Abstract
Background
We assessed the association between serum 25-hydroxyvitamin D levels and genital human papillomavirus (HPV) prevalence, incidence, and clearance among female participants in the HPV Infection and Transmission among Couples through Heterosexual activity (HITCH) Cohort Study.
Methods
We genotyped HPV DNA in vaginal samples and quantified baseline serum 25-hydroxyvitamin D levels using Roche’s Linear Array and Total vitamin D assay, respectively. We used logistic and Cox proportional hazards models, respectively, to estimate adjusted odds ratios (ORs) and hazard ratios (HRs) with 95% confidence intervals (CIs).
Results
There was no association between vitamin D levels (every 10-ng/mL increase) at baseline and HPV prevalence (OR, 0.88; 95% CI, .73–1.03) or incidence (HR, 0.88; 95% CI, .73–1.06), but we observed a modest negative association with HPV clearance (HR, 0.76; 95% CI, .60–.96). Vitamin D levels <30 ng/mL, compared with those ≥30 ng/mL, were not associated with HPV prevalence (OR, 0.98; 95% CI, .57–1.69) or incidence (HR, .87; 95% CI, .50–1.43), but they were associated with a marginally significant increased clearance (OR, 2.14; 95% CI, .99–4.64). We observed consistent results with restricted cubic spline modeling of vitamin D levels and clinically defined categories. HPV type-specific analyses accounting for multiple HPV infections per participant showed no association between vitamin D levels and all study outcomes.
Conclusions
This study provided no evidence of an association between low vitamin D levels and increased HPV prevalence, acquisition, or clearance.
Keywords: HPV DNA infection, HPV prevalence, HPV clearance, 25-hydroxyvitamin D, vitamin D
Analysis of data from female participants (baseline and follow-up visits at 4, 8, 12, 18, and 24 months) provided no evidence of an association between baseline serum 25-hydroxyvitamin D levels and vaginal human papillomavirus prevalence, incidence, and clearance.
Human papillomavirus (HPV) is the most commonly diagnosed sexually transmitted infection and a necessary etiological agent of cervical cancer [1, 2]. HPV also plays a central role in other anogenital cancers [3–5]. Worldwide, the prevalence of cervical HPV infection in cytologically normal women in the pre-HPV vaccination era was estimated at 11.7% [6]. Infection with oncogenic HPV genotypes and the burden of HPV-related diseases remain global public health concerns despite currently implemented primary prevention efforts, that is, prophylactic vaccination. Identifying modifiable risk factors could form the basis to simple prevention strategies to decrease the burden of HPV infections.
Low blood concentration of serum 25-hydroxyvitamin D, the major circulating vitamin D metabolite, was found to be associated with an increased prevalence of vaccine-preventable [7] and high-risk [7, 8] HPV types. Conversely, another recent study found no significant associations between serum 25-hydroxyvitamin D and prevalent high-risk HPV [9]. These studies, however, were limited by their cross-sectional nature, which cannot establish temporality between vitamin D levels and acquisition of HPV.
Research in various patient populations have suggested that vitamin D may play a range of roles beyond the regulation of bone metabolism, including cardiovascular health, cell growth and differentiation, immunomodulation, and infectious and inflammatory diseases [10–12]. Apart from HPV, other genital health conditions, such as bacterial vaginosis and Chlamydia trachomatis, are reported to be associated with low levels of serum 25-hydroxyvitamin D in some [13, 14], but with discordant findings in other studies [15, 16].
In light of the limited evidence on the association between serum 25-hydroxyvitamin D levels and genital HPV infection as well as the lack of research on its association with HPV incidence and clearance, our objective was to quantify this association in female participants of the HPV Infection and Transmission among Couples through Heterosexual activity (HITCH) Cohort Study. More specifically, we examined the association of vitamin D levels with vaginal HPV prevalence, incidence, and clearance.
METHODS
Study Design and Population
We analyzed data from female participants who were enrolled (2005–2011) and followed up for 24 months as part of a study of HPV transmission among recently formed couples, the HITCH Cohort Study [17]. The study population consisted of young women (18–24 years old) attending universities or junior colleges in Montreal and their male partners who were in a sexual relationship of ≤6 months; the present analysis was restricted to female participants. Women were instructed to self-collect a vaginal sample for HPV testing, and a nurse collected blood samples, at enrollment and 5 subsequent follow-up visits (at 4, 8, 12, 18, and 24 months). Women self-completed computer-assisted questionnaires that measured sociodemographic, lifestyle, and sexual behavior characteristics. The study was approved by the ethical institutional review boards of McGill and Concordia universities. All participants provided written informed consent.
HPV Testing and Genotyping
Vaginal specimens were tested by a polymerase chain reaction protocol based on amplification of a 450 bp segment in the L1 HPV gene using the Linear Array HPV genotyping assay (Roche Molecular Systems) [18]. This technique can detect 36 mucosal HPV genotypes: types 6, 11, 16, 18, 26, 31, 33, 34, 35, 39, 40, 42, 44, 45, 51, 52, 53, 54, 56, 58, 59, 61, 62, 66, 67, 68, 69, 70, 71, 72, 73, 81, 82, 83, 84, and 89.
Vitamin D Testing
Using blood samples collected at enrollment, total serum 25-hydroxyvitamin D concentrations (vitamin D2 and D3)—hereafter referred to as “vitamin D levels”—were quantified using the Roche Total vitamin D assay on the Roche Cobas e602. The lower detection level of the assay was 7.5 nmol/L. For our analyses, measurement results of <7.50 nmol/L (n = 18) and those >175 nmol/L (n = 1) were set to 3.77 (midpoint of 0–7.54; 7.54 being the lowest concentration detected) and 175, respectively.
