Safety of <7500 RE (<25000 IU) vitamin A daily in adults with retinitis pigmentosa

ABSTRACT

Background: Vitamin A supplementation is being used successfully to treat some forms of cancer and the degenerative eye disease retinitis pigmentosa. The daily biological need for vitamin A is estimated to be 800 retinol equivalents (RE)/d (2667 IU/d) for adult women and 1000 RE/d (3300 IU/d) for adult men; doses ≥7500 RE (≥25000 IU)/d are considered potentially toxic over the long term.

Objective: We assessed the safety in adults of long-term vitamin A supplementation with doses above the daily biological need but <7500 RE (<25000 IU)/d.

Design: Adults aged 18–54 y with retinitis pigmentosa but in generally good health (n = 146) were supplemented with 4500 RE (15000 IU) vitamin A/d for ≤12 y (group A) and compared with a similar group (n = 149) that received 23 RE (75 IU)/d (trace group). Mean total consumption of vitamin A in group A was 5583 RE (18609 IU)/d (range: 4911–7296 RE/d, or 16369–24318 IU/d) and that in the trace group was 1053 RE (3511 IU)/d (range: 401–3192 RE/d, or 1338–10638 IU/d).

Results: Patients in group A showed an 8% increase in mean serum retinol concentration at 5 y and an 18% increase at 12 y (P < 0.001); no retinol value exceeded the upper normal limit (3.49 μmol/L, or 100 μg/dL). Mean serum retinyl esters were elevated ≈1.7-fold at 5 y and remained relatively stable thereafter. No clinical symptoms or signs of liver toxicity attributable to vitamin A excess were detected.

Conclusions: Prolonged daily consumption of <7500 RE (<25000 IU) vitamin A/d can be considered safe in this age group.

Vitamin A, supplementation, triacylglycerols, liver enzymes, retinal degeneration, retinitis pigmentosa, retinol, adults

INTRODUCTION

Vitamin A has been used successfully as a therapeutic and chemopreventive agent for some types of cancer (1–3) and for the degenerative eye disease retinitis pigmentosa (4). Vitamin A supplementation is common in the United States (5, 6) and many persons consume amounts that exceed the estimated daily biological need of 800 retinol equivalents (RE) (2667 IU retinol) for women and 1000 RE (3300 IU retinol) for men (7). Because vitamin A is fat soluble and stored in the liver, deficiencies of this vitamin are seldom encountered in the industrialized world (8–12). Because excessive intake is potentially toxic to the liver and other tissues (5, 6, 13, 14), vitamin A supplementation is of concern to public health officials as well as to physicians who prescribe this vitamin.

Vitamin A exists in both the provitamin A form as β-carotene in plants and as preformed vitamin A, primarily as retinyl esters, in certain animal products (eg, meats, milk, and eggs). After ingestion and absorption, vitamin A is packaged by enterocytes as retinyl esters into chylomicron particles that enter the plasma through the lymphatic circulation. These particles are rapidly converted to chylomicron remnants that are taken up by the liver, where the retinyl esters are stored mostly in stellate (Ito) cells. Subsequently, retinyl esters are converted to retinol that is released into the plasma bound to retinol binding protein for use by tissues, including the retina (15–18). Because the liver is the major storage site and has an enormous capacity to store vitamin A, doses that moderately exceed the normal requirement for health, but that are not large enough to induce acute toxicity, ordinarily are not cause for concern. Long-term consumption of quantities that exceed biological need, however, could saturate the liver and result in toxicity.

Vitamin A hepatotoxicity, typically associated with elevated concentrations of liver enzymes such as aspartate transaminase (AST) and alanine transaminase (ALT) (19), has been documented in persons chronically supplemented with doses of vitamin A ≥7500 RE (≥25000 IU)/d (20–22). Only limited evidence is available to suggest that long-term supplementation with vitamin A (≤7500 RE/d, or ≤25000 IU/d) may lead to hepatotoxicity in certain individuals (20, 23, 24). For example, Krasinski et al (25) suggested that elevated liver enzymes were more prevalent in elderly patients with elevated plasma retinyl esters resulting from long-term supplementation with 151–7500 RE (5000–25000 IU) vitamin A/d and that in that group plasma retinyl esters ≥380 nmol/L might indicate potential hepatotoxicity. Ellis et al (26) suggested that individuals with severe hypertriacylglycerolemia (>4.5 mmol/L, or >400 mg/dL) are at greater risk of developing vitamin A toxicity.

