In 1982, Krause and co-workers reported abnormally reduced plasma concentrations of biotin in 74% of adults undergoing long-term anticonvulsant therapy (1). Carbamazepine, phenytoin, or primidone (or some combination of the three) were the particular anticonvulsants associated with reduced plasma concentrations of biotin (1,2). The mechanism was not determined. In a second study, four of seven subjects had a pattern of organic aciduria consistent with deficiency of two biotin-dependent enzymes. However, we have shown that biotin accounts for less than half of the total of biotin plus biotin metabolites in plasma(3). That the bioassay used by Krause and co-workers to measure serum biotin also detects an inactive biotin metabolite (biotind- sulfoxide) complicates interpretation of plasma levels. Furthermore, in a study of experimental biotin deficiency in normal adult subjects (4), the plasma concentration of biotin was not a good indicator of biotin deficiency, although decreased urinary excretion of biotin and increased urinary excretion of 3-hydroxyisovaleric acid were. Because of these recent findings, we sought to determine whether biotin breakdown is accelerated in children undergoing long-term therapy with certain anticonvulsants, especially carbamazepine or phenytoin or a combination, and, if so, whether these children had any indications of reduced biotin status.
Thirteen children (three girls) who had developmental disabilities and who were receiving long-term anticonvulsant therapy for seizure disorders were recruited to participate in this study. The ages of the children, identity, and current duration of treatment with anticonvulsant drugs are listed inTable 1. All children were reported to be consuming a mixed general diet or a general pureed diet. None was reported to be receiving any biotin supplementation. Other than prospectively excluding subjects whose parents stated that the subject was receiving biotin supplements, no attempt was made to assess or control dietary intake of biotin. The 13 children receiving anticonvulsant therapy were subdivided into two groups: (a) seven children who were receiving either carbamazepine, phenytoin, or both, regardless of other anticonvulsant therapy; and (b) six children who were receiving phenobarbital monotherapy.
Sixteen normal, healthy children (nine girls) who were consuming a mixed general diet and who were receiving neither anticonvulsants nor biotin supplements served as the control group. The mean age for the control group(Table 1) was not significantly different from that of the combined anticonvulsant-treated group (unpaired Student'st- test).
Sample Collection and Assay
To facilitate subject recruitment and because we have found that urinary excretion rates normalized by creatinine serve as a useful index of biotin status (4), untimed urine samples were obtained at a regular clinic visit. Samples were frozen at -20 °C until analysis. Determinations of urinary biotin and metabolites were performed as described previously (5) with highperformance liquid chromatography separation of biotin and biotin metabolites by reversed-phase chromatography. Urinary concentrations of 3-hydroxyisovaleric acid were determined by gas chromatography/mass spectrometry according to the method of Mock and Mock(6) with unlabeled and deuterated authentic 3-hydroxyisovaleric acid as the external and internal standards. Creatinine was determined by the standard picric acid method of Jaffe(7).
For group data, central tendency and variability were expressed as mean± 1 SD. Comparison of the means of the data from the control group, the group of children treated with carbamazepine and/or phenytoin, and the group of children treated with phenobarbital was by one-way analysis of variance. When significant differences were identified, interaction was determined by Fisher's post hoc testing (8).
Informed consent was obtained according to procedures approved by the Committee on Research Involving Human Subjects at the University of Iowa. Initial contacts with anticonvulsant-treated children and parents were made during regular visits for follow-up medical care provided by the Division of Developmental Disabilities of the Department of Pediatrics at the University of Iowa Hospitals and Clinics. This group was sought out because of their long-term treatment with anticonvulsants. Developmental disabilities of this group were not a factor in the study.
As shown in Fig. 1A, the mean urinary excretion rate of bisnorbiotin for the carbamazepine/phenytoin group was significantly increased compared with that of the control group (p < 0.02 by analysis of variance); for three of the seven children, the excretion rates were greater than the upper limit of the controls. For the phenobarbital group, the mean urinary excretion rate of bisnorbiotin was also significantly increased compared with that of the control group (p < 0.0005); for five of the six children, the excretion rates of bisnorbiotin were greater than the upper limit of the controls.
The mean urinary excretion rate of biotin sulfoxide (Fig. 1B) was significantly greater than that of the control group only for the carbamazepine/phenytoin group (p < 0.009); for three of the seven in this group, the excretion rates were greater than the upper limit of the control subjects. For the phenobarbital group, the biotin sulfoxide excretion was not significantly different from that of the control group. Biotin sulfoxide excretion was greater than the upper limit of the control group for only one child.