Statistical Analyses
We used 4 exposure definitions for vitamin D levels (expressed in nanograms per milliliter; 1 ng/mL = 0.4 nmol/L). First, vitamin D levels were modeled as a continuous variable (primary exposure) to assess the association of each additional 10 ng/mL increase in vitamin D levels with HPV prevalence, incidence, and clearance. We chose the 10-ng/mL increase to reflect a clinically meaningful unit increase in vitamin D levels that would also enable observing a potential effect on the various study outcomes (HPV prevalence, incidence, and clearance). Second, vitamin D levels were dichotomized into clinically relevant levels—≥30 ng/mL (sufficient) or <30 ng/mL (insufficient) [19]. Third, vitamin D levels were categorized according to more clinically discriminant categories for comparison purposes with those used by Shim et al [7]; levels ≥30 ng/mL were considered sufficient, 20–29 ng/mL insufficient, 12–19 ng/ deficient, and <12 ng/mL severely deficient. Finally, to account for possible nonmonotonic dose responses in the associations, vitamin D levels were modeled using restricted cubic splines with 3–5 interior knots [20]. The optimal number of knots was determined by fitting the models and estimating the Akaike information criterion for each model which provides a balance of model fit against parsimony. The number of knots resulting in the minimum Akaike information criterion was chosen [21].
We summarized the distribution of HPV types, species, and subgenus groups in the study population at enrollment by the above-mentioned clinically discriminant categories of vitamin D levels. We grouped HPV infections according to Alphapapillomavirus species clusters that exhibit similar tissue tropism and biological behavior concerning cancer risk [22–24]. Subgenus 1 group included HPV types 6, 11, 40, 42, 44, and 54, from species α-1, α-8, α-10, and α-13. Subgenus 2 included oncogenic HPV types 16, 18, 26, 31, 33, 34, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, 69, 70, 73, and 82, from species α-5, α-6, α-7, α-9, and α-11. Subgenus 3 included HPV types 61, 62, 71, 72, 81, 83, 84, and 89, from species α-3 and α-15.
We conducted a cross-sectional analysis examining the association between vitamin D levels (for each of the above 4 exposure definitions) and prevalence of HPV infection at baseline. We used multivariable logistic regression to model these associations, estimating odds ratios (ORs) and 95% confidence intervals (CIs). We also performed a prospective cohort analysis evaluating the association of baseline vitamin D levels (for continuous, dichotomous, and clinical categories) with (1) the incidence of HPV infection among HPV-negative women at enrollment and (2) the clearance of HPV infection among HPV-positive women at enrollment. We fitted Cox proportional hazards models to calculate hazard ratios (HRs) with 95% CIs. HPV clearance was defined as ≥2 consecutive HPV-negative results after a positive test result. In addition, we used restricted cubic splines to model the potential nonlinear association between vitamin D levels and HPV, using 30 ng/mL as the reference value.
The above analyses were performed for any HPV infection and HPV subgenera 1–3. Considering the incidence of grouped HPV infections, we also conducted a sensitivity analysis examining the risk of acquiring HPV infections of a given subgenus in participants who at baseline did not test positive for HPV types belonging to that specific subgenus.
We estimated crude and adjusted ORs and HRs. The adjusted models included the following a priori potential confounders: age (<22 or ≥22 years), ethnicity (French Canadian, English Canadian, or other), smoking status (never, former, or current smoker), age at first sexual intercourse (<17 or ≥17 years), age at menarche (<13 or ≥13 years), number of lifetime sex partners (<4, 4 to <7, ≥7 to <12, or ≥12), any history of pregnancy (yes or no), condom use (never, rarely, sometimes, most times, or always), and season of blood sample collection (spring, summer, fall, or winter). We used the month of examination to define the season of blood collection as spring (March–May), summer (June–August), fall (September–November), or winter (December–February).
We also conducted analyses to account for multiple HPV infections per participant (ie, unit of analysis was at the HPV level). In these analyses, each participant could contribute multiple observations (up to 36 different HPV types). We used hierarchical logistic models to examine the association between vitamin D levels (modeled as continuous, dichotomous, and categorical) and HPV prevalence, and log-normal frailty models for the association between vitamin D levels and HPV incidence and clearance at the HPV level while accounting for clustering of observations at an individual level. Analyses were carried out using Stata 16 software (StataCorp).
RESULTS
Baseline Participants’ Characteristics and HPV Prevalence
Of the 502 female participants enrolled in the HITCH study, 490 were included in the current analyses. We excluded from the analysis participants with incomplete data (ie, no baseline genital specimens collected from 4 women, invalid HPV test results in 2 samples, and no testing of vitamin D levels in 6 serum samples). Characteristics of this cohort have been previously described [17]. Specifically, in the current analysis sample (Supplementary Table 1), the median age was 21 years (standard deviation, 2.1 years; range, 18–26 years). A total of 132 women (27%) were of French ancestry, and 35% were English Canadian. The majority had never smoked (75%), and 13% were former smokers. Almost half (49%) had their first menstruation before age 13 years and 45% had their first vaginal intercourse before age 17 years. Only 10% were ever pregnant, 32% had <4 lifetime sex partners, and 41% reported using condoms most times or always. The mean (standard deviation) vitamin D levels fluctuated slightly by the season of blood sample collection (spring, 16.1 [10.4] ng/mL; summer, 22.1 [12.3] ng/mL; fall, 23.9 [15.4] ng/mL; and winter, 22.3 [12.6] ng/mL).
Cross-Sectional Analysis of Baseline Vitamin D Levels and HPV Infection
Table 1 shows genotype- and group-specific HPV prevalence according to clinically defined vitamin D categories at enrollment. Overall, there was no consistent pattern of variation in the prevalence of HPV infections by these categories. The prevalence of any HPV infection was 55% in the severely deficient vitamin D group, 57% in the deficient group, 57% in the insufficient group, and 60% in the sufficient group.
Table 1.
