Vitamin D Deficiency and Periodontal Clinical Attachment Loss in HIV Seropositive Women: A Secondary Analysis Conducted in the Women’s Interagency HIV Study (WIHS)

Abstract

Aim

To test a hypothesized positive association between low Vitamin D (VitD) serum levels and the severity of periodontal disease in HIV-infected women.

Materials and Methods

This is a cross sectional secondary analysis of data from an oral substudy conducted within the Chicago site of the Women's Interagency HIV Study. Serum VitD levels and clinical attachment loss measurements were available for 74 HIV-infected women. VitD levels were treated as both continuous and categorical variables in bivariate and multivariate analyses. Mean clinical attachment loss (mCAL) was determined for each subject by averaging measurements taken at 4 sites in each measured tooth.

Results

Average age of study participants (n=74) was 39.6 years (SD 7.2) and the majority were African-American (70.3%) and VitD deficient (58.1%). VitD deficiency was positively associated with higher mCAL (p=.012). With adjustment for race, age, smoking, HIV viral load, an association was found between VitD deficiency and mCAL (Beta 0.438, p=0.036).

Conclusions

We identified a previously unreported association between VitD deficiency and mCAL in HIV-infected women. Larger and more inclusive multisite longitudinal studies are warranted to investigate whether these findings generalize to all HIV infected individuals in the current treatment era and to determine causality.

Introduction

Human immunodeficiency virus (HIV) associated oral pathologies have long included candidiasis, hairy leukoplakia, and necrotizing periodontal diseases¹. The relationship of HIV infection with more common, and perhaps less severe forms of oral diseases, including chronic periodontitis is less clear. Early studies reported greater periodontal disease severity in HIV seropositive (HIV+) men² when compared to HIV seronegative controls, and more rapid periodontal disease progression in HIV infected patients with declining CD4 lymphocyte counts³. These findings were not confirmed by others⁴. Alves et al reported no differences in periodontal disease severity and progression when comparing HIV-infected with HIV-uninfected women in the Women Interagency Health Study (WIHS); with no association found between CD4 counts or HIV viral load and clinical attachment loss (CAL).

In the industrialized world, many HIV+ patients receive highly active antiretroviral therapy (HAART) and achieve excellent viremic control and immune reconstitution⁵. Following the introduction of HAART in 1996, life expectancy and quality of life began to increase ⁵; thus, many patients aging with HIV now present with age-related diseases including chronic periodontitis⁶. Vitamin D (VitD) deficiency has been recently reported as a risk factor of periodontal disease⁷ in the general US adult population. With increased longevity, non-HIV related factors such as VitD status, may emerge as potential modifiers of periodontal disease in individuals living with HIV.

It is estimated that 25–30% of the US population is VitD deficient (defined as <20ng/mL) ⁸, this deficiency is even more prevalent in people living with HIV, especially among HIV+ women⁹. The most prominent role of VitD in the human body is the regulation of calcium and phosphate homeostasis¹⁰. Beyond that role, VitD has also been reported to have antimicrobial¹¹ and anti-inflammatory activities^12,13. More specifically, VitD induces the expression of cathelicidin¹¹, which is a host peptide with antibacterial properties. VitD is also known to downregulate human dendritic cells and T-/B- cell activity¹⁴. Both 25(OH)D and 1,25(OH)²D³ have been found to inhibit the production of pro-inflammatory cytokines including IFNγ, IL-6, IL-17, IL-21 and TNF-α through the reduction of NFκB activity¹², to suppress prostaglandin pathways via inhibition of cyclooxygenase-2 production¹⁵, and to have an inverse association with CRP levels¹⁶. Possible associations between VitD status and periodontal health are likely mediated by these immune mechanisms.

According to Dietrich et al. ¹⁷, VitD serum levels are inversely associated with CAL and gingival inflammation. Also, circulating VitD levels have been found to be lower in patients with periodontal disease compared to healthy individuals^{18, 19}. VitD deficient patients do not seem to respond as well to periodontal surgery²⁰. Serum VitD was found to be inversely associated with incidence of tooth loss, an effect shown to be at least partially explained by less active periodontitis in patients who were not VitD insufficient²¹. Notwithstanding the reports summarized above, there is no consensus on the link between VitD deficiency and greater periodontal disease severity as other studies have not found such an association²¹. Recently, Ferreira et al reported no association between periodontitis and VitD status in a cohort of adult Brazilians living with HIV²². Dissimilarities in defining thresholds for VitD deficiency as well as significant differences in the racial and gender composition of the studied populations together with differences in comorbidities could explain the lack of consensus in the studies of VitD status and periodontal health.

