Carrie A. Redlich a, *, Joyce S. Chung a, Mark R. Cullen a, William S. Blaner b, Ariette M. Van Bennekum b, Lars Berglund b. 1.

Transcription

1 Atherosclerosis 143 (1999) Effect of long-term beta-carotene and vitamin A on serum cholesterol and triglyceride levels among participants in the Carotene and Retinol Efficacy Trial (CARET) Carrie A. Redlich a, *, Joyce S. Chung a, Mark R. Cullen a, William S. Blaner b, Ariette M. Van Bennekum b, Lars Berglund b a Yale Uni ersity School of Medicine, Occupational and En ironmental Medicine Program, Department of Medicine, 135 College St., New Ha en, CT 06510, USA b Department of Medicine, College of Physicians and Surgeons, Columbia Uni ersity, New York, NY, USA Received 13 May 1998; received in revised form 27 October 1998; accepted 6 November 1998 Abstract Objecti e: The Carotene and Retinol Efficacy Lung Cancer Chemoprevention Trial (CARET) ended prematurely due to the unexpected findings that the active treatment group on the combination of 30 mg -carotene and IU retinyl palmitate had a 46% increased lung cancer mortality and a 26% increased cardiovascular mortality compared with placebo. This study was designed when the CARET intervention was halted to evaluate the effects of long-term supplementation with -carotene and retinol on serum triglyceride and cholesterol levels, in an attempt to explore possible explanations for the CARET result. Methods: Serum triglyceride levels, and total, high-density lipoprotein (HDL), and low-density lipoprotein (LDL) cholesterol levels were determined in a subgroup of 52 CARET participants. Baseline and mid-trial levels were available on 23 participants on placebo and 29 on active treatment who were then serially followed for 10 months after trial termination. Results: Triglyceride, and total, HDL and LDL cholesterol levels were similar in the two groups at baseline. After a mean of 5 years on the intervention there was a small nonsignificant increase in serum triglyceride levels in the active group, but no difference in total, HDL, or LDL cholesterol levels. After stopping the intervention there was a decrease in triglyceride levels in the active intervention group, and no change in the other parameters. Conclusion: Based on a small convenience sample, CARET participants in the active treatment arm had a small nonsignificant increase in serum triglyceride levels while on the intervention, and a decrease in serum triglyceride levels after the intervention was discontinued. No significant changes in total or HDL cholesterol were noted. These results argue against a major contribution of treatment-induced changes in serum lipid and lipoprotein levels to the increased cardiovascular mortality in the active treatment group Elsevier Science Ireland Ltd. All rights reserved. Keywords: Vitamin A; -Carotene; Cholesterol; Triglyceride; High-density lipoprotein (HDL) 1. Introduction Abbre iations: CARET, The Carotene and Retinol Efficacy Trial; HDL, high-density lipoprotein; LDL, Low-density lipoprotein; SD, standard deviation; CI, confidence intervals. * Corresponding author. Tel.: ; fax: address: carrie.redlich@yale.edu (C.A. Redlich) The Carotene and Retinol Efficacy Trial (CARET) is a multicenter randomized cancer chemoprevention trial in which the active intervention ended prematurely due to the unexpected findings that the active treatment group on the combination of 30 mg -carotene and IU retinyl palmitate had a 46% increased lung cancer mortality and a 17% increased overall mortality /99/$ - see front matter 1999 Elsevier Science Ireland Ltd. All rights reserved. PII: S (98)

2 428 C.A. Redlich et al. / Atherosclerosis 143 (1999) compared with placebo [1]. This present study was undertaken at the time CARET was halted to determine the effects of long-term administration of the CARET intervention on serum total cholesterol, HDL and LDL cholesterol, and triglyceride levels in a subgroup of CARET participants. It was prompted by the unexpected finding of a significant 26% increase in cardiovascular mortality (46% increase with downweighting early end-points) in CARET participants who were on the active intervention. These findings were consistent with those in the -carotene arm of the Finnish -Tocopherol/ -Carotene Cancer Prevention Study [2], and more recently with those in the retinol arm of another lung cancer chemoprevention trial [3]. Why either -carotene and/or retinol would cause increased cardiovascular mortality in this population is unclear, but of serious concern, especially considering the magnitude of the effect and the widespread use of these dietary supplements [4,5]. Based