Baseline Prevalence of Human Papillomavirus Infections in Women (n = 490) According to Clinically Defined Vitamin D Categories
| HPV Infection Grouping | Prevalence by Vitamin D Category, No. (%) of HPV Infectionsa | |||
|---|---|---|---|---|
| Severe Deficiency (n = 129) | Deficiency (n = 143) | Insufficiency (n = 113) | Sufficiency (n = 105) | |
| HPV type | ||||
| 6 | 5 (3.9) | 4 (2.8) | 1 (0.9) | 8 (7.6) |
| 11 | 1 (0.8) | 2 (1.4) | 0 (0.0) | 0 (0.0) |
| 16 | 16 (12.4) | 25 (17.5) | 13 (11.5) | 28 (26.7) |
| 18 | 5 (3.9) | 6 (4.2) | 3 (2.7) | 3 (2.9) |
| 26 | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| 31 | 6 (4.7) | 2 (1.4) | 6 (5.3) | 10 (9.5) |
| 33 | 1 (0.8) | 2 (1.4) | 3 (2.7) | 1 (1.0) |
| 34 | 1 (0.8) | 0 (0.0) | 0 (0.0) | 2 (1.9) |
| 35 | 0 (0.0) | 2 (1.4) | 0 (0.0) | 2 (1.9) |
| 39 | 10 (7.8) | 10 (7.0) | 4 (3.5) | 10 (9.5) |
| 40 | 7 (5.4) | 0 (0.0) | 2 (1.8) | 3 (2.9) |
| 42 | 10 (7.8) | 10 (7.0) | 11 (9.7) | 5 (4.8) |
| 44 | 1 (0.8) | 3 (2.1) | 3 (2.7) | 3 (2.9) |
| 45 | 1 (0.8) | 3 (2.1) | 3 (2.7) | 1 (1.0) |
| 51 | 14 (10.9) | 12 (8.4) | 15 (13.3) | 8 (7.6) |
| 52 | 9 (7.0) | 11 (7.7) | 4 (3.5) | 11 (10.5) |
| 53 | 7 (5.4) | 7 (4.9) | 8 (7.1) | 14 (13.3) |
| 54 | 6 (4.7) | 10 (7.0) | 6 (5.3) | 8 (7.6) |
| 56 | 7 (5.4) | 8 (5.6) | 6 (5.3) | 4 (3.8) |
| 58 | 8 (6.2) | 4 (2.8) | 7 (6.2) | 4 (3.8) |
| 59 | 5 (3.9) | 12 (8.4) | 9 (8.0) | 4 (3.8) |
| 61 | 3 (2.3) | 3 (2.1) | 4 (3.5) | 2 (1.9) |
| 62 | 8 (6.2) | 11 (7.7) | 11 (9.7) | 10 (9.5) |
| 66 | 9 (7.0) | 6 (4.2) | 6 (5.3) | 10 (9.5) |
| 67 | 9 (7.0) | 6 (4.2) | 6 (5.3) | 6 (5.7) |
| 68 | 7 (5.4) | 1 (0.7) | 3 (2.7) | 3 (2.9) |
| 69 | 0 (0.0) | 0 (0.0) | 0 (0.0) | 0 (0.0) |
| 70 | 1 (0.8) | 0 (0.0) | 3 (2.7) | 0 (0.0) |
| 71 | 1 (0.8) | 1 (0.7) | 0 (0.0) | 0 (0.0) |
| 72 | 0 (0.0) | 0 (0.0) | 1 (0.9) | 1 (1.0) |
| 73 | 3 (2.3) | 6 (4.2) | 4 (3.5) | 5 (4.8) |
| 81 | 1 (0.8) | 1 (0.7) | 3 (2.7) | 3 (2.9) |
| 82 | 1 (0.8) | 7 (4.9) | 2 (1.8) | 4 (3.8) |
| 83 | 1 (0.8) | 3 (2.1) | 4 (3.5) | 1 (1.0) |
| 84 | 9 (7.0) | 14 (19.8) | 11 (9.7) | 14 (13.3) |
| 89 | 10 (7.8) | 16 (11.2) | 11 (9.7) | 16 (15.2) |
| Any HPV | 71 (55.0) | 82 (57.3) | 64 (56.6) | 63 (60.0) |
| Speciesb | ||||
| α-1 | 10 (7.8) | 10 (7.0) | 11 (9.7) | 5 (4.8) |
| α-3 | 29 (22.5) | 38 (26.6) | 33 (29.2) | 33 (31.4) |
| α-5 | 15 (11.6) | 18 (12.6) | 16 (14.2) | 12 (11.4) |
| α-6 | 23 (17.8) | 19 (13.3) | 16 (14.2) | 24 (22.9) |
| α-7 | 22 (17.1) | 27 (18.9) | 21 (18.6) | 18 (17.1) |
| α-8 | 7 (5.4) | 0 (0.0) | 2 (1.8) | 3 (2.9) |
| α-9 | 32 (24.8) | 38 (26.6) | 32 (28.3) | 40 (38.1) |
| α-10 | 7 (5.4) | 8 (5.6) | 4 (3.5) | 9 (8.6) |
| α-11 | 4 (3.1) | 6 (4.2) | 4 (3.5) | 7 (6.7) |
| α-13 | 6 (4.7) | 10 (7.0) | 6 (5.3) | 8 (7.6) |
| α-15 | 1 (0.8) | 1 (0.7) | 0 (0.0) | 0 (0.0) |
| Species groupc | ||||
| Subgenus 1 | 27 (20.9) | 26 (18.2) | 22 (19.5) | 20 (19.1) |
| Subgenus 2 | 59 (45.7) | 65 (45.5) | 55 (48.7) | 55 (52.4) |
| Subgenus 3 | 29 (22.5) | 38 (26.6) | 33 (29.2) | 33 (31.4) |
Abbreviation: HPV, human papillomavirus.
aCategories were defined as follows: severe deficiency, <12 ng/mL; deficiency, 12–19 ng/mL; insufficiency, 20–29 ng/mL; and sufficiency, ≥30 ng/mL.
bHPV species include α-1 (HPV42), α-3 (HPV types 61, 62, 72, 81, 83, 84, and 89), α-5 (types 26, 51, 69, and 82), α-6 (types 53, 56, and 66), α-7 (types 18, 39, 45, 59, 68, and 70), α-8 (HPV40), α-9 (types 16, 31, 33, 35, 52, 58, and 67), α-10 (types 6, 11, and 44); α-11 (types 34 and 73), α-13 (HPV54), and α-15 (HPV71).
cHPV groups include subgenus 1 (α-1, α-8, α-10, and α-13), 2 (α-5, α-6, α-7, α-9, and α-11), and 3 (α-3 and α-15).
