100/200 analyzer (Luminex Corporation, Austin,TX). Parathyroid (PTH) concentrations were analyzed frombanked samples (stored at –70°C) for participants from the Se-attle site (N # 40; missing # 1). PTH was not analyzed in samplesfrom the Hawaii site, due to concerns of potential sample deg-radation during overnight transit. Additional clinical parameterswere collected at baseline and again after 12-weeks: height,weight, heart rate and blood pressure (BP); body mass index(BMI); hemoglobin A1c (HbA1c); total cholesterol; low-densitylipoprotein (LDL); high-density lipoprotein (HDL); triglycer-ides; calcium; fasting insulin and glucose; and a comprehensivemetabolic panel.

Assessment of Safety, Dose Reduction Protocoland Trial Stopping Criteria

Safety was assessed by inquiry at clinical research visits, safetymonitoring telephone interviews at 6-weeks (ie, trial midpoint)and self-report throughout the trial via telephone or email. Ad-verse event (AE) data were collected using a standardized adverseevent reporting form, based on the DSHS form 10–334, whichquantifies severity on a 0–4 scale where 0 # no symptom and 4 #FDA serious AE. If participants reported a symptom associatedwith hypercalcemia or vitamin D toxicity at any time during thetrial, serum calcium and 25(OH)D were measured immediately.To ensure participant safety, a two-tiered dose reduction proto-col was implemented for participants who reported symptomsassociated with hypercalcemia for three days or longer, despitean absence of laboratory measured hypercalcemia. Dosage wasfirst reduced to 6,000 IU/d, and subsequently to 2,000 IU/d ifsymptoms did not resolve within 3 days. Trial stopping criteria

included: hypercalcemia occurring in ! 10% of participants,greater than three deaths from any cause, or one death attribut-able to a dietary supplement.

OutcomesThe primary, secondary and tertiary outcome comparisons

between groups were: change in mean serum 25(OH)D/mcg vi-tamin D3 administered; difference in proportion of participantswith D3 “insufficiency” (ie, 25(OH)D ! 30 ng/dL) reaching“sufficiency” (ie, 25(OH)D ! 30 ng/dL); and mean change inserum 1,25(OH)2D.

Measurement of 25(OH)DSerum 25(OH)D was measured at Pacific Physicians Labo-

ratory, LLC, Lynnwood, WA (PPL) using a chemi-luminescenceimmunoassay (LIAISON 25-OH Vitamin D Total Assay –310600; DiaSorin Inc. Stillwater, MN) (35). Serum 1,25(OH)2Dwas measured by liquid chromatography tandem mass spec, us-ing the extraction, chromatography, radioreceptor assay (QuestDiagnostics, Madison, NJ). Samples from Seattle, WA, were re-frigerated and transported on the same day to PPL via courier;samples from Kailua-Kona, HI, were stored on ice (-4°C) andshipped overnight directly to PPL for processing. All sampleswere stored for less than 24 hours.

Statistical Analysis and Data ManagementOur target sample size of 60 total participants allowed for

detection of a 25% differential response between-groups with80% power at a significance level of "#0.05. Baseline charac-teristics were compared between randomization groups by anal-ysis of variance (ANOVA) to confirm balanced allocation.ANOVA was conducted to determine if there was a significantdifference in 25(OH)D/mcg D3 administered (primary outcome)and/or 1,25(OH)2D (tertiary outcome) between groups, ie, H0#difference in mean change between groups # 0 ng/mL. If the nullhypothesis was not accepted, we compared the final distributionsof 25(OH)D and 1,25(OH)2D between group pairs usingTukey’s test, ie, H0# difference in mean change betweengroups # 0 ng/mL. Analysis of covariance (ANCOVA) was per-formed post hoc to determine if adjustment for baseline covari-ates (ie, age, gender, study site, BMI, baseline 25(OH)D, dietaryintake of vitamin D, sun protection behavior, sun exposure, andenrollment season) modified the primary results. For both ad-justed and unadjusted analyses we calculated our primary out-come based on the actual (ie, not label-claimed dose) of vitaminD administered to each group by calculating the change in25(OH)D in ng/ml for each participant and dividing by the actualvitamin D dose in micrograms (mcg) they received based on theresults of third party measurement of each supplement. Actualvitamin D dose was calculated by averaging the results frompretrial and post-trial triplicate measurements of D3 in each sup-plement, conducted by a third party laboratory. Because twoparticipants in the study reported adverse events and required adose reduction, we calculated the average dose over the course ofthe entire study for these participants based on the number ofdays they took the full dose of 10,000 IU daily, and the numberof days they took the reduced dose of 6,000 IU daily to generatea time-weighted average dose for these two participants. We thencalculated the “mean change in 25(OH)D ng/ml per mcg vit D3

administered” for the entire group. To determine the proportionof participants with baseline D3 “insufficiency” who reached

doi: 10.1210/jc.2013-3162 jcem.endojournals.org 3

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

“sufficiency” at follow-up (secondary outcome), we applied twodefinitions of “insufficiency, ie, both as defined by our clinicallaboratory (ie, 25(OH)D ! 33 ng/dL) and as defined by theInstitute of Medicine (ie, 25(OH)D ! 30 ng/dL). Proportionswere compared between groups using Fisher’s Exact test.

