Optimizing vitamin supplementation via reference interval update of vitamins A, E, B1, and B6 using HPLC

Abstract

Vitamins are essential micronutrients obtained from the diet, required by the body in small amounts daily for proper metabolism. Monitoring their levels is necessary for detecting deficiencies and guiding supplementation in certain clinical conditions. This study aimed to update the reference values for vitamins A, B1, B6, and E, and some related ratios, adjusted to the adult population of our health reference area using liquid chromatography in a direct approach calculation (n = 146, age: 21–64 years, 64% females). No significant differences in vitamin levels or ratios were observed based on age and sex. We obtained reliable and updated reference values: 1.1–2.8 ‍μmol/L and 18.9–42.2 ‍μmol/L for vitamins A and E respectively, 85.9–181.6 nmol/L and 57.0–165.7 nmol/L for vitamins B1 and B6 respectively; and related ratios of 246.2–561.1 ‍ng/g for vitamin B1 corrected by hemoglobin; 5.2–8.9 ‍μmol/mmol and 4.5–7.4 ‍μmol/mmol for vitamin E corrected by cholesterol and total lipids, respectively. These reference values significantly differ from those provided by the reagent manufacturer currently in use. While correcting vitamin E for lipids and vitamin B1 for hemoglobin is not recommended for the general population, these adjustments may be useful in interpreting results in certain pathological conditions.

Introduction

Reference intervals (RI) play an important role in clinical diagnosis, follow-up, treatment, and repeat testing. It is estimated that up to 70% of clinical decisions rely on laboratory results, making RI a vital component of patient management.⁽¹⁾ Laboratory professionals are responsible for providing reliable and accurate RI to ensure the optimal use of laboratory tests in patient care; however, this is not practically feasible for most clinical laboratories.^(2,3)

To overcome this issue, clinical laboratories adopt RI recommended by assay manufacturers, or from other laboratories or published data. These approaches are simple, cost-effective, and easily adaptable by clinical laboratories. However, they may not be suitable for all populations and may lead to misdiagnosis of diseases or clinical conditions, owing to the differences in reference population, analytical methods, and statistical approaches used to calculate them. If possible, RI can be calculated by clinical laboratories using a direct approach, involving studies on a representative sample of healthy individuals (reference subjects), or an indirect approach, requiring extensive routine analysis data from the Laboratory Information System (LIS).^(4,5)

Fat-soluble vitamins (such as vitamins A and E) and water-soluble vitamins (such as vitamins B1 and B6) play key roles as coenzymes in various physiological and metabolic processes.⁽⁶⁾ Therefore, the clinical symptoms of hypovitaminosis are diverse and vitamin-specific. For instance, vitamin A deficiency is associated with ophthalmological symptoms—such as night blindness and xerophthalmia,⁽⁷⁾ whereas vitamin E deficiency is often correlated with neurological symptoms—such as ataxia, peripheral neuropathy, and muscle weakness.⁽⁸⁾ Vitamin B1 deficiency can lead to severe neurological and cardiovascular complications,⁽⁹⁾ while vitamin B6 deficiency can cause microcytic and iron-refractory anemia and neurological disorders.⁽¹⁰⁾

Adequate vitamin intake is essential to achieve and maintain optimal serum vitamin levels. However, symptoms overlapping with other conditions can complicate the diagnosis of hypovitaminosis or hypervitaminosis. Therefore, monitoring of vitamin level is necessary to detect deficiencies and guide supplementation in some clinical conditions.⁽⁶⁾ Although there is a consensus on the cut-off values for determining vitamin deficiencies,^(7–10) no widely accepted range for these values exist. RI is a powerful tool for population-based screening, diagnosis, monitoring, and decision-making regarding supplementation, particularly in patients with chronic, polymorbid, and complex conditions. Several clinical nutrition guidelines recommend measuring vitamin levels and adjusting supplementation levels based on normal laboratory values. For example, serum levels of fat-soluble vitamins A and E are monitored in patients with cystic fibrosis, due to their impaired mechanism of rapid absorption, despite adequate pancreatic storage.⁽¹¹⁾ In adults with chronic kidney disease (CKD) stage 5D on maintenance hemodialysis or peritoneal dialysis, along with vitamin supplementation, serum vitamin A/E levels should be monitored to prevent toxicity.⁽¹²⁾ Another example is the prevalence of vitamin B1 deficiency (up to 49%) in post-weight-loss surgery (WLS) patients, which necessitates regular vitamin B1 measurements during follow-up for the first 6 months and then every 3–6 months, regardless of the bariatric procedure.⁽¹³⁾ Serum vitamin levels should also be measured at the start of home parenteral nutrition in patients with chronic intestinal failure, and then monitored annually to adjust doses as needed. Subclinical findings are usually observed in these patients; however, they need to be treated to correct the suboptimal vitamin levels.⁽¹⁴⁾ Monitoring vitamin B6 levels may help assess its direct effect on peripheral neuropathy.⁽¹⁵⁾ In our laboratory, which serves 1.5 million people, establishing RI for vitamin levels was essential for determining cut-off values associated with vitamin deficient, suboptimal, or excess status.⁽¹⁶⁾