on (1) the knowledge that retinoids can cause significant increases in triglyceride and LDL cholesterol levels and reductions in HDL cholesterol levels at routinely used pharmacologic and chemopreventive doses [6 11], (2) case reports of similar lipid changes following acute high doses of vitamin A [12,13], and (3) the well known strong association between reduced plasma HDL cholesterol levels and increased risk of ischemic heart disease [14], it was hypothesized that long-term supplementation with the CARET doses of vitamin A (retinyl palmitate) and -carotene might similarly raise triglyceride levels and lower HDL cholesterol levels. This effect would provide a possible explanation for the excess cardiovascular mortality seen in subjects on the active intervention. CARET investigators had previously reported a small increase in serum triglycerides with no significant change in serum total cholesterol [15], but did not report any data on HDL or LDL cholesterol levels. This present study was designed to determine the effects of long-term supplementation with the active intervention (30 mg -carotene and IU retinyl palmitate) on serum triglyceride levels, total cholesterol, and HDL and low-density lipoprotein (LDL) cholesterol in a subgroup of CARET participants. 2. Materials and methods 2.1. Study design and recruitment The unexpected sudden cessation of CARET placed constraints on study design, patient recruitment and other logistic aspects of the study. A sample of 65 CARET participants followed in the New Haven Center (total N=1043), were invited by phone to participate in the current study; this represented about 10% of the subjects who were still alive, living in Connecticut and currently participating. The participants were recruited based on their geographic proximity to the New Haven Center, before they or the investigators were informed of the participants intervention status (active vs placebo). Fifty-two of these 65 participants agreed to participate, 29 of whom were subsequently determined to be on the active agent and 23 on placebo. Fasting bloods were drawn within 1 week of discontinuing the CARET intervention (mean 2.5 days) and at 1- to 3-month intervals up to 14 months off the intervention. Participants were interviewed regarding their food intake prior to their blood draw to confirm fasting status. In addition, as part of the original CARET study design, serum had been obtained on all subjects prior to randomization (baseline), and at 2-year intervals thereafter, which had been stored at 70 o C at the CARET Coordinating Center. Relevant routinely collected stored samples on participants were obtained. Sixtyeight percent of all blood samples were fasting. The study was approved by the Human Investigations Committees at Yale University and Lawrence and Memorial Hospital, and written informed consent was obtained from each subject. Baseline and demographic information previously obtained was available from the CARET Coordinating Center. The use of lipid lowering medications and diagnosis of diabetes was ascertained by review of the patient s medical and medication records Laboratory analyses Venipuncture was performed under reduced light and samples stored in the dark. Serum was immediately separated by centrifugation and frozen at 70 o C. Serum levels of triglycerides and total cholesterol were determined using standardized enzymatic procedures (Boehringer Mannheim, Germany) in a Hitachi 705 automated spectrometer. HDL cholesterol levels were analyzed after precipitation of apo B-containing proteins with phosphotungstic acid [16]. LDL cholesterol levels were calculated using the Friedewald formula [17]. The laboratory participated in the Centers for Disease Control Lipid Standardization Program, and interassay coefficients of variation were 2% for cholesterol and triglycerides and 3% for HDL cholesterol. Serum levels of retinol and -carotene were measured by reverse phase high performance liquid chromatography (HPLC) as previously described [18,19] under reduced light Statistical analyses Values are expressed as the mean standard deviation (S.D.). Baseline characteristics of the ancillary