Study data were managed using REDCap v4.0 (ResearchElectronic Data Capture; Vanderbilt University) hosted at theUniversity of Washington (36). One of the principal investigators(PI) of this trial (R Bradley) did not have access to interim trialresults and was separated from all data entry. The other PI of thistrial (M Traub) did have access to individual participant’s resultsfor the purposes of ongoing clinical care at the Kona, HI site; MTraub did not have access to composite trial results and was notinvolved in data analysis. The data analysis plan was finalized byR. Bradley and J. Finnell before any data were analyzed. Theunmasking of both PIs to the allocation groups occurred onlyafter the results were considered final. STATA 11.0 (StataCorp,College Station, TX) was used to conduct statistical analyses.

Role of the Funding SourceThis project was supported by the National Institutes of

Health National Center of Complementary and AlternativeMedicine (NCCAM) grant T32AT00815 and National Centerfor Research Resources (NCRR) grant 1KL2RR025015–01.Additional support was provided by: the Diabetes Action Re-search and Education Foundation; and Bastyr University. Thefunding sources and manufacturers had no influence on any as-pects of the trial, including data collection, manuscript prepa-ration, or decisions regarding where to submit the manuscript forpublication, however the product manufacturers were allowedup to 30 days to review the manuscripts prior to submission andoffer their comments.

Results

A total of 302 individuals were screened for inclusionbased on their interest in the trial (Figure 1). One hundredand ninety-nine participants total (n # 199) were excludedfor reasons summarized in Figure 1. Of the one hundredand three (103) participants screened for eligibility basedon 25(OH)D status, thirty-seven (37) were sufficient inserum 25(OH)D. Sixty-six (66) participants met partici-pation criteria and were randomized. Loss to follow upwas evenly distributed among treatment groups and to-taled 11 (16.7% of randomized sample) participants, leav-ing a remaining sample of 55 (83% of randomized sample)participants in which to assess outcomes. Baseline char-acteristics (Table 1) did not differ significantly betweengroups (all P " .05 by ANOVA), including importantcovariates thought to correlate with vitamin D status, eg,baseline vitamin D status, age, gender, BMI, dietary in-take, sun protection, sun exposure, state and enrollmentseason.

Dose of Vitamin D3

Third party testing using LC-MS/MS determined eachmean dose as follows: TAB # 3,943 IU/dosage unit or 98.6mcg/dosage unit (ie, 197% label claim); DROP # 2,208IU/dosage unit or 55.2 mcg/dosage unit (ie, 110% labelclaim); and CAP # 4,131 IU/dosage unit or 103.3 mcg/dosage unit (ie, 207% label claim).

Figure 1. CONSORT flow diagram of three-arm comparative effectiveness trial

4 Response to Vitamin D3 Supplement Matrix J Clin Endocrinol Metab

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

Change in Serum 25(OH)DWithin group analysis unadjusted for vitamin D con-

tent (Table 2) demonstrated all treatments increased25(OH)D significantly between baseline and 12-week fol-low-up (P ! .0001 for all groups)

Changes in 25(OH)D per mcg D3 administered, basedon the results of third party analysis (TPA), were: TAB #0.068 $ 0.016 ng/ml 25(OH)D/mcg D3; DROP #0.125 $ 0.015 ng/ml 25(OH)D/mcg D3; and CAP #0.106 $ 0.017 ng/ml 25(OH)D /mcg D3, as shown inFigure 2. Between group ANOVA demonstrated signifi-cant between-group differences (P # .04) and subsequentTukey’s test revealed significant differences betweenDROP vs. TAB (P ! .05), but not between TAB vs. CAPor between DROP vs. CAP (P " .05 for both). The primaryresult did not change in post hoc ANCOVA analyses. Be-tween group ANCOVA demonstrated significant be-tween-groupdifferences (P# .03) and subsequentTukey’stest mirrored the unadjusted analyses with differences be-tween DROP and TAB showing significance (P ! .05), butnot TAB vs. CAP or DROP vs. CAP (P " .05 for both).