The data in Fig. 1 provide strong evidence of increased formation of biotin catabolites. However, if the subjects were inadvertently receiving biotin supplementation, the metabolite excretion might be the consequence of greater biotin in some subjects (rather than anticonvulsant treatment). Our studies using oral and intravenous administration of large amounts of biotin suggest that the ratio of metabolites to biotin remains roughly constant (5). Thus, normalizing urinary metabolite data by urinary biotin provides a more specific index of accelerated biotin breakdown. Depicted are the bisnorbiotin to biotin ratio(Fig. 1C) and the biotin sulfoxide to biotin ratio(Fig. 1D). Bisnorbiotin excretion remained significantly greater for both the carbamazepine/phenytoin group (p < 0.002) and for the phenobarbital group (0.04). The urinary biotin sulfoxide:biotin ratio was significantly greater for the carbamazepine/phenytoin group(p < 0.0005) but not for the phenobarbital group. Thus, normalizing by biotin did not change the interpretation of biotin sulfoxide data.
As shown in Fig. 2A, the mean urinary excretion rate of 3-hydroxyisovaleric acid for the carbamazepine/phenytoin group was significantly greater compared with the control group (p < 0.0001 by analysis of variance). For five of seven children, the 3-hydroxyisovaleric acid excretion rates were greater than the upper limit of the control group. For the phenobarbital group, the 3-hydroxyisovaleric acid excretion rate was not different from the control group.
Although an increased 3-hydroxyisovaleric acid excretion rate suggests biotin deficiency, the urinary excretion of biotin was not significantly decreased in the carbamazepine/phenytoin group. Urinary biotin excretion was less than the lower limit of the controls for only one child; this child was in the carbamazepine/phenytoin group. However, the number of abnormal values in the phenobarbital group was only one of six (rather than five of six), suggesting that part of the increased bisnorbiotin excretion was indeed attributable to increased (inadvertent) biotin intake rather than induction of the bisnorbiotin pathway by phenobarbital.
One of the children treated with carbamazepine/phenytoin who had an abnormally high 3-hydroxyisovaleric acid excretion rate was studied again 1 year later. The patient continued with carbamazepine therapy at the same dose; in this interval, he had not received biotin supplements by history. During the interval, the urinary excretion of 3-hydroxyisovaleric acid increased from 38 μmol/mol creatinine (340 nmol/mg creatinine) to 52 μmol/mol creatinine (463 nmol/mg creatinine). We instituted treatment with 4.1 μmol(1,000 μg) of biotin orally each day. Figure 3 depicts the effect on urinary excretion of 3-hydroxyisovaleric acid. After 10 days of biotin supplementation, 3-hydroxyisovaleric acid excretion had decreased to below the upper limit of the normal range, 23 μmol/mol creatinine (203 nmol/mg creatinine). However, the excretion of 3-hydroxyisovaleric acid did not remain within the normal range, rising to abnormal values after 20 days. This relapse of abnormal organic aciduria was not attributable simply to failure to take biotin. Compliance was confirmed by measuring total avidin-binding substances in each urine sample; values remained 10-fold greater than the upper limit of normal (3-13 μmol/mol creatinine).
Findings of Biotin Deficiency
Neither alopecia nor the characteristic periorificial, scaly dermatitis[the cutaneous signs of overt biotin deficiency (9)] were observed in any of the subjects. Because of the developmental delay and seizure disorders present in these children, the CNS dysfunction caused by biotin deficiency (anorexia, malaise, depression, and hallucinations in adults, and a peculiar withdrawn behavior and developmental delay in children) were not assessed for relevance to biotin status.
The data from this study provide strong evidence that the urinary excretion of bisnorbiotin and biotin sulfoxide was increased in a substantial percentage of children studied who were receiving carbamazepine, phenytoin, or a combination of these two drugs. To determine whether the increased excretion could simply be attributed to an inadvertent increased biotin intake, urinary excretion of each metabolite was normalized by urinary biotin. Rather than minimizing the abnormal metabolite excretion, abnormalities became even more statistically significant, and an additional child was identified as excreting increased amounts of bisnorbiotin.
Results from the phenobarbital group may be thought of as providing a control group in which the children also had developmental delays and seizure problems but received an anticonvulsant that produced only modest acceleration of biotin biotransformation. Thus, for the phenobarbital group, most of the increase of urinary excretion of bisnorbiotin might be principally the result of an inadvertent increase in biotin intake.