Because no systematic documentation exists on supervised, long-term supplementation with moderate amounts (4500–7500 RE/d, or 15000–25000 IU/d) of vitamin A, the present study was initiated to address outstanding questions concerning vitamin A supplementation for up to 12 y in adults with retinitis pigmentosa in generally good health. Long-term supplementation with vitamin A [4500 RE (15000 IU)/d, plus the usual dietary load] was assessed for its effect on fasting serum concentrations of retinol, retinyl esters, and triacylglycerols, as well as on the relation between serum retinyl esters and liver enzymes that signal the possibility of hepatic toxicity. We also analyzed the prevalence of symptoms potentially related to vitamin A excess.

SUBJECTS AND METHODS

Subjects and experimental design

The patients in this study represent a subset of 601 screened persons (aged 18–49 y; weighing greater than the lower fifth percentile for a given age, sex, and height; not pregnant; and with no systemic disease affecting vitamin A absorption or metabolism) participating in a 4–6-y randomized, double-masked treatment trial for retinitis pigmentosa at the Berman-Gund Laboratory, Harvard Medical School, Massachusetts Eye and Ear Infirmary, Boston (4). This work was done with the approval of the appropriate institutional review boards. The original study design encompassed 4 treatment groups receiving different supplements of vitamins A and E. The patients in group A (n = 146) were supplemented with vitamin A [4500 RE (15000 IU)/d in the form of retinyl palmitate] plus a trace amount of vitamin E [2.2 α-tocopherol equivalents (TE) (3 IU)/d as all-rac-α-tocopherol]. The patients in the trace group (n = 149; placebo control patients) received 23 RE (75 IU) vitamin A/d and 2.2 TE (3 IU) vitamin E/d. The other 2 groups were group A+E [n = 151; 4500 RE (15000 IU) vitamin A/d and 296 TE (400 IU) vitamin E/d] and group E [n = 155; 23 RE (75 IU) vitamin A/d and 296 TE (400 IU) vitamin E/d]. Using a food-frequency questionnaire, we estimated that the average daily contribution of preformed vitamin A from sources other than the study supplement in these patients was 1083 RE (3609 IU)/d in group A and 1053 RE (3511 IU)/d in the trace group. The mean total vitamin A intake (ie, from diet, multivitamins when taken, and study supplement) was 5583 RE (18609 IU)/d (range: 4911–7296 RE/d, or 16369–24318 IU/d) for group A and 1053 RE (3511 IU)/d (range: 401–3192 RE/d, or 1338–10638 IU/d) for the trace group. The mean alcohol intake in group A was 7.5 g/d (range: 0–73 g/d) and that in the trace group was 11.1 g/d (range: 0–83 g/d). Ninety-five percent of the patients in group A consumed <32 g/d and 95% of those in the trace group consumed <38 g/d (ie, about 2.5 drinks/d in both groups).

As part of an evaluation for safety, a series of blood analyses [including measurement of hemoglobin, hematocrit, white blood cell count, serum retinol, and serum triacylglycerols, and serum liver function tests (AST, ALT, and alkaline phosphatase)] was performed without knowledge of treatment group assignment on samples from patients who had fasted overnight. Separated serum samples were stored at −70°C before being analyzed. Routine urinalysis was also performed at each examination.

Initially, to determine the average longitudinal effect of supplementation with 4500 RE (15000 IU) vitamin A/d on serum retinol, serum triacylglycerols, and serum liver function, we evaluated sera at year 0 (pretreatment), year 3, and year 5 among those patients in group A (n = 115) and in the trace group (n = 106) for whom serum samples were available at these 3 time points.

In addition, serum retinol, triacylglycerol, and liver enzyme (AST, ALT, and alkaline phosphatase) values were scrutinized individually across the entire data set to identify those patients from group A (n = 146) and the trace group (n = 149) who were within the normal range initially, but who exceeded the upper limit of the normal range at the last 2 consecutive follow-up visits. A symptom questionnaire was also administered annually to monitor possible vitamin A toxicity. The frequency of symptoms recurrent at the last 2 visits was recorded.