Proportion Attaining 25(OH)D Sufficiency StatusWe report the proportion of participants attaining suf-

ficiency per the reference range of our clinical reference

lab, Pacific Physician’s Laboratories (PPL: Lynnwood,WA) and guidelines set by the Institute of Medicine. Perthe PPL defined reference range, the proportion of partic-ipants with D3 “insufficiency” (ie, 25(OH)D ! 33 ng/dL;n # 55) reaching “sufficiency” (ie, 25(OH)D ! 33 ng/dL)between treatment arms was 100% for TAB and CAP, and80% for DROP (P # .03). Per the IOM clinical definitions,the proportion of participants with D3 “insufficiency” (ie,25(OH)D ! 30 ng/dL; n # 46) reaching “sufficiency” (ie,25(OH)D ! 30 ng/dL) between treatment arms was 100%for TAB and CAP, and 82% for DROP (P # .07).

Change in Serum 1,25(OH)2DThere were no clinically or statistically significant

changes in 1,25(OH)2D within or between allocationgroups; P " .05 for all within group comparisons andbetween group ANOVA (Table 3).

Changes in Other Clinical ParametersChanges in serum calcium and plasma PTH levels were

not significant within or between groups (all P " .05).

Adverse EventsThree mild AEs occurred, including tachycardia, anx-

iety, itching, muscle cramping and nausea; all occurred

Table 1. Baseline characteristics

Group Characteristic (s)

All n ! 55Mean(SD)

or Frequency(%)

TAB n ! 18Mean(SD)

or Frequency(%)

DROP n ! 20Mean(SD)

or Frequency(%)

CAP n ! 17Mean(SD)

or Frequency(%)

Groupcomparison

p-value

Study Site (% from WA) 75% 72% 75% 76% 0.43

Age (years) 39.9 (13.7) 40.3 (13.5) 39.5 (13.6) 39.9 (14.7) 0.98

Gender (% female) 85% 83% 85% 88% 0.92

Systolic BP (mmHg) 111.3 (12.7) 112.4 (11.2) 107.7 (9.8) 114.4 (16.4) 0.25

Diastolic BP (mmHg) 73.1 (9.5) 75.9 (9.4) 69.8 (9.9) 73.8 (8.5) 0.13

Height (inches) 65.8 (3.1) 65.5 (3.0) 65.5 (3.1) 66.4 (3.4) 0.67

Weight (pounds) 161.5 (39.6) 168.9 (46.9) 154.8 (27.6) 161.6 (44.1) 0.56

Body mass index (kg/m2) 26.1 (5.3) 27.4 (6.5) 25.2 (4.0) 25.6 (5.2) 0.39

Dietary Vitamin D intake(IU/week)

1525 (1826) 1545 (1933) 1502 (1869) 1530 (1770) 0.10

Sun exposure(hours/week)

9 (10) 10 (13) 10 (10) 8 (5) 0.73

Sun protection (% using) 76% 83% 60% 88% 0.09

Parathyroid hormonea(pg/mL)

36.6 (51.9) 21.2 (19.3) 53.2 (78.3) 31.7 (25.7) 0.26

Baseline 25(OH)D(ng/mL)

22.6 (6.7) 21.9 (7.8) 24.2 (5.1) 21.5 (7.2) 0.41

aPTH measured in Seattle site samples only: total n # 40; n TAB # 12; n DROP # 15; n CAP # 13.

Table 2. Mean changes in total serum 25-hydroxycholecalciferol (25(OH)D)**

Group

25(OH)D

p-value forchange*

Baseline Mean(SD)(ng/ml)

Week 12 Mean(SD)(ng/ml)

Mean(SD)change (ng/ml)

95% ConfidenceInterval

TAB (n # 18) 21.9 (7.8) 55.1 (15.2) %33.3 (5.7) [21.9,44.6]

!0.0001

DROP (n # 20) 24.2 (5.1) 58.6 (27.4) %34.4 (5.4) [23.6,45.1]

!0.0001

CAP (n # 17) 21.5 (7.2) 75.1 (22.8) %53.6 (5.8) [41.9,65.2]

!0.0001

*p value corresponds to within group changes by 2-sided, paired t test

** Results are reported by randomization group, and are not standardized by actual dose administered.

doi: 10.1210/jc.2013-3162 jcem.endojournals.org 5

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

without hypervitaminosis D or hypercalcemia (verified bylaboratory tests per protocol). Two dose reductions to6,000 IU D3/d were required per protocol, and all AEsresolved.

