This pilot study included obese individuals participating in a multidisciplinary weight loss program (OPTIFAST®52) causing an average weight loss of 26.0 kg in men and 19.6 kg in women within one year 19. Briefly, the program consists of a five-phase lifestyle modification program designed for 52 weeks, including (i) a 1-week introduction during which a detailed nutrition analysis is performed; (ii) a 12 week period of low-calorie diet (LCD) (800 kcal/day) during which participants consume a formula diet exclusively (daily consumption of 5 sachets at 160 kcal each, Optifast 800® formula, Nestlé Inc.); (iii) a 6 weeks refeeding phase, (iv) a 7 weeks stabilization phase and (v) a 26 weeks maintenance phase. We previously reported more detailed information about the program 19. The five daily consumed formula meals in the second phase contain vitamins, minerals and trace elements according to DRI for healthy adults (Table 1), except for the flavor potato/leek, which contains less amounts of vitamin C, vitamin B₁₂, folate and calcium. However, flavor selection was documented, recording that the latter one was only consumed occasionally.

Flavor A: vanilla, chocolate, strawberry, coffee, and tomato; flavor B: potato/leek.

To determine micronutrient intake in obese individuals we analyzed dietary record data obtained during the introduction week from all participants enrolled from 02/2006 to 02/2010 (n = 104). A randomly assigned subgroup of subjects (n = 32) participated in a pilot study, and was followed up for micronutrient measurement before and after the three-month formula diet period (LCD intervention group). Patients with a history of bariatric surgery were excluded from study participation. The measurements were performed by intracellular micronutrient analysis in buccal mucosa cells (BMC) and by additional serum micronutrient analysis in 14 of the subjects. The latter subjects did not differ in initial body weight and weight loss to the whole LCD intervention group. For 9 subjects, long-term data on intracellular vitamin concentrations were also obtained at the end of the program year, 9 months after completion of the LCD phase. The nutrients evaluated in this study were vitamin C (BMC, serum, leukocytes), vitamin E (BMC, serum), lycopene (BMC), β-carotene (BMC), vitamin A, 25(OH)D, vitamin B₁₂, folate, selenium, iron, zinc and calcium (all serum). Deficiency was defined as a concentration below the reference interval: vitamin A 0.2-1.2 mg/ml, vitamin E 5–16 mg/ml (serum)/9.5-20.3 pmol/μg DNA (BMC), 25(OH)D, 20–70 ng/ml, vitamin C 28.3-85.1 μg/ml (serum)/57–114 nmol/10⁸ cells (leukocytes)/3.9-11.1 pmol/μg DNA (BMC), vitamin B₁₂ 191–663 pg/ml, folate 4.6-18.7 ng/ml, selenium 74–139 μg/ml, iron 40–170 μg/dl, zinc 70–150 μg/dl, calcium 2.2-2.6 mmol/l, β-carotene 0.1-05 pmol/μg DNA, and lycopene 0.1-05 pmol/μg DNA. Moreover, C-reactive protein (CRP) concentrations were analyzed in serum samples. Drug use was monitored and documented on standardized forms. We calculated body mass index (BMI) and relative weight loss in percent (RWL; = 100xΔweight loss in kg/initial body weight in kg). Body composition was analyzed using bioelectrical impedance analysis at all study visits (Data Input Nutriguard M, Darmstadt, Germany). The study protocol was part of a multicenter clinical trial, research project “Obesity and the gastrointestinal tract” (ClinicalTrials.gov identifier: NCT01344525), approved by the ethics committee of the University Hospital of Tübingen, Germany. Informed consent was obtained from every subject prior to participation.

Food intake was recorded using a predesigned daily journal. Subjects had to document time, amount (in gram) and situation, in which a food or beverage was consumed. They were instructed to: (1) document all consumed foods and beverages in as much detail as possible, (2) to weigh foods (or to estimate doses, if weighing was not possible in some situations), (3) to document food or beverage intake immediately after consumption and (4) not to change usual eating habits. Data analysis was performed using the nutrition software EBISpro, version 8.0 for Windows. In analyzing the eating records, nutrients were studied in relation to the DRI (D-A-CH reference values) 18. The DRI utilized corresponds to the age group and sex. Furthermore, data were compared with the Second National Nutrition Survey (NNSII), which is representative for the German population. The dietary assessment in this survey was performed using controlled interviews according to the diet history method, which were reinforced by dietary record collection and 24 hour-recalls in a representative sub-sample.