Overall, irrespective of the exposure definition, no associations were found between vitamin D levels and prevalence of any HPV infection (Table 2 and Figure 1). We did not observe an association between odds of HPV infection and each additional 10-ng/mL increase in vitamin D levels (adjusted OR, 0.97; 95% CI, .82–1.15). Similarly, vitamin D levels dichotomized at 30 ng/mL were not associated with HPV prevalence (adjusted OR, 0.98; 95% CI, .57–1.69). Likewise, we observed no associations between HPV prevalence and clinically defined categories of vitamin D levels, irrespective of taxonomic HPV grouping.
Table 2.
Association Between Vitamin D Levels and Prevalent Human Papillomavirus Infection at Baseline
| Exposure Definition for Vitamin D Levels | Odds Ratio (95% Confidence Interval) | |||||||
|---|---|---|---|---|---|---|---|---|
| Any HPV | Subgenus 1a | Subgenus 2a | Subgenus 3a | |||||
| Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | |
| Continuousc (n = 490) | 1.04 (.91–1.19) | 0.97 (.82–1.15) | 0.96 (.81–1.14) | 0.89 (.73–1.08) | 1.07 (.94–1.22) | 1.03 (.87–1.21) | 1.10 (.95–1.28) | 1.11 (.94–1.32) |
| Dichotomousd | ||||||||
| Sufficient (n = 105) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficient (n = 385) | 0.86 (.55–1.34) | 0.98 (.57–1.69) | 1.03 (.59–1.78) | 1.29 (.70–2.39) | 0.79 (.51–1.22) | 0.85 (.50–1.42) | 0.77 (.48–1.23) | 0.79 (.46–1.34) |
| Clinically defined categoriese | ||||||||
| Sufficiency (n = 105) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficiency (n = 113) | 0.87 (.51–1.49) | 0.94 (.50–1.80) | 1.03 (.52–2.02) | 1.21 (.58–2.54) | 0.86 (.51–1.47) | 0.87 (.47–1.62) | 0.90 (.51–1.60) | 0.97 (.51–1.83) |
| Deficiency (n = 143) | 0.90 (.54–1.50) | 0.89 (.48–1.66) | 0.94 (.50–1.80) | 1.15 (.56–2.36) | 0.76 (.46–1.26) | 0.74 (.41–1.35) | 0.79 (.45–1.37) | 0.72 (.39–1.35) |
| Severe deficiency (n = 129) | 0.82 (.48–1.38) | 1.19 (.62–2.30) | 1.13 (.59–2.15) | 1.62 (.77–3.43) | 0.77 (.46–1.28) | 0.97 (.51–1.83) | 0.63 (.35–1.13) | 0.67 (.34–1.32) |
Abbreviations: HPV, human papillomavirus; Ref, reference.
aSubgenus 1 group includes HPV types 6, 11, 40, 42, 44, and 54; subgenus 2, types 16, 18, 26, 31, 33, 34, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, 69, 70, 73, and 82; and subgenus 3, types 61, 62, 71, 72, 81, 83, 84, and 89.
bModels were adjusted for age, ethnicity, smoking, age at first sexual intercourse, age at menarche, number of lifetime sex partners, any history of pregnancy, condom use, and the season of blood sample collection.
cPer 10-unit increase in levels of vitamin D.
dDichotomized into ≥30 ng/mL (sufficient) or <30 ng/mL (insufficient).
eCategories were defined as follows: sufficiency, ≥30 ng/mL; insufficiency, 20–29 ng/mL; deficiency, 12–19 ng/mL; and severe deficiency, <12 ng/mL.
Figure 1.
Association between vitamin D Levels and the prevalence of human papillomavirus (HPV) infection. Serum vitamin D levels were modeled using restricted cubic splines for any HPV (A) and for subgenus 1 (B), 2 (C), and 3 (D). Solid lines represent the model estimates, and shaded regions correspond to the 95% confidence intervals.
A nonmonotonic relationship was found when modeling vitamin D levels using restricted cubic splines, although we did not observe statistical evidence for a higher HPV prevalence at levels <30 ng/mL (Figure 1A). Overall, vitamin D levels were not correlated with infection prevalence for types from individual HPV subgenera (Figure 1B, 1C, and 1D), with consistent results across exposure definitions.
Longitudinal Analysis of HPV Acquisition
The crude incidence rates per 1000 person-months of any HPV were 12.5, 22.2, 26.6, and 21.5 in severely deficient, deficient, insufficient, and sufficient vitamin D concentration categories, respectively (Supplementary Table 2). As shown in Table 3, among HPV-negative women at enrollment, we found no significant risk variation in HPV acquisition as a function of vitamin D levels based on the 10 ng/mL definition (adjusted OR, 1.14; 95% CI, .95–1.37) or when considering vitamin D levels on a dichotomous scale (0.87; .48–1.58). In general, there was a tendency for reduced risks of incident infection, irrespective of HPV grouping, for categories equated with deficient vitamin D levels. However, there was no consistent pattern in the associations. On a continuous scale, the best fit was for nonmonotonic associations between clinical categories of vitamin D levels and the risk of HPV acquisition (Figure 2). Modeling vitamin D levels using restricted cubic splines revealed a tendency for increased risk of HPV infection around the referent vitamin D concentration of 30 ng/mL. Findings were materially the same in secondary analyses restricted to participants without infection by a specific HPV subgenus at baseline (Supplementary Table 3 and Supplementary Figure 1).
Table 3.