3 C.A. Redlich et al. / Atherosclerosis 143 (1999) participants were compared to the entire New Haven population and between intervention arms using the Student s t-test. Triglyceride values and serum carotene levels were log-transformed to correct positive skewness. Lipid levels between intervention arms at three points in the study (baseline, on intervention, and off intervention) were tested using a Student s t-test. The relative risk of developing diabetes or using lipid lowering medication was tested for significance using the Fisher s exact test. Statistical significance was achieved at an level of Data Desk 6.0 PPC (Ithaca, NY) and SAS statistical programs were used for analysis. 3. Results 3.1. Characteristics of the participants The baseline characteristics of the 52 participants in the current study were very similar to the larger group of 1043 New Haven CARET subjects; no significant differences in age, race, duration in study, smoking status, or other parameters were observed (data not shown). A comparison of baseline characteristics of the 23 participants in the placebo group and 29 participants in the active intervention group is shown in Table 1. The two groups were well matched with no significant Table 1 Baseline characteristics of ancillary participants a Placebo Active No. in intervention Age at first visit (years) Mean S.D Sex % Male (c women) 91% (2) 90% (3) Duration in study (years) Mean S.D Race White 23 (100%) 28 (97%) Other 1 (3%) Weight (lbs) Mean S.D Smoking status Former 16 (70%) 16 (55%) Current Pack-year (years) 7 (30%) 13 (45%) Mean S.D Relevant health condition* Diabetics 0 (0%) 2 (7%) On lipid agents/meds 1 (4%) 1 (3%) Food frequency intake -Carotene ( g/day) Retinol ( g/day) Vitamin A ( g RE/day) a P 0.05 for all comparisons. * As reported in participant chart. Table 2 New Haven CARET lipid data a,b Point in study Placebo Active N Mean S.D. N Mean S.D. Cholesterol (mg/dl) Baseline On intervention Off intervention Triglycerides (mg/dl) Baseline On intervention Off intervention HDL (mg/dl) Baseline On intervention Off intervention LDL (mg/dl) Baseline On intervention Off intervention a Baseline, before intervention; On intervention, blood drawn within 1 week (mean 2.5 days) of stopping intervention; Off intervention, blood drawn 10 months off the intervention. b P 0.05 for all comparisons between active and placebo groups at each point in study. differences in age, duration in study, smoking status, baseline dietary intake, weight, or other baseline parameters Serum nutrient le els at the end of the inter ention As expected serum -carotene levels were markedly increased in participants on the active intervention, from a mean of 0.13 mol/l to 3.75 mol/l (P 0.001). Serum retinol levels were similar in both the placebo and active group (2.20 mol/l), consistent with the known tight regulation of serum retinol levels, with little change in serum levels despite vitamin A supplementation [21]. These results are consistent with prior serum nutrient data on the larger CARET population [20,21], and confirm participant compliance with the intervention. -Carotene and retinol levels were stable with storage, consistent with CARET quality control methods and prior CARET results [20,21] Serum triglyceride and cholesterol le els Serum triglyceride and total cholesterol levels at three time points are shown in Table 2: (1) baseline (before the intervention), (2) on the intervention (drawn within 1 week of stopping the intervention), and (3) 10 months off the intervention. One participant in the active group who did not return for the post-intervention visit is included in the analysis. Two outliers in the placebo

4 430 C.A. Redlich et al. / Atherosclerosis 143 (1999) group (levels 3 S.D. from the mean and highly variable between time points) were eliminated from further analysis, leaving 21 participants in the placebo group. The two outliers were not diabetic or on lipid lowering medication. There were no significant differences in serum triglyceride and total cholesterol levels between the active and placebo groups at baseline. There was a slight but nonsignificant increase in mean serum triglyceride levels (mean 32 mg/dl), in participants on the active intervention compared to baseline levels ( vs mg/dl). As seen in Table 2, after cessation of intervention, serum triglyceride levels decreased a mean of 86 mg/dl ( mg/dl vs ). This decrease was seen 6 months after cessation of therapy, and remained decreased at 10 months off the intervention (Fig. 1). No comparable changes were seen in the placebo group. Total cholesterol levels remained unchanged in both active and placebo groups at the three time points. Reanalysis of the data omitting all non-fasting samples (32% of all samples) did not alter the findings. Fig. 1. Serum triglyceride and total cholesterol levels (mg/dl) in the participants on placebo and active treatment over time starting at baseline, before the intervention. Cholesterol values are given as mean S.D. for each time point. Triglyceride values are given as geometric mean S.D.