Subsequently, to assess the even longer-term effect of vitamin A intake (up to 12 y) on serum retinol and serum retinyl esters as well as the relation, if any, between serum retinol and liver enzymes, fasting serum was analyzed from a subset of 36 patients who returned 12 y after baseline (ie, year 0) to participate in a different treatment program. Serum retinol, triacylglycerol, AST, ALT, and alkaline phosphatase analyses were performed at years 0, 5, and 12 on serum samples from patients taking 4500 RE (15000 IU) vitamin A/d (ie, patients who were originally members of group A or group A+E). Serum retinyl ester values were available for analysis at years 0, 5, and 12 for 20 of the 36 patients and in 9 randomly selected patients from the trace group at years 0 and 5. [No patients in the trace group continued beyond 5 y because all participants in the trial were prescribed 4500 RE (15000 IU) vitamin A/d at that point and told to cease supplementation with vitamin E.]

With respect to compliance in the randomized trial, capsule counts indicated that 94% of the capsules were consumed in any given year; monthly calendar reports indicated that 88% of patients consumed >90% of their capsules. All patients who returned at 12 y reported taking vitamin A capsules daily or missing only occasionally.

Blood and urine analyses

A commercial laboratory using a standard autoanalyzer protocol provided a normal range of values for all blood indexes, with the exception of retinol and retinyl esters, and all urine indexes obtained from the general screening profile.

Serum vitamin A analyses

Serum retinol was measured by reversed-phase HPLC according to the method of Bieri et al (27). A standard of all-trans-retinyl acetate was obtained from Hoffmann-La Roche Inc (Parsippany, NJ). A total of 200 μL serum was added to 100 μL of a standard solution of retinyl acetate and 100 μL ethanol and then extracted with 600 μL hexane. A 400-μL portion of the hexane extract was evaporated under a stream of nitrogen gas, redissolved in 100 μL mobile phase, and injected onto a C₁₈, reversed-phase, 150 mm × 4.6 mm HPLC column (Supelcosil LC-18, 5-mm particle size; Supelco Inc, Bellefonte, PA). The mobile phase (methanol:water, 95:5) was delivered at a flow rate of 2.1 mL/min (model 110A pump; Beckman Instruments Inc, Berkeley, CA). Retinol (eluting time: ≈2 min) was detected at 290 nm in a Beckman ultraviolet detector. The retinol concentration was quantified to the peak area of the internal retinyl acetate standard.

The concentration of serum retinyl esters was measured by reversed-phase HPLC based on the method of Broich et al (28). A serum sample of 300 μL was mixed with 300 μL ethanol and extracted with 900 μL hexane. A portion of the hexane extract (660 μL) was dried under a light stream of nitrogen gas and redissolved in 110 μL mobile phase; then 100 μL was injected onto a C₁₈, reversed-phase, 150 mm × 4.6 mm HPLC column (Supelcosil LC-18, 5-mm particle size; Supelco Inc). The mobile phase (methanol:acetonitrile:chloroform, 47:47:6) was delivered at a flow rate of 2.1 mL/min (model 110A pump; Beckman Instruments Inc). The retinyl ester peak (retention time: ≈10 min for retinyl palmitate) was detected at 325 nm in an ultraviolet detector (Lambda-Max model 480 spectrophotometer; Waters Associates, Milford, MA). The sample concentration of retinyl esters was calculated by comparing the sample peak area with that of the standard. The crystalline retinyl palmitate standard was purchased from Sigma Chemical Co (St Louis).

Retinol values are reported in μmol/L (1 μmol/L = 0.0349 μg/dL, or 1 μg/dL = 29 μmol/L); retinyl esters are expressed in nmol/L (1 nmol/L = 19 μg/dL, or 1 μg/dL = 0.0526 nmol/L).Triacylglycerol values are reported in mmol/L (1 mmol = 89 mg/dL).

Statistical analyses

The descriptive results are presented as means ± SDs except for triacylglycerol values, for which a geometric mean rather than an arithmetic mean and no SD is shown. To assess change over time, a longitudinal regression model was used with the general linear model with intraclass correlation procedure, which is equivalent to a compound symmetry longitudinal regression model (29, 30). Using this approach, we estimated and compared rates of change over time by treatment group. In addition, we compared rates of change by subgroup within the vitamin A treatment group (eg, stratified by initial total retinol intake, serum triacylglycerol concentration, or alcohol intake). Simple correlation analyses were performed to assess the relation between retinol and log triacylglycerol by treatment group. Using repeated-measures analysis of variance and Fisher's least-significant-difference criteria for comparing specific time points, we compared year-5 and year-12 values, respectively, with year-0 values for patients followed for 12 y and taking 4500 RE (15000 IU) vitamin A/d. In addition, a similar analysis was performed over 2 time points (years 0 and 5) for patients taking 23 RE (75 IU) vitamin A/d.