Association Between Baseline Vitamin D Levels and Incident Human Papillomavirus Infection at Follow-up
| Exposure Definition for Vitamin D Levels | Hazard Ratio (95% Confidence Interval) | |||||||
|---|---|---|---|---|---|---|---|---|
| Any HPV | Subgenus 1a | Subgenus 2a | Subgenus 3a | |||||
| Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | |
| Continuousc (n = 210) | 1.10 (.95–1.28) | 1.14 (.95–1.37) | 1.13 (.91–1.41) | 1.09 (.83–1.43) | 1.06 (.89–1.27) | 1.13 (.92–1.38) | 1.13 (.90–1.42) | 1.23 (.91–1.65) |
| Dichotomousd | ||||||||
| Sufficient (n = 42) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficient (n = 168) | 0.91 (.52–1.59) | 0.87 (.48–1.58) | 0.72 (.35–1.51) | 0.83 (.37–1.84) | 1.19 (.61–2.34) | 1.07 (.53–2.18) | 0.84 (.38–1.84) | 1.04 (.44–2.44) |
| Clinically defined categoriese | ||||||||
| Sufficiency (n = 42) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficiency (n = 49) | 1.23 (.65–2.33) | 1.18 (.57–2.45) | 1.04 (.45–2.41) | 1.46 (.53–4.05) | 1.78 (.84–3.77) | 1.43 (.63–3.27) | 1.24 (.51–3.01) | 1.12 (.41–3.12) |
| Deficiency (n = 61) | 1.03 (.54–1.95) | 0.81 (.38–1.75) | 0.55 (.22–1.35) | 0.44 (.14–1.38) | 1.49 (.71–3.12) | 0.98 (.41–2.32) | 0.74 (.29–1.89) | 0.66 (.22–1.97) |
| Severe deficiency (n = 58) | 0.55 (.27–1.14) | 0.51 (.22–1.16) | 0.64 (.26–1.58) | 0.69 (.22–2.12) | 0.51 (.21–1.26) | 0.33 (.12–0.94) | 0.59 (.21–1.62) | 0.50 (.14–1.74) |
Abbreviations: HPV, human papillomavirus; Ref, reference.
aSubgenus 1 group includes HPV types 6, 11, 40, 42, 44, and 54. Subgenus 2 includes HPV types 16, 18, 26, 31, 33, 34, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, 69, 70, 73, and 82. Subgenus 3 includes HPV types 61, 62, 71, 72, 81, 83, 84, and 89.
bModels were adjusted for age, ethnicity, smoking, age at first sexual intercourse, age at menarche, number of lifetime sex partners, any history of pregnancy, condom use, and season of blood sample collection.
cPer 10-unit increase in levels of vitamin D.
dDichotomized into ≥30 ng/mL (sufficient) or <30 ng/mL (insufficient).
eClinical values were defined as follows: sufficiency, ≥30 ng/mL; insufficiency, 20–29 ng/mL; deficiency, 12–19 ng/mL; and severe deficiency, <12 ng/mL.
Figure 2.
Association between baseline serum vitamin D levels and the incidence of human papillomavirus (HPV) infection at follow-up. Serum vitamin D levels were modeled using restricted cubic splines for any HPV (A) and for subgenus 1 (B), 2 C), and 3 (D), using 30 ng/mL as the reference value. Solid lines represents the model estimates, and dashed lines correspond to the 95% confidence intervals. Abbreviation: HR, hazard ratio.
Longitudinal Analysis of HPV Clearance
As shown in Table 4, among women with prevalent HPV infection at baseline (n = 280), we observed a negative association between vitamin D levels at baseline (defined as per 10-ng/mL increase in levels of vitamin D) and subsequent HPV clearance (adjusted HR, 0.76; 95% CI, .60–.96). When modeling exposure as a dichotomous variable, rates of clearance of any HPV tended to be higher among women with lower vitamin D levels (<30 vs ≥30 ng/mL) in an association that approached statistical significance (adjusted HR, 2.14; 95% CI, .99–4.64). There was also a tendency for increased clearance as vitamin D levels decreased based on clinically defined categories (Table 4). Figure 3 shows that the best fit for correlations with vitamin D as a continuous variable were determined by nonmonotonic spline formulations, using 30 ng/mL as the referent. The seemingly higher clearance rate of HPV infections at low concentrations was driven by the correlation seen in infections of subgenus 1 exclusively (Figure 1B).
Table 4.
Association Between Baseline Vitamin D Levels and Clearance of Human Papillomavirus Infection at Follow-up
| Exposure Definition for Vitamin D Levels | Hazard Ratio (95% Confidence Interval) | |||||||
|---|---|---|---|---|---|---|---|---|
| Any HPV | Subgenus 1a | Subgenus 2a | Subgenus 3a | |||||
| Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | Crude | Adjustedb | |
| Continuousc (n = 280) |
0.83 (.68–1.02) | 0.76 (.60–.96) | 0.82 (.65–1.03) | 0.70 (.54–.92) | 0.94 (.79–1.11) | 0.94 (.77–1.15) | 1.06 (.89–1.27) | 1.11 (.90–1.36) |
| Dichotomousd |
||||||||
| Sufficient (n = 63) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficient (n = 217) | 1.72 (.88–3.39) | 2.14 (.99–4.64) | 1.29 (.69–2.40) | 1.78 (.85–3.76) | 1.34 (.75–2.41) | 1.32 (.70–2.50) | 0.64 (.37–1.08) | 0.58 (.30–1.12) |
| Clinically defined categoriese |
||||||||
| Sufficiency (n = 63) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) | 1.00 (Ref) |
| Insufficiency (n = 64) |
1.33 (.58–3.09) | 1.56 (.62–3.93) | 1.10 (.49–2.44) | 1.38 (.53–3.59) | 1.32 (.65–2.72) | 1.17 (.55–2.52) | 0.50 (.23–1.06) | 0.55 (.24–1.31) |
| Deficiency (n = 82) | 1.72 (.79–3.73) | 2.29 (.96–5.50) | 0.90 (.41–1.98) | 0.93 (.34–2.47) | 1.42 (.72–2.82) | 1.65 (.78–3.49) | 0.63 (.31–1.26) | 0.57 (.25–1.32) |
| Severe deficiency (n = 71) | 2.07 (.97–4.39) | 2.90 (1.20–7.05) | 2.00 (.99–4.00) | 4.01 (1.68–9.92) | 1.29 (.65–2.54) | 1.17 (.53–2.55) | 0.78 (.40–1.49) | 0.61 (.27–1.38) |
aSubgenus 1 group includes HPV types 6, 11, 40, 42, 44, and 54; subgenus 2, types 16, 18, 26, 31, 33, 34, 35, 39, 45, 51, 52, 53, 56, 58, 59, 66, 67, 68, 69, 70, 73, and 82; and subgenus 3, types 61, 62, 71, 72, 81, 83, 84, and 89.