5 C.A. Redlich et al. / Atherosclerosis 143 (1999) Fig. 2. Serum HDL and LDL cholesterol levels (mg/dl) in the participants on placebo and active treatment over time starting at baseline, before the intervention. Values shown are mean S.D. for each time point. Triglyceride and cholesterol levels obtained on active and placebo participants at the various time points from pre-intervention (baseline), during the intervention, at the stop of the intervention, and up to 10 months off the intervention are shown longitudinally in Fig. 1. These additional longitudinal time points showed similar changes in serum triglyceride and cholesterol levels on and off the active intervention Serum HDL and LDL cholesterol le els Serum HDL and LDL cholesterol levels were also determined at the same three time points: (1) baseline (before the intervention), (2) on the intervention (drawn within 1 week of stopping the intervention), and (3) 10 months off the intervention (Table 2, Fig. 2). One participant s HDL level was not obtained at the stop of

6 432 C.A. Redlich et al. / Atherosclerosis 143 (1999) the intervention, and LDL cholesterol levels could not be calculated on six out of a total of 63 placebo blood samples and 17 out of a total of 86 active blood samples due to elevated serum triglycerides ( 400 mg/dl). There were no significant differences between the active and placebo groups at baseline, although HDL levels were slightly lower in the active participants compared with placebo ( mg/dl vs ; P=0.17). No significant changes were seen in HDL or LDL cholesterol over the three time points. HDL and LDL levels obtained on the active and placebo participants at the various time points from pre-intervention to up to 10 months off the intervention are shown longitudinally in Fig. 2. No significant changes were seen at any of the time points. Analysis of the data using only subjects with data at all time points did not alter the results. Reanalysis of the data omitting all non-fasting subjects and the diabetic subjects did not significantly alter the results Diabetes and the use of lipid lowering medications At the time of randomization into the CARET study, two of the 52 participants were diabetic (both in the active intervention group), and two participants were on lipid lowering medications (one each in the placebo and active groups), as determined from review of the participant s medical and medication records. At the end of the active intervention, a mean of 5 years later, there were more participants with clinically recognized diabetes and more participants on lipid lowering medications in the active group (N=5 and 9 respectively) compared to placebo group (N=1 and 4 respectively). It is in this context noteworthy that although the differences did not reach significance, the subjects in the active group had slightly higher triglyceride and lower HDL cholesterol levels at baseline than the placebo group (Table 2). This lipid profile is suggestive of increased insulin resistance, which could contribute to a higher relative risk of developing diabetes on the active intervention (OR 2.58, CI ). In addition, there was a greater tendency to start lipid lowering medication in subjects on the active intervention (OR 2.09, CI ). Neither of these was statistically significant, although the latter approaches significance. Reanalysis of the data using only subjects with data at all time points and eliminating all subjects on lipid lowering or diabetic medications did not significantly alter the results. 4. Discussion This study was prompted by the unexpected, unexplained, and important finding of a significant increase in cardiovascular mortality in CARET participants on the active intervention. Several studies have consistently shown no significant effect of -carotene supplementation on serum total cholesterol, HDL cholesterol or triglyceride levels [21 26]. However, retinoids such as isotretinoin (13-cis retinoic acid) and etretinate, as well as high doses of vitamin A are known to increase triglycerides and cholesterol levels and lower HDL levels [6 11], changes associated with an increased risk of coronary heart disease [14]. Whether chronic supplementation with retinol might have similar effects on lipids, especially HDL and LDL cholesterol, has not been well investigated, and was the focus of this study. The similar biologic activity of retinol and retinoids is felt to be mediated through their common metabolite retinoic acid [27]. The findings presented here show that in a subgroup of CARET participants those on the active intervention had a mean nonsignificant increase in serum triglycerides of 32 mg/dl, comparable to the previously reported findings in the larger CARET population [15], and a decrease in serum triglycerides off the intervention (86 mg/dl). No significant changes in HDL or LDL cholesterol levels were seen comparing baseline, on intervention, and off intervention time points. Of interest is the finding that more participants on the active versus placebo intervention started lipid lowering medication (relative risk 2.09) and developed diabetes (relative risk 2.58). The increased use of lipid lowering medication in the active participants suggests that the intervention may cause hyperlipidemia, and that the increased use of lipid lowering agents could mask an effect of the intervention on lipid levels. An increased risk of developing diabetes in the active participants is also intriguing as diabetes is a known risk factor for cardiovascular disease. However, we cannot at present exclude the possibility that the slight and nonsignificant difference in baseline triglyceride and HDL cholesterol levels might contribute to this. Thus, even a slight change of lipid parameters in this group might prompt a more active lipid-lowering approach among the caregivers. In addition, as mentioned above, the baseline lipid profile in the active treatment group compatible with a larger extent of insulin resistance than in the placebo group. It is therefore possible that an increased degree of insulin resistance over the 5-year treatment period might result in a higher development of diabetes mellitus, and that this was not necessarily causally related to the treatment. Whether or not these findings are real is unclear, but they raise important issues which merit further investigation, especially considering the markedly increased cardiovascular mortality seen in CARET, and the widespread use of such vitamin supplements in other ongoing chemopreventive trials [28 30], for routine dermatologic conditions [4], and as over-the-counter supplements [5].