Discussion

Our study demonstrates several novel and important clin-ical results. First, we identified considerable variabilitybetween label-claimed content and independently mea-sured content of two of the three supplements evaluated inthis trial, with one product containing double the label-claimed dose. We further compared the relative effective-ness of three different delivery matrices for vitamin D3 indietary supplements and measured statistically significantdifferences in clinical response, with the oil-emulsifiedproduct being the most effective of the products evaluated.We also measured an overall effectiveness of 93% for therepletion strategy tested here, compared to other repletionstrategies reported in the literature, ie, 5% to 89%. Fi-nally, all three vitamin D3 dietary supplements were safeto administer for 12 weeks at 10,000 IU/d (based on labelclaim), without hypercalcemia or other clinically signifi-cant adverse events.

StrengthsWe employed a rigorous, compar-

ative effectiveness trial design in or-der to compare commonly availablevitamin D3 dietary supplements rep-resenting three different delivery ma-trices. The trial was appropriatelypowered to measure a clinically sig-nificant 25% relative difference be-tween groups for changes in meanfor 25(OH)D, a difference whichwas exceeded by one product tested.Additionally, we assessed for alloca-tion imbalance between potentialconfounders, ie, BMI, dietary intakeof vitamin D, sun exposure, and sun-screen use-which did not differ be-tween randomization groups. To en-sure even minor residualconfounding did not impact our in-terpretation, we also adjusted our fi-nal results for covariates that maypredict response to vitamin D sup-plements, eg, baseline status, enroll-ment season, dietary intake, andBMI (37, 38). We also employedstandardized AE reporting, clinicallaboratory monitoring and dose re-

duction protocols to ensure safety of study participantsand detect of any cases of hypercalcemia. The generaliz-ability of our results is greatly improved by the conduct ofthe trial in two very different geographical regions, ie,Seattle, WA and Kona, HI. We also conducted our trialover an entire calendar year, recruiting participants duringall seasons, ie, participants were not enrolled during onlythe summer or the winter.

The greatest strength of this trial is that we contractedindependent measurement of the vitamin D3 content ofeach of three dietary supplements evaluated in the trial,before and after the intervention period, in order to ac-count for dosing variability and potential product degra-dation. In the absence of independent product testing, ourresults would be based solely on the product’s labelclaimed dose, which would have led to erroneous results(data not shown). Because this source of error is likelycommon, (38) and often not quantified, we recommendmanufacturer-independent testing of vitamin D3 dietarysupplements in all future clinical trials.

LimitationsWe did not employ an inactive or placebo control group

in this study because all participants had clinically low

Figure 2. a. Unadjusted between group changes in serum 25-hydroxycholecalciferolconcentration ng/ml/mcg vitamin D3 administered (ng/ml 25(OH)D/mcg D3) 2b. Adjustedbetween group changes in serum 25-hydroxycholecalciferol concentration ng/ml/mcg vitamin D3

administered (ng/ml 25(OH)D/mcg D3)

6 Response to Vitamin D3 Supplement Matrix J Clin Endocrinol Metab

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

vitamin D status warranting active treatment, and becausethe primary intention was to compare the efficacy of theseactive treatments. Similarly, participants were not maskedto their allocation group due to difficulties in masking theemulsified drops. There is little reason to suspect groupcrossover or other sources of bias due to their group as-signment because all supplements contained active D3, allwere provided free of charge, and each was dispensed inadequate supply for the entire duration of the trial. Al-though all other baseline characteristics were balancedduring randomization, we did not assess participant eth-nicity and thus were unable to measure any impact differ-ences in skin pigmentation may have had on our finalresults. Similarly, because we were primarily interested indetermining if the supplements tested would produce clin-ically-significant increases in 25(OH)D generalizable tomost patients, we did not measure the contribution ofhighly individualized, genetically-determined covariates,eg, DNA methylation levels of CYP2R1 and CYP24A1, ofvitamin D receptor (VDR) polymorphism status, whichmay have also predicted some of the observed variation inresponse (39–42).

There remains considerable debate about the “goldstandard” for the determination of serum 25(OH)D. Weused the DiaSorin method used by most clinical labora-tories. However, measuring serum 25(OH)D, using recentmethodology like liquid chromatography coupled withtandem mass spectroscopy (LC-MS/MS) may result in im-proved sensitivity and specificity in future clinical trialswhen absolute concentrations, rather than relativechanges, are critical to interpretation (43).

Although we attempted to control for sun exposure andseason by dispensing sunscreen to all participants, andadvising them to apply it daily to all sun-exposed surfaces,we were unable to determine participants’ adherence tothe use of sunscreen. However, because baseline sun pro-tection behaviors, eg, use of sun hats, long-sleeved shirtsand sunscreen, were equivalent between randomizationgroups, we would expect the use of sunscreen we providedto be equivalent between groups as well. Current researchhas not definitively demonstrated that the level of skinpigmentation affects changes in serum 25(OH)D inducedby supplementation with exogenous sources of vitamin D;

therefore the impact of this potential confounder was notassessed (15).