RESULTS

Retinol

Fasting serum retinol concentrations over a 5-y period for patients in group A and the trace group are presented in Table 1. Patients in group A showed an ≈8% increase in serum retinol concentrations over 5 y of supplementation; the absolute change from a slope analysis over 5 y was significantly different from zero. In contrast, no significant change was detected in the trace group. A significant group-by-time interaction was observed by comparing the differences between the 2 slopes (P < 0.001). The distribution of individual serum retinol values in group A at year 0 (pretreatment) and after 3 and 5 y of supplementation is illustrated in Figure 1; the distribution of serum retinol shifted to higher concentration values at years 3 and 5.

TABLE 1

Longitudinal values of serum retinol, triacylglycerols, and liver enzymes in group A and the trace group¹

Criteria	Year 0	Year 3	Year 5	Change2
Retinol (μmol/L)
Group A	1.79 ± 0.373	1.95 ± 0.36	1.92 ± 0.33	0.144
Trace group	1.73 ± 0.33	1.70 ± 0.33	1.75 ± 0.32	0.01
Triacylglycerols (mmol/L)
Group A	1.00	1.30	1.17	19.05
Trace group	0.92	1.08	0.97	7.5
AST (IU/L)
Group A	20.13 ± 6.73	19.20 ± 6.88	18.84 ± 6.11	−1.31
Trace group	20.79 ± 6.40	19.73 ± 7.84	20.02 ± 10.53	−0.85
ALT (IU/L)
Group A	26.42 ± 12.98	23.23 ± 14.01	20.52 ± 13.99	−5.855
Trace group	27.02 ± 13.68	26.15 ± 20.39	21.12 ± 12.78	−5.545
Alkaline phosphatase (IU/L)
Group A	70.74 ± 18.65	71.50 ± 18.84	68.75 ± 18.36	−1.73
Trace group	68.66 ± 18.94	70.85 ± 23.93	64.17 ± 19.71	−3.85

1

n = 115 in group A and 106 in the trace group (number of patients for whom samples were available for all 3 time points). AST, aspartate transaminase; ALT, alanine transaminase. Normal values are as follows: retinol, <3.49 μmol/L; triacylglycerols, ≤2.13 mmol/L; AST, ≤39 IU/L; ALT, ≤54 IU/L; and alkaline phosphatase, <115 IU/L.

2

Estimated absolute change from slope analyses over 5 y adjusted for sex, except for triacylglycerols, for which change is expressed as a percentage. For each index other than serum retinol there was no significant difference between the slope for group A and the slope for the trace group.

3

x̄ ± SD.

4

P < 0.001 for comparison of slope over 5 y versus zero.

5

P < 0.05 for comparison of slope over 5 y versus zero.

FIGURE 1.

Distribution of serum retinol values in patients with retinitis pigmentosa in group A (n = 115) at year 0 and at year 3 and year 5 of supplementation with 4500 RE (15000 IU) vitamin A/d.

Among patients in group A, serum retinol increased comparably in the subgroups with total retinol intake <6000 RE (<20000 IU)/d (ie, 4911–5999 RE/d, or 16369–19999 IU/d) and ≥6000 RE (≥20000 IU)/d (ie, 6000–7296 RE/d, or 20000–24318 IU/d), in the subgroups with serum triacylglycerols ≤2.13 and >2.13 mmol/L, and in the subgroups with daily alcohol intake ≤15 and >15 g/d (Table 2). Throughout this study, no patient had a serum retinol value equal to or above the upper limit of normal, ie, 3.49 μmol/L (100 μg/dL). Among 36 patients taking vitamin A and followed for 12 y, the mean serum retinol concentration was 1.7 μmol/L (48 μg/dL) at year 0 (range: 1.0–2.9 μmol/L, or 30–83 μg/dL), 1.8 μmol/L (51 μg/dL) at 5 y (range: 1.1–2.3 μmol/L, or 33–67 μg/dL), and 2.0 μmol/L (57 μg/dL) at 12 y (range: 1.0–3.1 μmol/L, or 28–89 μg/dL). Mean serum retinol was elevated by 8% above that at year 0 at 5 y (P = 0.002) and by 18% above that at year 0 at 12 y (P < 0.001).