BMC were collected using a kit from Day-med-concept GmbH, Berlin, Germany. Subjects first had to rinse their mouths with water thoroughly to remove food particles and then brush the inner lining of their cheeks with a soft toothbrush twenty times, twice on each side. The toothbrush was washed in 25 ml NaCl solution (0.9%) gently after each brushing. The samples were then centrifuged at 1,600 rpm for 3 minutes. The supernatant was discarded; the cells were completely resuspended in rinsing solution (phosphor float 0.15% w/v) and centrifuged again at 1,800 rpm for 3 minutes. After removal of the supernatant fraction, the stabilizing solution (heat-sensitive reducing agent 0.09% w/v) was added; and the cells were resuspended and stored. The micronutrients in BMC (pmol/μg DNA) of the samples were measured by BioTeSys GmbH (Esslingen, Germany) using an accredited RP-HPLC method according to DIN EN ISO/IEC 17025. The reference values are based on a statistical distribution reflecting the 25th and 75th percentile of a data set covering analysed BMC samples over one year.

Micronutrient concentrations were expressed in pmol/μg DNA. Micronutrient detection was possible when the amount of DNA in BMC was at least sufficient (>1 μg). If the amount of DNA extracted was < 1 μg, the data was excluded from analysis. Micronutrients below detection limit were defined as suboptimal cellular concentrations according to the analysis laboratory and therefore were included.

For serum micronutrient analysis, blood was collected by venipuncture between 8.00 a.m. and 10.00 a.m. after an overnight fast. Serum was separated by centrifugation at 2,000 x g for 15 min at 4°C. Aliquots were immediately stored at −80°C and sent to an external laboratory on the same day. HPLC was used to assay vitamin A and α-tocopherol. Measurement of vitamin C, calcium and iron was performed by photometry. Furthermore, serum was analyzed for 25(OH)D (radioimmunoassay), zinc and selenium (atomic absorption spectrometry) and vitamin B₁₂ and folate (luminescence immunoassay).

Isolation of leucocytes for vitamin C analysis was performed immediately after blood withdrawal. Briefly, sedimentation of erythrocytes was performed in a hermetic tube in dark environment using 6% Dextran according to the method of Denson et al.20. Supernatant was centrifuged for 10 min at 320 x g. After resuspension with HA buffer lysis of remaining erythrocytes was performed by 5 min incubation with 10 ml ammonium chloride. After centrifugation (10 min, 600 x g), leukocytes were manually counted and pellets frozen at −80 °C until analysis. Determination of total ascorbic acid concentrations in leukocytes was performed by HPLC. In summary, after thawing, leukocyte pellets reducing agent (20 μl 20% TCEP solution [tris (2- carboxyethyl) phosphine hydrochlorid]) was added, thoroughly mixed and incubated on ice for 5 min. For cell lysis 80 μl 10% meta-phosphoric acid was added, mixed and centrifuged (10 min, 13,000 x g, 4°C). Ascorbic acid concentrations in the supernatant were determined by UV/VIS detector at 245 nm.

All samples for vitamin C analysis were rapidly processed and shielded from light exposure.

Values are expressed as means ± SD if not indicated otherwise. For the comparison of dietary record parameters with the reference population, Gaussian distribution was assumed due to sample size (n = 15,371). Differences between the obese study population and the NNSII data were analyzed using unpaired t-tests. The paired t-test was applied to compare repeated measures of micronutrients. If Gaussian distribution could not be assumed Wilcoxon signed-rank test was used instead of t-tests. Frequencies were analyzed with cross tables according to the method of McNemar. Analysis of covariance (ANCOVA) was performed to control for the effect of sex, age, energy, and medication use on micronutrient concentrations. Relations between continuous variables were analyzed by calculating the Spearman´s rank correlation coefficient or, in case of normal distribution of the data, the Pearson’s correlation coefficient. P values <0.001 were interpreted as statistically highly significant, p-values between 0.001 and <0.01 as very significant and p-values between 0.01 and <0.05 as significant differences. All analyses were carried out by using the statistics software SPSS, version 19.0 (IBM®SPSS®, Chicago, IL), and Graph Pad Prism, version 5.0 (GraphPad Software, Inc., La Jolla, CA).