bModels were adjusted for age, ethnicity, smoking, age at first sexual intercourse, age at menarche, number of lifetime sex partners, any history of pregnancy, condom use, and season of blood sample collection.
cPer 10-unit increase in levels of vitamin D.
dDichotomized into ≥30 ng/mL (sufficient) and <30 ng/mL (insufficient).
eClinical values were defined as follows: sufficiency, ≥30 ng/mL; insufficiency, 20–29 ng/mL; deficiency, 12–19 ng/mL; and severe deficiency, <12 ng/mL.
Figure 3.
Association between baseline serum vitamin D levels and clearance of human papillomavirus (HPV) infection at follow-up. Serum vitamin D levels were modeled using restricted cubic splines for any HPV (A) and for subgenus 1 (B), 2 (C), and 3 (D), using 30 ng/mL as the reference value. Solid lines represent the model estimates, and dashed lines correspond to the 95% confidence intervals. Abbreviation: HR, hazard ratio.
HPV Type-Specific Infection-Level Analyses
Analyses at the HPV level showed consistent results with those conducted at the woman level. We found no association between vitamin D levels and HPV prevalence (Supplementary Table 4), incidence (Supplementary Tables 5 and 6), or clearance (Supplementary Tables 7 and 8). In particular, the adjusted HR was 0.98 (95% CI, .90–1.07), based on 1495 observations (Supplementary Table 7) for the association between vitamin D levels (continuous form) and HPV clearance.
DISCUSSION
We quantified associations between vitamin D levels and HPV infection prevalence, incidence, and clearance among young women in Montreal, Canada. Overall, we did not find an association between vitamin D levels and either HPV prevalence or incidence. At most, our data are suggestive of low vitamin D levels being associated with increased clearance of HPV in analyses conducted at the individual level. However, given the low precision of our estimates (ie, wide CIs), these findings should be interpreted with caution. Moreover, analyses at the HPV level, which increased statistical precision, confirmed the absence of an association between vitamin D levels and the study outcomes.
Corroborating our findings, a recent cross-sectional analysis of 404 women aged 30–50 years who were enrolled (2011–2012) in an HPV natural history study found no associations between vitamin D serum levels (considered as a continuous and categorical variable) and vaginal high-risk HPV prevalence [9]. On the other hand, our results are at variance with respect to those of 4 other studies that reported a putative protective effect of circulating vitamin D levels against HPV infection. A study of 82 patients with abnormal Papanicolaou test results admitted to a single outpatient clinic found lower mean levels of 25-hydroxyvitamin D3 among HPV DNA–positive women compared with those who tested negative [25]. A study of 67 patients with systemic lupus erythematosus found that those with 25-hydroxyvitamin D levels of <20 ng/mL had slightly elevated prevalence of cervical HPV infection relative to those with levels of ≥20 ng/mL (30.7% vs 25.8%), with a smaller difference observed for high-risk HPV infections (36.8% vs 31.5%) [26].
Shim and colleagues [7] explored the association between vitamin D levels and prevalent vaginal HPV infection using 2003–2006 data for 2353 sexually active women from the National Health and Nutrition Examination Survey. Roche Linear Array HPV genotyping tests for self-collected vaginal swab specimens were performed, which is the same assay we used. Quantification of serum vitamin D levels was done using the Diasorin radioimmunoassay kit (Diasorin). The authors reported 2 outcomes; high-risk HPV (presence of ≥1 of the following HPV types: 16, 18, 31, 33, 35 39, 45, 51, 52, 56, 58, 59, 68), which corresponds to what we defined a subgenus 2 group, and vaccine-type HPV infection (≥1 of the following HPV types: 6, 11, 16, 18). For high-risk HPV, the adjusted ORs were 1.46 (95% CI, .96–2.23), 1.41 (1.00–1.99), and 1.19 (.87–1.62) in women who had, respectively, severe deficiency, deficiency, and insufficiency compared with women with sufficient vitamin D. Equivalent risk estimates for vaccine-type HPV were 2.90 (95% CI, 1.32–6.38), 2.19 (95% CI, 1.08–4.45), and 2.19 (95% CI, 1.22–3.93) [7].
Also based on National Health and Nutrition Examination Survey data (2009–2014), another study among 4343 women aged 18–59 years found that serum vitamin D deficiency (<20 ng/mL) relative to sufficiency (≥20 ng/mL) was associated with an increased risk of high-risk HPV prevalence (risk ratio, 1.25; 95% CI, 1.04–1.49) [8]. A major limitation of the above-mentioned studies was the cross-sectional study design, which cannot establish temporality between vitamin D levels and acquisition of HPV. Two studies were also limited by small sample size and lack of adjustment for confounders [25, 26].
The underlying biological mechanism for an association between vitamin D deficiency and infections is unclear. Some have suggested that vitamin D may be involved in regulating the immune system and may up-regulate the interleukin 37 peptide and human β-defensin 2 [27]. However, the mechanisms by which vitamin D may exert antiviral actions have not been established. A recent systematic review has indicated high prevalence of vitamin D deficiency globally [28]. Thus, given the postulated association between vitamin D and immune response and high prevalence of vitamin D deficiency, numerous randomized controlled trials and observational studies have examined the effect of interventions on vitamin D levels (including vitamin D supplementation) and the risk of acquiring infections (including upper respiratory, tuberculosis, and human immunodeficiency virus) [29]. However, the results from these studies have been mixed partly due to heterogeneous study populations, vitamin D intervention strategies, type of infections, and methodological approaches [29]. There is sparse evidence for a mediating effect for vitamin D levels in cervicovaginal infections.