Several studies of biotin-deficient patients have indicated that increased urinary excretion of 3-hydroxyisovaleric acid can indicate reduced tissue activity of methylcrotonyl-CoA carboxylase. This has been the case whether the reduced enzyme activity resulted from isolated inborn methylcrotonyl-CoA carboxylase deficiency, methylcrotonyl-CoA carboxylase deficiency secondary to holocarboxylase synthetase deficiency, or biotin deficiency secondary to biotinidase deficiency. Moreover, increased 3-hydroxyisovaleric acid excretion has been observed in patients with biotin deficiency due to total parenteral nutrition without biotin. In both rats (6) and humans(4), experimental feeding of egg white and the resulting biotin deficiency has produced increased 3-hydroxyisovaleric acid excretion. In the eggwhite-fed rat, the expected decrease in hepatic methylcrotonyl-CoA carboxylase activity has been directly demonstrated (10). In human studies, the tissue deficits have been demonstrated in carboxylase activities of lymphocytes (11) and inferred from carboxylase activities in cultured fibroblasts (12). Indeed, in the only published study that examined indicators of biotin status during progressive biotin deficiency in human subjects, increased urinary excretion of 3-hydroxyisovaleric acid was an early and sensitive indicator of decreased biotin status (4). Decreased urinary excretion of biotin was also an early and sensitive indicator of decreased biotin status.
For these reasons, we chose to measure the urinary excretion of 3-hydroxyisovaleric acid and biotin to assess biotin status. The urinary excretion of 3-hydroxyisovaleric acid was significantly and strikingly increased. In this observation, we confirm the finding of Krause and co-workers (2) that treatment with anticonvulsant in this group caused increased urinary excretion of 3-hydroxyisovaleric acid. These excretion rates not only exceeded the values of the pediatric control group but also values for the adult control group (4.4-15 μmol/mol creatinine) and the abnormal values observed after 20 days of experimental egg-white feeding (15-58 μmol/mol creatinine) (4).
In contrast, normal urinary excretion of biotin indicates adequate biotin status. How can the apparent conflict between the two indicators of biotin status be resolved?
One possible explanation is that the urinary excretion of biotin was paradoxically increased from the expected deficient range back into the normal range. Theoretically, such a paradoxically normal urinary excretion of biotin could have resulted from impaired tubular reabsorption of biotin from the glomerular filtrate. If the renal transporter for biotin is indeed structurally specific, as suggested by previous studies(13,14), then increased concentrations of bisnorbiotin and biotin sulfoxide in the glomerular filtrate might have interfered with reclamation of biotin. Moreover, carbamazepine, phenytoin, and phenobarbital all incorporate a ureido group (-NH-CO-NH-) into their respective chemical structures as does biotin. Theoretically, one or more of the anticonvulsants could interfere with biotin reabsorption from glomerular filtrate and paradoxically increase biotin excretion, at least transiently. In accord with this possibility, Said and co-workers (15) demonstrated that physiologic concentrations of carbamazepine specifically and directly inhibit biotin uptake by the biotin transporter in the brushborder membrane of the human intestine.
In contrast to these potential explanations of paradoxical biotin excretion, the increased urinary excretion rates of bisnorbiotin and biotin sulfoxide cannot be attributed to a change in renal function per se. Any change in renal handling can produce only transient changes in excretion rates. The steady-state excretion rates must reflect net metabolite production. A steady state with respect to induction of biotin biotransformation is likely, because the anticonvulsant therapy had been continued in all subjects for at least 3 years.
An alternate explanation to the apparent conflict between 3-hydroxyisovaleric acid and urinary biotin is that increased biotin breakdown had not progressed to the point of depleting biotin at the tissue level. Such a situation might arise if the anticonvulsants interfered with renal handling or metabolism of 3-hydroxyisovaleric acid. Inconsistent with this explanation is the observation that abnormally increased 3-hydroxyisovaleric acid did not result from the treatment with phenobarbital. Also inconsistent with this explanation is the observation of Krause and co-workers(2) that four of seven adults receiving long-term treatment with carbamazepine, phenytoin, primidone, or phenobarbital alone or in combination had reduced plasma levels of biotin and excreted increased amounts of a constellation of organic acids typically seen in biotin deficiency.
In an attempt to gain insight into the mechanisms that might produce conflicting indicators of biotin status, we conducted an intervention in a single subject. The initial normalization of 3-hydroxyisovaleric acid excretion with biotin supplementation provided evidence that biotin status was impaired and that the impaired biotin status was producing abnormal organic aciduria, which responded to biotin supplementation. However, despite continued biotin supplementation, the patient again developed an erratic pattern of increased 3-hydroxyisovaleric acid excretion. Thus, after the first few days of intervention, increased 3-hydroxyisovaleric acid excretion could not have been reflecting biotin deficiency.
