Calcified Tissue International

, Volume 92, Issue 2, pp 191–206

Is High Dose Vitamin D Harmful?

Authors

    • Department of Medicine, Western HealthNorthWest Academic Centre, University of Melbourne
  • Geoffrey C. Nicholson
    • Department of Medicine, Western HealthNorthWest Academic Centre, University of Melbourne
    • Rural Clinical School, School of MedicineUniversity of Queensland
  • Peter R. Ebeling
    • Department of Medicine, Western HealthNorthWest Academic Centre, University of Melbourne
Original Research

DOI: 10.1007/s00223-012-9679-1

Cite this article as:
Sanders, K.M., Nicholson, G.C. & Ebeling, P.R. Calcif Tissue Int (2013) 92: 191. doi:10.1007/s00223-012-9679-1

Abstract

With the potential to minimize the risk of many chronic diseases and the apparent biochemical safety of ingesting doses of oral vitamin D several-fold higher than the current recommended intakes, recent research has focussed on supplementing individuals with intermittent, high-dose vitamin D. However, two recent randomized controlled trials (RCTs) both using annual high-dose vitamin D reported an increase, rather than a decrease, in the primary outcome of fractures. This review summarises the results from studies that have used intermittent, high doses of vitamin D, with particular attention to those finding evidence of adverse effects. Results from observational, population-based studies with evidence of a U- or J-shaped curve are also presented as these findings suggest an increased risk in those with the highest serum 25D levels. Speculative mechanisms are discussed and biochemical results from studies using high-dose vitamin D are also presented. Emerging evidence from both observational studies and RCTs suggests there should be a degree of caution about recommending high serum 25D concentrations for the entire population. Furthermore, benefit of the higher doses commonly used in clinical practice on falls risk reduction needs to be demonstrated. The safety of loading doses of vitamin D should be demonstrated before these regimens become recommended as routine clinical practice. The current dilemma of defining vitamin D insufficiency and identifying safe and efficacious repletion regimens needs to be resolved.

Keywords

Steroid hormonesVitamin DOsteoporosisFracturesAgeAging

The need for vitamin D supplementation has evolved because it is widely recognized that a significant proportion of many populations has inadequate vitamin D status [1]. Serum 25-hydroxyvitamin D (25[OH]D) levels of 25 nmol/L or less are considered “deficient,” while the definition of “insufficiency” varies, with some experts regarding less than 50 nmol/L as the cut-off [2] and others considering less than 75 nmol/L as “insufficient” [35]. “Insufficient” in this review refers to 25(OH)D levels in the range 25–50 nmol/L, “intermittent” dosing refers to at least 1-week dosing intervals, and “high dose” refers to an intermittent bolus dose of at least 20,000 IU or a daily dose of 4,000 IU. Although the risk of many chronic disorders may be reduced by an upward shift in the community’s vitamin D status, daily dosing has proven to be problematic, particularly for older people, the group most likely to directly benefit from an improvement in vitamin D status [6]. Many randomized trials have reported poor compliance with daily regimens [7]. Furthermore, some people require substantial doses of vitamin D to achieve serum 25(OH)D levels within the target range [8, 9]. An intermittent, larger dose of vitamin D reduces this compliance issue in a simple and cost-effective manner and reduces the likelihood that the target group will remain below the threshold of 25(OH)D regarded as “insufficient,” although significant controversy exists regarding what level of serum 25(OH)D is “sufficient” [3, 1012]. Both the public and practitioners want to make informed decisions regarding both the target level of 25(OH)D to optimize health and the appropriate dosing regimen to achieve this target level.

According to current evidence from biochemical, observational, and randomized controlled trials (RCTs), serum 25(OH)D levels of at least 50 nmol/L are required for normalization of PTH levels, to minimize the risk of osteomalacia, and for optimal bone and muscle function [2, 13, 14], with many experts regarding 75 nmol/L as the threshold for optimal bone health [12, 1517]. The skeletal consequences of 25(OH)D insufficiency include secondary hyperparathyroidism, increased bone turnover and bone loss, and increased risk of low-trauma fractures. From a skeletal perspective, evidence from RCTs suggests that vitamin D may be considered a threshold nutrient with little bone benefit observed at levels of 25(OH)D above that at which parathyroid hormone (PTH) is normalized [2]. However, molecular studies have demonstrated that vitamin D plays a role in cell differentiation, function, and survival [18, 19]. Adequate calcium intake is imperative to gain optimal benefit from improving vitamin D status in those with insufficient 25(OH)D levels. The relative contributions of vitamin D and calcium to reducing fracture risk remain unclear [20], and improving calcium intake is also associated with suppression of PTH levels [21, 22]. Observational studies have shown a decreased risk of many disorders, including certain types of cancer, mental disorders, cardiovascular disease, and skin and autoimmune disorders, associated with serum 25(OH)D levels greater than 70–80 nmol/L [9, 12, 16]. It has therefore been argued that 25(OH)D levels should be in the range of 70–100 nmol/L to maximize these nonskeletal benefits.

With the potential to minimize the risk of many chronic diseases and the apparent safety of ingesting doses of oral vitamin D severalfold higher than the current recommended intakes, recent research has focused on supplementing individuals with intermittent, high-dose vitamin D. There is an urgent need to determine the efficacy and safety of these regimens. Using biochemical parameters of safety, particularly plasma and urine calcium, there are numerous studies reporting that a single oral dose of 300,000–600,000 IU of D2/D3 rapidly enhances serum 25(OH)D and reduces PTH in people with deficiency [2325]. Although dosing intervals of greater than 2–3 months and/or intermittent bolus doses (>200,000 IU) are not regarded as physiological [26], such an approach offers a realistic and pragmatic public health measure to target at-risk populations and addresses the emerging public health issue of widespread vitamin D insufficiency [6, 27].