TABLE 2

Subgroup analyses: longitudinal values of serum retinol by initial total retinol intake, serum triacylglycerol concentration, or daily alcohol intake¹

Criteria	Year 0	Year 3	Year 5	Change2
		mmol/L
Retinol intake (RE/d)3
<6000 (n = 84)4	0.53 ± 0.115	0.59 ± 0.10	0.58 ± 0.10	0.15
≥6000 (n = 31)	0.54 ± 0.11	0.58 ± 0.10	0.58 ± 0.11	0.11
Serum triacylglycerols (mmol/L)
≤2.13 (n = 104)	1.77 ± 0.35	1.93 ± 0.35	1.91 ± 0.33	0.14
>2.13 (n = 11)	1.96 ± 0.46	2.20 ± 0.34	2.05 ± 0.35	0.12
Alcohol intake (g/d)
≤ 15 (n = 47)	1.71 ± 0.38	1.89 ± 0.37	1.87 ± 0.32	0.17
>15 (n = 68)	1.85 ± 0.34	2.00 ± 0.34	1.96 ± 0.33	0.12

1

Normal values are as follows: retinol, <3.49 μmol/L; triacylglycerols, ≤2.13 mmol/L.

2

Estimated absolute change in serum retinol values from slope analyses over 5 y adjusted for sex; change was not significantly different between each pair of subgroups.

3

To convert retinol equivalents (RE)/d to IU/d, multiply by 3.333.

4

n value is the number of patients for whom samples were available for all 3 time points.

5

x̄ ± SD.

Triacylglycerols

The mean fasting serum triacylglycerol concentration rose 19% over 5 y in group A and 7.5% in the trace group (Table 1). Among patients in group A, serum triacylglycerol concentrations rose comparably among patients with total retinol intake <6000 RE (<20000 IU)/d and ≥6000 RE (≥20000 IU)/d and with daily alcohol intake ≤15 and >15 g/d. Serum triacylglycerol concentrations, however, rose significantly more in those patients with initial triacylglycerol concentrations ≤2.13 mmol/L (P = 0.01). At year 5 a positive correlation was observed between serum retinol and log_e triacylglycerol values in both group A and the trace group (Figure 2). Also, significant correlations were observed between triacylglycerols and retinyl esters (r = 0.40, P < 0.02) as well as between retinyl esters and retinol (r = 0.41, P < 0.02) (data not shown).

FIGURE 2.

Correlation between serum retinol and log_e triacylglycerols in patients with retinitis pigmentosa from group A (left: r = 0.48, P < 0.001) and the trace group (right: r = 0.29, P < 0.01) for samples collected at 5 y of follow-up. There was no significant difference between groups (P = 0.10).

Among 36 patients taking vitamin A and followed for 12 y, mean serum triacylglycerol was 0.84 mmol/L (75 mg/dL) at year 0 (range: 0.33–2.13 mmol/L, or 29–190 mg/dL), 1.08 mmol/L (96 mg/dL) at year 5 (range: 0.42–3.45 mmol/L, or 37–307 mg/dL), and 0.93 mmol/L (83 mg/dL) at year 12 (range 0.40–3.09 mmol/L, or 36–275 mg/dL). Mean triacylglycerol was elevated 27% above year 0 at year 5 (P = 0.001) and 10% above year 0 at year 12 (NS).

In the entire data set, individual cases of initially normal triacylglycerol concentrations that subsequently exceeded the upper norm of 2.13 mmol/L (190 mg/dL) during the last 2 consecutive follow-up visits were more prevalent among supplemented patients in group A than in the trace group (Table 3). In Group A, individual elevated triacylglycerol values at the last 2 consecutive follow-up visits ranged from 2.15 to 4.71 mmol/L (191–419 mg/dL) among those with initially normal serum triacylglycerol concentrations.

TABLE 3

Percentage of patients in group A and the trace group with a consistent, above-normal serum chemistry profile at their last 2 consecutive follow-up visits ≈1 y apart¹

Criteria	Consistency
	%
Retinol (μmol/L)
Group A (n = 146)2	0 [0]3
Trace group (n = 149)	0 [0]
Triacylglycerols (mmol/L)
Group A (n = 133)	11 [15]4
Trace group (n = 138)	1 [2]
AST (IU/L)
Group A (n = 144)	0 [0]
Trace group (n = 149)	0 [0]
ALT (IU/L)
Group A (n = 140)	0 [0]
Trace group (n = 142)	1 [1]
Alkaline phosphatase (IU/L)
Group A (n = 141)	1 [1]
Trace group (n = 145)	0 [0]

1

n = 146 in group A and 149 in group trace at the start of the study. AST, aspartate transaminase; ALT, alanine transaminase. Normal values are as follows: retinol, <3.49 μmol/L; triacylglycerols, ≤2.13 mmol/L; AST, ≤39 IU/L; ALT, ≤54 IU/L; alkaline phosphatase, <115 IU/L.