The interpretation of our results regarding the apparentclinical effectiveness of 10,000 IU of D3/d as an appropri-ate repletion strategy needs to be qualified with our find-ing that only the emulsified drop had a D3 concentrationrelatively consistent with its label claim of 2,000 IU perdosage unit, while the tablet and the capsule containedhigher concentrations of vitamin D3 than were claimed onthe label. This finding exemplifies the need for improvedstandardization in quantification of the potency of D3 di-etary supplements for both clinical practice and research.In fact, a recent study of the quality of overthe-counter D3

dietary supplements showed similar variability in productquality (44). Sources of dosing variability to consider indietary supplements include the USP-NF and FDA(21CFR101.9) criteria allowing overages in D3 dietarysupplements to account for current good-manufacturingpractices, degradation of raw material and/or to ensure thelabel accuracy upon the expiration date, and the variabil-ity inherent in the analytical methodologies used to mea-sure D3 (26, 28). The fact that the oil-emulsified dropresulted in the greatest mean change in 25(OH)D despitebeing the lowest dosed product in the trial is surprising andfurther supports the apparent improved effectiveness ofthis preparation.

Although we had only three self-reported adverseevents, no cases of overt hypercalcemia, and no significantchanges in either 1,25(OH)2D or parathyroid hormone,we did not assess 24-hour urinary calcium excretion,which is considered a more sensitive indicator of vitaminD toxicity. Thus conclusions regarding the safety of therepletion strategy in this trial must be interpretedaccordingly.

Finally, our finding that only 82% of participants whoreceived the oil-emulsified drop achieved D3 “sufficiency”is surprising considering the high dose administered, theexcellent overall effectiveness of this product, and partic-ipants were instructed on proper mixing of the liquidproduct as instructed on the label prior to dosing. We areunable to explain this finding, however speculative expla-nations for this observation may include limited adher-ence, poor mixing of the product, intestinal malabsorp-

Table 3. Mean changes in serum 1,25-dihydroxycholecalciferol (1,25(OH)2D)**

Group

1,25 (OH)2 D

p – value*Baseline Mean (SD) (ng/ml) Week 12 Mean (SD) (ng/ml) Difference Mean (SD) (ng/ml) 95% Confidence IntervalTAB (n # 16) 41.4 (2.5) 44.9 (4.3) % 3.5 (3.9) [-11.9, 4.9] "0.05DROP (n # 19) 43.4 (3.2) 43.2 (2.9) & 0.2 (4.0) [-8.2, 8.5] "0.05CAP (n # 17) 42.8 (3.7) 39.0 (2.5) & 3.8 (4.3) [-5.4, 12.9] "0.05

*p value corresponds to within group 2-sided paired t-tests. Between group ANOVA P " 0.05.

** Results are reported by randomization group and not standardized by actual dose administered.

doi: 10.1210/jc.2013-3162 jcem.endojournals.org 7

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

tion, genetic polymorphisms in vitamin D receptor and/orvitamin D binding protein (39–41).

SignificanceThe Endocrine Society Clinical Practice Guidelines rec-

ommends a prescribed dosage of 50,000 IU Vitamin D2per week for 8 weeks as a safe, effective and well toleratedmethod to correct vitamin D deficiency (45). We concludethat the strategy used in this PCT, dosing commerciallyavailable vitamin D3 dietary supplements at 10,000 IU/dfor 12 weeks, is another safe, effective, and well toleratedstrategy for clinical vitamin D3 repletion in people withsuboptimal 25(OH)D. There remains inherent variabilityin the manufacture and acceptable potency in D3 dietarysupplements based on current regulatory frameworks(26–28); therefore, periodic clinical monitoring is recom-mended when providers and/or their patients use D3 di-etary supplements in high-dose strategies to increase se-rum 25(OH)D concentration.

Acknowledgments

Author Contributions: Study concept and design: Traub, Brad-ley, Finnell, Oberg

Acquisition of Data: Traub, Suhaila, Bhandiwad, FinnellAnalysis and interpretation of data: Bradley, FinnellDrafting of the manuscript: Traub, Bradley, FinnellCritical revision of the manuscript for important intellectual

content: Traub, Bradley, Finnell, ObergStatistical analysis: Bradley, FinnellObtained funding: Bradley, Finnell; Traub (donations of

products)Administrative, technical, or material support: Traub, Su-

haila, Bradley, FinnellStudy supervision: Traub, Bradley, FinnellFinal approval of the article: Traub, Bradley, FinnellConflict of Interest Disclosure: All authors have completed

and submitted forms for Disclosure of Potential Conflicts of In-terest. No conflicts of interest were disclosed.