However, two recent RCTs, both using annual high-dose vitamin D, reported an increase, rather than a decrease, in the primary outcome of falls [27] and fractures [27, 28]. These findings highlight the need for a better understanding of different dosing regimens before pragmatic, population-based interventions are implemented. This review summarizes results from studies that have used intermittent, high doses of vitamin D, with particular attention to those finding evidence of adverse effects. Results from observational, population-based studies with evidence of a U- or J-shaped curve are also presented as these findings suggest an increased risk in those with the highest serum 25(OH)D levels. We do not attempt to present a comprehensive, systematic review of the extensive number of observational and intervention studies reporting the health benefits of improving vitamin D status in adults with insufficient status [7, 2932] or to present a “balanced” view of the potential risk/benefit ratio of vitamin D supplementation.

Biochemical Outcomes of Single, Large Doses of Vitamin D

The immediate concern of hypervitaminosis D is hypercalciuria and hypercalcemia [2]. However, a large therapeutic window exists for vitamin D–related hypercalcemia, which has not been reported at serum 25(OH)D levels below 220 nmol/L and generally not reported below 500 nmol/L [2]. Based on these biochemical parameters, Vieth and colleagues [17] conducted a 6-month safety and efficacy study and concluded that consumption of more than 4,000 IU/day causes no harm and effectively raises 25(OH)D levels to “high-normal” concentrations (<140 nmol/L) in practically all adults. The 2011 Institute of Medicine report on dietary intake of vitamin D recommended an upper limit of 4,000 IU/day, although it also stated that up to 10,000 IU/day is safe [33].

Of the studies included in this review, the cases of hypercalcemia and/or hypercalciuria are few and their incidence in the randomized trials is rarely different from that observed in the placebo group (Table 1). The study by Grimnes and colleagues [34], where one group was given 6,500 IU/day and another was given 400 IU/day, reported a significant difference between groups in serum ionized calcium at 12 months. In another trial, two of 33 patients receiving a single bolus dose of 300,000 IU vitamin D3 had mild hypercalcemia [35]. Participants in this study were older, recruited from a rheumatology clinic, and likely to have reduced kidney function compared with other trial participants who were, on average, younger (Table 1). However, some of the larger trials did not specifically investigate for hypercalcemia or hypercalciuria (Table 1).
Table 1

Biochemical outcomes of studies using intermittent high-dose vitamin D

Reference

Dosing regimen

Serum 25(OH)D profile

Vitamin D2 or D3

Dose (IU) oral or IM

Regimen for dosing

25(OH)D at baseline

Peak 25(OH)D

Time point at peak 25(OH)D

25(OH)D end of study

Time point end of study

Assay for 25(OH)D

Diamond et al. [25]

D3

600,000

IM, single bolus

32 ± 8

114 ± 35

4 months

73 ± 13

12 months

RIA IDS

Bacon et al. [24]

D3

3 groups:

(1) 500,000

(2) 500,000 ± 50,000/month

(3) 50,000

Oral:

(1) single bolus

(2) bolus ± monthly

(3) monthly

58 ± 25

130

(SD not given)

1 month

69 ± 5

Group 1,

91 ± 4

Group 2,

80 ± 5

Group 3

9 months

RIA

DiaSorin

Premaor et al. [70]

D3

300,000

group 1, 800

group 2

Oral: ±500calcium daily;

group 1: single bolus

group 2: daily

27 ± 15

80 in single-bolus dose group 48 in daily dose (SD not given)

1-month peak in bolus group, 2-month peak in daily group

30

9 months

LIAISON

Chemiluminescence

DiaSorin

Ilahi et al. [47]

D3

100,000

Oral: single bolus

68 ± 19

105 ± 23

Day 9

74 ± 10

4 months

RIA IDS

Smith et al. [28]

D2

300,000

IM: once a year

57 ± 24

Reported 21 % increase but only the 36 % increase in 1,25(OH)D was significantly different

4 months

N/A: by 9 months postdose 25(OH)D but not 1,25(OH)D “drifted toward basal values”

3 years

RIA

Nicholls

Sanders et al. [27]

D3

500,000

Oral: once a year

49 (IQR 40–63)

120

1 month

74 ± 20

3–5 years

RIA

DiaSorin

Cipriani et al. [23]

D3

600,000

Oral: single bolus

40 ± 16

190

Day 3

156 ± 65

30 days

RIA

DiaSorin

Rossini et al. [48]

D3

600,000

300,000

100,000

Oral: single bolus

60 ± 17

167 ± 43 in 600,000 IU group

Day 3

88 ± 14

Day 90

ELISA

IDS

Von Restorff et al. [35]

D3

300,000

Oral: single bolus

15 ± 6

81 ± 30

3 months

69 ± 18

6 months

RIA

DiaSorin

Pekkarinen et al. [71]

D3

97,333 group 1,

800 group 2

Oral: all had 1 g calcium/day, group 1: 4-monthly dose, group 2: daily dose

55 (range 17–84)

72 in 4-monthly group,

78 in daily dose group

10 months in 4-monthly group,

3 months in daily group

74 in 4-monthly group, 62 in daily dose group

12 months

HPLC

Trivedi et al. [46]

D3

100,000

Oral: 4 monthly

N/A

N/A

N/A

74 nmol/L in vitamin D group 3 weeks postdose, 53 nmol/L in placebo group

4th year of 5-year study

Not stated

Giusti et al. [8]

