Study area and subjects

The present analysis was carried out in the context of a study entitled “Vitamin A status in children of 36–59 months of age in a malaria-endemic rural area in Burkina Faso”, which aimed to assess and monitor VA status of preschool children susceptible to infections. This study was conducted in two villages, Sourkoudougou and Banakeledaga, located at the health district of Dandé, in South West Burkina Faso. Dandé district is at a distance of 30 Km from Bobo Dioulasso, the second largest city in the country. The region is known for its agricultural activity that benefits from the Soudanian climate zone. During the rainy season, that lasts between May and September, staple foods such as corn and rice are grown for self-consumption while fruits and vegetables are produced as cash crops using water from a river which crosses the two villages [18]. The dry season corresponds to the lean season, when the population depends mainly on stored grains and small livestock for their subsistence. During the rainy season, stagnant water promotes the breeding of mosquitos, the vector of malaria parasites. Malaria is endemic in the region; however, its incidence increases from June through October. In addition to malaria, diarrheal diseases and respiratory infections are the main reasons for medical consultation and hospitalization of children under 5 years of age in this area. Primary health care is provided at a primary health facility called “Centre de Santé et de Promotion Sociale (CSPS)”.

The study included 115 apparently healthy children, from both sexes, 36–59 months of age who live in one of the two villages of Sourkoudougou and Banakeledaga. After a sensitization of the two communities, all participants and their caregivers were invited to come to the health center of Sourkoudougou for screening, enrollment and follow up activities. Children were enrolled in two separate surveys; the first one was conducted during the dry season from November 2016 to January 2017, and the second survey was carried out during the rainy season between August and September 2017 (Fig. 1). Children were included in the study after obtention of a written informed consent from one or both their parents. Participants whose birthdate could not be determined and those who: (1) were diagnosed with clinical signs of VAD including night blindness, conjunctival xerosis, Bitot’s spots, or corneal xerosis, ulceration or scars; (2) were severely acute malnourished (weight-for-height z-score [WHZ] < -3 standard deviations [SD], according to the 2006 WHO Child Growth Standards [19]); (3) had severe anemia with hemoglobin concentration < 70 g/L [20]; (4) had fever or reportedly had fever in the previous 24 h (fever was defined as an uncorrected axillary temperature above 37.5 °C); or (5) suffered from serious illness that necessitates hospitalization including coma, clinical severe dehydration, severe vomiting, or severe respiratory illness, were excluded.

Baseline data collection

Data were collected for eligible children during the first day of enrollment. Data included the child’s sex and age, father’s occupation, maternal education, marital status, and household assets including building material, radio, television, telephone, refrigerator, bicycle, and motorcycle. Dietary intake data and information on breastfeeding practices were collected using adapted 24-hour dietary recall questionnaire [21].

On the same day, children’s height and weight were measured in duplicate. Height, in cm, was measured to the nearest 0.1 cm using a portable length board (Seca 213, Hamburg, Germany), and weight, in kg was assessed with 50 g precision using an electronic balance (Seca 899, Hamburg, Germany). A third measurement was completed in case the two measurements differed by > 0.5 cm and by > 0.1 kg for height and weight, respectively. The mean of the two closest values was used in the analysis.

Blood sample collection and 13C2-retinyl acetate administration

We have followed the standard operating procedures in assessing VA status using stable isotope techniques [15]. All the procedures were carried out at the health center of Sourkoudougou. Briefly, during the first day (D0), 6 ml of venous blood were collected in plain tubes by a trained phlebotomist. Blood samples were protected from light by covering the tubes with aluminum paper. The protected tubes were immediately stored in an ice box and transported the same day to the laboratory of the “Institut de Recherche en Sciences de la Santé” (IRSS, Bobo Dioulasso). On the same day of their reception, blood samples were centrifuged at 3000 rpm for 10 min with a Universal 320R centrifuge (Hettich Zentrifugen, D-78,532 Tuttlingen). Serum was transferred into brown 2 mL cryotubes (Eppendorf, Hamburg, Germany) under yellow light and stored at -80 °C. Shortly after the D0 venous blood draw, 1.0 µmol 13C2-retinyl acetate [22], dissolved in 211.6 µL food grade soybean oil, was administered orally using a positive displacement pipette (Gilson Microman E M250 E, 50–250 µL). After making sure that the children swallowed the oil containing the stable isotope, they were given 1 ml of unfortified vegetable oil and a fat-containing snack to improve absorption and asked to eat it immediately. On day 14 after dose administration, another venous blood sample was drawn and processed in the same way as described above for the venous blood sample drawn at baseline.

