Introduction

Biotinidase deficiency (BTD) is one of the most common autosomal recessive diseases globally, with an incidence of 1:40,000–1:60,000 per live birth [1]. However, in Turkey, where consanguineous marriages are widespread, the frequency reaches up to 1:7116. Over 200 variants have been identified in the BTD gene [2]. BTD can cause neurodevelopmental delay, seizures, visual loss, sensorineural hearing loss, and rash in children of any age, from birth to adulthood. Biotinidase activity is frequently measured using the calorimetric technique. Based on residual enzyme activity, three biochemical phenotypes are categorized as follows: profound (≤ 10%), partial (10.1–30%), and normal (> 30.1%). However, the biochemical phenotypes can exhibit variability. Biotinidase activity does not act along a continuous spectrum [3]. Consequently, phenotypes may not always match the expected genotype when phenotypes are identified using predetermined cut-off values. Biotinidase activity has biphasic kinetics, which can lead to clinical heterogeneity [4]. In addition, enzyme maturity develops with age [5]. Screening tests for biotinidase are typically carried out three to five days after birth, at the lowest point of the enzyme’s activity. Therefore, biochemical phenotypic groups identified in the early neonatal period can change. These factors make it challenging to evaluate neonatal screening results, and newborns are frequently overdiagnosed with BTD. This complexity can make it challenging for each patient to choose a suitable therapy strategy and raise questions about cost-effectiveness [6].

The BTD gene, located on chromosome 3 and consisting of 4 exons and 543 amino acids, encodes biotinidase. BTD is caused by pathogenic variants in the BTD gene, resulting in a variety of clinical phenotypes [7]. Over 200 pathogenic variants have been identified, with the number of newly reported variants increasing. The frequency and diversity of disease-causing variants vary by country and region. The most common variants are generally: p.D444H (c.1330G > C), p.R157H (c.470G > A), p.C33Ffster*36 (c.198_104delGCGGCTGinsTCC), p.(T532M) (c.1595C > T), and p.Q456H ( c.1368A > C) (7). However, the correlation between genotype and phenotype is not well-established. The diversity of variant allele combinations also affects clinical and biochemical phenotypes. Despite some severe pathogenic variants exhibiting predictable clinical phenotypes, the genotype–phenotype relationship in BTD is not well-established.

This large cohort study aimed to investigate the relationship between BTD variants and biochemical phenotypes.

Materials and methods

This study was conducted at the Department of Pediatric Metabolism, Gaziantep University, as a single-center, observational, cross-sectional study. Approval for the study was obtained from the Clinical Research Ethics Committee of Gaziantep University (Decision no/date: 2019/194–19.6.2019). Written consent was obtained from all parents. Blood samples were collected on filtered paper three to five days after birth, following the national screening protocol. Patients who tested positive for biotinidase screening sought confirmation at our center within the first month of birth. Confirmation tests were conducted using quantitative measurements and the colorimetric method. Molecular analysis was performed to identify pathogenic variants of the BTD gene for each subject.

Subjects

The study included 711 nonrelative newborns referred to the Neonatal Biotinidase Screening Program. The majority of initial confirmatory tests were conducted within the first month of life. Oral biotin treatments of 10 mg/day (Profound) and 5 mg/day (Partial) were administered based on enzyme activity levels.

Enzyme activity measurement and determination of biochemical phenotype groups

Quantitative enzyme activity was measured using the colorimetric method. The normal reference range was 3.5–13.8 nmol/min/mL. Biochemical phenotyping was divided into three groups based on mean enzyme activity (7.5 nmol/min/mL): profound (≤ 10% of mean activity), partial (10.1–30%), and normal (> 30.1%).

Molecular analysis

Blood sample was sequenced using Sanger sequence method (AB 3130 XL) after amplification with PCR (Maxwell RSC Whole Blood DNA kit Promega, USA). DNA density was adjusted to 10–30 ng/ml using a spectrophotometer (Biotech USA). Sanger sequencing was conducted to analyze all exons, introns, and junctions of the BTD gene using the AB3130XL device. The genomic reference for the BTD gene in Ensembl was NM_000060(NM_001370658.1) (Transcript ID: ENS00000303498.5; Reference sequence F: ATGGCGCAT R:TACCGCGTA).