7 C.A. Redlich et al. / Atherosclerosis 143 (1999) Although the present study is small in scale, several strengths are notable. The 52 participants in the current study were remarkably similar to the much larger group (N=1043) of New Haven CARET subjects, with no significant differences noted in any of the baseline characteristics. Further, the placebo and active groups appear to be well matched with no significant differences in baseline parameters noted, and they were recruited before they or the investigators were informed of their intervention status, greatly reducing the risk of bias. Bloods were obtained at multiple time points over many years on the same participants, including at baseline, during the intervention, and after stopping the intervention, further increasing the strength of the findings. Given our sample size (N=52 subjects) our study should have been able to detect a 6.4 mg/dl or greater change in HDL, and 80 mg/dl or greater change in triglycerides. The study did not have sufficient power to detect smaller changes in HDL and triglycerides, which could be clinically important. It was of interest that HDL cholesterol levels in both active and placebo groups were low at baseline pre-intervention. Mean levels in the active group were only slightly higher than the HDL cholesterol cut-off level associated with increased cardiovascular risk (35 mg/dl) [14]. This suggests that at least some CARET subjects were at increased risk of ischemic heart disease. Although there was no significant treatment effect on HDL cholesterol levels, we cannot at present exclude the possibility that an additional small reduction in HDL levels, below the power of this study to detect, might contribute to the increased cardiovascular mortality found in the entire CARET active intervention group. The major concern with the current study is the small sample size. The unexpected and sudden cessation of the CARET study limited our ability to recruit subjects to participate in this study. However, as noted above, although a small subgroup, the 52 ancillary participants appear to be a representative sample of the much larger New Haven CARET population, and the active and placebo groups were well matched. In addition, it should be underscored that the present study represented a unique opportunity to prospectively evaluate lipid levels during cessation of intervention. In summary, these studies found no significant effect of the CARET intervention on total, HDL or LDL cholesterol levels. However, given several suggestive findings and the small sample size, further research is warranted on the effects of the CARET intervention on lipid levels, and on other potential explanations for the increased cardiovascular mortality seen in the active intervention group. The widespread use of vitamin A supplements and retinoids warrants further investigation into the effects of long-term use of retinoids and vitamin A on cardiovascular disease, diabetes and lipid levels. Acknowledgements This work was supported by: NIH KO8 HL03129 NIH NCI UOI CA 48200, NIH NCI ROI CA and the Yale Comprehensive Cancer Center. References [1] Omenn GS, Goodman GE, Thornquist MD, Balmes J, Cullen MR, Glass A, Keogh JP, Meyskens FL Jr, Valanis B, Williams JH Jr, Barnhard S, Hammar S. Effects of the combination of beta-carotene and vitamin A on lung cancer and cardiovascular disease. New Engl J Med 1996;334: [2] The Alpha-Tocopheral, Beta Carotene Cancer Prevention Study Group. The effects of vitamin E and beta carotene on the incidence of lung cancer and other cancers in male smokers. New Engl J Med 1994;330: [3] De Klerk NH, Musk AW, Ambrosini GL, Eccles JL, Hansen J, Olsen N, Watts VL, Lund HG, Pang SC, Beilby J, Hobbs MST. Vitamin A and cancer prevention II: Comparison of the effects of retinol and -carotene. Int J Cancer 1998;75: [4] Orfanos CE, Zouboulis CC, Almond-Roesler B, Geilen CC. Current use and future potential role of retinoids in dermatology. Drugs 1997;53: [5] Meyers DG, Maloley PA, Weeks D. Safety of antioxidant vitamins. Arch Intern Med 1996;156: [6] Marsden J. Hyperlipidaemia due to isotretinoin and etretinate: possible mechanisms and consequences. Br J Dermatol 1986;114: [7] Murray JC, Gilgor RS, Lazarus GS. Serum triglyceride elevation following high-dose vitamin A treatment for pityriasis rubra pilaris. Arch Dermatol 1983;119: [8] Zech LA, Gross EG, Peck GL, Brewer HB. Changes in plasma cholesterol and triglyceride levels after treatment with oral isotretinoin. A prospective study. Arch Dermatol 1983;119: [9] Bershad S, Rubinstein A, Paterniti JR, Le NA, Poliak SC, Heller B, Ginsberg HN, Fleishmajer R, Brown WV. Changes in plasma lipids and lipoproteins during isotretinoin therapy for acne. New Engl J Med 1985;313: [10] Vahlquist C, Michaëlsson G, Vahlquist A, Vessby B. A sequential comparison of etretinate (Tigason ) and isotretinoin (Roaccutane ) with special regard to their effects on serum lipoproteins. Br J Dermatol 1985;112: [11] Tangrea JA, Edwards BK, Taylor PR, Hartman AM, Peck GL, Salasche SJ, Menon PA, Benson PM, Mellette JR, Guill MA, et al. Long-term therapy with low-dose isotretinoin for prevention of basal cell carcinoma: A multicenter clinical trial. J Natl Cancer Inst 1992;84: [12] Pastorino U, Chiesa G, Infante M, Soresi E, Clerici M, Valente M, Belloni PA, Ravasi G. Safety of high-dose vitamin A. Oncology 1991;48: [13] Infante M, Pastorino U, Chiesa G, Bera E, Pisani P, Valente M, Ravasi G. Laboratory evaluation during high-dose vitamin A administration: a randomized study of lung cancer patients after surgical resection. J Cancer Res Clin Oncol 1991;117: [14] NIH Consensus Conference. Triglyceride, high-density lipoprotein, and coronary heart disease. J Am Med Assoc 1993;263: [15] Omenn GS, Goodman GE, Thornquist M, Brunzell JD. Longterm Vitamin A does not produce clinically significant hypertriglyceridemia: Results from CARET the -Carotene and retinol efficacy trial. Cancer Epidemiol Biomarker Prev 1994;3:711 3.

