This study characterized the clinical and molecular profile of Brazilian patients with reduced biotinidase activity and, thus, at risk of BD. Before 2012, neonatal screening for BD was unavailable through the public Unified Health System in the majority of Brazilian states, and was only provided by private laboratories located mainly in the South and Southeast regions of the country, which may have contributed to its underdiagnosis. Paradoxically, studies based on neonatal screening suggest that the overall frequency of BD (profound + partial BD) in Brazil may be among the highest ever reported, ranging from 1:6,843 to 1:62,500 newborns (NB) [17–20]. To the best of our knowledge, only one prior study conducted DNA analysis of Brazilian patients with BD (n = 21) .
Two studies conducted in the U.S., in which 92 patients with profound BD and 19 with partial BD were assessed, showed that the most common BTD variants in the study population were c.98_104del7ins3, c.1612C > T (p.R538C), c.1368A > C (p.Q456H), p.[A171T;D444H] and p.D444H, which, overall, accounted for approximately 60% of abnormal alleles found in the sample [12,21]. Taking into account this five-variant panel, p.[A171T;D444H] and p.D444H account for approximately 55% of alleles detected in the present study, with the occurrence of variants c.98_104del7ins3 and p.R538C in only one patient each. In our sample, we did not detect the variant p.Q456H, the profound BD allele most commonly found among children diagnosed by neonatal screening in the U.S. (allele frequency = 28%) , nor did we find any changes in exon 3. Furthermore, p.D252G appeared to occur more commonly (allele frequency = 9.4%) in our sample than in the U.S. population . Our findings corroborate those of Neto et al. (2004) , who analyzed the BTD gene in 21 Brazilian patients with reduced biotinidase activity detected by neonatal screening and also found that p.D444H, c.98_104del7ins3, p.[A171T;D444H], and p.D252G were the most prevalent variants; the authors also did not detect variant p.Q456H or changes in exon 3 among their sample. Therefore, we suggest that the profile of genetic variants found among Brazilian patients with BD differs from that found in U.S. patients, although these differences may be partly attributable to the relatively small number of patients with profound BD included in our sample (c.98_104del7ins3, p.R538C and p.Q546H are classically associated with profound BD).
Since the p.D444H variant - usually classified as a partial DB allele - was the most common variant found in our study, our findings also suggest that partial BD is the most common form of BD in Brazil. We found that allele C of the variant c.1330G > C or p.D444H constituted 4% of alleles in healthy individuals, a frequency close to that found in the U.S. population [22,23], where the incidence of profound and partial BD has been estimated at 1:80,000 NB and 1:31,000–1:40,000 NB respectively , and in Western Hungary, where the incidence of profound and partial BD is 1:97,000 NB and 1:23,000 NB respectively .
As previously reported, the association between BTD genotype and biotinidase activity is not absolute [18,25]. In the Greek study conducted by Thodi et al. (2013) , for instance, homozygosity for p.D444H was suggested to be associated with partial BD, and not with a 45–50% reduction of biotinidase activity, as expected. In the present study, among the seven cases of disagreement between the observed and expected biochemical phenotype, we highlight patient 10 because her genotype p.D444H/c.595G > A (p.V199M) and p.V199M/p.V199M have previously been reported in patients with residual biotinidase activity levels of 32% and 11.5% respectively , and it is possible that the p.V199M variant is not as damaging as other profound BD alleles. Furthermore, patient 31, from Northeast Brazil, and patients 32, 34 and 35, from the Southeast region, underwent testing at a laboratory in the South region, so their biochemical test results may have been artifactually low due to time constraints or transportation of samples to the laboratory under inadequate conditions, as suggested by Neto et al. (2004) . However, we cannot rule out the presence of changes in a region of the BTD gene not covered by our analyses causing decreased enzyme activity.
In three patients, we were unable to define the cis/trans status of variant p.D444H in relation to the other variant found. On the basis of enzyme activity, trans status is most probable for patient 16. For patient 17, who presented with a biotinidase level suggestive of heterozygosity (but very close to the upper limits of partial DB), only gDNA from her mother, who also presented low biotinidase activity in a dried spot blood sample (data not shown), was available for analysis; as she was found to be a carrier of both p.D444H and c.643C > T (p.L251F) variants, the patient can be heterozygous if both variants were inherited on the maternal side (if they were in cis), or have partial BD if p.D444H was inherited from the father and only p.L215F from the mother (in this case, the mother should also exhibit partial DB). Similarly, both parents of patient 18, who is heterozygous for variants p.D444H and c.1629C > A (p.D543E), are heterozygous for p.D444H, but the father is also a carrier of p.D543E; therefore, the patient can be heterozygous if p.D444H and p.D543 are in cis in the paternal allele, or can have partial DB if they are in trans (in which case the father should also exhibit partial DB). A new enzyme activity test in the child and testing of parents (or even analysis of grandparent DNA) could help characterize BD in this case.
Although the enzyme activity of symptomatic patients 32, 34 and 35 was not consistent with profound or partial BD, biotin treatment was started after other genetic conditions had been ruled out, as the patients’ clinical manifestations are consistent with BD and biotin has a good safety profile. Unfortunately, we are not aware whether there was any improvement with treatment. These patients may represent cases of delayed-onset BD, but the absence of variants in exons 2–4 of the BTD gene suggest they do not have BD. Patient 33 is also very interesting, since his biotinidase activity levels were very low on neonatal screening, and gradually approached the normal range thereafter, as would be expected to his genotype. We believe the wide variation in biotinidase levels presented by this patient is mainly explained by premature birth and neonatal jaundice. However, as the first test was performed when he was only 5 days old and the last at age 1 year, some of this variation could also be due to increasing age.
On the basis of the allele frequency of novel variants p.L40P and p.C160Y in controls and of the in silico prediction of pathogenicity, and taking observed enzyme activity into account, these alleles were considered probably damaging, and patients 36 and 37, who are compound heterozygous for the p.D444H variant, probably have partial BD. Variant p.L40P does not affect the peptide portion that constitutes the mature protein, but affects the signal peptide, which may impair targeting of the enzyme to secretory vesicles, thus resulting in reduced enzyme activity in plasma. Alleles associated with profound BD which affect this region have been described elsewhere [27,28]. The effect of variant p.D222N on biotinidase function could not be established clearly from bioinformatics, as we used two different software programs which presented contradictory conclusions regarding this variant. However, the absence of this variant in controls suggests it is probably damaging.
Regarding the other two nonpathogenic polymorphic variants found in controls, c.1171C > T or p.P391S (rs35034250) and c.1413 T > C or p.C471C (rs3817641), both were found at frequencies similar to those reported in populations of USA and African origin respectively (available at: http://www.ncbi.nlm.nih.gov/snp/).