Pathogenic variants were classified according to their pathogenicity using the PubMed and BTD databases (ClinVar, Varsome, ARUP, LOVD, Ensemble, MutationTaster, ACMG). The clinical significance and variant pathogenicity of the BTD gene were assessed using BTD databases [8,9,10,11,12,13]. Variant pathogenicity was classified according to the ACMG criteria [14]. Novel pathogenic variants were defined as those not present in the mentioned databases.

Genotypes were classified as homozygous (HMZ), heterozygous (HTZ), and compound heterozygous (C/H) based on heterozygosity.

Statistics

Statistical analyses were performed using the Statistical Package for Social Sciences version 22.0 (IBM Inc., Chicago, IL, USA). The Shapiro–Wilk test was used to confirm the normal distribution of continuous variables. The chi-square test was used to determine the significance between the expected and observed enzyme activity groups.

Results

Among participants (n = 711), 325 (45.8%) were female. The distribution of biochemical phenotype groups based on biotinidase activity was as follows: profound 22 (3%), partial 95(13%), and normal 594 (84%), (Table 1). The mean age at the time of the test was 18.53 ± 8.54 days. Table 1 shows the mean levels of biotinidase activity for each phenotype group.

Table 1 Clinical and genetic characteristics of newborns with biotinidase deficiency
Full size table

Of the total alleles (n = 1422), 386, 612 and 314 were from patients based on zygosity of 193 HMZ (27.1%), 306 HTZ (43.1%) and 157 C/H (22.1%), respectively. No pathogenic variant was detected in 55 (7.7%) patients (Table 1). Of these, 53 were in normal biochemical phenotype, while the remaining two were in the partial (n = 1) and profound (n = 1) groups.

The most common alleles among the total alleles (n​ = 1422) were p.D444H (c.1330G > C) (n = 526,37%). The next two alleles with the highest frequencies were p.R157H (c.470G > A) (n = 172, 12.1%) and p.C33Ffster*36 (c.198_104delGCGGCTGinsTCC (n = 84, 6%) (Table 2). The other most frequent alleles are shown in Table 2. There were 11 cases with the p.D444H HMZ genotype that presented with Profound (n = 3) and Partial (n = 8) BTD. Less frequent alleles are given in Table 2. Six new previously unreported mutations were detected in the BTD databases (Table 2).

Table 2 Distribution of BTD variants based on pathogenicity and zygosity between the biochemical phenotype groups
Full size table

According to BTD databases, the pathogenicity classification of variants was as follows: pathogenic: 481 (33.8%), variant of uncertain significance (VUS): 538 (37.8%), 4 likely pathogenic (0.3%) (Table 1). The distribution of newborns with pathogenic variants in both alleles into biochemical phenotypes was as follows: profound: 15 (68%), partial: 29 (30.5%), and normal: 43 (7%). The distribution of pathogenic variants (n = 481) into profound, partial, and normal biochemical groups was as follows: pathogenic variant 43 (7%), 105 (22%), and 342 (71%). VUS (n = 538): 6 (1.1%), 70 (13%), and 462 (85.9%).

Discussion

This large cohort study examines the relationship between BTD variants and biochemical phenotypes during the neonatal period. Most of the reports related to BTD are cross-sectional studies and involve the neonatal or early infant period [15,16,17].

Previous studies on early infancy have reported a higher frequency of profound and partial BTD [15, 17, 18]. Biotinidase enzyme activity levels are reported to be the lowest in the first days of life, but increases within the following days [15]. Therefore, low levels of enzyme activity can be expected in early neonatal confirmation tests with screening performed in the first postnatal period. Biotinidase activity maturation later in life should be taken into account when assessing biochemical phenotyping based on biotinidase levels in newborns with BTD. However, there is not enough research on the course of low activity levels later in life. Considering this pattern of enzyme activity that has settled to normal levels at a later age, the relatively high frequency of BTD may reflect the effort to adapt to the transition from the intrauterine period to the extrauterine period rather than a disease. This variability in biotinidase activity is influenced by factors arising from various clinical conditions and the natural course of the disease. These include premature birth, environmental temperature, transfer conditions of the samples, hyperbilirubinemia, and perinatal asphyxia [11,12,13]. These findings indicate that a single measurement in the neonatal period is not sufficient to determine the current enzyme activity level, and follow-up with sequential tests is needed to assign biochemical phenotypes in BTD.