In the present study, we aimed at elucidating the possible association between VitD status and periodontal health among women living with HIV. VitD deficiency is highly prevalent among US women living with HIV²³ with 60% of HIV-seropositive WIHS subjects reportedly VitD deficient. VitD deficiency is associated with HIV disease progression²⁴, with AIDS defining events²⁵, and with an increased prevalence of opportunistic infections including oral candidiasis^26,27. In this study, we investigated the association between VitD status and periodontal disease severity as measured by mean Clinical Attachment Loss (mCAL) in a convenience sample of HIV+ women in the Chicago WIHS. We hypothesized a positive and independent association between that VitD deficiency and worse periodontal disease.

Materials and Methods

We conducted a secondary cross-sectional analysis of data available from HIV-seropositive Chicago women participating in both an oral substudy and a biomarker study of oral candidiasis and VitD²⁶ of the Women's Interagency HIV Study (WIHS). The WIHS is an ongoing prospective cohort study of HIV-infected and uninfected at-risk women. A subset of WIHS participants were enrolled in a nested study of oral health. More information on the design and conduct of the WIHS, including the oral substudy ^28,29 and VitD assessments can be found elsewhere²⁷. For each subjects with ten or more natural teeth, 2 random quadrants (left or right) were selected whereas subjects with fewer than ten natural teeth had all their teeth examined by a calibrated examiner. The teeth were assessed at four sites per tooth: i.e., three on the buccal (distal [D], midbuccal [MB], and mesial [M]) and one on the lingual (midlingual [L]) using a mirror and a NIDCR probe. The mCAL for each subject was calculated by averaging all the measurements from all assessed teeth. Additional methodology and data on baseline periodontal health in the WIHS cohort have been described elsewhere ⁴.

Participants

Written informed consent was obtained from all participants in accordance with the US Department of Health and Human Services guidelines and the institutional review boards of Cook County Health and Hospital Systems and the University of Illinois Institutional Review Board. This analysis consisted of 74 non-edentulous HIV-infected women previously enrolled in a correlation study of VitD status and oral candidiasis²⁶ and for whom there were both VitD serum levels and periodontal disease measurements from the same visit. The visit dates spanned from 1995 to 2004. Available data with recoding of original variables, included (i) age at visit, (ii) race (African-American/Non-African American), (iii) current smoking status (yes/no), (iv) education (high school or above/no high school completion, (v) CD4 cell count (<200, 200–499, or ≥500 cells/mL), (vi) HIV Viral Load (≤4,000, >4,000 copies/mL) and (vii) HAART (yes/no).

Statistical analysis

Statistical analysis was performed using IBM SPSS (version 19.0). Bi-and multi-variate analyses were conducted using variables focusing on the associations between the continuous and two categorical thresholds of VitD³⁰ categorical [insufficiency (<30 ng/mL) and deficiency (<20 ng/mL)] with mCAL. VitD was analyzed using three approaches: continuous as reported from Quest Diagnostics laboratory, VitD insufficiency stratified at <30 ng/ml versus above 30, and VitD deficiency stratified at <20 ng/ml versus above 20. Potential confounders were assessed for statistical significance in the dataset and were considered from the context of biological plausibility. Bi-variate categorical assessments included Fisher Exact Test (FET) for 2 × 2 tables or Chi Square or linear-by-linear associations for variables with greater than 2 categories. For continuous variable bi-variate assessments, the means of continuous variables versus categorical variables were tested for difference by using independent student’s t-test in pairs or for pairs of continuous variable, the variables were assessed for correlation by bi-nominal Pearson’s correlation. Strong co-linearity (Pearson correlation r=−0.506, p<0.001) was found between age and number of teeth such that only age was considered in the multivariate analysis. Multivariate assessment was conducted via linear regression using mCAL as the outcome variable, considering crude and adjusted relationships with the three VitD variables (continuous, insufficiency and deficiency).