D3

300,000 group 1,

1,000 group 2

Oral: group 13-monthly,

group 2: daily dose

23 ± 12

Both groups had ? higher 25(OH)D at 6 months compared to 3 months

 

58 ± 19 in 1,000 IU/day group 80 ± 30 in 300,000 months group mean ± SD IU/3

6 months

RIA

DiaSorin

Jorde et al. [72]

D3

3 groups: D3 400,000 group 1, 20,000 group 2, placebo group 3

Oral: weekly dosing, all given calcium 500 mg/day

53 (range 11–112)

81 ± 30

3 months

Group 1: 112,

group 2: 88,

group 3: 50

12 months

Chemiluminescence

Modular E170

Grimnes et al. [34]

D3

6,500 group 1,

800 group 2

Oral:

daily doses

71 ± 23

N/A

N/A

Group 1: 185

(34),

group 2: 89 (17)

12 months

RIA DiaSorin

Vieth et al. [73]

D3

4,000

group 1,

600

group 2

Oral:

daily doses

48 ± 9 39 ± 9

(two

recruitment periods)

112 ± 41

4,000 IU group, 79 ± 30

600 IU group

~6 months (winter)

N/A

≤15 months

RIA DiaSorin

Heaney et al. [37]

D3

4 groups: placebo, 1,000, 5,000, 10,000

Oral: daily doses

70

200 in 10,000 IU group, 140 in 5,000 IU group (SD N/A)

120 days

Approximately the same as at 120 days

4 months

~IDS method

Law et al. [74]

D2

100,000

Oral:

3 monthly

47 (35–102:

90th percentile)

82 (67–185)

5th/95th percentiles)

1 month

N/A

10 months

ELISA IDS

Aloia et al. [75]

D3

4,000 (± 1,200 mg calcium)

Oral: daily doses

66 (24)

N/A

N/A

111 (30)

3 months

RIA DiaSorin

Reference

Participant characteristics

Outcomes

Number of participants

Age (years)

Gender

Primary outcome

Hypercalcemia reported (yes/no)

Proportion of hypercalcemia

Diamond et al. [25]

50

66, range 32–87

F, M

Serum 25(OH)D status

Yes

Mild hypercalcemia (<2.70 nmol/L) in 2/50, hyperparathyroidism in 1/50

Bacon et al. [24]

63

82 ± 7

F, M

Serum 25(OH)D and PTH and bone formation marker P1NP

Yes

0/63

Premaor et al. [70]

28

81 ± 9

F, M

Serum 25(OH)D and reversion of secondary hyperparathyroidism, higher bolus dose more effective in the short term

Yes

0/28

Ilahi et al. [47]

30

Range 61–84 group 1, 27–47 group 2

F, M

Serum 25(OH)D recommended dosing interval ≤2 months to ensure continuous 25(OH)D above baseline

Yes

0/30

Smith et al. [28]

43

79 (IQR 77–83)

F, M

Fractures combined F/M no difference between groups

No

N/A

Sanders et al. [27]

137 (n = 75 from vitamin D group and n = 58 placebo group)

76 (IQR 73–80)

F

Falls & fracture15 % increased falls and 26 % increased fractures in vitamin D groups

Yes

Specifically assessed in subgroup only

0/137 with biochemistry, 0/2, 256 reported in entire group

Cipriani et al. [23]

48

36 ± 8 (range 25–56)

F, M

Serum 25(OH)D and calcium metabolism

Yes

0/48

Rossini et al. [48]

37

75 ± 3

F, M

BTM: CTX in 600,000 IU and 300,000 IU groups, BAP unchanged in these groups but increased in 100,000 IU group

Yes

0/37

Von Restorff et al. [35]

33

81 ± 6

F, M

Serum 25(OH)D

Yes

N = 2/33 at 3 months

Pekkarinen et al. [71]

40

74 (range 69–79)

F

Serum 25(OH)D and bone formation marker (P1NP), P1NP decreased in both groups at 12 months

Yes

n = 0/40

Trivedi et al. [46]

n = 124 with biochemistry in vitamin D group

75 ± 5

F, M

Fracture rate and total mortality 22 % reduction in fractures in vitamin D group vs. placebo

No

N/A

Giusti et al. [8]

60

73 ± 5

F

Serum 25(OH)D and PTH

Yes

0/60 Mean corrected serum calcium significantly increased at 6 months

Jorde et al. [72]

441

21–70 range (median N/A)

F, M

Beck Depression Inventory score

Yes

5/441 (n = 1 placebo, n = 3 group 1 and n = 1 group 2)

Grimnes et al. [34]

297

50–80

F

BMD and bone turnover markers, no difference between groups in change in BMD

Yes

Modest hypercalcemia 9/149 in high-dose group, 4/148 in standard-dose group

Vieth et al. [73]

64

54 ± 11y

F, M

Serum 25(OH)D and PTH and measures of well-being (seasonal health questionnaire by Thomson and Cowan)

Yes

0/64

Heaney et al. [37]

67

39 ± 11

M

Serum 25(OH)D

No

N/A

Law et al. [74]

18/3,717

85  (SD N/A)

F, M

Falls and fractures: no difference between vitamin D and control groups

Yes

0/118

Aloia et al. [75]

35 (on vitamin D)

55 (20–80)

F, M

Parathyroid hormone and BTM, BTM increased in vitamin D alone group

Yes

0/35

Vitamin D doses expressed in international units (IU). D2 vitamin D2 (ergocalciferol), D3 vitamin D3 (cholecalciferol), IM intramuscular injection, 25(OH)D serum 25-hydroxyvitamin D levels (nmol/L), 1,25(OH)D serum 1,25-dihydroxyvitamin D (levels quoted as mean and standard deviation unless stated otherwise), IQR interquartile range, N/A not available, RIA radioimmunoassay, HPLC high-pressure liquid chromatography, BTM bone turnover markers, CTX C-terminal telopeptides (bone resorption marker), BAP bone-specific alkaline phosphatase, P1NP amino-terminal propeptide of type 1 procollagen (bone formation markers), F female, M male

A 1995 review of the safety and effectiveness of different regimes of vitamin D supplementation in the elderly suggested that daily low-dose supplementation is the regimen of choice for prevention of hypovitaminosis D but that intermittent high-dose regimens would be a safe and effective alternative in patients with poor compliance [36]. As reported in individual trials [8, 9, 37], there appears to be significant variation in the level of serum 25(OH)D reached when individuals are given the same dose and form of the vitamin.

