Study design and recruitment
This cross-sectional study was conducted on 173 postmenopausal women with T2D, aged between 50 and 87 years, living in the western region of Saudi Arabia (Jeddah). We assessed VitD status (25(OH)D) in the participants and its association with: (1) glycaemic parameters (HbA1c, fasting glucose, fasting insulin, fasting C-peptide and insulin sensitivity indices and high sensitivity C-reactive protein (hsCRP)); (2) bone related parameters (intact PTH, Ca, albumin, phosphorus (PO4) and magnesium (Mg)); (3) anthropometric measures (weight, height, waist hip ratio (WHR) and BMI); (4) lifestyle factors (physical activity, smoking, dietary VitD intake, veiling and sun exposure); (5) Socio-demographic factors including skin tone and ethnicity.
Subjects for this study were sequentially recruited from seven primary health care centers (PHCCs) distributed in Jeddah (a PHCC from each of the seven geographical sectors of Jeddah area to guarantee that the average health status of the participating women will represent a randomly selected adult population). A multi-stage sampling technique was implemented. In stage 1, one PHCC was chosen from each of the seven sectors of the Jeddah area. In stage 2, random selection of samples was conducted from the selected PHCCs to select female files of the registered population. In stage 3, all women in the selected age group (≥ 50 years) among selected files were contacted for possible recruitment to the study based on the predefined criteria of inclusion. The number of women randomly selected from each center was proportionally identified according to the number of the registered women in each center. Subjects willing to participate in this study were referred to a clinic at the Centre of Excellence for Osteoporosis Research (CEOR) in King Fahd Medical Research Centre (KFMRC), King Abdulaziz University (KAU), Jeddah. Each participant provided written informed consent relating to participation in this study. Following the ethical standards in Declaration of Helsinki, ethical approval of this study was obtained from the Research Ethics Committee, the Faculty of Medicine, KAU (ref no.179 − 16, Oct/2017).
The recruitment and selection of patients was based on specific inclusion and exclusion criteria. Inclusion criteria included postmenopausal women: Last menstrual period (LMP) ≥ 1year and follicular stimulating hormone (FSH) > 15 IU/L), previously diagnosed with T2D according to the criteria of the American Diabetes Association ( HbA1c ≥ 48 mmol/mol or fasting plasma glucose ≥ 7 mmol/L) . Women with history of chronic liver or renal disease, cancer, malabsorption syndrome, rheumatoid arthritis, hyperthyroidism, other endocrinal disorders that might affect bone (e.g. Hyperparathyroidism) or history of intake of medications with possible effects on VitD (e.g. VitD supplements, glucocorticoids and anticonvulsants) were excluded from the study. Following multiple stages of exclusion (Fig. 1), a sample size of 173 was included in this study.
Initially, all participants answered a questionnaire (completed by the researcher), which requested information including socio-demographic factors, dietary VitD intake (semi-quantitative Food Frequency Questionnaire (SFFQ) ), lifestyle history including smoking habits and physical activity, medical history, menstrual history and drug history. Each participant underwent anthropometric and blood pressure measurements.
Skin tone was recorded for each participant based on the Fitzpatrick skin tone classification . Duration (number of hours) of weekly exposure to outdoor sunlight in the last month was noted in the participants’ questionnaires as well as the use of sunscreen. Due to cultural or religious reasons, most women in Saudi Arabia, especially the elderly, wear a veil and cape. Women participating in the study who cover their head and body, with face and hands exposed were considered as partially covered, while participating women covering their whole body and face, with only the eyes and hands exposed were considered as totally covered.
VitD daily intake from food was estimated using a semi-quantitative Food Frequency Questionnaire (SFFQ). The SFFQ used in the study was adapted from a validated SFFQ in Saudi Arabia . Items included the most commonly VitD rich food consumed in the region; salmon, tuna, sardines, milk, laban (buttermilk), yogurt, egg and beef liver. The frequency of this food intake was expressed as number of servings per day/week/month. The daily VitD intake in IU was then calculated and compared to their estimated average requirement (EAR) (600–800 IU/day based on the IOM recommendation for women aged 50 y and over ).