Several strengths of the current study and analyses need to be acknowledged. First, the HITCH Cohort Study is a unique molecular epidemiologic resource with a relatively large sample size and long follow-up in a young population. This allowed us to account for important potential confounders such as the season of blood collection. We used highly sensitive methods for vitamin D testing, as well as for HPV detection and genotyping. Second, we used 4 exposure definitions including a continuous and dichotomous categorization of vitamin D levels (<30 or ≥30 ng/mL), clinically defined categories, and modeling of vitamin D levels using restricted cubic splines. The latter exposure definition allowed us to examine the nonlinear relationship between vitamin D levels and incidence and clearance of HPV, analyses that were not performed in previous studies. Third, in addition to examining the association between vitamin D levels and prevalence of HPV, the longitudinal study design of HITCH allowed us to also assess the association between baseline vitamin D levels and the incidence of HPV infection in HPV-negative women at baseline as well as HPV clearance among those who were initially HPV positive. Finally, we conducted HPV type-specific analyses, which largely corroborated those conducted with women as units of observation, albeit with more statistical precision.
Our study also had limitations. As in previous observational studies, we measured vitamin D levels only at baseline. This may have introduced exposure misclassification if serum vitamin D levels fluctuated between subsequent visits. Second, our estimates of associations using clinically defined categories of vitamin D were imprecise. Thus, the results from these analyses should be interpreted with caution. Finally, our results may not be generalizable to other populations of women as our sample consisted of young university students.
In conclusion, we did not corroborate the previously reported association between low serum vitamin D concentration and increased risk of genital HPV infection. At most, our findings are mildly suggestive of an association between low vitamin D levels and increased clearance, particularly of low-risk HPV types, a seemingly paradoxical finding that should be verified in larger studies.
Supplementary Data
Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.
Notes
Acknowledgments. We thank the volunteering participants and the following employees of the HPV Infection and Transmission among Couples through Heterosexual activity (HITCH) Cohort Study for study promotion: Vicky D’Anjou-Pomerleau, Jennifer Selinger, Elizabeth Montpetit-Dubrule, Jessica Sammut, Emilie Lapointe, Johanna Bleecker, and Shady Rahayel. We also thank Melanie Drew (Student Health Services Clinic, Concordia University) and the staff of the Student Health Services Clinics at McGill and Concordia universities for their collaboration with HITCH research nurses, and we acknowledge the research assistants Sheila Bouten and Lina Sobhi Abdrabo for aliquoting of blood specimens.
Additional HITCH study group members. Affiliated with the Division of Cancer Epidemiology, McGill University, Montréal, Canada: Allita Rodrigues (study coordinator); Gail Kelsall, Suzanne Dumais, Natalia Morykon, and Amelia Rocamora (management of subject participation and specimen collection); Nathalie Slavtcheva (study management); and Veronika Moravan and Michel Wissing (data management). Affiliated with the Département de Microbiologie Médicale et Infectiologie, Centre Hospitalier de l’Université de Montréal, Montréal, Québec, Canada: Michel Roger (collaborator).
Author contributions. A. N. B. and E. L. F. designed the parent HITCH study. P. P. T. was responsible for clinical oversight of the study. F. C. conducted the human papillomavirus (HPV) assays and their interpretation. M. E. Z. and E. L. F. formulated the research question. S. E. and E. M. contributed to the laboratory analyses of serum specimens. M. E. Z. and F. K. K. performed statistical analyses and drafted the manuscript. All authors commented on interim drafts and read and approved the final manuscript.
Disclaimer. The funders played no role in study design, data collection and analysis, preparation of the manuscript, or the decision to publish.
Financial support. The HITCH Cohort Study was supported by the Canadian Institutes of Health Research (grant MOP-68893 and team grant CRN-83320 to E. L. F.) and the US National Institutes of Health (grant RO1AI073889 to E. L. F.). This work was also supported by the Réseau FRQS SIDA-MI (support for HPV testing) and by Merck-Frosst Canada and Merck & Co.
Potential conflicts of interest. E. L. F. reports grants and personal fees from Merck outside of the submitted work. M. E. Z. and E. L. F. hold a patent related to the discovery “DNA methylation markers for early detection of cervical cancer,” registered at the Office of Innovation and Partnerships, McGill University, Montreal, Quebec, Canada (October 2018). A provisional utility patent application before the United States Patent and Trademark Office was also filed (November 2018) and a patent cooperation treaty application (PCT/IB2020/050885), filed in February 2020, has been published (no. WO 2020/115728; June 2020). All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.
Presented in part: International Papilloma Virus (IPV2017) conference, Cape Town, South Africa, March 2017.