All the pre-dose and post-dose samples were transported on dry ice to the University of Wisconsin-Madison for serum retinol and 13 C-retinol analysis. Serum extracts of 100 µL were transported on dry ice to the VitMin Lab, Willstaett, Germany for analysis of RBP and inflammatory biomarkers.

Serum retinol analysis

This study reports retinol analysis on all the participants from D0 (Fig. 1). Serum samples were processed following a previously described procedure with minor modifications as described elsewhere [23]. A volume of 100 µL serum sample was denatured with 150 µL of ethanol and 75 µL of C23-β-apo-carotenol was added as an internal standard. Compounds of interest were extracted three times with 500 µL hexanes. The three supernatant fractions were pooled and dried under nitrogen gas, reconstituted in 75 µL methanol:dichloroethane (75:25 vol:vol), and 3 µL injected into the Waters Acquity H-Class ultra-pressure liquid chromatograph (UPLC) system, under chromatography conditions that were reported in details elsewhere [24].

Fig. 1

Schematic diagram of the study flow and data collection tools

Total liver vitamin a reserves and total body stores

Method of extraction, purification, and analysis of 13 C using gas chromatography combustion isotope ratio mass spectrometry (GC/C/IRMS) was previously described [15, 25]. Briefly 0.3 to 1.0 mL serum sample was denatured with 2 ml of ethanol. C23 β-apo-carotenol was added as an internal standard. Retinol was extracted 3 times with 1–2 mL hexanes. The supernatant fractions were pooled, dried under nitrogen, resuspended in 100 µL methanol, frozen at -80˚C for 5 min, centrifuged at 1380 x g briefly, and injected into High-Performance Liquid Chromatography (HPLC) system 1 for quantification and purification. The retinol fraction, thus separated was collected, dried under nitrogen, and resuspended in 100 mL methanol for injection into HPLC system 2. The retinol fraction was dried in a Thermo Savant Speed-Vacuum centrifuge (Thermo Scientific, Waltham, MA), and reconstituted in 10 µl hexanes. Synthetic retinol, prepared by quick retinyl acetate saponification (Sigma-Aldrich, St. Louis, MO), was purified twice similarly to serum retinol and used as an external standard. Retinol extracts and the external standard (each 1.5 µl) were injected into the GC/C/IRMS with the programmed temperature vaporizing (PTV) injector. Atom percentage (At %) was directly calculated (Isodat version 2.0; Thermo Scientific) in reference to carbon dioxide, which was calibrated against a sucrose standard (National Institute of Standards and Technology, 8542).

Tracer-to-tracee ratio (TTR), which is an analogous to specific activity for isotopes was calculated using the following formula [26, 27]:

Where Fa is the 13C-abundance of the tracer dose [0.11 in 13C2- retinol (2 of 20 carbons labeled with 13C during synthesis [22] plus naturally abundant 13C at 1.1% in the remaining 18 carbons)]; Fb is the baseline abundance of 13C in retinol measured in the baseline serum sample D0; and Fc is the final 13C-abundance of serum retinol after the 14-d mixing period D14.

Total body VA stores (TBS, µmol) were calculated as previously described [25, 28] using the mass balance equation that includes the TTR:

$$TBS = a \times \frac} \times (factors}for}absorption}and}storage)$$

Where a is the amount of 13C retinyl acetate in the dose (1.0 µmol) and the fraction of 13C to 12C in the dose is 0.11, which accounts for natural abundance. TLR (µmol/g) is calculated using the formula:

$$TLR = \left( } \right)x\,liver\,fraction\,of\,BW}}} \right) \times 1000 \times fraction\,of\,TBS\,in\,liver$$

Where BW is body weight (kg), liver fraction of BW was estimated as 3% in preschool children, and 80% of TBS were assumed to be in the liver storage pool of children with adequate VA to hypervitaminosis [25, 28]. VAD and hypervitaminosis A were defined as TLR < 0.1 µmol/g of liver and TLR > 1.0 µmol/g of liver, respectively.