Funding Support: NIH NCCCAM grant T32AT00815; Na-tional Center for Research Resources grant 1KL2RR025015–01; Diabetes Action Research and Education Foundation; BastyrUniversity; Flora Research Laboratories

Role of the Sponsor: This was an investigator-initiated studythat was funded by the aforementioned entities and led by theco-PIs. Drs Bradley and Finnell independently performed all sta-tistical analyses and resolved any discrepancies.

Additional Contributions: The authors thank all productmanufacturers for providing the vitamin D; Pacific PhysicianLaboratories and Flora Research Laboratories for analyticaltesting; SolRx and MyChelle Dermaceuticals for providing sun-block; Michael D. Levin, Health Business Strategies, for expertassistance. AB and LS would like to acknowledge the Bastyr

University Research Institute and Lokahi Health Center, respec-tively, for hosting them during the conduct of this trial.

As corresponding author, I, Michael Traub, ND, affirm thatI have listed everyone who contributed significantly to the workof this study.

Address all correspondence and requests for reprints to: Mi-chael L. Traub, ND, Email: mtraubnd@me.com, 75–169 Hua-lalai Road, Suite 301, Kailua Kona, HI 96740, Phone: 808–329-2114, Fax: 808–326-2871.

This work was supported by Grants supporting the writing ofthe paper: National Institutes of Health National Center of Com-plementary and Alternative Medicine (NCCAM) grantT32AT00815 and National Center for Research Resources(NCRR) grant 1KL2RR025015–01. Additional support wasprovided by: the Diabetes Action Research and EducationFoundation.

Disclosure Summary: JF, AB, EO, LS, and RB have nothing todeclare. MT is a consultant for Nordic Naturals Inc., NutritionalFundamentals for Health Inc., and Kamedis, Inc.

Clinical Trial Registration Number: NCT01524874

References

1. Hypponen, E., Laara, E., Reunanen, A., Jarvelin, M.R., and Vir-tanen, S.M. Intake of vitamin D and risk of type 1 diabetes: a birth-cohort study. Lancet. 2001;358:1500–1503.

2. Pittas, A.G., Dawson-Hughes, B., Li, T., Van Dam, R.M., Willett,W.C., Manson, J.E., and Hu, F.B. Vitamin D and calcium intake inrelation to type 2 diabetes in women. Diabetes Care. 2006;29:650–656.

3. Levin, A., Bakris, G.L., Molitch, M., Smulders, M., Tian, J., Wil-liams, L.A., and Andress, D.L. Prevalence of abnormal serum vita-min D, PTH, calcium, and phosphorus in patients with chronic kid-ney disease: results of the study to evaluate early kidney disease.Kidney international. 2007;71:31–38.

4. Taskapan, H., Wei, M., and Oreopoulos, D.G. 25(OH) vitamin D3in patients with chronic kidney disease and those on dialysis: redis-covering its importance. International urology and nephrology.2006;38:323–329.

5. Holick, M.F. Sunlight and vitamin D for bone health and preventionof autoimmune diseases, cancers, and cardiovascular disease. TheAmerican journal of clinical nutrition. 2004;80:1678S–1688S.

6. Giovannucci, E. The epidemiology of vitamin D and cancer inci-dence and mortality: a review (United States). Cancer causes, control: CCC. 2005;16:83–95.

7. Pilz, S., Tomaschitz, A., Obermayer-Pietsch, B., Dobnig, H., andPieber, T.R. Epidemiology of vitamin D insufficiency and cancermortality. Anticancer research. 2009;29:3699–3704.

8. Giovannucci, E. Epidemiology of vitamin D and colorectal cancer:casual or causal link? The Journal of steroid biochemistry and mo-lecular biology. 2010;121:349–354.

9. Wei, M.Y., Garland, C.F., Gorham, E.D., Mohr, S.B., and Giovan-nucci, E. Vitamin D and prevention of colorectal adenoma: a meta-analysis. Cancer epidemiology, biomarkers, prevention : a publica-tion of the American Association for Cancer Research, cosponsoredby the American Society of Preventive Oncology. 2008;17:2958–2969.

10. Autier, P., and Gandini, S. Vitamin D supplementation and totalmortality: a meta-analysis of randomized controlled trials. Archivesof internal medicine. 2007;167:1730–1737.

11. Melamed, M.L., Michos, E.D., Post, W., and Astor, B. 25-hy-

8 Response to Vitamin D3 Supplement Matrix J Clin Endocrinol Metab

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

droxyvitamin D levels and the risk of mortality in the general pop-ulation. Archives of internal medicine. 2008;168:1629–1637.