The most common variants found in this study were p.D444H, p.R157H, and p.C33Ffs. Although p.D444H is considered a mild variant, it was found to be associated with profound activity in three HMZ cases (p.D444H-p.D444H), partial enzyme activity in eight cases with HMZ (p.D444H-p.D444H), and partial enzyme activity in three HTZ cases (p.D444H-No). The p.D444H variant has, against expectations, also been documented in previous studies, where it has been correlated to more severe biochemical phenotypes [17,18,19,20,21]. This could be the result of double pathogenic variants, which are combinations of severe variants that are undetectable, or the additional contribution of epigenetic factors [22]. Given that frequent consanguineous marriages are common in our region, complex pathogenic variants appear to be more frequent [23, 24]. p.R157H was one of the most common severe variants. However, biotinidase activity was normal level in two patients with the HMZ p.R157H genotype. It was found that the p.C33Ffs variant was seen more frequently in this study, compared to other reports [17, 18]. It has been reported that the p.C33Ffs variant is one of the most common reasons for profound BTD in the USA [25]. Contrary to expectations, the normal enzyme activity group was detected in 12 p.C33Ffs heterozygous genotype (p.C33Ffs-No) cases. Unlike the expected phenotype, partial enzyme activities were observed in severe mutant alleles, such as p.Q456H and p.T532M.

The genotype–biochemical phenotype compatibility in this study was lower than previously reported data [7, 18]. This may be due to increased enzyme maturation and activity with age.

Many asymptomatic cases with severe pathogenic variants have been reported in the literature, and some viewpoints have been proposed for genotype–phenotype incompatibility [26]. Genotype–phenotype incompatibility suggests that there is sufficient residual enzyme activity in the presence of a substrate alongside epigenetic factors. Another view suggests that biotinidase is caused by individual differences in Km kinetics. Adequate dietary biotin intake may also help to prevent the appearance of clinical findings [27]. These asymptomatic cases may present with clinical findings such as acute hearing loss as a result of exposure to metabolic stress at later ages.

Preanalytical factors, on the other hand, such as blood sample transportation or storage conditions until analysis, may also affect the measurement of enzyme activity. These variables are limitations of these studies involving genotype–phenotype correlation in BTD. This study supports the hypothesis that variants identified as pathogenic in BTD databases alone cannot predict clinical phenotypes. Of the cases with normal enzyme activity, 7.2% had pathogenic variants in both alleles. These cases require long-term follow-up. Likewise, pathogenic variants in both alleles were present in 30.5% of the partial group. Similar cases have been reported in the literature [28]. Asymptomatic cases are described, despite the fact that they have pathogenic variants. This large cohort will contribute to the implications of these cases. Another important contribution of the study is that it reports six new pathogenic variants that were not previously identified in BTD databases. One of the study’s major strengths is that it is one of the largest cohorts in the literature dealing with the genotype–phenotype relationship in BTD patients. Another significant contribution is the identification of six new pathogenic variants.

Fifty-three of the 55 cases without any pathogenic variant had normal enzyme activity. These cases were patients who applied because of screening positivity. It was thought that the false positivity in these cases was due to the lower enzyme activity in the neonatal period. Confirmatory tests from these newborns were performed at the end of the first month. It was interpreted as the presence of an undetected variant in the other two newborns, one of which was partial and the other profound.

Conclusion

The most common alleles found in this large cohort in Southeastern Anatolia were p.D444H (c.1330G > C), p.R157H (c.470G > A), and p.C33Ffster*36. The results of this study show that there is a low concordance between genotype and biochemical phenotype in newborns with BTD in this region. This study confirms that it is not sufficient to establish a predictable correlation between BTD variant pathogenicity and biochemical phenotype in newborns with biotinidase deficiency. Additionally, this study emphasizes the importance of considering clinical features and addressing analytical difficulties in enzyme activity measurements and molecular analyses to explain the mismatch between genotype and biochemical phenotype.