Contributor Information
for the HITCH study group:
Allita Rodrigues, Gail Kelsall, Suzanne Dumais, Natalia Morykon, Amelia Rocamora, Nathalie Slavtcheva, Veronika Moravan, Michel Wissing, and Michel Roger
References
- 1. Walboomers JM, Jacobs MV, Manos MM, et al. Human papillomavirus is a necessary cause of invasive cervical cancer worldwide. J Pathol 1999; 189:12–9. [DOI] [PubMed] [Google Scholar]
- 2. IARC Monographs on the evaluation of carcinogenic risks to humans. Human papillomaviruses, Vol. 90. World Health Organization. Lyon, France: International Agency for Research on Cancer; 2007. [Google Scholar]
- 3. Forman D, de Martel C, Lacey CJ, et al. Global burden of human papillomavirus and related diseases. Vaccine 2012; 30(suppl 5):F12–23. [DOI] [PubMed] [Google Scholar]
- 4. Chaturvedi AK. Global burden of human papillomavirus-positive head and neck cancers. Lancet Oncol 2014; 15:1282–3. [DOI] [PubMed] [Google Scholar]
- 5. Cutts FT, Franceschi S, Goldie S, et al. Human papillomavirus and HPV vaccines: a review. Bull World Health Organ 2007; 85:719–26. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Bruni L, Diaz M, Castellsagué X, Ferrer E, Bosch FX, de Sanjosé S. Cervical human papillomavirus prevalence in 5 continents: meta-analysis of 1 million women with normal cytological findings. J Infect Dis 2010; 202:1789–99. [DOI] [PubMed] [Google Scholar]
- 7. Shim J, Pérez A, Symanski E, Nyitray AG. Association between serum 25-hydroxyvitamin D level and human papillomavirus cervicovaginal infection in women in the United States. J Infect Dis 2016; 213:1886–92. [DOI] [PubMed] [Google Scholar]
- 8. Gupta A, Villa A, Feldman S, Citow B, Sroussi H. Site and sex-specific differences in the effect of vitamin D on human papillomavirus infections: analyses of NHANES 2009–2014. Sex Transm Infect 2021; 97:75–76. doi: 10.1136/sextrans-2020-054466. [DOI] [PubMed] [Google Scholar]
- 9. Troja C, Hoofnagle AN, Szpiro A, Stern JE, Lin J, Winer RL. Serum concentrations of emerging vitamin D biomarkers and detection of prevalent high-risk HPV infection in mid-adult women. Cancer Epidemiol Biomarkers Prev 2020; 29:1468–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Holick MF, Chen TC. Vitamin D deficiency: a worldwide problem with health consequences. Am J Clin Nutr 2008; 87:1080S–6S. [DOI] [PubMed] [Google Scholar]
- 11. Dankers W, Colin EM, van Hamburg JP, Lubberts E. Vitamin D in autoimmunity: molecular mechanisms and therapeutic potential. Front Immunol 2017; 7:697. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Zittermann A, Pilz S, Hoffmann H, März W. Vitamin D and airway infections: a European perspective. Eur J Med Res 2016; 21:14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Hensel KJ, Randis TM, Gelber SE, Ratner AJ. Pregnancy-specific association of vitamin D deficiency and bacterial vaginosis. Am J Obstet Gynecol 2011; 204:41.e1–9. [DOI] [PubMed] [Google Scholar]
- 14. He Q, Ananaba GA, Patrickson J, et al. Chlamydial infection in vitamin D receptor knockout mice is more intense and prolonged than in wild-type mice. J Steroid Biochem Mol Biol 2013; 135:7–14. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Turner AN, Carr Reese P, Chen PL, et al. Serum vitamin D status and bacterial vaginosis prevalence and incidence in Zimbabwean women. Am J Obstet Gynecol 2016; 215:332.e1-10. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Turner AN, Carr Reese P, Fields KS, et al. A blinded, randomized controlled trial of high-dose vitamin D supplementation to reduce recurrence of bacterial vaginosis. Am J Obstet Gynecol 2014; 211:479.e1–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. El-Zein M, Coutlée F, Tellier PP, Roger M, Franco EL, Burchell AN; HITCH Study Group . Human papillomavirus infection and transmission among couples through heterosexual activity (HITCH) Cohort study: protocol describing design, methods, and research goals. JMIR Res Protoc 2019; 8:e11284. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Coutlée F, Rouleau D, Petignat P, et al. ; Canadian Women’s HIV study Group . Enhanced detection and typing of human papillomavirus (HPV) DNA in anogenital samples with PGMY primers and the Linear array HPV genotyping test. J Clin Microbiol 2006; 44:1998–2006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Institute of Medicine Food and Nutrition Board. Dietary reference intakes for calcium and vitamin D. In: Ross AC, Taylor CL, Yaktine AL, Valle HBD, eds. Washington, DC: National Academy Press; 2011. [PubMed] [Google Scholar]
- 20. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med 1989; 8:551–61. [DOI] [PubMed] [Google Scholar]
- 21. Harrell F. Regression modeling strategies: with applications to linear models, logistic and ordinal regression, and survival analysis. Springer International; 2015. [Google Scholar]
- 22. Schiffman M, Wentzensen N, Wacholder S, Kinney W, Gage JC, Castle PE. Human papillomavirus testing in the prevention of cervical cancer. J Natl Cancer Inst 2011; 103:368–83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Schiffman M, Clifford G, Buonaguro FM. Classification of weakly carcinogenic human papillomavirus types: addressing the limits of epidemiology at the borderline. Infect Agent Cancer 2009; 4:8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24. Schiffman M, Herrero R, Desalle R, et al. The carcinogenicity of human papillomavirus types reflects viral evolution. Virology 2005; 337:76–84. [DOI] [PubMed] [Google Scholar]
- 25. Özgü E, Yılmaz N, Başer E, Güngör T, Erkaya S, Yakut Hİ. Could 25-OH vitamin D deficiency be a reason for HPV infection persistence in cervical premalignant lesions? J Exp Ther Oncol 2016; 11:177–80. [PubMed] [Google Scholar]
- 26. García-Carrasco M, Mendoza-Pinto C, Munguía-Realpozo P, et al. Lack of association between serum 25-hydroxyvitamin D levels and cervical human papillomavirus infection in systemic lupus erythematosus. Lupus 2015; 24:606–12. [DOI] [PubMed] [Google Scholar]
- 27. Gombart AF. The vitamin D–antimicrobial peptide pathway and its role in protection against infection. Futur Microbiol 2009; 4:1–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Palacios C, Gonzalez L. Is vitamin D deficiency a major global public health problem? J Steroid Biochem Mol Biol 2014; 144 Pt A:138–45. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Yamshchikov AV, Nirali DS, Blumberg HM, Ziegler TR, Tangpricha V. Vitamin D for the treatment of infectious diseases: a systematic review. Endocr Pr 2009; 15:438–49. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.