12. Ross, A.C., Manson, J.E., Abrams, S.A., Aloia, J.F., Brannon, P.M.,Clinton, S.K., Durazo-Arvizu, R.A., Gallagher, J.C., Gallo, R.L.,Jones, G., et al. The 2011 report on dietary reference intakes forcalcium and vitamin D from the Institute of Medicine: what clini-cians need to know. The Journal of clinical endocrinology and me-tabolism. 2011;96:53–58.

13. Brannon, P.M., Yetley, E.A., Bailey, R.L., and Picciano, M.F. Over-view of the conference “Vitamin D and Health in the 21st Century:an Update”. The American journal of clinical nutrition. 2008;88:483S–490S.

14. Medicine, I.o. Dietary reference intakes for calcium and vitamin D.Washington D.C.: Institute of Medicine. 2011;

15. Aloia, J.F., Patel, M., Dimaano, R., Li-Ng, M., Talwar, S.A.,Mikhail, M., Pollack, S., and Yeh, J.K. Vitamin D intake to attain adesired serum 25-hydroxyvitamin D concentration. The Americanjournal of clinical nutrition. 2008;87:1952–1958.

16. Talwar, S.A., Aloia, J.F., Pollack, S., and Yeh, J.K. Dose response tovitamin D supplementation among postmenopausal African Amer-ican women. The American journal of clinical nutrition. 2007;86:1657–1662.

17. Bischoff-Ferrari, H.A., Giovannucci, E., Willett, W.C., Dietrich, T.,and Dawson-Hughes, B. Estimation of optimal serum concentra-tions of 25-hydroxyvitamin D for multiple health outcomes. TheAmerican journal of clinical nutrition. 2006;84:18–28.

18. Vieth, R., Ladak, Y., and Walfish, P.G. Age-related changes in the25-hydroxyvitamin D versus parathyroid hormone relationshipsuggest a different reason why older adults require more vitamin D.The Journal of clinical endocrinology and metabolism. 2003;88:185–191.

19. Vieth, R. Vitamin D supplementation, 25-hydroxyvitamin D con-centrations, and safety. The American journal of clinical nutrition.1999;69:842–856.

20. Giovannucci, E., Liu, Y., Hollis, B.W., and Rimm, E.B. 25-hy-droxyvitamin D and risk of myocardial infarction in men: a pro-spective study. Archives of internal medicine. 2008;168:1174–1180.

21. Heaney, R.P., Davies, K.M., Chen, T.C., Holick, M.F., and Barger-Lux, M.J. Human serum 25-hydroxycholecalciferol response to ex-tended oral dosing with cholecalciferol. The American journal ofclinical nutrition. 2003;77:204–210.

22. Chel, V., Wijnhoven, H.A., Smit, J.H., Ooms, M., and Lips, P. Ef-ficacy of different doses and time intervals of oral vitamin D sup-plementation with or without calcium in elderly nursing home res-idents. Osteoporosis international : a journal established as result ofcooperation between the European Foundation for Osteoporosisand the National Osteoporosis Foundation of the USA. 2008;19:663–671.

23. Pietras, S.M., Obayan, B.K., Cai, M.H., and Holick, M.F. VitaminD2 treatment for vitamin D deficiency and insufficiency for up to 6years. Archives of internal medicine. 2009;169:1806–1808.

24. Saadi, H.F., Dawodu, A., Afandi, B.O., Zayed, R., Benedict, S., andNagelkerke, N. Efficacy of daily and monthly high-dose calciferol invitamin D-deficient nulliparous and lactating women. The Ameri-can journal of clinical nutrition. 2007;85:1565–1571.

25. Leventis, P., and Kiely, P.D. The tolerability and biochemical effectsof high-dose bolus vitamin D2 and D3 supplementation in patientswith vitamin D insufficiency. Scandinavian journal of rheumatol-ogy. 2009;38:149–153.

26. United States Pharmacopeial Convention. USP 35-NF 30: DietarySupplements Compendium. Rockville, MD: United States Pharma-copeial Convention. 2012;1468–1510 pp.

27. Miller, R.K., Celestino, C., Giancaspro, G.I., and Williams, R.L.FDA’s dietary supplement CGMPs: standards without standardiza-tion. Food and drug law journal. 2008;63:929–942.

28. FDA, U.S. Code of Federal Regulations. In Title 21, Volume 2, Part101.9. D.o.H.a.H. Services, editor. 2012;

29. Ilahi, M., Armas, L.A., and Heaney, R.P. Pharmacokinetics of asingle, large dose of cholecalciferol. The American journal of clinicalnutrition. 2008;87:688–691.

30. Gough, H., Goggin, T., Bissessar, A., Baker, M., Crowley, M., andCallaghan, N. A comparative study of the relative influence of dif-ferent anticonvulsant drugs, UV exposure and diet on vitamin D andcalcium metabolism in out-patients with epilepsy. The Quarterlyjournal of medicine. 1986;59:569–577.