8 434 C.A. Redlich et al. / Atherosclerosis 143 (1999) [16] Lopes-Virella MF, Stone P, Ellis S, Colwell JA. Cholesterol determination in high-density lipoproteins separated by three different methods. Clin Chem 1977;23: [17] Friedewald WT, Levy RI, Frederickson DS. Estimation of the concentration of low-density lipoprotein cholesterol in plasma without use of the preparative ultracentrifuge. Clin Chem 1972;18: [18] Redlich CA, Grauer JN, van Bennekum AM, Clever SL, Ponn RB, Blaner WS. Characterization of carotenoid, vitamin A, and alpha-tocopherol levels in human lung tissue and pulmonary macrophages. Am J Respir Crit Care Med 1996;154: [19] Mills JL, Tuomilehto J, Yu KF, Colman N, Blaner WS, Koskela P, Rundle MS, Forman M, Toivanen L, Rhoades GG. Maternal vitamin levels during pregnancies producing infants with neural tube defects. J Pediatr 1992;120: [20] Goodman GE, Metch BJ, Omenn GS. The effect of long-term beta-carotene and vitamin A administration on serum concentrations of alpha-tocopherol. Cancer Epidemiol Biomarker Prev 1994;3: [21] Goodman GE, Thornquist M, Kestin M, Anderson G, Omenn GS, and CARET Coinvestigators. The association between participant characteristics and serum concentrations of betacarotene, retinol, retinyl palmitate, and alpha-tocopherol among participants in the carotene and retinol efficacy trial (CARET) for prevention of lung cancer. Cancer Epidemiol Biomarker Prev 1996;5: [22] Nierenberg DW, Bayrd GT, Stukel TA. Lack of effect of chronic administration of oral -carotene on serum cholesterol and triglyceride concentrations. Am J Clin Nutr 1991;53: [23] Ribaya-Mercado JD, Ordovas JM, Russell RM. Effect of betacarotene supplementation of the concentrations and distribution of carotenoids, vitamin E, vitamin A, and cholesterol in plasma lipoprotein and non-lipoprotein fractions in healthy older women. J Am Coll Nutr 1995;14: [24] Hughes GS Jr, Ringer TV, Francom SF, Means LK, DeLoof MJ. Lack of effects of beta-carotene on lipids and sex steroid hormones in hyperlipidemics. Am J Med Sci 1994;308: [25] Van Poppel G, Hospers J, Buytenhek R, Princen HM. No effects of -carotene supplementation on plasma lipoproteins in healthy smokers. Am J Clin Nutr 1994;60: [26] Ringer TV, DeLoof MJ, Winterrowd GE, Francom SF, Gaylor SK, Ryan JA, Sanders ME, Hughes GS. Beta-carotene s effects on serum lipoproteins and immunologic indices in humans. Am J Clin Nutr 1991;53: [27] Blaner WS, Olson JA. Retinol and retinoic acid metabolism. In: Sporn MG, Roberts AB, Goodman DS, editors. The Retinoids: Biology, Chemistry and Medicine, 2nd edn. New York: Raven Press, 1994: [28] Sankaranarayanan R, Mathew B. Retinoids as cancer preventive agents. IARC Sci Publ 1996;139: [29] Veronesi U, De Palo G, Costa A, Formelli F, Decensi A. Chemoprevention of breast cancer with fenretinide. IARC Sci Publ 1996;136: [30] De Vries N, Pastorino U, van Zandwijk N. Chemoprevention of second primary tumours in head and neck cancer in Europe: EUROSCAN. Eur J Cancer 1994;30B: