cia- and all-trans-/3-carotene. The method also detects Materials and Methods Chromatography

Transcription

1 CLIN. CHEM. 40/3, (1994) #{149} Automation and Analytical Techniques Retinol, a-tocopherol, Lutein/Zeaxanthin, /3-Cryptoxanthin, Lycopene, a-carotene, trans-p-carotene, and Four Retinyl Esters in Serum Determined Simultaneously by Reversed-Phase H PLC with M ultiwavelength Detection Anne L. Sowell, Daniel L. Huff, Patricia R. Yeager, Samuel P. Caudill, and Elaine W. Gunter We describe the use of HPLC with muitiwavelength detection to measure retinol, a-tocopherol, Iutein/zeaxanthin, f3-cryptoxanthin, lycopene, a-carotene, trans-p-carotene, /3-carotene, and the linoleate, oleate, palmitate, and stearate esters of retinol in a single 200-L serum sample. The method is sensitive enough to detect individual retinyl esters in fasting serum from a nonhyperlipidemic population and requires only 12 mm for each sample. Serum concentration ranges and means are reported for retinol, a-tocopherol, lutein/zeaxanthin, /3-cryptoxanthin, lycopene, a-carotene, trans-/3-carotene, and the sum of the retinyl esters from serum analyses of 3480 participants from several different studies. IndexIng Terms: nutritional status/vitamin A/vitamin E/carotenoids Serum concentrations of the fat-soluble nutrients retinol, a-tocopherol, /3-carotene, and other carotenoids are measured in studies of general nutrition as well as in studies of their potential use as chemopreventive agents for some cancers. They are also measured in clinical trials as adjuncts to conventional cancer therapy or to improve the outcome of non-cancer-related diseases. Thus there is a need for a rapid and sensitive method to measure the serum concentrations of these nutrients and, for retinol, to determine whether hypervitarninosis A, which can cause hepatotoxicity, is present. Moreover, because serum retinol is not a good indicator of high or toxic body concentrations of vitamin A (1), measurement of serum retinyl esters is becoming increasingly important. Among the methods used for simultaneous measurement of retinol and a-tocopherol is the quantification of one or more carotenoids (2-9). Other methods measure some carotenoids, retinol, and retinyl pahnitate and may include a-tocopherol (10-12). However, our review of the literature has found no other method that simultaneously quantifies retinol, a-tocopherol, and the most prevalent retinyl esters and carotenoids in serum. We describe an isocratic HPLC method in which an octadecylsilane column (5-pm particles) and simultaneous detection at three different wavelengths are used to measure concentrations in serum of retinol and a-to- Division of Environmental Health Laboratory Sciences, National Center for Environmental Health, Centers for Disease Control and Prevention, Public Health Service, US Department of Health and Human Services, Atlanta, GA. Address correspondence to this author at: Nutritional Biochemistry Branch, Centers for Disease Control and Prevention, Mail Stop F-18, 4770 Buford Hwy. NE, Atlanta, GA Fax ; ALS1@CEHEHL1.EM.CDC.GOV. Received May 20, 1993; accepted November 11, copherol, four retinyl esters (retinyl stearate, retinyl palmitate, retinyl oleate, retinyl linoleate), lutein/zeaxanthin, /3-cryptoxanthin, lycopene, a-carotene, and 13- cia- and all-trans-/3-carotene. The method also detects y-tocopherol, 5-tocopherol, 2,3 -anhydrolutein, a-cryptoxanthin, and a carotenoid peak with the same retention time as canthaxanthin but containing additional carotenoids. Materials and Methods Chromatography We used a chromatographic system consisting of a WISP 712 autosampler with a refrigeration unit, a Model 590 pump, a temperature-control module with a RCM-100 column heater, a Model 490 detector, and a Model 820 full-control data station with Maxima software (all from Waters Chromatography Div. of Millipore, Milford, MA). Extracts were held at 4#{176}C in the autosampler until injection of a 30-giL sample onto