The Influence of Skin Pigmentation on UVB Induced Vitamin D Increase

Background. Melanin molecules (pigment) in skin, absorb UVB and presumably reduce the photo-conversion of 7-dehydrocholesterol. Epidemiological studies have reported poor vitamin D status in relatively darker-skinned immigrants residing in countries located at high latitudes. It is therefore commonly believed that skin pigmentation is a major influencer of vitamin D status. However, controlled UVB intervention studies have yielded conflicting results regarding the influence of skin pigmentation on UVB induced vitamin D increase.

Study hypothesis. The investigators hypothesized that objectively measured skin pigmentation (assessed as Pigment Protection Factor) and/or pigment genes influence UVB-induced 25-hydroxyvitamin D (25(OH)D) increase significantly in combination with other influential parameters. These other influential parameters include sex, age, weight, height, BMI, number of fatty fish meals per week, Fitzpatrick Skin Type and 25(OH)D start level. The genetic parameters include 33 Vitamin D receptor- and pigment-SNPs.

Study design. This is a single-centre, open and non-blinded clinical trial.

Intervention. The study was planned so that all included participants would receive identical UVB doses gradually increasing over nine weeks from November to January. The total UVB dose during the nine weeks was planned to be 26 kilo Joules/m2 (56 standard erythema doses). Irradiation time would be determined and regulated by measuring UV intensity with a Sola-Hazard spectroradiometer at study start, after five weeks and at the end of the exposure period. To avoid erythema, the standard erythema dose must be below the lowest measured UVB tolerance in the participants included. Based on previous experience of measurement on light-skinned individual, the UVB treatment dose was set to a maximum of 1.4 kilo Joules/m2 (3 standard erythema doses).

Methodological aspects.

Skin pigmentation. The content of melanin in epidermal biopsies is significantly correlated to measurement of pigment protection factor (PPF). This measurement of PPF can be performed non-invasively with a skin reflectance meter (UV-Optimize Scientific, Chromo-light, Denmark). In the following, PPF is therefore used as a measure of skin pigmentation and referred to as such. The measuring range of PPF is 1 to 25, meaning that a UVB dose between 1 standard erythema dose and 25 standard erythema doses is needed to elicit a just perceptible erythema. The constitutive PPF was measured on the buttocks, a body area not normally exposed to UVB. The facultative PPF was assessed as a mean of measurements on the chest, midriff, back of shoulder, and the medial and lateral sides of the arm. These are body areas influenced by prior lifelong solar exposure. Both constitutive PPF and facultative PPF were measured at study start, after 4½ weeks, and at study end. Mean values at these time-points for each parameter were used for further investigation. Furthermore, subjects self-reported their skin photo-type according to Fitzpatrick's criteria was assessed.

Blood analysis. Serum 25(OH)D was analyzed on a liquid chromatography tandem mass spectrometer (LC-MS/MS), which is the current most accurate method for this analysis. This method has been cross checked against another clinical laboratory and proven reliable.

Analytical variability. To minimize analysis variance, at least triplet analyses (technical replicates) were performed of PPF measurement and 25(OH)D analysis. All 25(OH)D samples from the same subject were analyzed in one batch. An internal standard solution with 25(OH)D was used in each run.

The total relative standard deviation (SD) varied between 4.9% at 20 nmol/l and 14.1% at 222 nmol/l reflecting experimental variability.

Biological variability. By including 10 measuring time points for serum 25(OH)D per participant, the intra-individual variability was reduced. Serum 25(OH)D is known to variate with season and solar exposure. This study was therefore conducted during winter half-year, when ambient UVB radiation and solar-exposed body areas are negligible at this time of the year (17).

Genetic parameters. SNP genotyping. A total of 33 SNPs were investigated. SNPs located in the Vitamin D Receptor (VDR) gene. The influence of the vitamin D receptor gene was investigated by genotyping the two single nucleotide polymorphisms (SNP), rs1544410 (BsmI) and rs2228570 (FokI), located in the gene (ENSG00000111424, Chromosome 12q13) as previously described. These two SNPs were selected due to their association with bone diseases as well as many other diseases, thereby indicating that polymorphisms in those regions of VDR were important for the functioning of VDR. These two SNPs were investigated together with the other investigated demographic parameters.

SNPs located in genes with influence on pigment. Thirty-one SNPs were selected according to potential influence on pigment.

All SNPs were genotyped. SNPs with genotype subgroups containing less than five subjects were merged with other allele-sharing subgroups. Heterozygote subgroups with less than five subjects was merged with an allele-sharing subgroup displaying insignificant (P > 0.05) difference in influence on the 25(OH)D increase rate. SNPs with no allele dose effect or dominant allele effect on 25(OH)D were excluded. For SNPs with dominant allele effect, subgroups with no significant differences in influence on 25(OH)D increase rate were merged according to allele sharing.