Results

Characteristics of the study participants included in the analysis and by VitD grouping are presented in Table 1. Participants had a mean age of 39.6 (SD ±7.2) years and a mean VitD level of 20.3 ng/mL (SD ±12.9). The study group was predominantly African-American (70.3%); most of the group had less than high school education (56.8%), listed as a current smoker (52.7%) and tested as VitD deficient (58.1%). Few individuals (12%) reported using HAART (expected, given that study visits occurred before and during early availability of HAART). African-Americans compared to non-African- Americans were more likely to be VitD insufficient (88.5% vs 68.2%. p=.049) and deficient (67.3% vs. 36.4%, p=0.014). Current smokers were found to be more likely to be VitD deficient (69.2% vs. 44.1%, FET=0.030). Average mCAL for the cohort was 1.24 mm (SD ±0.88). VitD deficient subjects (VitD ≤20 ng/mL) had a statistically significant higher mCAL of 1.45 mm (SD ±0.98) versus a mCAL of 0.96 mm (SD ±0.65) for VitD non-deficient subjects (p= 0.012).

Table 1.

Characteristics of the study sample by Vitamin D status: Chicago WHIS 1995-2004

		Continuous		Below Sufficient (Insufficient)			Below Deficient (Deficient)
Characteristic	Total	Vit D ng/mL Mean (SD)	p- value	Vit D ≤30 ng/mL N (%)	Vit D >30 ng/mL N (%)	p- value	Vit D ≤20 ng/mL N (%)	Vit D >20 ng/mL N (%)	p- value
Total	74 (100%)	20.3 (12.9)	−	61 (82.4%)	13 (17.6%)	−	43 (58.1%)	31 (41.9%)	−
Mean Clinical Attachment Loss (SD)	1.24 (0.88)	Corr - 0.220	0.060	1.31 (0.90)	0.95 (0.78)	0.194	1.45 (0.98)	0.96 (0.65)	0.012
Demographics
Mean Age, years (SD)	39.6 (7.2)	Corr - 0.226	0.055	40.2 (7.2)	36.7 (6.7)	0.108	40.9 (7.3)	37.8 (6.8)	0.065
Race: African-American	52 (70.3%)	18.5 (13.0)	0.065	46 (88.5%)	6 (11.5%)	0.049 FET	35 (67.3%)	17 (32.7%)	0.014
Non-African-American	22 (29.7%)	24.5 (11.9)		15 (68.2%)	7 (31.8%)		8 (36.4%)	14 (63.6%)
Education: High School or above	32 (43.2%)	21.4 (11.8)	0.552	23 (74.2%)	8 (25.8%)	0.125	15 (48.4%)	16 (51.6%)	0.174
Below High School	42 (56.8%)	19.6 (13.8)		37 (88.1%)	5 (11.9%)		27 (64.3%)	15 (35.7%)
Smoker: Current	39 (52.7%)	18.4 (13.8)	0.180	34 (87.2%)	5 (12.8%)	0.233	27 (69.2%)	12 (30.8%)	0.030
Non/Non-current	35 (46.7%)	22.5 (11.7)		26 (76.5%)	8 (23.5%)		15 (44.1%)	19 (55.9%)
Dental Status
Mean Number of Teeth (SD)	20.7 (6.8)	Corr 0.316	0.006	19.9 (6.9)	24.6 (4.8)	0.007	19.5 (6.8)	22.3 (6.5)	0.081
HIV Status
Mean CD4 (SD) (cells/mL)	355 (289)	Corr 0.228	0.053	330 (276)	473 (327)	0.106	323 (286)	399 (291)	0.271
CD4 Cat – 3 levels (cells/mL)
1. <200	25 (33.8%)	16.9 (10.6)	0.282 ANOVA	22 (91.7%)	2 (8.3%)	0.198 FET 1 vs. 2,3	16 (66.7%)	8 (33.3%)	0.529
2. 200-499	29 (39.2%)	22.5 (15.9)		23 (79.3%)	6 (20.7%)		15 (51.7%)	14 (48.3%)
3. ≥500	20 (27.0%)	21.1 (10.4)		15 (75.0%)	5 (25.0%)		11 (55.0%)	9 (45.0%)
Mean Viral Load (SD), 1,000 copies/mL	124 (377)	Corr - 0.093	0.433	126 (386)	116 (342)	0.931	80 (193)	187 (540)	0.303
Viral Load Cat – 2 levels (copies/mL)
1. ≤4,000	30 (40.5%)	20.1 (10.6)	0.642	24 (80.0%)	6 (20.0%)	0.534 FET	14 (46.7%)	16 (53.3%)	0.076
2. >4,000	44 (58.7%)	18.9 (11.0)		37 (86.0%)	6 (14.0%)		29 (67.4%)	14 (32.6%)
HAART: Yes	9 (12.0%)	14.6 (8.1)	0.158	9 (100%)	0 (0%)	0.346 FET	7 (77.8%)	2 (22.2%)	0.288 FET
NO	65 (88.0%)	21.1 (13.2)		52 (80.0%)	13 (20.0%)		36 (55.4%)	29 (44.6%)

