Clinical, biochemical & molecular spectrum of adrenoleukodystrophy: A single centre experience

Materials & Methods

This was a cross-sectional study conducted between February 2014 and May 2020 at the Pediatrics Genetics & Research Laboratory, department of Pediatrics, Maulana Azad Medical College and associated Lok Nayak Hospital, a tertiary care hospital in North India, after obtaining clearance from the Institutional Ethical Committee. After obtaining informed consent, the authors enrolled children, adolescents, and adults with clinical and/or radiological evidence of ALD, as well as individuals carrying pathogenic variations in the ABCD1 gene identified by screening first-degree relatives of individuals with ALD. The study participants were further subclassified into five clinical subtypes of ALD based on their clinico-radiological phenotype. Healthy controls were also recruited for VLCFA level estimation after obtaining informed consent, as normative data for VLCFAs are scarcely available in the Indian population. The control population comprised of age and sex-matched healthy individuals selected from paediatric and medicine clinics, general health check-up camps in colleges, and workplace wellness programmes. Participants with no history of neurological or metabolic disorders or a family history of ALD were included after obtaining informed consent from participants (adults) or their parents/ guardians (children) for the same.

Results

A total of 35 study participants with suspected ALD were screened for eligibility using a combination of clinical features, neuroimaging findings, and biochemical parameters. Of these, 29 study participants with clinical and/or radiological evidence of ALD were included in this study. Loes scoring was done based on neuroimaging findings⁸. In addition, six asymptomatic individuals who were identified by screening of first-degree relatives of individuals with ALD were also enrolled. The majority (65.7%) of the enrolled subjects presented at an age below 18 yr. However, three of the asymptomatic individuals were less than 10 yr old, one was between 18-25 yr, and two of them were more than 25 yr old at the time of enrolment. Among this entire male cohort, most of the study participants were found to have adolescent ALD (13/35, 37.1%) or adrenomyeloneuropathy (AMN; 9/35, 25.7%). There were six study participants with childhood cerebral ALD, and one study participant with Addison type ALD. Of these 35 study participants, four with the AMN phenotype and three with the ALD phenotype were born out of a consanguineous mating. Four of the patients included had a history of similar illnesses in one or more family members. Gonadal mosaicism may explain cases in which mothers were not obligate carriers⁹.

Clinical presentation is shown in table I reveals muscle weakness and vision loss as the most common features; in addition, dysarthria, cataracts, adrenal insufficiency, and facial abnormalities were present in one study participant each.

A total of 35 patients and 383 healthy controls were studied for the evaluation of reference ranges for VLCFAs. The levels of all the VLCFA metabolites (C20:0, C22:0, C24:0, and C26:0 LPC) and their ratios were significantly greater among the patients as compared to the controls as shown in table II. However, no significant difference was detected in any of the VLCFA analytes except for the C24:0/C20:0 ratio among symptomatic and asymptomatic study participants. The VLCFA levels of individuals with various subtypes of ALD are presented in table III.

The diagnostic performance of C24:0 LPC and C26:0 LPC was evaluated using receiver operating characteristic (ROC) curves (Fig. 1). At cut-off values of 0.907 and 0.604 (µmol/L) C24:0 LPC and C26:0 LPC, respectively, were found to have a sensitivity and specificity of 100 per cent each, for the identification of ALD, as depicted in table IV.

Close

Fig. 1.

ROC curve for cut-off values of very long chain fatty acids for the identification of ALD. (A) At cut-off values of 0.69, C20:0 LPC was found to have a sensitivity of 66 per cent and specificity of 98 per cent. (B) At cut-off value of 0.8, C22:0 LPC was found to have a sensitivity of 66 per cent and specificity of 100 per cent. (C) At cut-off values of 0.907 C24:0 LPC was found to have a sensitivity and specificity of 100 per cent each. (D) At cut-off values of 0.604 (µmol/L) C26:0 LPC was found to have a sensitivity and specificity of 100 per cent.

Export to PPT

Magnetic resonance imaging (MRI) of the brain was performed for 22 of the recruited subjects (19 symptomatic individuals and 3 asymptomatic siblings). All symptomatic individuals had a typical MRI picture of X-ALD, in the form of hyper intensities involving the parieto-occipital areas and the splenium of the corpus callosum. Other common areas of involvement were the visual pathway in 89.5 per cent (17 of 22) and the auditory pathway in 63 per cent (12 of 22). Global atrophy was observed in 10 out of 22 study participants (52.6%), who also had disease onset in early childhood, indicating higher severity of illness. A majority of study participants had a mean Loes score between 5 and 10, but the mean Loes score of symptomatic individuals was 13.7 (minimum score: 5, maximum score: 30). An abnormality involving the splenium of the corpus callosum was significantly more common in symptomatic patients than in asymptomatic study subjects with ALD (P< 0.05). All the asymptomatic individuals had a Loes score ≤ 2.

Sequencing of the ABCD1 gene revealed that the most common mutations involved exon 1, as shown in supplementary table. We identified 4 novel variants in the ABCD1 gene in our cohort. Figure 2 shows the electropherogram of one of the novel variants identified in this study. While several mutations were observed throughout the genomic region of the ABCD1 gene, we further elucidated the novel mutations identified within our cohort. To assess the impact of these novel mutations on the protein structure, a three-dimensional visualisation of the ABCD1 gene (PDB id: 7RR9) was conducted. Of the four novel mutations, three resulted in alterations in the protein structure, while no changes at the protein level were reported for the g.11476 [G>A] mutation. All the mutations were modelled using the approaches elucidated in methodology. As depicted by multiple sequence alignment in supplementary figure 1, the mutants exhibit a significant degree of sequence similarity with the wild-type ABCD1 gene sequence (Uniprot ID: P33897). However, structural comparison revealed a significant degree of variability, resulting in the emergence of a different structure of the mutants. The protein structural models, herein described for c.904 [G>T]; p.E302*, as shown in supplementary figure 2, signify that the translation process is prematurely halted at position 302, leading to the production of a truncated protein when G-to-T replacement occurs at the 904 position in the gene DNA sequence. The generated model revealed 89.6 per cent residues in the most favoured region, suggesting its high quality. Since proteins are not static, their residues are in continuous motion inside the system. Therefore, a dynamic cross-correlation study was employed to analyse and visualise the dynamic motions and correlations between atoms or residues among the protein structures caused by mutations. The correlated fluctuations in the wild type and mutant systems have been graphically presented in supplementary figures 3. Positive correlations are mapped in the upper left triangle, while negative correlations are mapped in the lower right triangle. A deeper colour indicates positively or negatively correlated motions between the structural patterns. As shown in supplementary figures 2 and 3, it can be easily seen that the identified novel variants tend to affect protein flexibility, and major structural regions affected by the mutation are highlighted in both instances.

Close

Fig. 2.

Electropherogram of a novel mutation c.493_494 [ins CAC]; p.165Pro; identified in a study participant with adrenomyeloneuropathy. The insertion of 3 nucleotides CAC results in generation of one extra protein i.e., Proline.

Export to PPT

Discussion

The present study enrolled individuals with ALD subtypes, with adolescent ALD (13/35, 37.1%) and AMN (9/35, 28.1%) being the most common. This finding is consistent with published literature showing that ALDs can present with a wide range of phenotypes, from slowly progressive myelopathy to rapid demyelination of the brain white matter¹². A study from southern India revealed the childhood cerebral form of ALD to be the most common phenotype, followed by the adolescent cerebral form, with 37.9 per cent of the symptomatic patients between 10-18 yr of age⁵. According to a retrospective evaluation of patients with ALD from Turkey, the mean age at diagnosis was 10.5 yr. Eight out of 12 study participants had the childhood cerebral ALD subtype, two of them had the cerebral form of AMN, and one patient each presented with the adult form of cerebral ALD, and the Addison-only phenotype¹³.