Simultaneous measurement of other compounds related to these vitamins and ratios (to correct for bias) have been proposed to avoid post-analytical errors that mask hypo- or hypervitaminosis in the assessment of nutritional status. For instance, serum vitamin E concentration can be expressed as a ratio to cholesterol, triglycerides, or their sum, because of its association with lipoproteins such as chylomicrons, very low-density lipoproteins (VLDL), and low-density lipoproteins (LDL).⁽¹⁷⁾ Moreover, approximately 90% of the total thiamine (vitamin B1) content in erythrocytes is in its biologically active form, thiamine pyrophosphate (TPP), which needs to be corrected for hemoglobin.⁽¹⁸⁾

This study aimed to update RI for vitamins A, B1, B6, and E in the adult population of our health reference area using liquid chromatography, by employing a direct approach calculation. Additionally, we reviewed available ratios that support the decision-making process regarding vitamin test results which can be easily estimated in clinical laboratories.

Materials and Methods

Subjects

This study included an analytic cohort of 146 healthy Caucasian adult volunteers, aged between 18–65 years, following Mediterranean diet habits. Volunteers were enrolled over a period of one year, from 1 June 2023 to 31 May 2024, at the Vall d’Hebron University Hospital, Barcelona, Spain. The Clinical and Laboratory Standards Institute (CLSI) EP28-3c guideline⁽¹⁹⁾ was followed for enrolling participants. This study was approved by the Vall d’Hebron Hospital Institutional Review Board [approval number: PR(AG)188-2023] and was conducted in accordance with Spanish and European legislation, and the principles of the Declaration of Helsinki. Written informed consent was obtained from all participants.

The volunteers were hospital staff belonging to different departments and estates, who were informed about their participation in the study, and were recruited by the hospital’s Occupational Risk Prevention Department. A questionnaire was used by the departmental physicians to verify the health status and daily habits of the volunteers before phlebotomy. The inclusion criteria were as follows: age between 18–65 years, body mass index (BMI) less than 30 ‍kg/m², positive evaluation on the checklist, and fasting for at least 8 ‍h before sample collection (Supplemental Table 1*). The exclusion criteria were presence of medical comorbidities, such as malabsorption diseases (Crohn’s disease or ulcerative colitis), short bowel syndrome or other abdominal surgeries, hypercholesterolemia, CKD, use of chronic medication, or intake of vitamin supplements.

Sample preparation

Blood samples were collected in BD Vacutainer^® spray-coated K2EDTA and BD Vacutainer^® Serum tubes with a Hemogard closure (Becton Dickinson, Milan, Italy). Samples were transported and centrifuged at room temperature, protected from light, and stored at −20°C upon arrival at the laboratory, until analysis. The samples were processed in a single replicate together with the remaining routine patient samples. The acceptance criterion for the results was the same as that for the routine samples.

Retinol and alpha-tocopherol (alpha-T) were measured in human serum samples using a simple precipitation step; TPP and pyridoxal 5'-phosphate (PLP) were measured in human whole blood samples using both precipitation and derivatization steps. Both sample preparations were performed based on the Chromsystems^® protocols (https://chromsystems.com/en/).

Reagents and standard samples

We used the Chromsystems^® Vitamins A and E in Serum/Plasma HPLC Reagent kit to simultaneously measure vitamin A (measured as retinol) and vitamin E (measured as alpha-T); TPP (vitamer of vitamin B1) and PLP (vitamer of vitamin B6), were analyzed using the Chromsystems^® Vitamin B1 in Whole Blood and Vitamin B6 in Whole Blood/Plasma - UHPLC Reagent kit (Chromsystems Instruments & Chemicals, Munich, Germany). Calibrators, internal standards, and controls were provided by the manufacturer for both reagent kits. The methods were traceable to certified products and HPLC reference tests, and were validated according to IVDR 2017/746.

Serum total cholesterol (CH) and triglyceride (TG) levels were measured using Siemens reagents and analyzer (Atellica^TM CH 930; Siemens Healthineers, Erlangen, Germany). Hemoglobin (Hb) levels were examined using a Sysmex XN-1000 analyzer (Sysmex Corporation, Kobe, Japan).

HPLC analysis

A Nexera X2 UHPLC system (Shimadzu Corporation, Kyoto, Japan) was employed for the analysis. Chromatographic separation was performed in one isocratic HPLC run for retinol and alpha-T (REF 34000-BK) (Supplemental Fig. 1*). After performing the pre-treatment, 5 ‍μl of supernatant were injected. Chromatographic separation was achieved in 4 ‍min on a reversed phase HPLC C18 column through an isocratic delivery mobile phase at a flow rate of 0.6 ‍ml/min. Column temperature was set at 35°C and absorbance was recorded at 325 ‍nm for retinol and at 295 ‍nm for alpha-T in a UV/SPD-20AD detector. In order to measure TPP/PLP, chromatographic separation was carried out with two mobile phases provided by the manufacturer (REF 52952/UHPLC/F) (Supplemental Fig. 1*). After performing the pre-treatment, 15 ‍μl of supernatant were injected. Chromatographic separation was accomplished on a reversed phase HPLC C18 column. The separation was conducted using gradient mode mobile phase delivery for 3.5 ‍min at a 0.7 ‍ml/min flow rate. Column temperature was fixed at 25°C and fluorescence was measured by RF-20A xs detector (excitation/emission wavelengths: 320/415 ‍nm for PLP, 367/435 ‍nm for TPP). Vitamins were identified and quantified by comparing their retention times and peak areas with those of standards. Internal standards were also employed in order to correct possible losses or deviations.

Quality control

The assay method was subjected to internal quality control using materials provided by the manufacturer in each run, as well as external quality assessment (EQA) using the INSTAND program on a quarterly basis (https://www.instand-ev.de/en/instand-eqas/eqa-program/). EQA was performed under two schemes: Vitamins 01-290 (serum, for retinol and alpha-T) and Vitamins 02-291 (whole blood, for TPP and PLP). The target values were defined by the provider and Z-scores were calculated with respect to the other participants.

We estimated the total analytical error, which included both imprecision (coefficient of variation, CV) and trueness (bias). The acceptable total error was set according to the biological variability data obtained from the European Federation of Clinical Chemistry and Laboratory Medicine (EFLM) biological variation website (https://biologicalvariation.eu/).

Statistical analysis

All data were analyzed using STATA ver. 15.1 and GraphPad Prism 10.2.3. RI were calculated per test following the workflow presented in Fig. 1. Outliers were detected and excluded using the Tukey’s exclusion test. Visual inspection of the box plots and descriptive statistics were used to assess the need for partitioning (based on sex and age). To test for normality, the Shapiro–Wilk test was preferred because of the sample size. A p value of <0.05 was considered statistically significant. If a nonparametric distribution was observed, a Box–Cox transformation was performed. The lower and upper limits were estimated to be the 2.5th and 97.5th percentiles of the test results, respectively. We calculated 90% confidence intervals (CI) for both the lower and upper reference limits. The bias ratios (BR) at the lower and upper limits of the RI were calculated to assess the differences between RI (Equation 1–3).

Fig. 1.

Workflow used for calculating reference intervals, including the selection process for the reference sample group from the reference population. BMI, body mass index; CI, confidence interval; LRL, low reference limit; URL, upper reference limit.

$${\mathrm{S}\mathrm{D}}_{\mathrm{R}\mathrm{I}}=\frac{{\mathrm{U}\mathrm{R}\mathrm{L}}_{\mathrm{R}\mathrm{E}\mathrm{F}}-{\mathrm{L}\mathrm{R}\mathrm{L}}_{\mathrm{R}\mathrm{E}\mathrm{F}}}{3.92}$$ (1)

$${\mathrm{B}\mathrm{R}}_{\mathrm{L}\mathrm{R}\mathrm{L}}=\frac{{\mathrm{L}\mathrm{R}\mathrm{L}}_{\mathrm{R}\mathrm{E}\mathrm{F}}-\mathrm{L}\mathrm{R}\mathrm{L}}{{\mathrm{S}\mathrm{D}}_{\mathrm{R}\mathrm{I}}}$$ (2)

$${\mathrm{B}\mathrm{R}}_{\mathrm{U}\mathrm{R}\mathrm{L}}=\frac{{\mathrm{U}\mathrm{R}\mathrm{L}}_{\mathrm{R}\mathrm{E}\mathrm{F}}-\mathrm{U}\mathrm{R}\mathrm{L}}{{\mathrm{S}\mathrm{D}}_{\mathrm{R}\mathrm{I}}}$$ (3)

where BR is the bias ratio, SD_RI is standard deviation of the reference interval, RI is the reference interval, LRL is the lower reference limit, URL is the upper reference limit, and REF is the reference.

Alpha-T corrected for lipid levels was calculated from the alpha-T level divided by total cholesterol (alpha-T/CH ratio) or total lipid level (cholesterol + triglycerides; alpha-T/TL ratio) (Equation 4 and 5). TPP was corrected for hemoglobin using the ratio of TPP concentration divided by hemoglobin (Equation 6).

$$\begin{array}{c}\text{alpha-T}/\mathrm{C}\mathrm{H}\mathrm{r}\mathrm{a}\mathrm{t}\mathrm{i}\mathrm{o}\mathrm{}\left(\mathrm{\mu }\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{m}\mathrm{m}\mathrm{o}\mathrm{l}\right)=\hfill \\ \frac{\text{Serum alpha-T concentration}\left(\mathrm{\mu }\mathrm{M}\right)}{\text{Serum cholesterol concentration}\left(\mathrm{m}\mathrm{M}\right)}\hfill \end{array}$$ (4)

$$\begin{array}{c}\text{alpha-T}/\text{TL ratio}\left(\mathrm{\mu }\mathrm{m}\mathrm{o}\mathrm{l}/\mathrm{m}\mathrm{m}\mathrm{o}\mathrm{l}\right)=\hfill \\ \frac{\text{Serum alpha-T concentration}\left(\mathrm{\mu }\mathrm{M}\right)}{\text{Serum cholesterol concentration}\left(\mathrm{m}\mathrm{M}\right)+\text{serum triglycerides concentration}\left(\mathrm{m}\mathrm{M}\right)}\hfill \end{array}$$ (5)

$$\begin{array}{c}\text{TPP}/\text{Hb ratio}\left(\mathrm{n}\mathrm{g}/\mathrm{g}\right)=\hfill \\ \frac{\text{Whole blood TPP concentration}\left(\mathrm{n}\mathrm{g}/\mathrm{L}\right)}{\text{Whole blood hemoglobin concentration}\left(\mathrm{g}/\mathrm{L}\right)}\hfill \end{array}$$ (6)

Results

The results of the EQA program for 2023 and 2024 were satisfactory (Z-scores <2 for all results). The bias and CV values obtained were −0.4 and 5.8% for alpha-T, 8.2 and 3.9% for retinol, −2.0 and 5.5% for PLP, and −1.5 and 4.7% for TPP, respectively. The total analytical errors were 7.4%, 13.3%, 2.9%, and 2.2% for alpha-T, retinol, PLP, and TPP, respectively; the total allowable errors were 10.5%, 15.7%, 9.2%, and 3.7%, respectively.

The demographic data and analytical tests of the participants included in this study are presented in Table 1. The results confirmed the absence of acute (negative inflammatory markers) or chronic (normal kidney and liver markers) pathological conditions. In addition, no evidence of malnutrition was found among the volunteers. To assess the effect of age, participants were divided into four age groups: 18–29 years (n = 33, 22.6%), 30–39 years (n = 40, 27.4%), 40–49 years (n = 38, 26.0%), and 50–65 years (n = 35, 24.0%). The percentage of values excluded by the test was a maximum of 6% and <4%, if outliers were stratified by subclass.

Table 1.

Descriptive and comparative statistics of laboratory analytes between female and male subjects

	Female	Male	p value
Gender	63.7% (n = 93)	36.3% (n = 53)	0.0014
Age (years)	39.2 (11.5)	41.6 (10.7)	0.1595
18–29	26.9%	15.1%	0.1269
30–39	25.8%	30.2%
40–49	26.9%	24.5%
50–65	20.4%	30.2%
Liver function
ALT (U/L)	15 (6)	24 (15)	<0.0001
AST (U/L)	18.0 (6)	22 (7)	0.0004
GGT (U/L)	14 (6)	21 (14)	0.0186
ALP (U/L)	56 (20)	62 (21)	0.0250
Renal function
Creatinine (mg/dl)	0.70 (0.1)	0.9 (0.2)	<0.0001
Urea (mg/dl)	32 (9)	37 (10)	0.0013
Nutrition factors
Erythrocytes (×1012/L)	4.5 (0.4)	4.9 (0.3)	<0.0001
Hb (g/dl)	13.3 (1.2)	14.9 (0.9)	<0.0001
Lymphocytes (×109/L)	1.8 (0.7)	1.8 (0.6)	0.9700
Lipid metabolism
CH (mg/dl)	178 (35)	189 (50)	0.4453
HDL-CH (mg/dl)	68 (14)	55 (21)	<0.0001
LDL-CH (mg/dl)	101 (32)	109 (44)	0.0162
Triglyceride (mg/dl)	64 (37)	79 (41)	0.0056
Inflammation markers
Leukocyte count (×109/L)	5.5 (1.8)	5.8 (1.8)	0.3973
Neutrophil count (×109/L)	3.0 (1.2)	3.0 (1.2)	0.6306
Platelet count (×109/L)	239 (66)	233 (73)	0.8118

Continuous variables are expressed as median and interquartile range (in parentheses). The Mann–Whitney test was used to examine the differences between continuous variables. The chi-squared test was used to assess the effects of sex and age (grouped by range). Statistical significance was set at p<0.05. ALT, alanine aminotransferase; AST, aspartate aminotransferase; GGT, gamma glutamyl transferase; ALP, alkaline phosphatase; Hb, hemoglobin; CH, cholesterol; HDL, high-density lipoprotein; LDL, low-density lipoprotein.

The four vitamins included in this study showed a parametric distribution after discarding the outliers; however, the ratios deviated from the statistical normal (p>0.05) (Fig. 2). No stratification based on age or sex was performed (Fig. 3 and 4).

Fig. 2.

Normal QQ plots of vitamins and ratios. P value corresponds to Shapiro–Wilk test. Alpha-T, alpha-tocopherol; Hb, hemoglobin; CH, cholesterol; PLP, pyridoxal 5'-phosphate; TL, total lipids; TPP, thiamine pyrophosphate.

Fig. 3.

Distribution of data by sex for all analytes and ratios. L(λ) value = 1.727214 for TPP/Hb, −1.14114 for alpha-T/TL, and −1.142716 for alpha-T/CH. Alpha-T, alpha-tocopherol; Hb, hemoglobin; CH, cholesterol; PLP, pyridoxal 5'-phosphate; TL, total lipids; TPP, thiamine pyrophosphate.

Fig. 4.

Distribution of data by age for all analytes and ratios. For each subclass, the mean and standard deviation are plotted (green line). Alpha-T, alpha-tocopherol; Hb, hemoglobin; CH, cholesterol; PLP, pyridoxal 5'-phosphate; TL, total lipids; TPP, thiamine pyrophosphate.

The calculated reference intervals and 90% CI are presented in Table 2, which detail the final sample size for each vitamin and the ratio after outlier elimination. All lower and upper limits were statistically different from the RI currently reported by the reagent manufacturers, determined using the BR calculation with a threshold of 0.375, except for the lower limits of alpha-T and PLP. In all reference limits, the width of the 90% CI was less than 0.2 times the width of the RI.

Table 2.

Currently reported and calculated reference intervals for vitamins and related ratios

Test/Ratio	RIreported	n	LRL	URL
Retinol [Vitamin A] (μmol/L)	1.6–2.5	137	1.1 (1.1–1.3)	2.8 (2.7–2.9)
Alpha-Tocopherol [alpha-T, Vitamin E] (μmol/L)	18.6–46.2	139	18.9 (17.6–21.4)	42.2 (41.6–43.2)
Alpha-T/CH ratio (μmol/mmol)	NA	138	5.2 (4.9–5.3)	8.9 (8.6–9.3)
Alpha-T/TL ratio (μmol/mmol)	NA	137	4.5 (4.2–4.6)	7.4 (7.3–7.5)
Thiamine pyrophosphate [TPP, Vitamin B1] (nmol/L)	78–144	141	85.9 (78.1–95.8)	181.6 (173.3–187.8)
TPP/Hb ratio (ng/g)	NA	135	246.2 (237.4–269.7)	561.1 (535.0–588.8)
Pyridoxal 5'-phosphate [PLP, Vitamin B6] (nmol/L)	51–183	138	57.0 (48.2–60.9)	165.7 (155.2–176.6)

Numbers in parentheses indicate 90% confidence intervals of each limit. RI_reported, reference intervals currently reported in our laboratory; LRL, lower reference limit; URL, upper reference limit; NA, not applicable.

Discussion

In this study, the RI were calculated using a direct approach for several vitamins that are routinely analyzed in a clinical laboratory, as well as related ratios that may facilitate their clinical interpretation. The direct approach requires a minimum of 120 healthy individuals for each test and partition, to ensure that the sample is sufficiently representative of the general population and that possible biases are mitigated. The statistical analysis of these data is straightforward, and the RI is defined as the central 95% of the data range obtained.⁽²⁰⁾ However, the direct approach has some well-known limitations: it requires a sufficient number of samples from well-defined subjects to be collected and analyzed, and is time-consuming and expensive.^(5,21) Although the calculation of RI using a direct approach is becoming increasingly obsolete due to these limitations, it is still considered in studies as the applicability of the indirect approach remains limited. In our hospital, these limitations were decisive for the choice of method, as vitamin tests can only be requested by specialists (in-patients and out-patients, excluding primary care). Thus, the available LIS data would not fulfil the basic requirement for the calculation of the RI by indirect methods, which is a high proportion of healthy individuals in the database. In this study, a total of 146 healthy individuals were enrolled using well-established inclusion and exclusion criteria.

All the lower and upper limits were statistically different from the currently used RI, except for the lower limits of alpha-T and PLP. The RI for vitamin A was wider than the current one, whereas that of vitamin E was slightly narrower. The RI values for TPP and PLP showed moderate differences in the lower limits, but significant differences were observed in the upper limits, when compared with the RI recommended by manufacturers. Our results are consistent with those of previous studies that did not recommend stratification based on sex or age.^(18,22–24)

The lower limit for vitamin A was comparable but the upper limit was lower than that reported in a multicenter study from five European countries: 1.07–3.55 ‍μmol/L for men and 1.07–3.32 ‍μmol/L for women.⁽²⁵⁾ In another study, vitamin A RI found for men were 1.05–3.39 ‍μmol/L (>20 years), and for women were 0.73–2.66 ‍μmol/L (10–30 years) and 0.83–2.84 ‍μmol/L (>30 years).⁽²⁶⁾ The observed differences could be due to differences in the selected populations, including dietary, environmental, or genetic factors. The RI for alpha-T and alpha-T/CH ratio observed in the present study were found to be narrower than the previously reported values of 24.62–54.67 ‍μmol/L and 5.11–11.27 ‍μmol/mol.⁽²⁷⁾ A recent study carried out in a Spanish population using an indirect approach reported reference values for alpha-T and showed similar results to our study, ranging from 17.3 to 50.9 ‍μmol/L.⁽²⁴⁾ Compared with our results, the upper limit of both studies was higher, supporting the idea that not only the reference sample and population selected but also the methodology used and the statistical treatment of the data have an important influence on the reference values obtained. We obtained a similar but slightly narrower range for TPP compared with previously reported values of 85–203 nmol/L.⁽²²⁾ Low RI values were obtained by Lu and Frank,⁽²⁸⁾ 70–179 nmol/L. In patients with atypical clinical manifestations of Wernicke’s encephalopathy or those without a history of alcohol consumption, the diagnosis can be confirmed by determining blood thiamine concentrations.⁽²⁹⁾ In such cases, it is of the utmost importance to correctly define population-specific reference values, particularly the lower reference limit. Van Zelst et al.^(30,31) used two reference ranges for PLP in whole blood, 35–110 nmol/L and 51–183 nmol/L: the first ones, probably their own, but not published; the second ones, calculated by Steen et al.⁽³²⁾ using Chromsystems^® reagents. Other reference values for PLP have been reported for plasma⁽²²⁾ and erythrocytes.⁽³³⁾

A strong dependence of alpha-T levels on serum lipids has been previously observed.⁽³⁴⁾ The calculation of these ratios involving vitamin E, could benefit patients with impaired lipid metabolism, improving the interpretation of the results and clinician decision-making.^(35,36) Thurnman et al.⁽³⁷⁾ evaluated several lipid ratios in a group of controls (n = 40) and alcoholic patients (n = 85). Using these cohorts and regression analysis, they calculated the cutoff points for vitamin E deficiency at different ratios. They used the tocopherol: cholesterol + triglyceride + phospholipid ratio as a reference. They concluded that the alpha-T/TL ratio, followed by the alpha-T/CH ratio, had the highest diagnostic ability for detecting vitamin E deficiency (sensitivity and specificity of 95% and 99% vs 86% and 94%, respectively). Recent publications have used one or the other, the alpha-T/TL ratio or the alpha-T/CH ratio.^{(34,35,38,39)} In this study, the cholesterol and total lipid ratios showed excellent agreement (Pearson’s r = 0.8950). They can be interchangeably used in healthy patients. Ford et al.⁽³⁵⁾ recommended the use of the alpha-T/CH ratio to assess vitamin E status in patients with conditions that lead to elevated LDL cholesterol levels, such as cholestasis, pancreatic insufficiency, or liver transplantation, but did not recommend a general estimation for all patients. We recommend their addition at the discretion of the physician, but always for patients with hypo- or hypertriglyceridemia or cholesterolemia. Depending on additional patient information such as statin use or cholestasis, these may also be reported.

In terms of the ratio of water-soluble vitamin B1 to hemoglobin, Evliyaoglu et al.⁽⁴⁰⁾ published a retrospective analysis in Germany, in which they considered the inclusion of hemoglobin correction in the measurement of vitamin B1 in non-anemic patients because of the possibility of post-analytical errors leading to falsely elevated results. Talwar et al.⁽¹⁸⁾ recommended whole blood analysis only when accompanied by hemoglobin estimation in a prospective analysis in UK.

There are some limitations in the present study. First, although the age range chosen is that recommended by the CLSI, the world population is becoming increasingly older, and the upper limit of 65 years is falling short. Second, factors such as smoking status and pregnancy, which could affect vitamin levels, were not considered in this study. However, to the best of our knowledge, no pregnant women were included.

In conclusion, in this study, we updated the RI of vitamins A, E, B1, and B6, and their related ratios. Some published works suggest a benefit to the analysis of additional ratios since they may also be useful for certain patients. In this context, each laboratory should consider their inclusion depending on the available resources and the types of patients seen in their geographical area. In addition, we obtained a reliable and updated RI of water- and fat-soluble vitamins that can be used by other laboratories analyzing vitamins in populations similar to ours, thereby improving patient care management and optimizing vitamin supplementation guidance.

Author Contributions

Recruitment of participants: AA and MC; Study concept and design: AA, AB-G, AC, LC, MC, PG-M, GG-S, VM-R, FM, LP-S, and YV; Acquisition and analysis of data: AB-G, GG-S, and FM; Drafting the manuscript: AC, VM-R, LP-S, and YV; Statistical analysis: SG-E, CS-G, LC, and PG-M; Critical revision of the manuscript: AA, AB-G, AC, LC, MC, PG-M, SG-E, GG-S, VM-R, FM, LP-S, and CS-G.

Ethics Approval Statement

The study was approved by the Vall d’Hebron Hospital Institutional Review Board (PR(AG)188-2023) and was conducted in accordance with Spanish and European legislation and the tenets of the Declaration of Helsinki.

Abbreviations

alpha-T

alpha-tocopherol

BMI

body mass index

BR

bias ratios

CH

cholesterol

CI

confidence intervals

CKD

chronic kidney disease

CLSI

Clinical and Laboratory Standards Institute

CV

coefficient of variation

EFLM

European Federation of Clinical Chemistry and Laboratory Medicine

EQA

external quality assessment

Hb

hemoglobin

LDL

low-density lipoproteins

LIS

Laboratory Information System

TL

total lipids

PLP

pyridoxal 5'-phosphate

RI

reference intervals

TG

triglyceride

TPP

thiamine pyrophosphate

VLDL

very low-density lipoproteins

WLS

weight-loss surgery

Conflict of Interest

No potential conflicts of interest were disclosed.