Associations between mCAL and subjects’ characteristics are presented in Table 2. A positive correlation between age and mCAL was demonstrated (r=.409, P<0.001). Current smokers had an almost 50% higher mCAL that nonsmokers (1.47 ± 0.89 mm vs.1.01 ± 0.81 mm, p=0.024). While no association was found between mCAL and ethnicity or CD4 counts, higher HIV viral load (above the 4,000 copies/mL) was significantly associated with higher mCAL (0.92 ± 0.59 mm for viral load ≤4,000 versus 1.41 ± 0.95 mm for viral load >4,000; p=0.008).

Table 2.

Association Between Characteristics of the Study Sample and mean Clinical Attachment Loss (mCAL): Chicago WHIS 1995–2004

Characteristic	mCAL Mean (SD)	p
Total	1.24(0.88)	−
Demographics
Age	Corr 0.409	<0.001
Race:		0.415
African-American	1.19 (0.83)
Non-African-American	1.37 (1.01)
Education:		0.931
High School or above	1.24 (.85)
Below High School	1.22 (.95)
Smoke:		0.024
Current	1.47 (.89)
Non/Non-Current	1.01 (.81)
HIV Status
CD4 (SD) (cells/mL)	Corr -0.101	0.393
CD4 Cat – three levels (cells/mL)		0.270
1. <200	1.47 (1.09)	ANOVA
2. 200–499	1.08 (0.73)
3. ≥500	1.20 (0.80)
Viral Load (SD) × 1,000 copies/mL	Corr 0.239	0.041
Viral Load Cat – two levels copies/mL		0.008
1. ≤4,000	0.92 (0.59)
2. >4,000	1.41 (0.95)
HAART:		0.602
YES	1.10 (0.92)
NO	1.26 (0.88)

Corr = 2-tailed Spearman correlation

The multivariate linear regression models assessing VitD and mCAL are presented in Table 3. The fully adjusted models considered the following variables as potential confounders: current smoking status, viral load (in two categories), age, and race. VitD was included in the models as a continuous variable (Model 1) or a dichotomous variable based on insufficiency (Model 2) and deficiency (Model 3). After adjusting for all the variables described above if significant at P<0.10, VitD as a continuous variable (beta -0.021, p=0.028) and VitD deficiency (beta 0.438, p=0.036) were associated with mCAL.

Table 3.

Linear regressions on mCAL by Vit D definition (Crude and Adjusted): Chicago WHIS 1995–2004

Vitamin D Definition		Crude Beta (p-value)	Adjusted Model Beta (p-value)
Vitamin D Continuous		−0.006 (0.432)	Model 1	−0.021 (0.028)
Age at visit (years)				0.030 (0.022)
Race (AA/non-AA)				−0.464 (0.025)
Viral load (dichotomized)				0.402 (0.230)
Current Smoker				0.306 (0.104)
Vitamin D Insufficiency		0.354 (0.194)	Model 2	0.420 (0.109)
Age at visit (years)				0.032 (0.015)
Race (AA/Non-AA)				−0.417 (0.044)
Viral load (dichotomized)				0.393 (0.038)
Current Smoker				0.363 (0.054)
Vitamin D Deficiency		0.389 (0.061)	Model 3	0.438 (0.036)
Age at visit (years)				0.031 (0.017)
Race (AA/non-AA)				−0.449 (0.029)
Viral load (dichotomized)				0.330 (0.081)
Current Smoker				0.302 (0.111)

Discussion

This analysis investigated the association between Vitamin D serum levels and the severity of periodontal disease as measured by mean Clinical Attachment Loss (mCAL) in HIV+ women. Vitamin D deficient subjects had a 50% greater mCAL compared to those who were not Vitamin D deficient. The effect of Vitamin D deficiency on mCAL was similar in scale in our sample to that of the well-known deleterious effect of current tobacco smoking ^31,32 or that seen in association with higher HIV viral load³³. This emphasizes the potential clinical significance of Vitamin D status to periodontal health.