Inflammation indicators and retinol binding proteins determinations

Acute phase proteins including CRP, AGP, and RBP were analyzed in serum samples by ELISA (DBS-Tech in Willstaett, Germany) [29]. The coefficients of variation (CVs) of the CRP, AGP and RBP for a pooled serum sample were respectively 5.84%, 8.09%, and 3.61%.

Hemoglobin concentration, malaria and intestinal parasites

Using the remaining blood drops from the venous blood draw, hemoglobin concentration using point-of-care Hemocue (Hemocue HB 301, HemoCue® AB, Ängelholm, Sweden) was measured, and rapid diagnosis test (RDT; SD-Bioline Malaria-Ag-Pf/Pan™) for malaria [30] and malaria blood smear [31] were performed to assess malaria status. Thirty grams (30 g) of stool were collected on the spot from the children, and later a direct examination on a slide after preparation was performed by a qualified laboratory technician to search for the presence of intestinal parasites at the Centre Muraz (Bobo-Dioulasso). The test was considered negative for intestinal parasites if no parasite was identified.

Data processing

Z-scores of height-for-age (HAZ), weight-for-age (WAZ) and weight-for-height (WHZ) were calculated using the WHO Child Growth Standards and the STATA macros [19]. Dietary diversity score ranging from 0 to 9, was constructed based on the reported child’s consumption of one of nine food groups during the 24-h dietary recall: starchy foods, dark green leafy vegetables, VA-rich fruits and vegetables, other fruits and vegetables, organ meat, meat and fish, eggs, legumes, nuts and seeds, and dairy products [32]. The household asset index was constructed based on the information on baseline ownership of a set of assets (radio, television, telephone, refrigerator, bicycle, or motorcycle) and building materials, using multiple correspondence analysis [33].

Vitamin A deficiency was defined by serum retinol or RBP concentrations < 0.7 µmol/L [6, 7]. Cutoffs of > 5.0 mg/L for CRP and > 1.0 g/L for AGP were used to define elevated acute phase proteins. Inflammation phases were defined as: (1) no inflammation if neither of acute phase proteins was elevated (CRP ≤ 5.0 mg/L and AGP ≤ 1.0 g/L), (2) incubation if CRP was elevated (CRP > 5.0 mg/L and AGP ≤ 1.0 g/L), (3) early convalescence if both CRP and AGP were elevated (CRP > 5.0 mg/L and AGP > 1.0 g/L), and (4) late convalescence when only AGP was elevated (CRP ≤ 5.0 mg/L and AGP > 1.0 g/L). Serum retinol and RBP concentrations were adjusted for the presence of inflammation based on elevation of one or both acute phase proteins (AGP and CRP) or no inflammation using the ratio between the arithmetic mean of the indicator in the respective inflammation category and its mean in the reference group, that is with no inflammation [13, 34]. Hemoglobin concentrations below 11.5 g/dL categorize children as anemic [20].

Data analysis

All statistical analyses were carried out using Stata software, version 15.1 (Stata Corp, TX, USA). Descriptive statistics [geometric mean (95% confidence interval, 95% CI), means (± standard deviation, SD), and proportions] were performed to assess VA status, inflammation indicators and baseline sociodemographic and socio-economic characteristics and dietary intake.

Additionally, the accuracy of serum retinol and RBP were evaluated by calculating the sensitivity [True Positive / (True Positive + False Negative)], specificity [True Negative / (False Positive + True Negative)], positive predictive value [True Positive / (True Positive + False Positive)], and negative predictive value [True Negative / (False Negative + True Negative)] using TLR as the reference standard indicator. The diagnosis accuracy as a proportion of correctly classified subjects ([True Positive + True Negative] among all subjects) was also calculated [35].

With:

True positive: subjects with serum retinol or RBP concentrations < 0.7 µmol and TLR < 0.1 µmol /g of liver.

False positive: subjects with serum retinol or RBP concentrations < 0.7 µmol and TLR > 0.1 µmol /g of liver.

True negative: subjects with serum retinol or RBP concentrations > 0.7 µmol and TLR > 0.1 µmol /g of liver.

False negative: subjects with serum retinol or RBP concentrations > 0.7 µmol and TLR < 0.1 µmol /g of liver.

The effects of season, inflammation (CRP, AGP) and helminth infection on sensitivity and specificity were also assessed by estimating sensitivity and specificity in the presence versus absence of inflammation or helminth infection, and during dry season versus rainy season after participants classification in the above categories (true positives, false positives, true negatives, and false negatives).