There is a caveat that earlier studies were reliant on 25(OH)D assays that have shown considerable intra- and intersample variation in the assessment of serum 25(OH)D concentrations and were unreliable in measuring serum D2 levels [38]. Although the performance of radioimmunoassay and enzyme-linked assays is acceptable, the bias and imprecision of many automated methods may be problematic at the lower, clinically and analytically important range (<50 nmol/L) of the assay [2].

Evidence from Observational Studies

A majority of observational studies have reported that vitamin D is associated with a beneficial effect on risk of colon, breast, prostate, and ovarian cancers [39]. Since vitamin D synthesis and serum 25(OH)D levels are inversely correlated with latitude and positively correlated with sunlight, some studies have “mapped” disease incidence rates with latitude to investigate a possible protective effect of vitamin D status and risk of disease. Encouragingly, there is a consistency of findings between geographic studies and “serum” studies where samples of the population have had biochemistry assessments, ideally with the blood collection point several years prior to any diagnosis of cancer or other disease of interest [39]. Vitamin D and its metabolites are thought to reduce the incidence of many types of cancer by inhibiting tumor angiogenesis and hyperproliferation as well as stimulating cellular apoptosis [40]. Since vitamin D regulates a gamut of physiological processes, including immune modulation, resistance to oxidative stress, and modulation of other hormones, it is not surprising that low vitamin D has been associated with increased risk of several cancers and chronic diseases [41] as well as cancer mortality [42]. Nevertheless, there are now several observational studies reporting a U- or J-shaped association between disease and serum 25(OH)D and latitude and/or ultraviolet B radiation levels, where those in the highest percentiles have an inverse risk compared with those in the lowest (Table 2) [43]. While cross-sectional data have many limitations, the findings are hypothesis-generating [44] and can be used to develop protocols for RCTs. The findings from prospective case–control cohort studies where blood collection occurred many years prior to diagnosis add another dimension to the evidence. The results from these studies generally support vitamin D supplementation in those with “low” vitamin D status. However, the findings argue for caution before increasing 25(OH)D levels and associated dosing regimens beyond evidence clearly supported by RCTs and meta-analyses [45].
Table 2

Observational studies with evidence of harm at higher end of vitamin D status

Reference

Study design

Population characteristics

Outcome

Methodology for 25(OH)D

Comments

Ensrud et al. [53]

Observational: frailty categorized as robust, intermediate, or frail

N = 6,307 women aged >69 years, Caucasian, USA (SOF)

Odds of frailty in those 25(OH)D >75 nmol/L compared to referent of 50–74 nmol/L group, odds ratio 1.32 (1.06–1.63)

LC–MS/MS method for D3

As supplements during study period 1992–1998 were all vitamin D2 not D3, the influence of supplements was not included in vitamin D status

Chen et al. [76]

Case–cohort study nested within the General Population Trial of Linxian, China

Measured pretrial serum 25(OH)D levels in 979 cases and 1,105 cohort participants, cases were diagnosed with cancer during 5.25 years of follow-up

Esophageal and gastric cancers, in men there was an increased risk of developing esophageal cancer for those in the highest (4th) quartile of 25(OH)D level at baseline: HR 1.77 (1.16–2.70), p = 0.0033

Enzyme immunoassay (IDS)

Prospectively examines the relationship between pretrial 25(OH)D status and risk of developing esophageal or gastric cancer, no association found in women

Michaelsson et al. [54]

Uppsala Longitudinal Study of adult men

n = 1,194 Swedish men, 12.7 years follow-up, 49 % died

Both low (<46 nmol/L) and high (>98 nmol/L) serum 25(OH)D associated with increased total mortality and cancer risk, for higher 25(OH)D risk HR (95 % CI): total mortality 1.67 (1.12–2.49) cancer mortality 2.64 (1.46–4.78)

HPLC mass spectrometry

 

Ahn et al. [77]

Prospective case–control study nested within the Prostate, Lung, Colorectal and Ovarian Cancer Screening Trial

n = 749 cases and n = 781 controls, blood used for serum 25(OH)D level was drawn at least 8 years prior to diagnosis of cancer

“Higher serum 25(OH)D status was not associated with lower risk of prostate cancer; indeed higher 25(OH)D levels may be associated with an increased risk of aggressive cancer”

RIA

 

Tuohimaa et al. [78]

Longitudinal nested case–control study, data from men who donated blood samples stored in biobanks in Finland, Norway, and Sweden

n = 622 cases and 1,451 matched controls

Both low (≤19 nmol/L) and high ≥80 nmol/L serum 25(OH)D associated with higher prostate cancer risk

RIA (Incstar)

The authors suggest that a high vitamin D level might lead to vitamin D resistance through increased inactivation via enhanced expression of 24-hydroxylase

Stolzenberg-Solomon et al. [79]

Pancreatic cancer risk

200 cases and 400 matched controls, nested case–control using prediagnostic 25(OH)D male smokers aged 50–69 years