To this day, there is no consensus as to the threshold of serum VitD levels that would predict significantly inferior health outcomes. It is likely to differ in different populations and depend on the health outcome(s) being studied. The multivariate analysis presented here supports the relevancy of Vitamin D deficiency rather than insufficiency for its association with worse periodontal health. This is consistent with what was seen in oral candidiasis²⁶ and bacterial vaginosis²⁷ in two studies conducted in the WIHS.

Our analysis describes an association between Vitamin D status and periodontal disease in HIV infected women in the Chicago WIHS. Similar findings of an association between Vitamin D deficiency and periodontal disease have been reported in non-HIV cohorts. Millen et al. reported that Vitamin D sufficient women had a 33% lower odds of periodontal disease than women with insufficient Vitamin D serum levels³⁴. Jabbar et al. compared women with and without periodontal disease (active or past) and observed that the serum concentrations of Vitamin D were significantly lower in those with either active or past disease¹⁸.

In a recently reported study, no association between VitD status and periodontal disease was noted in Brazilian adults living with HIV²². This study was conducted in a cohort of women and men subjects who were mostly on HAART (94.8%). This could explain, in-part, the apparent discrepancy between our study and that of Ferreira et al. It should be noted that latest data from the Center for Disease Control and Prevention (CDC) indicate that approximately a third of the people living with HIV in the United States are receiving antiretroviral therapy³⁵.

Our report has strengths and limitations. While statistically significant differences were seen in comparing mCAL in different Vitamin D status groups, the mCAL numbers were small and the clinical significance of their differences remained uncertain. As noted above, differences in mCAL between Vitamin D groups were similar in magnitude to what was found between tobacco smokers and non-smokers and in people with lower or higher plasma HIV viral load. As tobacco smoking has a well-established and generalizable effect on periodontal health^{31,32,36,37,38}, the clinical relevancy of the Vitamin D related observations of this analysis cannot be disregarded.

The availability of the WIHS cohort data collected for other purposes provided an opportunity to do this analysis. The sample was however of convenience and of HIV infected adult women, mostly African-American, living in Chicago. Therefore, it is not representative of persons living with HIV throughout the US particularly since geographic location is a known predictor of Vitamin D status. The study was based on a split mouth design and therefore may not provide an accurate representation of the periodontal status of each patient³⁹. The subjects received dental care because of their participation in the WIHS²⁹. Since dental therapy includes extraction of hopeless teeth and periodontal treatment, it is possible that this study underestimated the severity of periodontal disease.

As data collection took place between 1995 and 2004, a small percentage of individuals were receiving HAART. This highly beneficial pharmaceutic regimen became available in 1996 in the United States but uptake among women in WIHS was not immediate⁴⁰. While this is a limitation of our study, the HIV care cascade from the latest data made available by the CDC³⁵ indicate that more than 60% of people living with HIV in the United States remain untreated, making our study not irrelevant to today’s conditions.

HAART regimen effectiveness and administration simplicity were different compared to what is available today⁵. Extrapolation of these results to HIV seropositive individuals of today is thus challenging. None the less, the findings reported in our study are novel and are of significant clinical interest especially for inspiring and contrasting with future studies done in the modern HAART era.

In conclusion, Vitamin D status may be relevant to the periodontal health of people living with HIV. As HIV+ women have a statistically significant higher mCAL when Vitamin D deficient, it would be important to monitor whether mCAL progresses at a faster rate and becomes clinically meaningful in that population. Future research should aim at assessing prospectively the association of Vitamin D deficiency in periodontal disease in today’s HIV population. Possible racial/ethnic based differences should also be carefully evaluated. Ultimately, the effects of Vitamin D supplementation on periodontal health in prospective controlled clinical trials should be considered.

Clinical Relevance.

In this study, we show that low serum vitamin D levels are independently associated with more loss of clinical attachment in women living with HIV. It is therefore possible that vitamin D deficiency contributes to periodontal disease in that population.