According to the analysis of VLCFA levels, the positive predictive values ranged from 72.4 per cent to 100 per cent, specificity from 97.7 per cent to 100 per cent, sensitivity from 65.6 per cent (C20:0) to 100 per cent (C26:0LPC), with a diagnostic accuracy of 95 per cent for C20:0, 97.1 per cent for C22:0, and 100 per cent for C24:0/C26:0. The cut-off values for C24:0 LPC, C26:0 LPC, the C26:0/C22:0 ratio, and the C26:0/C20:0 ratio, which yielded the best sensitivity and specificity in the study population, were 0.907 µmol/3.2mm punch, 0.604 µmol/3.2mm punch, 1, and 1.108, respectively, as obtained by ROC analysis. According to a study by Natarajan et al⁵, the median concentrations of C24:0 and C26:0-LPCs were 7- and 18-fold greater, respectively, in ALD patients (0.56[0.20–1.48] μmol/L and 0.55[0.14–1.34] μmol/L) than in healthy controls (0.08[0.03–0.16] μmol/L and 0.03[0.01–0.05] μmol/L). The reference intervals of C22:0, C24:0, and C26:0 were found to be 32.0-73.4 μmol/L, 30.3-72.0 μmol/L, and 0.20-0.71 μmol/L, respectively. Additionally, the C24:0/C22:0 and C26:0/C22:0 ratios were 0.75-1.28 and 0.005-0.0139, respectively, in the Chinese population¹⁴. Rattay et al¹⁵ reported that the best differentiation of non-ALD and ALD individuals was obtained with cut-off values of <1.0 and <0.02 for C24:0/C22:0 ratio and C26:0/C22:0 ratio, respectively, among the German population. Wu et al¹⁶ found 26:0 LysoPC to be superior for diagnosing Japanese X-ALD, as compared to 24:0 LysoPC and 20:0 LysoPC. Huffnagel et al¹⁷ compared the sensitivity of C26:0-carnitine and C26:0-lysoPC in DBS to screen neonates for ALD, and found that C26:0-carnitine was elevated in only 83 per cent of babies who screened positive for ALD.

The present study highlights the importance of considering VLCFA ratios, along with the individual values for establishing the diagnosis, as including the ratios especially C24/C22 and C26/C20 would improve the sensitivity from 65.6 to 96.9 per cent, which is of significant additive prediction for identification of patients with suspected ALD. However, identification of female heterozygotes must rely on genetic testing, as normal VLCFA levels may be found in 15 per cent cases⁹. Our study also highlights the importance of neuroimaging and calculation of Loes score, which could direct the clinicians in decision-making on the appropriateness of instituting Haematopoietic stem cell transplant (HSCT), as well as follow up of these patients. In our series, five individuals (26.3%) had a Loes score <9, and two of them had a score <4. These individuals would therefore be ideal candidates for HSCT if the resources were available for the same. In addition, elivaldogene autotemcel (SKYSONA) gene therapy has been recently approved by the FDA for asymptomatic or mildly symptomatic boys 4-17 yr of age with early, active cerebral adrenoleukodystrophy (CALD) who have Loes scores of 0.5-9 on neuroimaging to slow the progression of neurologic dysfunction in these individuals¹⁸. Asymptomatic individuals with ALD identified by cascade screening are expected to have better outcomes as they can be identified before there is significant neurological damage and can be enrolled for definitive therapy. This can have significant survival benefits and prognostic implications by arresting and interrupting the demyelination process¹⁹.

Our study identified a mutational hotspot in the ABCD1 gene, which was exon 1 in our cohort, accounting for 54.3 per cent of patients. Of our patients, 20 per cent, were found to have a pathogenic variant in exon 5. This may strategise the priority of exon 1 analysis for the preliminary screening of ABCD1 gene, followed by exon 5. No pathogenic variant was detected in the exons 8, 9, and 10 of ABCD1 in our study. No genotype-phenotype correlation could be elucidated in our cohort. Marketa et al²⁰ also did not find any correlation between the disease severity and the genotype. In the present study, most of the patients had a pathogenic variant in exon 1 (54.3%, n=19), followed by exon 5 (20%, n=7), and none in exons 8-10. However, a study by Boehm et al²¹ reported maximum mutations in exon 1 followed by exons 8 and 9.

Overall, the present study emphasises the importance of clinical suspicion, combined with biochemical and radiological investigations, in the diagnosis of X-linked ALD. Our cohort of 35 individuals with ALD from a single northern Indian centre, provides normative data for VLCFA and the cut-off values that offer best sensitivity and specificity in diagnosing patients with ALD. The study also highlights the importance of considering VLCFA ratios, in addition to individual values, to improve the sensitivity of diagnostic tests. This study also identifies a hotspot in exon 1 of the ABCD1 gene, which could be used for the preliminary screening of the ABCD1 gene in patients with suspected ALD. Additionally, the study highlights the importance of early diagnosis through cascade screening to identify asymptomatic individuals who may benefit from early intervention and have better clinical outcomes. The six asymptomatic individuals found to have investigation findings suggestive of ALD were advised for close follow up with a neurologist. However, as this study included subjects with ALD from a single tertiary care hospital in North India, the number of subjects recruited is not very large. As ALD is a heterogenous disease with various subtypes, the number of individuals falling in each subgroup was quite small, precluding any further analysis according to the disease subtype. There is a need for further similar, larger, multicentric studies on ALD patients from various ethnic groups to validate our results.