Serum measurements of 25(OH)D and other hormones
Serum 25(OH)D and intact PTH levels were measured by direct competitive chemiluminescence immunoassay (CLIA), using a LIASON auto-analyzer (DiaSorin Inc, Stillwater, MN, USA). The intra and inter-assay coefficient of variations (CV) of serum samples were < 8%. VitD deficiency was defined based on Institute of Medicine (IOM) guidelines  as the 25(OH) D level below 12 ng/ml (< 30 nmol/l) and VitD insufficiency as the 25(OH)D level of 12–19 ng/ml (30–49 nmol/l), and VitD sufficiency between 20–50 ng/ml (50–125 nmol/L). FSH and Thyroid function test (TFT) including thyroid-Stimulating Hormone (TSH), free thyroxin (T3) and free triiodothyronine (T4) were measured in serum by immunoassays, using VITROS ECiQ (Ortho-Clinical Diagnostics Inc., Rochester, NY, USA) to exclude any women with hyperthyroidism or not postmenopausal.
Serum measurements of liver enzymes, renal function, high-sensitivity C-reactive protein, lipid and bone profile
Liver enzymes (including Aspartate Aminotransferase (AST), Alanine Aminotransferase (ALT), and Alkaline Phosphatase (ALP)), creatinine, total cholesterol, direct High Density Lipoprotein (HDL), triglycerides, Low Density Lipoprotein (LDL), Very Low Density Lipoprotein (VLDL) albumin, Ca, PO4 and Mg) were all measured in serum by reflectance spectrophotometry, employing the colorimetric method using a VITROS 250 Clinical Chemistry Auto-analyzer (Ortho-Clinical Diagnostics Inc., Rochester, NY, USA). The intra and inter-assay CV of serum samples were 4.1% and 4.5% respectively. Low Density Lipoprotein (LDL) concentrations in serum were directly calculated by the analyzer, based on standardized calculations (Friedewald equation ) dependent on the level of total cholesterol, direct HDL and triglyceride measured by the same analyzer. Very Low Density Lipoprotein (VLDL) serum levels were estimated by dividing the triglyceride by 2.2. Subjects with high liver enzymes were excluded (the normal clinical level of serum being AST ˂45 U/L; ALT ˂50 U/L and ALP between 80 and 280 U/L). Samples with creatinine levels higher than normal were excluded (i.e. a normal level of serum creatinine in females ˂105µmol/L).
Hs-CRP was measured in serum by immunoassay, using a VITROS 5,1 FS chemistry auto-analyzer (Ortho-Clinical Diagnostics Inc., Rochester, NY, USA). The intra-assay and inter-assay CV of serum samples were 3.5% and 4% respectively.
Measurements of glycaemic control parameters
Glycosylated hemoglobin (HbA1c) was determined using a VITROS 5,1 FS chemistry auto-analyzer (Ortho-Clinical Diagnostics Inc., Rochester, NY, USA). Fasting glucose in serum was measured by means of a colorimetric method, using a VITROS 250 Clinical Chemistry Auto-analyzer (Ortho-Clinical Diagnostics Inc., Rochester, NY, USA). The intra and inter-assay CV for HbA1c and fasting glucose samples were both < 5%. Insulin and c-peptide (a consequent product produced when insulin is secreted) were measured in serum by a sandwich CLIA using a LIAISON autoanalyzer (DiaSorin Inc, Stillwater, MN, USA). The intra and inter-assay CV for insulin and C-peptide serum samples were both < 6%. Fasting insulin and Homeostasis Model Assessment 2 (HOMA2)  were measured for all women, with the exception of those on insulin therapy, due to the influence of insulin intake on these measures.
HOMA2 was used to estimate insulin resistance and β-cell function. HOMA2 indices  (HOMA2-IR and HOMA2-%β) were calculated from fasting glucose, fasting insulin and fasting c-peptide in a steady-state condition (fasting glucose: 3–25 mmol/L, fasting insulin: 2.88–43.16 mIU/L and fasting c-peptide: 0.6–10.5 µU/ml) using an updated computer HOMA calculator software (version 2.2.3) issued by University of Oxford Diabetes Trials Unit, available at https://www.dtu.ox.ac.uk/homacalculator/ .
The statistical analysis was performed using SPSS program (v.20 SPSS Chicago Inc). Normality of data was tested by Kolmogorov-Smirnov test. All numerical parametric results were expressed as means ± SD, while numerical non-parametric results were presented as median (IQR). Descriptive results were expressed as a percentage of the total sample number. Correlations between different parameters were obtained using Pearson correlation for normally distributed data and Spearman correlation for non-normally distributed data. The non-parametric test, Kruskal-Wallis H test, was used based on non-normal distribution of data to compare between groups. A probability value ≤ 0.05 was considered statistically significant. Multiple linear regression analysis (stepwise) was used for independent variables that showed significant correlations at the bivariate level (P ≤ 0.05) to determine independent predictors of the dependent variable.