31. Taylor, C., Lamparello, B., Kruczek, K., Anderson, E.J., Hubbard,J., and Misra, M. Validation of a food frequency questionnaire fordetermining calcium and vitamin D intake by adolescent girls withanorexia nervosa. J Am Diet Assoc 2009;109:479–485, 485 e471–473.

32. Douglass, J.P.J. Bowes and Church’s Food Values of Portions Com-monly Used. Philadelphia: Lippincott Williams, Wilkins. 2004;

33. Nutrient-Data-Laboratory. USDA Nutrient Database for StandardReference, Release 16. A.R.S. U.S. Department of Agriculture, ed-itor. 2003;

34. Glanz, K., Yaroch, A.L., Dancel, M., Saraiya, M., Crane, L.A.,Buller, D.B., Manne, S., O’Riordan, D.L., Heckman, C.J., Hay, J.,et al. Measures of sun exposure and sun protection practices forbehavioral and epidemiologic research. Archives of dermatology.2008;144:217–222.

35. Ersfeld, D.L., Rao, D.S., Body, J.J., Sackrison, J.L., Jr., Miller, A.B.,Parikh, N., Eskridge, T.L., Polinske, A., Olson, G.T., and MacFar-lane, G.D. Analytical and clinical validation of the 25 OH vitaminD assay for the LIAISON automated analyzer. Clinical biochemis-try. 2004;37:867–874.

36. Harris, P.A., Taylor, R., Thielke, R., Payne, J., Gonzalez, N., andConde, J.G. Research electronic data capture (REDCap)–a meta-data-driven methodology and workflow process for providingtranslational research informatics support. Journal of biomedicalinformatics. 2009;42:377–381.

37. Zhao, L.J., Zhou, Y., Bu, F., Travers-Gustafson, D., Ye, A., Xu, X.,Hamm, L., Gorsage, D.M., Fang, X., Deng, H.W., et al. Factorspredicting vitamin D response variation in non-Hispanic white post-menopausal women. The Journal of clinical endocrinology and me-tabolism. 2012;97:2699–2705.

38. Wortsman, J., Matsuoka, L.Y., Chen, T.C., Lu, Z., and Holick,M.F. Decreased bioavailability of vitamin D in obesity. The Amer-ican journal of clinical nutrition. 2000;72:690–693.

39. Elnenaei, M.O., Chandra, R., Mangion, T., and Moniz, C. Genomicand metabolomic patterns segregate with responses to calcium andvitamin D supplementation. Br J Nutr. 2011;105:71–79.

40. Larcombe, L., Mookherjee, N., Slater, J., Slivinski, C., Singer, M.,Whaley, C., Denechezhe, L., Matyas, S., Turner-Brannen, E., Nick-erson, P., et al. Vitamin D in a northern Canadian first nation pop-ulation: dietary intake, serum concentrations and functional genepolymorphisms. PLoS One. 2012;7:e49872.

41. Muindi, J.R., Adjei, A.A., Wu, Z.R., Olson, I., Huang, H., Groman,A., Tian, L., Singh, P.K., Sucheston, L.E., Johnson, C.S., et al. Serumvitamin D metabolites in colorectal cancer patients receiving chole-calciferol supplementation: correlation with polymorphisms in thevitamin D genes. Horm Cancer. 2013;4:242–250.

42. Zhou, Y., Zhao, L.J., Xu, X., Ye, A., Travers-Gustafson, D., Zhou,B., Wang, H.W., Zhang, W., Lee Hamm, L., Deng, H.W., et al. DNAmethylation levels of CYP2R1 and CYP24A1 predict vitamin Dresponse variation. The Journal of steroid biochemistry and molec-ular biology. 2013;

43. Phinney, K.W. Development of a standard reference material forvitamin D in serum. The American journal of clinical nutrition.2008;88:511S–512S.

44. LeBlanc Es, P.N.J.J.D.B.A.H.T. Over-the-counter and com-pounded vitamin d: Is potency what we expect? JAMA InternalMedicine. 2013;1–2.

45. Holick, M.F., Binkley, NC, Bischoff-Ferrari, H.A., Gordon, C.M.,Hanley, D.A., Heaney, R.P., Murad, M.H., and Weaver, C.M. Eval-uation, treatment, and prevention of vitamin D deficiency: an En-

doi: 10.1210/jc.2013-3162 jcem.endojournals.org 9

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.

docrine Society clinical practice guideline. The Journal of clinicalendocrinology and metabolism. 2011;96:1911–1930.

10 Response to Vitamin D3 Supplement Matrix J Clin Endocrinol Metab

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 04 April 2014. at 12:31 For personal use only. No other uses without permission. . All rights reserved.