2

n value is the number of subjects with normal values at year 0.

3

n (number of subjects with elevated values at their last 2 consecutive visits) in brackets.

4

Significantly different from the trace group, chi-square = 9.52, P < 0.01 (chi-square test).

Liver enzymes

The mean activity of AST, ALT, and alkaline phosphatase showed no significant increase in either group throughout the study (Table 1). As shown in Table 3, of patients with normal liver enzyme activity at year 0 (ie, AST 7–39 IU/L, ALT 2–54 IU/L, and alkaline phosphatase 30–115 IU/L), one patient in group A and one patient in the trace group had elevated values for any of these tests at the last 2 follow-up evaluations.

None of the patients in either group with initial AST or ALT values slightly above normal had elevated values at the last 2 consecutive visits. Of the 9 patients who initially had slightly elevated alkaline phosphatase activity at year 0, 1 patient had a slightly elevated value at the end of 5 y.

In 36 patients taking vitamin A and followed for 12 y, mean AST activity was 21 IU/L at year 0 (range: 13–40 IU/L), 19 IU/L at year 5 (range: 10–33 IU/L), and 22 IU/L at year 12 (range: 14–38 IU/L). Mean ALT activity was 25 IU/L at year 0 (range: 12–54 IU/L), 17 IU/L at year 5 (range: 6–36 IU/L), and 21 IU/L at year 12 (range: 8–50 IU/L). Mean alkaline phosphatase activity was 69 IU/L at year 0 (range: 34–122 IU/L), 64 IU/L at year 5 (range: 36–111 IU/L), and 59 IU/L at year 12 (range: 36–99 IU/L). Mean AST, ALT, and alkaline phosphatase values were not significantly elevated above year 0 at 12 y. No significant relation was detected between AST, ALT, and alkaline phosphatase and serum retinyl esters at year 12.

Retinyl esters

Serum retinyl esters rose significantly over 12 y of supplementation with 4500 RE (15000 IU) vitamin A/d (Figure 3). The 20 patients taking vitamin A and followed for 12 y had a mean baseline retinyl ester value of 94 nmol/L (range: 36–207 nmol/L). Retinyl esters were significantly elevated at 5 y (mean: 161 nmol/L; range: 70–300 nmol/L; P = 0.0001) and at 12 y (mean: 137 nmol/L; range: 65–420 nmol/L; P = 0.002). The one patient whose retinyl ester concentration exceeded 380 nmol/L at 12 y had normal AST, ALT, and alkaline phosphatase activity. The mean serum retinol concentrations for this group of 20 patients were 1.61 μmol/L at year 0, 1.72 μmol/L at year 5, and 1.96 μmol/L at year 12. Compared with concentrations in patients receiving 23 RE (75 IU) vitamin A/d, retinyl ester concentrations in patients receiving 4500 RE (15000 IU) vitamin A/d increased ≈1.7-fold. The average circulating serum retinyl esters constituted 5.6% of the total serum vitamin A pool (retinol plus retinyl esters) at year 0, 8.5% at year 5, and 6.5% at year 12. Those patients taking 23 RE (75 IU) vitamin A/d revealed no significant change from year 0 to year 5.

FIGURE 3.

Mean (±SD) longitudinal profile for serum retinyl esters from a subset of patients with retinitis pigmentosa supplemented with 4500 retinol equivalents (RE) (15000 IU) vitamin A/d at 0, 5, and 12 y (n = 20) or with 23 RE (75 IU) vitamin A/d (n = 9) at 0 and 5 y [at 5 y all patients were prescribed 4500 RE (15000 IU) vitamin A/d]. Means at years 5 and 12 for those consuming 4500 RE (15000 IU) vitamin A/d were significantly different from the mean at year 0 (pretreatment) (P < 0.05, repeated-measures ANOVA) but were not significantly different from each other.