a 150 x 4.6 mm Ultramex octadecylsilane (C18) column packed with 5-Lm particles (Phenomenex, Torrance, CA) or a 150 x 4.6 mm octadecylsilane 5-nm-particle column (Burdick and Jackson Labs., Muskegon, MI); the column temperature was controlled at 29#{176}C. The mobile phase was an equivolume solution of ethanol and acetonitrile containing 0.1 ml of diethylamine per liter of solvent; the flow rate was 0.9 mldmin. Chromatograms were obtained for absorbances measured at 450, 325, and 300 nm. A standard curve was prepared daily from two injections of each standard at a single concentration; the standard curve was forced through zero. Quantification was based on peak-height measurements rather than peak area to allow quantification of the trans-only carotenoid isomers. Because commercial standards are not available for 13-cis-f3-carotene, we quantified this compound by using the response factor for the all-trans isomer, as is done in the National Cancer Institute/National Institute of Standards and Technology (NCI/NIST) fat-soluble vitamin quality-assurance program.2 Serum Samples Specimens analyzed in this study were sera collected from fasting subjects as part of several studies: the Third National Health and Nutrition Examination Sur- 2NOnsJdard abbreviations: NCI/NIST, National Cancer Institute/National Institute of Standards and Technology; NHANES Ill, Third National Health and Nutrition Examination Survey; HHANES, Hispanic Health and Nutrition Examination Survey. CLINICAL CHEMISTRY, Vol. 40, No. 3,

2 vey (NHANES III) (civilian, noninstitutionalized adults and children older than 1 year), the Eye Disease Case! Control Study (adults, ages years), the Proposed Surgical Control of Hypertension study (adults with hyperlipidemia, some of whom had had partial ileal bypass surgery), and the Dietary Intervention Study in Children (children, ages 8-10 years, with hyperlipidemia). All of these studies were approved by the appropriate institutional human subjects review board. The specimens were stored in polypropylene vials at - 70#{176}C for to 12 months before analysis. Sample Preparation Fat-soluble compounds were extracted from human serum according to the following procedure: In a glass 12 x 75 mm culture tube combine 200 L of serum with 200 tl of ethanol containing nonapreno-j3-carotene and retinyl butyrate as internal standards and vortex-mix for los. Add 1000 tl of hexane and vortex-mix for 30s. Separate the phases by centrifugation for 5 miii at 1500g. Transfer the hexane layer (900 L) to a 12 x 75 mm tube and dry under reduced pressure without heating in a Speedvac concentrator (Savant Instrument Co., Farmingdale, NY). Dry the extract to a waxy or glassy consistency, not to dryness. Add 100 tl of ethanol to dissolve the residue, then add 100 j.l of acetonitrile. Vortex-mix the extract and immediately filter it through a 0.45-am pore-size filter (no. SJHVOO4NS; Millipore, Bedford, MA) into an injection vial. Carry out all sample preparation at 27#{176}Cunder gold fluorescent lights (we used General Electric no. F4OGO) to avoid photooxidation of the analytes. Chemicals Standards of luteun, zeaxanthin, and 13-cryptoxanthin were provided by Hoffmann-LaRoche (Nutley, NJ). Standards of retinol, a-tocopherol, retinyl palmitate, lycopene, a-carotene, and 13-carotene were purchased from Sigma Chemical Co. (St. Louis, MO). Nonapreno-13-carotene, a-cryptoxanthin, and 2,3 -anhydrolutein were generously contributed by Frederick Khachik (Nutrient Composition Laboratory, US Department of Agriculture, Beltsville, MD). HPLC-grade acetonitrile and hexane were from Burdick and Jackson Labs. Absolute ethanol (USP), stored in glass, was obtained from various sources and checked before use for interfering substances by HPLC. Standard solutions were prepared by dissolving small amounts (1-2 L of liquids; for solids, an amount that occupied a similar volume) of the standards in ethanol, injecting the solutions onto the column, and collecting