Prior to intervention all subjects had at least grade I obesity (BMI ≥ 30 kg/m²), about 50% even obesity grade III (BMI ≥ 40 kg/m²). The obese study group (dietary records) and the reference population were well matched in respect to age, but the obese group comprised more females (Table 2). BMI of the reference population was significantly lower compared to the obese study group (26.5 kg/m² versus 40.9 kg/m², p < 0.001). 32 out of 104 obese individuals (31%) underwent the LCD intervention. On average, these subjects lost 16.2% of initial body weight during the three-month LCD period. The LCD intervention subgroup (n = 9), which was followed up until program end (12 months after start of intervention), did not differ in weight loss during the LCD period compared with the whole LCD intervention group (Table 3).

Data are presented as mean ± SD.

^afollow-up until program end.

Abbreviations: WL Weight loss, RWL Relative weight loss.

86% of the subjects undergoing LCD took medication on a regular basis, which may have resulted in drug-nutrient interaction effects, thus influencing micronutrient bioavailability: ACE inhibitor (21%), proton pump inhibitor (8%), antidepressants (14%), L-thyroxine (14%), biguanide (7%), insulin (7%), loop diuretics (8%), NSAID (7%) and xanthine oxidase inhibitor (14%). Frequencies and dosage of medication decreased slightly in 36% of the cases during the LCD period. Controlling for sex, age, energy intake, and medication use, did not reveal significant effects on micronutrient concentrations.

Micronutrient intakes in obese subjects are shown in Table 4. Micronutrient intake in female subjects was significantly lower for five micronutrients compared to the reference population, in male subjects for six micronutrients, respectively. Women consumed nine micronutrients in amounts below DRI whereas with men this applied to seven micronutrients. Lowest micronutrient intakes compared to DRI were observed for retinol, ß-carotene, vitamin D, folate, iron and iodine (>75% of the obese study population below DRI) and vitamin E, C and calcium (>50% respectively).

Abbreviations: DRI = Dietary Reference Intake, NNS II = National Nutrition Survey II.

Data are presented as mean ± SE.

^a Recommendations for nutrient intake in Germany, Austria and Switzerland (D-A-CH reference values)18.

^b Obese study population compared to German reference population (NNS II).

Micronutrient intake did not differ between the whole study group (n = 104) and the subset of obese individuals undergoing LCD, which were further subjected to intracellular and serum micronutrient analysis.

Baseline deficiencies in serum micronutrient concentrations were observed for 25(OH)D, vitamin C, selenium and iron (Table 5). Except for 25(OH)D, the number of cases with deficiencies either remained (selenium, iron, both n.s.) or tended to further increase, which was the case for vitamin C (+15% of subjects with deficiency, n.s.). After the formula period, additional deficiency in zinc (7.7%) and calcium (53.8%) occurred. Figure 1 shows serum concentrations that decreased (calcium (p < 0.01) and iron (p < 0.05)) or increased (25(OH)D (p < 0.01), folate (p < 0.05) and zinc (p < 0.01)), or remained unchanged (vitamin C (n.s.)) in the course of intervention using formula diet.

^a p-values were calculated in relation to absolute values using paired t-tests.

Intracellular antioxidant status in BMC is displayed in Figure 2. Before intervention, deficiencies were observed for vitamin C (63.3%), ß-carotene (20.0%) and lycopene (10.0%). After the LCD period BMC concentrations of vitamin C (p < 0.05) and lycopene (p < 0.05) decreased, resulting in a higher percentage of subjects with deficiencies (vitamin C +30.2%, p < 0.01 and lycopene +6.1%, n.s.). At the end of the weight loss program (9 months after termination of the formula diet) antioxidative status was optimal or even above reference range, except for vitamin C (deficiencies in 77.8% of subjects), which also tended to improve (n.s.).

No correlation between vitamin C serum, BMC and leukocyte concentrations was observed. Serum and BMC vitamin C levels tended to decrease during the formula period (serum 59.44 ± 22.73 μmol/l before intervention, 37.29 ± 13.48 μmol/l after formula diet (n.s.), BMC 6.62 ± 1.52 pmol/μg DNA and 2.95 ± 1.10 pmol/μg DNA (p < 0.001), respectively), whereas leukocyte vitamin C concentrations increased (81.64 ± 16.30 ng/10⁸ cells to 111.2 ± 36.32 ng/10⁸ cells, p < 0.05) (Figure 3).

CRP concentrations positively correlated with body weight (r = 0.6558, p < 0.0001). There was a negative correlation between iron and CRP (r = −0.456, p < 0.05). Lipophilic vitamins were negatively associated with body fat in percent of total body weight (25(OH)D: r = −0,6369, p < 0.0001, vitamin A: r = −0,4663, p < 0.05, vitamin E (serum): r = −0,4378, p < 0.05).