Selection of participants. Participants were included and allocated into a light-skinned group and a darker-skinned group based on Fitzpatrick Skin Type (self-reported skin photo-type) and ethnic origin. The light-skinned included Fitzpatrick Skin Types I-IV originating from Northern countries (Denmark, the Faroe Islands and the UK). The darker-skinned included Fitzpatrick Skin Types V-VI originating from countries located at latitudes below 50 degrees N. Thus, it could be ensured that the participants represented a wide range of skin pigmentation. All participants must reside in Denmark. Furthermore, an equal sex distribution was attempted.

Statistical considerations.

Randomization. No randomization was used, as the participants were allocated into two groups based on other criteria at inclusion. Both groups received identical intervention and blinding was therefore not used.

Size of study. SD was 14.1 nmol/l for high-end 25(OH)D samples. On the basis of a previous study lasting two weeks (10), the expected detectable difference was 15 nmol/l. The sample size corresponding to number of subjects required to complete the study was 16 subjects in each group, given a significance level of 5% and 80 % power to detect a difference of 15 nmol/l. Due to the relatively long study period, a minimum of 21 subjects were included in each group allowing five drop-outs. Three subjects dropped out from the darker-skinned group, and none from the light-skinned group.

Data collection. All participants were assigned with a unique study number prior to study start. Demographic data (gender, age, weight, height, Fitzpatrick Skin Type, measured constitutive and facultative skin pigmentation (PPF)) was collected/measured and registered in prior to study start. The number of daily consumed fatty fish meals was recorded in a questionnaire. All forms were marked with the participants unique study number. 25(OH)D samples were analyzed after study end. The participant study number linked to participant contact details were held in a separate sheet kept securely and in line with the regulation of the Danish Data Protection Agency.

Background diet and change in diet during intervention. The number of daily consumed fatty fish meals was recorded in a questionnaire during the study period as this parameter potentially can influence the primary outcome measure, 25(OH)D increase over time. The questionnaire was examined for errors weekly and to ensure compliance.

Background health status and lifestyle, and changes in health status and lifestyle during intervention.

The health status of the participants and start of new medication was monitored weekly throughout the study.

Adverse events. As the intervention included UVB radiation, all participants were assessed for and requested to report skin erythema.

Compliance. To optimize compliance in this long-term study, the participants were requested to schedule weekly intervention dates and complete all planned intervention throughout the study period. Furthermore, it was required that all participants had weekly blood samples and measurement of skin pigmentation three times during the study period. Also it was required that a diet questionnaire was completed and checked for errors on a weekly basis. All participants were informed thoroughly of the study requirements prior to inclusion. The acceptable level of compliance was defined at a completion of the entire intervention planned, assessment of all objective measurements and completion of all questionnaires.

Statistical analysis. Descriptive statistics of participants by group will be presented to characterize the participants included in the study. Individual data were tested with the Kolmogorov-Smirnov test to assess whether the data were normally distributed. Normally distributed data will be tested by 2-tailed t test. Otherwise non-parametric tests will be used such as 2-tailed Mann-Whitney's test or Kruskal-Wallis test. Comparison of 25(OH)D levels at different time points will be performed using paired t test (2-tailed). Categorical data will be tested with Fisher's exact test or Pearson's Chi-square test (2-tailed).

As the light-skinned group and the darker-skinned group may differ, these two groups will initially be investigated separately to examine whether there is basis to merge these two groups for common analysis. Therefore, the increase in 25(OH)D over time will be investigated separately in the light-skinned group and the darker-skinned group the following models: linear, inverse, quadratic, cubic, power, sigmoid and exponential. The derivate function was defined as the average daily change in 25(OH)D between two sample time-points (Δ25(OH)D per day). Selection of the best suitable model will be performed based on the accordance between the investigations of the 25(OH)D increase over time (individual- and group-based) and the derivate function.

The inter-individual variations in the combined group in a linear transformation of the increase of 25(OH)D over time will be explored by comparing general linear models (GLMs) with: (1) common increase rate (25(OH)D increase per day and slope) and common 25(OH)D start level (intercept), (2) common increase rates and individual intercepts (measured 25(OH)D start levels) and (3) individual increase rates (i.e. an individual constant) and measured 25(OH)D start levels.

The influence of parameters (secondary outcome measures) on the variation of the 25(OH)D increase rate will be examined separately. As the influence of separate significant parameters may not be independent, separate significant parameters will subsequently be investigated by a stepwise backward elimination of a combined GLM according to P-value (25).

Data will be statistically analyzed using SPSS 24.0 for Windows (SPSS Inc., Chicago, U.S.A.). P < 0.05 will be considered significant. No interim analysis will be performed.