This study is limited in several ways. The number of subjects studied was not large enough to permit generalization to larger patient groups. The functional significance of the marginal biotin status (if any) is not clear. Severe biotin deficiency causes hair loss, a characteristic dermatitis, and CNS dysfunction. CNS dysfunction includes anorexia, malaise, and a peculiar withdrawn behavior and developmental delay in children(16). To our knowledge, no patient in the current study exhibited any of the cutaneous signs of biotin deficiency; because of the existing developmental disabilities of these patients, it was not possible to separately evaluate the CNS symptoms specific to biotin deficiency. Whether patients receiving long-term anticonvulsant therapy might have subtle deleterious effects caused by marginal biotin deficiency such as impaired immune function, retarded growth, or teratogenesis is not known.
The data of this study, combined with those of Krause and co-workers, are of sufficient concern that we believe additional studies are justified. First, we hope to establish the validity of other organic acids (e.g., methylcitric acid or 3-hydroxypropionic acid) and putative indicators that do not depend on renal function (e.g., lymphocyte carboxylase activities) in experimental egg-white feeding studies. Next, we plan to assess these indicators and their responses to biotin therapy in individuals receiving long-term anticonvulsant therapy. If these individuals are genuinely biotin deficient, we would expect the indicators to return to normal.
Acknowledgment: This study was supported by the National Institute of Diabetes and Digestive and Kidney Diseases, R01 DK36823(D.M.M.).
1. Krause K-H, Berlit P, Bonjour J-P. Impaired biotin status in anticonvulsant therapy. Ann Neurol
2. Krause K-H, Kochen W, Berli PT, Bonjour J-P. Excretion of organic acids associated with biotin deficiency in chronic anticonvulsant therapy. Int J Vitam Nutr Res
3. Mock DM, Lankford GL, Mock NI. Biotin accounts for only half of the total avidin-binding substances in human serum. J Nutr
4. Mock N, Malik M, Stumbo P, Bishop W, Mock D. Increased urinary excretion of 3-hydroxyisovaleric acid
and decreased urinary excretion of biotin are sensitive early indicators of decreased stauts in experimental biotin deficiency. Am J Clin Nutr
5. Mock DM, Heird GM. Urinary biotin analogs increase in humans during chronic supplementation: the analogs are biotin metabolites.Am J Physiol: Endocrinol Metab
6. Mock NI, Mock DM. Biotin deficiency in rats: disturbances of leucine metabolism are detectable early. J Nutr
7. O'Brien D, Ibbott FA, Rodgerson DO. Laboratory manual of pediatric micro-biochemical techniques,
4th ed. New York: Hoeber Medical Division, Harper & Row, 1968.
8. StatView. Berkely, CA: Abacus Concepts, 1996.
9. Mock DM. Skin manifestations of biotin deficiency. In: Roe DA, ed. Cutaneous manifestations of nutritional disease
. Philadelphia: WB Saunders, 1991:296-302.
10. Mock DM. Evidence for a pathogenic role of ω6 polyunsaturated fatty acid in the cutaneous manifestations of biotin deficiency. J Pediatr Gastroenterol Nutr
11. Velazquez A, vonRaesfeld D, Gonzalez-Noriega A, et al. Pyruvate carboxylase responsive to ketosis in a multiple carboxylase deficiency patient. J Inherit Metab Dis
12. Wolf B. Fibroblasts-disorders of biotin metabolism. In: Scriver CR, Beaudet AL, Sly WS, Valle D, eds. The metabolic and molecular bases of inherited disease
. New York, NY: McGraw-Hill, 1995:3161.
13. Podevin R-A, Barbarat B. Biotin uptake mechanisms in brushborder and basolateral membrane vesicles isolated from rabbit kidney cortex. Biochim Biophys Acta
14. Spencer PD, Roth KS. On the uptake of biotin by the rat renal tubule. Biochem Med Metab Biol
15. Said HM, Reyadh R, Nylander W. Biotin transport and anticonvulsant drugs. Am J Clin Nutr
16. Mock DM: Biotin. In Ziegler EE, Filer LJ Jr, eds.Present knowledge in nutrition
. Washington, DC: International Life Sciences Institutes-Nutrition Foundation, 1996:220-35.