In the field of rare diseases, prospective observational studies are vital for establishing patterns of disease progression and to help inform and shape relevant outcome measures for future clinical trials . In this prospective, observational cohort study of NPC, the disease severity of the study population worsened from Visit 1 to Visit 2; this applies to all the clinical scores assessed (5-domain NPCCSS, 17-domain NPCCSS and NPC-cdb). The mean (± SD) change on the 5-domain NPCCSS at 6 months was 0.75 (± 1.58), corresponding to an annualized disease progression of 1.5 within the cohort. The mean (± SD) change on the 17-domain NPCCSS at 6 months and the mean (± SD) overall change on the NPC-cdb were 1.47 (± 2.25) and 5.0 (± 7.9), respectively. The findings from the NPC-001 study therefore support the characterization of NPC as a heterogenous, progressive neurodegenerative disease.
The use of the 5-domain NPCCSS has previously been supported by Cortina-Borja et al. 2018 . The study reported that these five domains (ambulation, cognition, fine motor skills, speech and swallowing) were determined by individuals with NPC, caregivers and clinicians, to be the most clinically relevant and were highly correlated with the total 17-domain NPCCSS (Spearman’s correlation coefficient = 93%) . The findings from the Cortina-Borja et al. cohort indicate that the 5-domain NPCCSS reflects disease progression measured by the 17-domain NPCCSS; further work is underway to validate and to confirm the patients and clinicians view on clinical meaningful changes on the 5-domain NPCCSS and is documented in the NPC Patient-Focused Drug Development report to the Food and Drug Administration .
Disease progression observed on the 17-domain NPCCSS was consistent with other NPC populations. In the population studied by Ory et al. , the change in disease progression per 12 months on the overall 17-domain NPCCSS was 2.92 (standard error of the mean [SEM] = 0.27), which is consistent with the mean (± SD) reported progression rate per 6 months in this current study of 1.47 (± 2.25) of a heterogeneous study population. As mentioned in the introduction, a potential confounder of the 17-domain NPCCSS is its complexity and the inherent difficulty of its implementation in everyday clinical practice. The 5-domain abbreviated scale has practical benefits in improving time efficiency, minimizing statistical variability, and reducing influence from disease manifestations that can cause general clinical fluctuations such as seizures and respiratory complications.
NPC disease severity and complications may also fluctuate day to day in individuals with NPC, although this confounder is minimized by the format of the 5-domain NPCCSS, which means that no single domain can be deemed the most important. As a total score, the 5-domain NPCCSS captures the impact of NPC on patients’ functioning, so that each step on the score reflects a considerable change in function. The 5-domain NPCCSS total score differs from tools developed to assess subtle changes in different domains, such as neuropsychological batteries used to assess potential cognitive impairment, the Scale for Assessment and Rating for Ataxia , the 9-Hole Peg Test for fine motor changes , and the 3-minute stair climb test for ambulation. Although these cognitive evaluations have been validated in other disease populations, they are not disease specific and there are not enough data on their specificity and sensitivity regarding rates of change in NPC populations.
Missense mutations in NPC1 are the most common type of mutation reported in NPC (75% of patient alleles) ; this was also the case in this study. As in most lysosomal storage diseases, a clear genotype/phenotype relationship is difficult to establish in NPC [47, 48], and it was not possible to correlate the biomarker and progression values reported here to specific mutations in a meaningful way. One exception was a patient homozygous for the A1108fs frameshift mutation, who exhibited a disease progression of 14 points in 13 months, confirming that double functional null genotypes are highly predictive of a severe and early progressive disease course [12, 47]. Other NPC cohorts of similar sizes have also reported individuals with double functional null genotypes occurring at similar rates within representative, heterogenous populations [49, 50]; therefore, the occurrence of this genotype in the current study is not unexpected. A detailed report on the mutations found in this study population, and in that of the double-blind interventional NPC-002 study, will be explored further in a separate publication.
This study aimed to characterize novel NPC disease biomarkers in order to establish a set of biomarkers for NPC that are focused on interventions targeting the HSP system in NPC and the underlying cellular pathology of NPC, including altered NPC1/2 protein function and lipid metabolism. During the time of this study and its publication, it is worth noting that other biomarkers such as lyso-sphingomyelin-509 and bile acids have also shown potential as biomarkers of NPC disease [41, 51-53].
Determination of unesterified cholesterol in skin biopsy samples showed significantly increased levels in individuals with NPC (p = 0.0006) compared with a cohort of healthy individuals whose biomarker levels were representative of the wider population. Hence, the findings confirm that LC-MS/MS-based measurements of skin biopsy samples are a valid method to support the biochemical diagnosis of NPC , and to measure the cholesterol storage burden in individuals with NPC. No evidence of changes in the storage burden over time were observed, nor were any correlations with the 5-domain NPCCSS found (Fig. 5a). A limitation of these data was that at Visit 2, skin unesterified cholesterol assessments were performed on fewer individuals than at Visit 1; a minimum of 4 mg skin was required to perform the analysis, and 16 of 36 skin biopsy samples from Visit 2 did not fulfil this requirement. As a result, the repeat assessment for these individuals could not be confirmed. Nevertheless, the levels in the NPC individuals here are similar to those reported previously using similar LC-MS/MS methods . These findings also demonstrate that unesterified cholesterol may be measured directly in skin biopsy samples in lieu of isolating and culturing fibroblasts for subsequent filipin staining, which is both time and labour intensive. This therefore has the potential to shorten the time for diagnosis . To potentially monitor longitudinally unesterified cholesterol levels in individuals with NPC, the PBMC method has benefits over skin biopsy sampling because it eases the sampling burden on the individual, especially within the young NPC demographic, and may reduce variability.