Higher vitamin D status associated with 3-fold increased risk for pancreatic cancer, odds ratio (95 % CI) 2.9 (1.6–5.5), >65 vs. <32 nmol/L groups

RIA (DiaSorin)

Smoking duration and smoking intensity did not differ between cases and controls

Stolzenberg-Solomon et al. [80]

Pancreatic cancer risk within the Prostate, Lung, Colorectal and Ovarian Screening Trial

n = 184 cases and 2 to 1 matched control to case, nested case–control study using prediagnostic serum 25(OH)D, men and women aged 55–74 years

25(OH)D levels were not associated with risk of pancreatic cancer overall, report positive associations in those with low solar UBV exposure

RIA

Did not confirm the strong positive findings observed in the earlier study in male Finnish smokers

McGrath et al. [55]

Used Danish national registers and neonatal biobank

n = 424 individuals with schizophrenia, 424 matched without schizophrenia

Both low and high levels of neonatal vitamin D status associated with increased risk of developing schizophrenia

LS-MS

U-shaped curve was not explained by a wide variety of variables

Vitamin D doses expressed in international units (IU). D2 vitamin D2 (ergocalciferol), D3 vitamin D3 (cholecalciferol), IM intramuscular injection, 25(OH)D serum 25-hydroxyvitamin D (nmol/L), 1,25(OH)D serum 1,25-dihydroxyvitamin D (levels quoted as mean and standard deviation unless stated otherwise), IQR interquartile range, N/A not available, RIA radioimmunoassay, HPLC high-pressure liquid chromatography, LCMS/MS liquid chromatography coupled with tandem mass spectrometry, HR hazard ratio (95 % confidence interval), RR relative risk (95 % confidence interval), BTM bone turnover markers, CTX C-terminal-telopeptide (bone resorption marker), BAP bone-specific alkaline phosphatase, P1NP amino-terminal propeptide of type 1 procollagen (bone formation markers), F females, M males

RCTs Demonstrating Harm

The evidence of harm relating to high-dose vitamin D centers on the findings of two RCTs that used annual high-dose vitamin D (Table 3), although results from RCTs using lower, more frequent dosing regimens have not been consistently clear. The different forms of the vitamin used in the studies and the different delivery modes demonstrate that the adverse outcomes are not restricted to one form of the vitamin. Neither study included calcium supplementation as part of the protocol. In the British “Wessex” study, 9,440 community-dwelling participants (4,354 men and 5,086 women) aged 75–100 years were randomly allocated to receive an annual injection of 300,000 IU vitamin D2 or matching placebo every autumn over 3 years [28]. In the entire cohort the risk of any first fracture was not different in the two treatment groups. However, the vitamin D group showed an increased risk of hip/femur fracture (hazard ratio [HR] = 1.49, 95 % confidence interval [CI] 1.02–2.18) and hip/femur/wrist fracture (HR = 1.40, 95 % CI 1.07–1.82). Analysis of the female subjects showed that vitamin D treatment was associated with a borderline increased risk of any nonvertebral fracture (HR = 1.21, 95 % CI 1.00–1.47) and increased risk of hip/femur (HR = 1.80, 95 % CI 1.12–2.90) and hip/femur/wrist fracture (HR = 1.59, 95 % CI 1.17–2.16). However, vitamin D treatment was not associated with increased risk of any fracture in males. No effect on falls was observed, although this was not a primary outcome and falls were ascertained by 6-monthly recall. The other study, of 2,256 community-dwelling Australian women aged 70–92 years randomly allocated to receive an annual oral dose of 500,000 IU vitamin D3, demonstrated a 15 % (95 % CI 1.02–1.30) increased rate of falls and a 26 % (95 % CI 1.00–1.59) increased rate of fractures [27]. A temporal pattern was observed, with the greatest increase occurring in the first 3 months after dosing (falls: p for homogeneity = 0.02). A temporal pattern of risk was not demonstrated in the Wessex study, although the 6-monthly ascertainment of fractures did not optimize this post hoc analysis (unpublished).
Table 3

Clinical trials of vitamin D with intermittent dose intervals between 1 week and 1 year

Reference

Sample size

Study subjects

Treatment dose and duration

Primary outcome

Type of control (double-blind unless stated otherwise)

Baseline 25(OH)D status (nmol/L)

Reported adverse effect attributable to treatment

Outcomes

Comments

Heikinheimo [50]

n = 799

Age >75 years, 79 % female

Single annual dose IM 150,000 IU or 300,000 IU

3 years

Fracture

Unblinded control with no placebo

Outpatient group: 49 ± 14 municipal home group: 45 ± 16

None reported

Fracture rate 16 % vs. 22 %, p = 0.034; vitamin D and control groups, respectively

Some design and protocol flaws to this study

Smith et al. [28]

n = 9,440

Age 79 years median, IQR 77–83, 54 % female

Single annual IM dose D2 300,000 IU 3 years

Risk of first fracture (nonvertebral fractures)

Placebo

57 ± 24

Females only: HR (95 % CI) hip or femur 1.80 (1.12–2.90), any nonvertebral 1.21 (1.0–1.47)

Fractures—all nonvertebral; men and women HR = 1.09 (0.93–1.28), p = 0.29 for any first fracture

 

Sanders et al. [27]

n = 2,256

Women, median age 76 years (IQR 73–-80), community-dwelling

Single annual oral dose 500,000 IU, intervention 3–5 years

Fractures and falls

Placebo

49 (IQR 40–63), subgroup n = 137

Falls RR = 1.15 (1.02–1.30), p = 0.03; fractures RR = 1.26 (1.00–1.59), p = 0.047