Other evaluations

In the entire data set, the frequency of potential vitamin A–induced symptoms reported by patients in group A at the last 2 follow-up visits was not significantly different from that reported by patients in the trace group (Table 4). In addition, the clinical evaluations based on serum samples (eg, hemoglobin, hematocrit, white blood cell count, mean corpuscular volume, calcium, phosphorus, glucose, serum urea nitrogen, creatinine, uric acid, cholesterol, total protein, albumin, globulin, total bilirubin, lactic dehydrogenase, potassium, sodium, chloride, and carbon dioxide) were all comparable. Similarly, evaluations based on urine (eg, white blood cell and red blood cell counts, protein, glucose, bile, and blood) did not reveal any significant differences between the 2 groups during the course of the study (data not shown).

TABLE 4

Prevalence of symptoms relevant to vitamin A toxicity reported by patients in group A and the trace group during the last 2 consecutive follow-up visits ≈1 y apart¹

Symptoms	Prevalence
	%
Joint or bone pain
Group A	1 [2]2
Trace group	2 [3]
Headaches
Group A	0 [0]
Trace group	1 [2]
Increased hair loss
Group A	0 [0]
Trace group	2 [3]
Erythema or scaly skin
Group A	0 [0]
Trace group	0 [0]
Unusually dry skin
Group A	0 [0]
Trace group	1 [1]
Desquamation, exfoliation, or eruption
Group A	0 [0]
Trace group	1 [1]
Red rash
Group A	0 [0]
Trace group	0 [0]
Unusually brittle or soft nails
Group A	1 [1]
Trace group	1 [1]
Lip fissures or chapping
Group A	0 [0]
Trace group	1 [2]
Nausea
Group A	0 [0]
Trace group	0 [0]
Diarrhea
Group A	0 [0]
Trace group	0 [0]

1

n = 146 in group A and 149 in the trace group except for diarrhea, for which n = 138 in group A and 140 in the trace group. None of these subjects reported these symptoms at year 0.

2

n (number of subjects who reported symptoms at year 0 and at their last 2 consecutive visits) in brackets.

DISCUSSION

Vitamin A has been reported to have a beneficial effect in the treatment and prevention of some types of cancer, such as lung and breast cancers (2, 3), and on the course of the degenerative eye disease known as retinitis pigmentosa (4). Because long-term consumption of this vitamin [generally >15000 RE (>50000 IU)/d] may lead to hypervitaminosis A, with a broad spectrum of clinical abnormalities (31, 32), systematic documentation of the response to long-term treatment with optimal dosages is desirable. This report evaluated the carefully documented consumption of vitamin A in persons administered 4500 RE (15000 IU) vitamin A/d in addition to their usual dietary load for up to 12 y, an amount and time not evaluated previously. The present analysis of fasting serum retinol, retinyl esters, liver enzymes, and triacylglycerols in this longitudinal study provides evidence that oral supplementation with 4500 RE (15000 IU) vitamin A/d causes no toxic manifestation suggestive of chronic hypervitaminosis A in young and middle-aged adults with retinitis pigmentosa in generally good health.

Previous studies presented an inconsistent picture of the serum retinol response, which was either elevated or unaffected depending in part on the level of vitamin A supplementation (25, 33–35). Nonetheless, highly elevated serum retinol in excess of 3.49 μmol/L (100 μg/dL) usually is a diagnostic indicator of excess vitamin A intake and hypervitaminosis A (6, 8). Some of this abnormal concentration of retinol circulates nonspecifically in association with plasma lipoproteins, as opposed to normal transport bound to retinol binding protein, and thus has a potentially destructive effect on cellular membranes leading to cell and tissue damage (22, 36, 37). During the course of the present study, no significant changes were noted in the serum retinol concentrations of unsupplemented patients in the trace group, whereas significant increases of ≈8% and 18% were observed at years 5 and 12, respectively, among those supplemented with 4500 RE (15000 IU) vitamin A/d. The bulk of individual serum retinol values in group A remained within the initial range of 1.41–2.45 μmol/L (40–70 μg/dL), with no individual serum retinol concentration reaching ≥3.49 μmol/L (100 μg/dL).

Vitamin A supplementation also resulted in an increase in mean serum retinyl esters to ≈150 nmol/L (or 1.7-fold) within 5 y; these concentrations did not increase significantly more over 12 y of supplementation. This elevation in serum retinyl esters, which constituted 8.5% of the total vitamin A pool (retinol plus retinyl esters) at year 5 and 6.5% at year 12 compared with 5.6% in the trace group at year 5, was lower than the generally accepted concentration of retinyl esters indicative of chronic hypervitaminosis A (>10% of the total plasma vitamin A pool) (16). By way of clarification, note that the percentage of retinyl esters in the total plasma vitamin A pool represents the molar ratio of plasma retinyl esters to the sum of plasma retinol and retinyl esters (ie, the total plasma vitamin A pool). In previous reports, persons who were diagnosed with chronic hypervitaminosis A displayed highly elevated retinyl esters that contributed ≈40–70% of the total plasma vitamin A pool (37).

In the study by Krasinski et al (25), plasma concentrations of retinyl esters ≥380 nmol/L after vitamin A supplementation, recorded in 5 of 77 elderly and 2 of 77 young adults, were associated with elevated liver enzymes: AST in 4 and ALT in 2 of the elderly patients (60–98 y old). Such elevations in liver enzymes may indicate hepatic toxicity caused by supplementation. In the study by Stauber et al (35), retinyl esters were significantly correlated with AST in elderly females (≥60 y old) but not in elderly males. Our younger adult population revealed no relation between biochemical markers of liver toxicity, such as elevated AST, ALT, and alkaline phosphatase, and elevated serum retinyl esters. Furthermore, the average concentration of serum liver enzymes remained in the normal range throughout the study. Whether a high concentration (eg, ≥380 nmol/L, or 200 μg/dL) of circulating retinyl esters directly initiates the pathologic changes in liver morphology and function described in chronic hypervitaminosis A is still unclear and requires further investigation. This seems unlikely, though, because carnivorous species such as cats and dogs normally transport most of their vitamin A as retinyl esters (38).

Although hypertriacylglycerolemia secondary to high doses [≥7500 RE (≥25000 IU)/d] of vitamin A has been reported by several investigators (23, 34, 39, 40), the effect of sustained supplementation with a moderate dose [<7500 RE (<25000 IU)/d] has not been described in previous human studies. A significant difference in rate of change could be detected over 5 y between a subset of 104 patients consuming 4500 RE (15000 IU) vitamin A/d with initial triacylglycerol concentrations ≤2.13 mmol/L and in 11 patients with initial concentrations >2.13 mmol/L. Although a subset of patients had values outside the normal range at 5 y, the mean value of group A remained within the normal range (0.45–2.13 mmol/L, or 40–190 mg/dL). Elevated triacylglycerols might be attributed to the competition for lipoprotein lipase between triacylglycerols, the preferred substrate, and retinyl esters, both of which are hydrolyzed by this enzyme (41). As reported by others (33), this possibility was supported by the direct correlation between fasting triacylglycerols and retinyl esters in our data. A correlation was also found between serum retinol and triacylglycerols. Furthermore, the correlation was higher in the supplemented group than in the unsupplemented one, suggesting that the relation between triacylglycerols and retinol was affected by supplementation.

No relation was found between alcohol intake and change in serum retinol concentrations in patients consuming 4500 RE (15000 IU) vitamin A/d. Note that most of these patients reported that they consumed <32 g/d, ie,<2.5 alcoholic beverages/d. These data are therefore consistent with the idea that intake of alcohol at this level can be tolerated by patients aged 18–54 y in generally good health and supplemented with 4500 RE (15000 IU) vitamin A/d.

No significant differences were observed between group A and the trace group in the prevalence of transiently elevated AST, ALT, or alkaline phosphatase. On the other hand, individual serum triacylglycerol values became elevated >2.13 mmol/L (190 mg/dL) and remained elevated more frequently among patients in group A than in the trace group, suggesting that hypertriacylglycerolemia might be one consequence of supplementation with 4500 RE (15000 IU) vitamin A/d.

The frequency of symptoms (for example, hair loss, persistent headaches, or joint pain) reported by patients on the symptom questionnaire was comparable in group A and the trace group and in normal and hypertriacylglycerolemic patients. None of the few symptomatic changes reported could be attributed to excessive consumption of vitamin A according to an examination by a consulting internist.

The status of serum retinol, retinyl esters, liver enzymes, and triacylglycerols presented in this study leads us to conclude that long-term supplementation with 4500 RE (15000 IU) vitamin A/d in healthy adults aged 18–54 y elicited no adverse side effects, including no evidence of hepatic toxicity. Thus, this duration and amount of supplementation can be considered safe in this age group.

FOOTNOTES

2

Supported by NEI grants U10-EY02014 and U10-EY11030 and by the Foundation Fighting Blindness, Baltimore.