the effluent corresponding to the middle third of the peak. The purified material from several injections was pooled and diluted with mobile phase. The concentration of the standards was determined by ultraviolet or visible spectrophotometry. Table 1 shows typical concentrations of the standard solutions and the wavelengths and molar absorptivities used to determine the solution concentrations. The purified retinoid and tocopherol standard solutions were stored in glass vials under nitrogen at -70#{176}Cfor 6 months. The carotenoid standards were up Table 1. TypIcal concentrations of standard solutions, and wavelengths and molar absorptlvltles (In ethanol) used to determine analyte concentrations. Wavelength,, Target conc, Analyt. nm L mor cm pmol/l Retinal a-tocopherol Lutein/zeaxanthin Cryptoxanthin Lycopene a.carotene a-carotene Retinyl linoleate Retinyl oleate Retinyl palmitate Retinyl stearate stored in polypropylene vials under the same conditions. All standards were allowed to equilibrate to room temperature before use. Under these conditions we have not observed irreversible precipitation of the standards during storage. Retinyl butyrate, retinyl linoleate, retinyl oleate, and retinyl stearate standards were synthesized from retinol and the corresponding acid anhydrides (purchased from Sigma) as follows: Add 3 ml of triethylamine, 20 ml of n-hexane, and 1.00 mmol of the appropriate acid anhydride to 0.87 mmol of retinol in a 5O-mL round-bottom flask. Stir the reaction mixture for 3.5 h at 60#{176}C, cool, and remove the hexane and triethylamine under reduced pressure. Separate the ester from the residue by column chromatography on a column of grade III alumina, using the procedure described by Ross (13). The ester is isolated as a colorless material with yellow fluorescence under ultraviolet light. The retinyl esters are stored in glass vials at -70#{176}C. Comparison of Methods for Retinol and a-tocopherol To compare this method with the HPLC method used to measure retinol and a-tocopherol for the Hispanic Health and Nutrition Examination Survey (HHANES) (14, 15), we analyzed 209 specimens from the NHANES III, Tampa, FL, pilot study in single measurements by both methods on the same days. Results Chromatography Figure 1 is a set of overlaid chromatograms from different standard mixtures containing the internal standards. Retinol and the retinyl esters are detected at 325 nm, the tocopherols at 300 nm, and the carotenoids at 425 nm. Fig. 2 is a set of chromatograms from a typical serum sample. There is no absorbance at 450 run from any of the retinoids or tocopherols. There is some absorbance at 325 nm from lycopene and f3-carotene, which may interfere with quantification of retinyl iinoleate and retinyl oleate, respectively, especially in serum from people who have taken )3-carotene supplements. However, retinyl palmitate and retinyl stearate 412 CUNICAL CHEMISTRY, Vol. 40, No. 3, 1994

3 have been the predominant retinyl esters in virtually all of the samples we have analyzed for NHANES Ill. Retinyl palmitate and retinyl stearate are present in most of the specimens we have analyzed; the other retinyl esters are found less often. In serum samples, lycopene is a combination of cis and trans isomers and its peak is therefore much broader than in the mixed standards. We observed no broadening and shortening of the lycopene peak in specimens analyzed at both the beginning and the end of analytical runs when we used a refrigerated autosampler. Similarly, no increase in the height of the cis-/3-carotene peak, with or without concurrent decrease in the trans-$-carotene peak, was observed. These chromatographic changes, which indicate isomerization of lycopene or 13-carotene, were usually present in chromatograms at the end of runs where the autosampler was not refrigerated. 4 Minutes Fig.1. Overlaid chromatograms of mixed standard solutions (each normalized to highest peak) at 450, 325, and 300 nm. Peaks: 1, lutein/zeaxanthin; 2, p-ciyptoxanthin: 3, lycopene; 4, a-carotene; 5, p-carotene; 6, nonapreno-/3-carotene; 7, c!s-nonapreno-/3-carotene; 8, retinol; 9, retinyl butyrate; 10, a-tocopherol; 11, retinyl linoleate; 12, retinyl oleate; 13, retinyl palmitate; 14, relinyl stearate Linearity, Detection Limits, and Precision of the Method The method is linear for all analytes over their physiological ranges. Specimens with analyte concentrations outside of the linear range of the method are diluted with isotonic saline and reanalyzed. Table 2 contains the interrun CVs, intrarun CVs, and mean concentra- Table 2. Inter- and Intrarun CV5 for retlnol, cv-tocopherol, five carotenoids, and retlnyl palmltate In three serum pools analyzed In >370 runs over 45 months. Pool 1788 Pool 1888 Pool 1988 CV,% Mean, Mean, Mean, Analyte,umol/L Interrun lntrarun 1imol/L lnterrun Intrarun pmol/l Interrun lntrarun Retinol a-tocopherol Lutein/zeaxanthin Cryptoxanthin Lycopene a-carotene a-carotene Retinyl palmitate (0 E Fig. 2. (a) Chromatograms of a typical human serum specimen analyzed at 450, 325, and 300 nm; (b) enlargement of the 450 nm and 325 nm chromatograms, showing the retinyl esters. Peaks land serum concentration (1anol/L), when quantified] are as follows: 1, lutein/zeaxanthin (0.27); 2, 2,3 -anhydrolutein; 3, carotenoids,includingcanthaxanthln; 4, a-cryptoxanthin; 5, (3-cryptoxanthin (0.19); 6, lycopene (0.11);7,a-carotene (0.07);8, trans- /3-carotene (0.19); 9, 13-cis-p-carotene; 10, nonapreno-/3- carotene; 11, retinol (2.66);12, retinyl butyrate;13, y-tocopherol; 14, a-tocopherol (65.32);15, retinyl oleate (0.05); 16, retinyl palmitate (0.22); 17, retlnyl stearate(0.15). 0, E IS 0 Minutes CLINICAL CHEMISTRY, Vol. 40, No. 3,

4 Table 3. TypIcal retention times and lower detection limits. Retinol Analyte a-tocopherol Lutein/zeaxanthin p-cfyptoxanthin Lycopene a-carotene p-carotene Retinyl linoleate Retinyl Retinyl Retinyl oleate palmitate stearate a In 3O-. injection. Ret.ntlon time, mln Mass ng Conc, imol/l tions for retinol, a-tocopherol, the carotenoids, and retinyl palrnitate, analyzed in three quality-control pools run in duplicate over 45 months. Typical detection limits are shown in Table 3. Detection limits for retinyl linoleate, retinyl oleate, and retinyl stearate are comparable with that of retinyl palmitate. The sensitivity of this method for retinyl esters allows the measurement of these analytes in fasting serum from normal subjects. More than samples have been analyzed by this method. Table 4 shows the ranges and means of serum concentrations of retinol, a-tocopherol, the five carotenoids, and the sum of the retinyl esters in samples from 3480 subjects of all ages, analyzed as part of several different studies. Comparison with HHANES Method for Retinol and a-tocopherol The retinol and a-tocopherol results were compared with the results obtained with the method used in the Table 4. Concentration means and ranges of analytes (prnoiil) from 3480 free-living male and female subjects, ages 4-93 years. Analyte Mean Lutein/zeaxanthin 0.36 p-cryptoxanthin 0.22 Lycopene 0.40 a-carotene 0.08 p-carotene 0.34 Retinyl ester sum 0.18 Retinol 1.91 a-tocopherol percentile range HHANES study. Using linear regression analysis with an error-in-both-variables technique (16) to compare the results for retinol by the two methods, we obtained values for slope, intercept, and correlation coefficient of 1.064, 1.8, and 0.987, respectively, for 209 singlet analyses. For a-tocopherol these values were 0.959, 78, and 0.976, respectively, for 205 singlet analyses in which a-thcopherol was 70 mo1/l. The inter- and intrarun CV ranges for retinol and a-tocopherol for both methods are shown in Table 2. Except for a-carotene at the low concentration, the CVs obtained with this method are comparable with those reported by Milne and Botnen (5) and Thurnham et al. (11). The CVs for the carotenoids are higher than those reported by Cantilena and Nierenberg (9); however, their CVs are based on an extremely limited number of analyses, which may not accurately reflect the performance of the method over time. Our values are for >370 analytical runs over almost 4 years and demonstrate the true long-term precision of our method. The results for retinol and a-tocopherol obtained with this method are comparable with results by the method used in the HHANES study, but our method has several advantages: shorter run time, so more samples can be run each day, and detecting a-tocopherol at 300 nm to avoid interference from plasticizers (4), cholesterol, and materials produced by degradation of serum in longterm storage (17). Our laboratory has been participating in the NCI/ NIST fat-soluble vitamin quality-assurance program for vitamins A and E and 13-carotene analyses since When this method was used for Round Robins 20 through 26 (28 specimens), our laboratory had a mean bias of 3% for retinol, 2.7% for a-tocopherol, and a mean bias from the mean value for the laboratories participating for >1 year of 6.9% for total (3-carotene. The concentrations of retinol, a-tocopherol, total (3-carotene, and all-trans-(3-carotene in NIST Standard Reference Material 968a as measured by our laboratory are shown in Table 5. Discussion In studies such as NHANES III, where many analyses must be carried out on a large number of samples over several years, it is necessary to maximize the amount of information that can be obtained from each analytical method and to select methods at the beginning of the study that are rugged, highly precise for analytes in the physiological range, and state-of-the-art (18). By using Table 5. A esu Its of analysis of NIST Standard Reference Mate Mean (SD) conc, pmol/l rial 968a. Low MedIum High Analyte Retinol a-tocopherol Total p-carotene All-trans p-carotene Our results (0.013) (0.58) (0.052) 0.46 (0.05) NIST 0.642(0.056) (0.51) 0.492(0.147) 0.37 Our results NIST (0.022) (0.073) (0.36) (0.60) 1.70 (0.097) 1.83 (0.153) 1.57 (0.11) 1.7 Our results (0.050) (0.25) 4.08 (0.153) 3.79 (0.15) NIST 2.263(0.091) (1.30) 4.94 (0.41) CLINICAL CHEMISTRY, Vol. 40, No. 3, 1994

5 a multichannel detection system and a chromatographic system that separates the carotenoids as well as the tocopherol isomers and the predominant retinoids, we could measure in a single analysis several micronutrients that are important to general health as well as some that are of interest in specialized applications. With the use of automated equipment for analysis and data handling, we analyzed an average of 300 serum samples per week per instrument, in addition to standards and controls. Refrigeration of the samples between preparation and analysis allowed us to prepare a large number of samples without detectable isomerization of the carotenoids between sample preparation and analysis. Heating the column to slightly above ambient temperature did not significantly affect the separation but did ensure that fluctuations in room temperature did not affect the chromatographic conditions. Because of our high sample throughput, and to ensure maximum quality of our survey data, the chromatographic columns are replaced every 3 months to ensure consistent performance. The columns we use do not undergo significant deterioration during 3 months of -95 injections per day, 4 days per week. The Maxima software limits the choice of internal standards because the system requires that an internal standard be present on each channel with the components to be quantified. Therefore, at least two internal standards must be used, one for the visible wavelength and one for the ultraviolet wavelengths. The use of a retinyl butyrate instead of tocol or tocopherol acetate allows the same material to be used as the internal standard for retinol, the retinyl esters, and a-tocopherol. Our performance on the NCIJNIST fat-soluble nutrient Round Robin analyses and our own recovery studies indicate that, in this assay, retinyl butyrate is a good internal standard for retinol and a-tocopherol, but its use may lead to underestimation of the retinyl ester concentrations of lipemic serum. We measure the retinoids at 325 nm rather than at 300 nm so that we can quantify the low concentrations of retinyl esters present in fasting serum. The tocopherols are detected on a separate channel to avoid baseline fluctuations associated with changing the wavelength in the middle of a chromatogram. The tocopherols are measured at 300 nra instead of nm, to eliminate interferences from plasticizers leached from serum storage bags (4) and breakdown products present in aged serum (17). We use nonapreno-(3-carotene as the internal standard for the carotenoid analysis because it does not elute near any of the other analytes. One objection to the use of nonapreno-j3-carotene is that it is retained longer than (3-carotene and lengthens the chromatographic run. This is not a problem with our method because retinyl stearate elutes after nonapreno-(3-carotene. Because nonapreno-(3-carotene has physical properties quite similar to a-carotene and (3-carotene, it is an excellent internal standard for these analytes; however, the xanthophylls are considerably more hydrophilic than the carotenes and therefore are more completely extracted than the later-eluting carotenoids. This can lead to overestimation of the xanthophylls in lipemic specimens. The use of tocopherol acetate, echinenone, or other material eluting before a-carotene may lessen the problem of overestimation of the xanthophylls but can lead to underestimation of a-carotene and (3-carotene in lipemic specimens. It has not been possible to automate sample preparation for this analysis; however, the use of a multitube vortex-mixer and automated dispensers have made sample processing less time consuming and labor intensive. We use only one hexane extraction of the serum; using two hexane extractions increased the amount of analyte recovered by only -6%-----too little to compensate for the increased effort in sample preparation. The use of internal standards with extraction recoveries similar to those of the analytes minimizes the effects of incomplete recovery. This method is being used to determine serum concentrations of retinol, a-tocopherol, five carotenoids, and the sum of four retinyl esters for the NHANES III survey, which will provide normative values of these analytes for the US population. We are also obtaining information about serum concentrations of three additional carotenoids (2,3 -anhydrolutein, a-cryptoxanthin, and 13-cis-/3-carotene) and y-tocopherol. We routinely perform the analysis with 200 L of serum, although we have successfully used as little as 75 L of serum when analyzing pediatric specimens. This method is especially useful for large clinical studies or surveys because of its short chromatographic run time and its analytical precision being similar to that of methods with longer run times. In addition to the NHANES III survey, this method has been used to measure serum and tissue concentrations of retinol and retinyl esters in rats to study the effect of glucocorticosteroids on retinol and retinyl ester concentrations (19), to measure serum retinol in children with measles (20, 21), and to determine serum carotenoid and retinoid concentrations of children at risk for vitamin A deficiency in the US, Belize, and Micronesia. This work was supported in part by the National Center for Health Statistics, Centers for Disease Control and Prevention, Public Health Service. The Eye Disease Case/Control Study is supported by an intraagency agreement with the National Eye Institute; analyses for the Proposed Surgical Control of Hypertension study were performed for the National Heart, Lung, and Blood Institute; and the Dietary Intervention Study in Children is supported by an intraagency agreement with the National Heart, Lung, and Blood Institute. The retinyl ester standards were prepared by Vik Reddy. Use of trade names is for the purpose of identification only and does not constitute endorsement by the Public Health Service or the US Department of Health and Human Services. References 1. Smith FR, Goodman DWS. Vitamin A transport in human vitamin A toxicity. N Engi J Med 1976;294: Mifier KW, Lorr NA, Yang CS. Simultaneous determination of plasma retinol, aipha-tocopherol, lycopene, alpha-carotene, betacarotene by high-performance liquid chromatography. Anal Biochem 1984;138: Miller KW, Yang CS. An isocratic high-performance liquid chromatography method for the simultaneous analysis of plasma CLINICAL CHEMISTRY, Vol. 40, No. 3,

6 retinol, alpha-tocopherol, and various carotenoids. Anal Biochem 1985;145: Nierenberg DW, Lester DC. Determination of vitamins A and E in serum and plasma using a simplified clarification method and high-performance liquid chromatography. J Chromatogr 1985; 345: Mine DB, Botnen J. Retinol, a-tocopherol, lycopene, and a- and p-carotene simultaneously determined by isocratic liquid chromatography. Cliii Chem 1986;32: Sam DC, Santa-Cruz MC. Simultaneous measurement of retinol and alpha-tocopherol in human serum by high-performance liquid chromatography with ultraviolet detection. J Chromatogr 1986;380: MacCrehan WA, Schonberger E. Determination of retinol, a-tocopherol, and p-carotene in serum by liquid chromatography with absorbance and electrochemical detection. Clin Chem 1987; 33: Nierenberg DW, Nann SL. A method for determining concentrations of retinol, tocopherol and five carotenoids in human plasma and tissue samples. Am J Cliii Nutr 1992;56: Cantilena LR, Nierenberg DW. Simultaneous analysis of five carotenoids in human plasma by isocratic high performance liquid chromatography. J Micronutr Anal 1989;6: Kalman DA, Goodman GE, Omenn GS, Bellamy G, Rollins B. Micronutrient assay for cancer prevention clinical trials: serum retinol, retinyl palmitate, alpha-carotene and beta-carotene using high performance liquid chromatography. J Natl Cancer Inst 1987;79: Thurnham DI, Smith E, Flora PS. Concurrent liquid-chromatographic assay of retunol, a-tocopherol, p-carotene, a-carotene, lycopene, and /3-cryptoxanthin in plasma with tocopherol acetate as internal standard. Clin Chem 1988;34: Olmedilla B, Granado F, Rojas-Hidalgo E, Blanco I. A rapid separation of ten carotenoids, three retinoids, alpha-tocopherol and d-alpha-tocopherol acetate by high performance liquid chromatography and its application to serum and vegetable samples. J Liq Chromatogr 1990:13: Ross AC. Separation of long-chain fatty acid esters of retinol by high-performance liquid chromatography. Anal Biochem 1981; 115: Gunter EW, Miller DT. Laboratory procedures used by the Division of Environmental Health Laboratory Sciences, Centers for Disease Control, for the Hispanic Health and Nutrition Examination Survey (HHANES) Atlanta, GA: Centers for Disease Control, 1986: Driskell WJ, Neese JW, Bryant CC, Bashor MM. Measurement of vitamin A and vitamin E in human serum by highperformance liquid chromatography. J Chromatogr 1982;231: DemingWE. Statistical adjustment of data. New York: Wiley, 1943: Gunter EW, Driskell WJ, Yeager PR. Stability of vitamin E in long-term stored serum. Clin Chim Acta 1988;175: McQuillan GM, Gunter EW, Lannom L. Field issues for the plan and operation of the laboratory component of the Third National Health and Nutrition Examination Survey. J Nutr 1990;120: Georgieff MK, Radmer WJ, Sowell AL, Yeager PR, Blaner WS, Gunter EW, Johnson DE. The effect of glucocorticosteroids on serum, liver, and lung vitamin A and retinyl ester concentrations. J Pediatr Gastroenterol Nutr 1991;13: Frieden TR, Sowell AL, Henning KJ, Huff DL, Gunn RA. Vitamin A levels and severity of measles: New York City. Am J Dis Child 1992;146: Butler JC, Havens PL, Sowell AL, HUff DL, Peterson DE, Day SE, et al. Measles severity and serum retinol (vitamin A) concentration among children in the United States. Pediatrics 1993;91: CLINICAL CHEMISTRY, Vol. 40, No. 3, 1994