Falls and fractures, results per previous column

Temporal pattern of increased risk in first 3 months postdose, no differences between groups in indices of mental well-being (Sanders et al. [49])

Rossini et al. [48]

n = 37

Adults aged 75 ± 3 years

Single oral dose D3 randomized to 600,000, 300,000, or 100,000 IU; follow-up 90 days

BTM and vitamin D status

Unblinded, no placebo, controls n = 24

Baseline 25D level: 60 ± 17 nmol/L

Dose-dependent acute increases in CTX

Mean serum PTH level decreased in all 3 groups, remained < baseline in 600,000 group, returned to baseline by 60 days in 300,000 group and within a few days in 100,000 group; no changes were observed in BAP in 600,000 and 300,000 groups but rose by 15–23 % in 100,000 group

Results may explain fracture reduction benefit in RCT using 100,000 IU/4 months (Trivedi et al. [46]) and unexpected increase in RCT using 500,000 IU (Sanders et al. [27])

Jorde et al. [72]

n = 441

282 women, overweight and obese;

BMI 28–47, aged 21–70 years

3 groups: weekly dose D3 group 1: 40,000 IU, group 2: 20,000 IU, group 3: placebo; all given calcium 500 mg/day intervention 1 year

Symptoms of depression in overweight/obese adults

Placebo

53 (range 11–112)

N/A

Symptoms of depression, using BDI score, significant reduction (improvement) in high-dose group, although serum 25(OH)D levels did not correlate with BDI scores; high-dose group had improved scores at 1 year

Overweight subjects with serum 25D levels <40 nmol/L had higher (more depressive) scores compared to subjects with 25D levels >40 nmol/L

Lyons et al. [56]

n = 3,440 (2,624 women)

Adults in residential care, aged 84 ± 8 years

4-monthly dose D2 oral 100,000 IU, intervention 3 years

Risk of first fracture

Placebo

Unknown

N/A

Risk of first fracture HR = 0.95 (0.79–1.15)

End of study 25(OH)D only (DiaSorin RIA) 80 vs. 54 nmol/L (n = 102, vitamin D vs. controls, respectively); 55 % died during the intervention with no difference between groups

Trivedi et al. [46]

n = 2,686

(76 % men)

Community-dwelling adults, aged 75 ± 5 years

4-monthly dose

D3

100,000 IU,

intervention 5 years

Risk of first fracture

Placebo

N/A

N/A

RR (95 % CI) any first fracture 0.78 (0.61–0.99, P = 0.04), first hip, wrist or forearm or vertebral fracture: 0.67 (0.48, 0.93; P = 0.02)

No significant effects of vitamin D on total mortality or incidence of cancer or cardiovascular disease

Law et al. [74]

N = 3,717 (76 % women)

Residential aged care,

average age 85 years (SD N/A)

Vitamin D2 100,000 IU 3- monthly

dose,

follow-up 10 months (IQR 7–14)

Non-vertebral fractures

Cluster

randomized, no

placebo given

47 (35–102,

90th percentile)

in n = 18 (1 % of vitamin D group)

No significant difference

between groups but fracture rate was “directionally higher in treated group”

RR (95 % CI) all nonvertebral fractures: 1.48 (0.99–2.20),

Hip fractures: 1.36 (0.80–2.34), falls: 1.09 (0.95, 1.25)

 

Serial biochemistry was performed only in a very small proportion of participants in both these RCTs (0.04 % and 6.1 % participants; Smith et al. [28] and Sanders et al. [27], respectively). Neither study recruited participants based on low 25(OH)D levels at screening. We are unable to infer that the adverse effects are confined to participants whose 25(OH)D levels were either deficient/insufficient or replete at baseline. It is well documented that the incremental increase in serum 25(OH)D is likely to be lower in those already replete prior to supplementation [24], and there is substantial variation in dose–response curves between individuals [8, 9]. There is therefore no evidence base to justify large annual loading doses of vitamin D to specific groups based on their baseline 25(OH)D level. Based on the reduction in fractures using 4-monthly dosing regimens in the Trivedi et al. [46] RCT and the biochemical results by Bacon et al. [24] and Ilahi et al. [47], it seems prudent to restrict intermittent higher doses to intervals not greater than 2–4 months. However, the reasoning is speculative, and RCT evidence with physical outcomes using a variety of dosing regimens is urgently needed. The fall characteristics from the Australian study do not suggest that the increased falls were attributable to one subgroup of participants experiencing the most falls. The proportion of participants falling multiple times did not vary between the vitamin D and placebo groups (unpublished data), and Kaplan–Meier plots of time to first fall show significant differences between the groups (p = 0.003). In another recent publication, a small group (n = 12) of older (mean age 73 years) subjects was treated with a single oral dose of 600,000 IU vitamin D3 [48]. Serum 25(OH)D increased from 54 ± 14 nmol/L at baseline to 168 ± 43 nmol/L at day 3 when the bone turnover markers C-terminal telopeptide (CTX) and N-terminal telopeptide (NTX) peaked at over 50 % above baseline. PTH decreased and 1,25-dihydroxyvitamin D (1,25[OH]D) increased by 25–50 % [48]. Rossini and colleagues [48] suggest that this transient increase in bone turnover markers may explain the negative clinical results obtained in studies using intermittent high-dose vitamin D. Sanders and colleagues [49] have also reported increased bone turnover among a sample of participants who underwent biochemistry assessments and had a very high incremental rise in serum 25(OH)D levels. While increased bone turnover may contribute to the demonstrated increase in fracture risk, this does not explain the clear evidence of increased falls in the Australian study (Table 3).