In clinical practice, cholestane-triol is frequently used as a biomarker to support diagnosis . In this trial, cholestane-triol levels are similar to those reported in other NPC cohorts at diagnosis [23, 41]. An increase over 6 months in cholestane-triol levels was observed with a mean (± SD) change per 6 months of 2.64 ng/mL (± 10.69), and cholestane-triol levels were significantly correlated (Spearman’s correlation coefficient = 0.265, p = 0.0411) at the population level with disease severity on the 5-domain NPCCSS (Fig. 5c). These results confirm previous studies demonstrating an increased cholestane burden with NPC disease severity . However, unlike previous studies, the CT-ORZY-NPC-001 results reported here do not support a correlation between serum cholestane-triol levels and age of NPC disease onset (data on file) [22, 41]. Together these data support the use of cholestane-triol as a biomarker of disease severity in the NPC population as a whole and may be useful in the follow-up of an individual patient to potentially monitor disease.
Based on the Vanier 1988 method , a new analytical method for cholesterol esterification in PBMC was established, aiming to shorten the time to diagnosis and reduce the burden on the individual, while also giving a direct measurement of NPC protein function. The findings demonstrate that cholesterol esterification is significantly decreased in the PBMCs of individuals with NPC compared with healthy individuals (p < 0.0001). Mutations in the NPC genes either result in reduced protein function because of limited binding capacity, or reduced abundance because of a more rapid degradation of the dysfunctional protein [2, 4, 54, 55]; therefore, NPC function is not expected to change with disease progression. Accordingly, no significant change in cholesterol esterification was observed during the study, nor did cholesterol esterification correlate with disease progression (Fig. 5d). Although variation of cholesterol esterification levels in PBMC was high, both within and between individuals, this was likely owing to the limited viability of PBMCs for the analytical method. Nonetheless, the current study confirms that PBMCs are a valid matrix on which to assess NPC protein function and to potentially support NPC diagnosis.
Establishing the basal levels of HSP70 in the PBMCs of individuals with NPC and assessing its variability during the study was the key intention of measuring HSP70 expression levels. We have established that HSP70 expression levels are decreased in individuals with NPC compared with those in the cohort of healthy individuals, with relatively stable levels over time. These findings are consistent with those demonstrated in murine models where Npc1-/- mice exhibit lower levels of HSP70 expression and HSF1 activation compared with wildtype mice; drug-induced elevated expression of HSP70 in the Npc1-/- mouse model was associated with reduction in glycosphingolipid accumulation, improvement in CNS myelination, improvement in measurable manifestations of ataxia and reduced loss of motor coordination . Additionally, insufficient induction of HSPs has been associated with a variety of chronic neurological diseases . The data presented here on HSP70 expression levels in the PBMCs of healthy individuals are in accordance with those previously published in Madden et al. 2010 .
With regard to safety, this was a non-interventional study with no investigational medicinal product, and so AEs were recorded to inform the background occurrence of adverse events in individuals with NPC on routine clinical care. Overall, nine of the 120 AEs (7.5%) were considered definitely related to NPC disease and eight AEs (6.7%; all cases of diarrhoea) were considered to be related to miglustat use by the investigators. This highlights the importance of ensuring stable routine clinical care prior to enrolment.
Among the potential limitations of the NPC-001 observational study was the high proportion of individuals receiving miglustat (30/36; 83.3%). Miglustat is likely to modify the clinical progression of NPC . However, the nearly ubiquitous use of miglustat across the study population meant that subgroup analysis of those not receiving miglustat could not be done, and so the effect of miglustat use on clinical evolution and disease progression could not be determined. With regard to the baseline disease characteristics as reported by referring clinicians as current medical conditions, the specific term of ataxia (reported in 2 of 36 individuals [5.6%]) appeared to be lower than in other reported NPC populations. This is likely because many of the study participants were defined non-specifically at baseline as clumsy or uncoordinated, rather than given the specific term of ataxic. However, from the NPC-cdb tool during baseline study assessments, ataxia was found to be present in 28 of 36 individuals (77.8%); this is in line with both the ambulation domain scores of the current study population and with the literature of other reported populations, in which it may be present in up to 76% of individuals with NPC [1, 10, 58]. Ideally, to optimize an assessment of NPC disease progression, the study could have been improved by a longer duration and more frequent or intermediary assessments (e.g. 6-monthly) than over the 6–14-month observation period employed here.
Another study limitation was the lack of standardization of skin punch biopsy depth across the clinical study sites. Although all analyses were performed by a single central laboratory, the biopsy samples were collected by different clinicians at different study sites and some biopsy samples were not large enough to apply the analysis method. Additionally, this variability was likely also influenced by the challenge of acquiring skin biopsy samples from children, the quality of which is dependent on skin thickness and elasticity, which is determined by the precise site of biopsy and age of individual.
The reliability study of the 5-domain NPCCSS, conducted to verify the accuracy of the scoring method between and within raters, provided additional support for the use of the 5-domain NPCCSS. The results support the use of the 5-domain NPCCSS score as a reliable endpoint in future studies and suggest that clinicians were able to rate participants similarly and consistently, as well as utilize each of the categorical severity ratings in the manner intended. The supportive Many-Facets Rasch analysis illustrated that each of the items contributing to the 5-domain NPCCSS were locally independent (i.e. not dependent on responses to other items), and that the instrument, in its brevity, is able to target participants along the severity continuum with precision.