Speculative Mechanisms

The mechanism by which high-dose vitamin D might increase falls and fracture is uncertain. The opposing outcomes of two studies [28, 46] that used the same total annual dose (300,000 IU intramuscularly) suggest that the dosing regimen (i.e., 4-monthly vs. annually) rather than the total dose might determine the outcome. While a dosing interval of 12 months is equivalent to four biological half-lives of vitamin D, at the time these two studies were conducted there was biochemical evidence of safety and preliminary evidence that these intermittent bolus doses may be efficacious at decreasing fracture risk in older women [36, 46, 50]. In addition, the Australian study team was specifically addressing the drop in vitamin D and increased fractures that occurs during winter [51]. The line of reasoning regarding the dosing interval is supported by the temporal risk pattern observed in the study of Sanders et al. [27] and the fact that harm has not been reported in the numerous studies that have used more frequent dosing [52]. However, the lower-level evidence of a U-shaped dose–response curve reported in some observational studies [5355] is not consistent with a temporal pattern since it is unlikely that those in the highest quintile of vitamin D status in the community use high intermittent doses of supplemental vitamin D. However, it is possible that seasonal fluctuations in 25(OH)D levels may contribute to this apparent phenomenon. Vieth [26] contends that a U- or J-shaped curve of risk is observed only in populations residing farther away from the equator and who, therefore, have greater seasonal fluctuations. It is argued that the annual downward phase in seasonal cycles almost definitely creates a non-steady-state situation for the paracrine production of 1,25(OH)D responsible for the noncalcemic effects of vitamin D. It is also possible the adverse mechanism may be associated with gender since Smith and colleagues [28] did not demonstrate an increased fracture risk among men and the majority of participants in the Trivedi et al. [46] study demonstrating a reduced fracture risk using 100,000 IU every 4 months were men (76 %). A Welsh RCT using the same dosing regimen as Trivedi et al. [46] but with 76 % women reported no difference in fracture outcome [56] (Tables 1, 3). In addition, the Australian RCT recruited only women [27]. There is also weak evidence from RCTs using more frequent dosing regimens that the mechanism is not a single aspect but may be more complex (Table 4).
Table 4

Clinical trials of vitamin D with daily dose regimens and evidence of possible adversity

Reference

Sample size

Study subjects

Treatment dose and duration

Primary outcome (ascertainment)

Type of control (double-blind unless stated otherwise)

Baseline 25(OH)D status nmol/L

Reported adverse effect attributable to treatment

Comments

Flicker et al. [81]

n = 117

n = 117 participants with mild cognitive impairment

1,000 IU D3 intervention 18 months

Cognitive function (CAMCOG)

Placebo

49 ± 12 nmol

Vitamin D group more likely to deteriorate than placebo group; overall change in CAMCOG score was –0.70 (–1.43 to 0.02)

Serum 25D: 6-month 78 ± 22 nmol/L, 12-month 84 ± 21 nmol/L.

Stein et al. [82]

n = 32

Aged >60 years, with mild–moderate Alzheimer disease,

titrated high dose for 8 weeks then immediately randomized to nasal insulin for 48 h

All participants had 8-week “run-in” period of daily low-dose vitamin D2 (1,000 IU), then were group-randomized to high-dose vitamin D2 or placebo;

high dose consisted of 2 weeks of 36,000 lU/day, followed by 0–36,000 lU/day titrated to aim for serum 25D level maintained in the range 130–175 nmol/L;

titrations were done at 2, 4, and 6 weeks;

intervention 16 weeks

Cognitive function and dementia-induced disability

(Alzheimer disease assessment scale-cognitive subscale (ADAS-cog); disability assessment in dementia (after high-dose vitamin D)

Placebo

49 (median) IQR N/A

60 (56–70) at

commencement of high-dose regimen

The authors conclude “the RCT found no benefit from adding high-dose vitamin D to ongoing low-dose vitamin D supplementation. In the RCT, serum 25D concentrations were generally supra-physiologic and may have had no additional effect.”

Median 25(OH)D 60 nmol/L at 8 weeks on 1,000 IU/day group and 187 nmol/L after a further 8-week high-dose vitamin D

Grimnes et al. [34]

n = 297

Women aged 50–80 years,

T score BMD < –2.0 at either site

Daily dose 6,500 or 800 IU D3,

both groups given 1,000 mg calcium/day,

intervention 1 year

BMD at hip and spine

Low (800 IU) vs. high (6,500 71 ± 23 IU), no placebo

71 ± 23

Authors suggest that standard dose (800 IU/day) was more efficient as bone formation marker P1NP had more pronounced reduction in this group

BMD, no significant differences between groups; bone turnover reduced in both groups

Bischoff-Ferrari et al. [83]

n = 173 (79 %

women)

Adults with acute hip fracture aged 65+ years

Either 2,000 or 800 IU D3,

factorial design: extended vs. standard physiotherapy and vitamin D 2,000 vs. 800 IU/day, intervention 12 months

Rate of falls

High (2,000 IU) vs. standard fewer hospital re- care lower dose (800 IU)

32 ± 20

Falls rate shows trend to be higher in 2,000 vs. 800 IU daily dose rate of falls per observed patient-year 1.63 vs. 1.25. adjusted relative rate difference 28 % (–4 to 68)

Higher dose group had fewer hospital readmission during 12-month follow up: Adjusted RR (95 %C.I) −39 (−62, −1)

Age expressed as mean ± SD or median and interquartile range (IQR). Oral dose expressed in international units (IU). D2 vitamin D2 (ergocalciferol), D3 vitamin D3 (cholecalciferol), IM intramuscular injection, 25(OH)D serum 25-hydroxyvitamin D (nmol/L), 1,25(OH)D serum 1,25-dihydroxyvitamin D (levels quoted as mean and standard deviation unless stated otherwise), BDI Beck Depression Inventory, CAMCOG Cambridge Cognitive, BTM bone turnover markers, CTX C-terminal telopeptides (bone resorption marker), BAP bone-specific alkaline phosphatase (bone formation marker), F female, M male, RR relative risk (95 % confidence interval), N/A not available/applicable

It has been hypothesized that the increased numbers of falls and fractures may have, ironically, resulted from the benefits of vitamin D in that the older women randomized to vitamin D felt better and consequentially engaged in more “at-risk” falls behavior [52]. However, the Australian authors subsequently published mental well-being outcomes of this study. No significant differences were detected in any of the measured outcomes of mental health [57], making this explanation less likely. No differences between the groups relating to the circumstances or activity of the fall events has been identified (unpublished).

In their editorial, Dawson-Hughes and Harris [52] also hypothesized that the 500,000 IU dose may have triggered a “short-term protective” reaction in which CYP24 (25-hydroxyvitamin D-24-hydroxylase), the enzyme that catabolizes 1,25(OH)D, was upregulated, resulting in decreased blood and tissue levels of 1,25(OH)D. Although this hypothesis is consistent with results from an animal study [58], both the Wessex [28] and Rossini et al. [48] studies demonstrated increases (25–50 %) in serum 1,25(OH)D in those who had serial biochemistry assessments. From an evolutionary approach, Vieth [59] presents the argument that oral supplementation of vitamin D is needed to improve health outcomes by lessening the destabilizing effect of annual fluctuations in serum 25(OH)D. He argues that the paracrine regulation of 1,25(OH)D in many tissues is disrupted by unstable 25(OH)D levels and that this adversely affects bone mineral density, mental well-being, infection, and cancer risk. The profile of 25(OH)D levels from the two studies showing harm does not appear distinctly different from a range of high-dose biochemical studies (Table 1). Although there is no uniformity in the time points of 25(OH)D assessment, peak 25(OH)D levels from these studies tend to be around 120–140 nmol/L. Ilahi and colleagues [47] suggest that the dosing interval of intermittent dosing regimens be not greater than 70 days to ensure that 25(OH)D levels do not decline below a target of 70 nmol/L.

The increased risk of falls in the Australian study demonstrates that the adverse mechanism is not confined to the skeleton. Post hoc analysis of changes in muscle strength in a nested substudy of these older women who underwent annual physical functioning assessments suggests a decline in muscle strength in those whose 25(OH)D level showed the greatest fluctuation from baseline [60]. Since there are vitamin D receptors in muscle, a sudden increase in vitamin D receptor occupancy could have an adverse effect on muscle function [61]. Vitamin D receptors are also present in the central nervous system [62], so an adverse effect on balance or coordination is also possible. Another recent Australian RCT of 686 ambulant women aged at least 70 years reported neither a beneficial nor an adverse effect on falls or physical function using a 3-monthly dosing of 150,000 IU cholecalciferol compared to placebo [63]. The study intervention period was 9 months, and the baseline 25(OH)D level measured in a subgroup of 40 participants was 66 ± 23 nmol/L. A review by Stockton and colleagues [64] concluded that vitamin D supplementation does not have a significant effect on muscle strength in vitamin D–replete adults.

Concluding Summary

While epidemiological studies provide evidence that vitamin deficiencies are associated with an increased risk of chronic disorders and/or cancer, the consequent philosophy that higher doses of the vitamin are protective and confer a reduced risk of these diseases is flawed [65, 66]. Two recent editorials on high-dose vitamin D have drawn analogies from the “hard” lessons learned from RCTs on high-dose vitamins A, B, C, and E [65, 66]. Supraphysiological levels of the vitamin taken as supplements do not emulate the apparent benefits of diets high in food that contain those vitamins and other lifestyle factors [67]. The findings from two recent high-dose RCTs [27, 28] identify a potential harm associated with high-dose vitamin D and support the notion that vitamin D could be now added to this list. Thus, in addition to evidence from enzyme kinetics relating to vitamin D metabolism [44], there is now high-level RCT evidence that vitamin D supplementation has potential toxicities other than simply hypercalcemia/-uria. As our understanding of the pharmacokinetics of vitamin D metabolism becomes more sophisticated, clinical trials with novel dosing regimens should apply the principles of conventional pharmacology and vitamin D metabolism to the study design.

Interpretation of findings from many large RCTs has been limited by the lack of assessment of 25(OH)D status in the majority of participants. Future studies of supplementation should be adequately funded to allow comprehensive or universal measurement of serum 25(OH)D and related biochemical parameters [65, 68], with particular attention to large and rapid fluctuations in vitamin D status. Future studies should not base toxicity solely on the risk of hypercalcemia/-uria. There is an urgent need for dose-ranging studies with physical function outcomes [61, 64]. Emerging evidence from both observational studies and RCTs suggests that there should be a degree of caution about recommending high serum 25(OH)D concentrations for the entire population. Furthermore, a benefit of the higher doses commonly used in clinical practice on falls risk reduction needs to be demonstrated [69]. While it is recognized that intramuscular high-dose vitamin D preparations may be the only way of ensuring adequate vitamin D status in specific “at-risk” groups of patients, such as those suffering fat malabsorption, the safety of loading doses of vitamin D administered to the general population should be demonstrated before these regimens become recommended as routine clinical practice [65]. The current dilemma of defining vitamin D insufficiency and identifying safe and efficacious repletion regimens needs to be resolved.

Copyright information

© Springer Science+Business Media New York 2012