Due to the limited amount of information, modeling longitudinal rare-disease data can benefit from integrating clinical knowledge. Yet, elicitation of expert knowledge and formalization for model fitting is challenging, in particular due to limited time of clinical experts. To nevertheless make domain knowledge accessible during model fitting, we use large language models (LLMs) as synthetic clinical experts to supervise a variational-autoencoder-based approach that learns low-dimensional latent summaries of visit-level observations. Specifically, LLMs are queried offline on textual descriptions of patient observations to obtain judgments, e.g., the suspected clinical category. To improve the variational autoencoder fit, we train a differentiable surrogate model on these judgments and augment the loss function to encourage reconstructions that preserve the clinical-label distribution of their corresponding input profile. In an application to longitudinal motor-function assessments from children with spinal muscular atrophy, we map visit-level clinical profiles to low-dimensional representations that are linked by a multivariate mixed-effects model. The synthetic expert loss discourages reconstructions that remain numerically close in data space but alter the clinical interpretation of the reconstructed motor function profile, such as by crossing a disease-type boundary. We thus reduced disagreement between original and reconstructed SMA type labels from about 11 to 7 percent. Furthermore, informing the latent representation by the synthetic expert improved prediction of motor function milestones compared with unsupervised latent representations and a data-level baseline. These results suggest that incorporating LLMs into model fitting can make clinical knowledge available to representation learning and improve clinical faithfulness for longitudinal rare-disease data.

ObjectivePLOD1-related kyphoscoliotic Ehlers-Danlos syndrome (kEDS) is a rare autosomal recessive connective tissue disorder characterized by generalized joint laxity, severe congenital hypotonia, progressive kyphoscoliosis, hyperextensible and easily bruised skin, ocular abnormalities, and significant vascular complications.MethodsWe report on nine patients from seven families, eight of them carrying the common homozygous duplication of exons 10-16 in PLOD1. Longitudinal clinical assessments included muscle ultrasound (n=6) and vascular ultrasound (n=4). Genetic diagnostics varied, with most patients ultimately undergoing trio exome or genome sequencing. Urine pyridinoline analysis was performed in seven of nine patients. A literature review and age-stratified recalculation of vascular incidence, including our cohort, were conducted.ResultsDiagnosis was challenging in five families, as the exon 10-16 duplication often escaped detection due to its high allele frequency. Seven patients were initially diagnosed with congenital myopathy. Muscle ultrasound revealed abnormalities in five of six sonographically examined cases. Severe vascular events included neonatal intracranial hemorrhage, fatal aortic aneurysm rupture at the age of 13 years, multiple aneurysms/dissections (ages 14-19 years), and mesenteric dissection at the age of 10 years. Four younger patients (aged 3-9 years have had no vascular complications to date. Urine pyridinoline analysis was abnormal in all tested cases.DiscussionPLOD1-related kEDS often presents with a phenotype of congenital myopathy, complicating genetic diagnosis and potentially leading to underdiagnosis - especially in cases where the common PLOD1 duplication may be missed by strict frequency filters in exome or genome sequencing data. Literature and our data indicate a vascular event incidence from childhood age of ∼25%.

Many rare diseases offer limited established treatment options, leading patients to switch therapies when new medications emerge. To analyze the impact of such treatment switches within the low sample size limitations of rare disease trials, it is important to use all available data sources. This, however, is complicated when the use of measurement instruments changes during the observation period, for example when instruments are adapted to specific age ranges. The resulting disjoint longitudinal data trajectories complicate the application of traditional modeling approaches like mixed-effects regression. We tackle this by mapping observations of each instrument to an aligned low-dimensional temporal trajectory, enabling longitudinal modeling across instruments. Specifically, we employ a set of variational autoencoder architectures to embed item values into a shared latent space for each time point. Temporal disease dynamics and treatment switch effects are then captured through a mixed-effects regression model applied to latent representations. To enable statistical inference, we present a novel statistical testing approach that accounts for the joint parameter estimation of mixed-effects regression and variational autoencoders. The methodology is applied to quantify the impact of treatment switches for patients with spinal muscular atrophy. Here, our approach aligns motor performance items from different measurement instruments for mixed-effects regression and maps estimated effects back to the observed item level to quantify the treatment switch effect. Our approach allows for model selection as well as for assessing effects of treatment switching. The results highlight the potential of modeling in joint latent representations for addressing small data challenges.

Real-world treatments for 5q-spinal muscular atrophy (SMA) have evolved rapidly following the sequential approval of three disease-modifying treatments (DMTs): nusinersen, onasemnogene abeparvovec and risdiplam. The aim of this study was to map the sequence and timing of SMA treatments accurately using the SMArtCARE registry, a disease-specific registry for patients with SMA across 84 participating centres in Germany, Austria and Switzerland. All patients registered in SMArtCARE were included in the analysis. Patients were grouped based on their treatment regimen: those who remained on the first DMT versus those who switched DMT. The impacts of clinical and genetic factors on treatment decisions were evaluated, including age at initiation of treatment, SMN2 copy number, motor function status, the need for ventilator support or tube feeding, and the presence of scoliosis. A total of 2140 patients were included. Of these, 1294 patients (60.5%) initiated treatment with nusinersen, 514 patients (24.0%) with risdiplam and 243 patients (11.4%) with onasemnogene abeparvovec. Overall, 1366 patients (63.8%) remained on the first DMT. Most treatment switches occurred shortly after approval of a new DMT. Notably, most patients who switched showed no change in motor milestone status between the start of the first and the second DMT. In this large real-world cohort, we present the first comprehensive analysis of SMA treatment patterns across all age groups and disease severities. Although most patients remained on the first DMT, switches were observed, mainly after DMT approvals. Decisions to switch appear multifactorial and are not related directly to motor function effectiveness.

BACKGROUND:Systematic protocols for long-term surveillance of children with spinal muscular atrophy treated with onasemnogene abeparvovec are lacking. Together with best practice recommendations for safety monitoring and management, such recommendations should increase patient safety and confidence levels of healthcare providers. OBJECTIVE:Based on systematic literature review and evidence grading from part 1, this initiative aims to develop a structured treatment plan applicable across all treatment centers in Germany, Austria and Switzerland. Additionally, it seeks to establish consensus recommendations for the clinical management of safety alerts. METHODS:Part 2 describes the methodology used to formulate Delphi consensus statements, the development of a structured treatment plan for OA treatment, and the anonymous consensus voting process with standardized follow-up in case of disagreement. RESULTS:A total of 12 consensus statements were developed, addressing diagnostic work-up, safety evaluation, and best practice management of common adverse drug reactions associated with OA gene therapy. All statements achieved >95% consensus in the anonymous Delphi voting. Additionally, two consensus recommendations for handling of positive newborn screening result achieved consensus of 97% and 87%, respectively. A structured treatment plan for gene therapy was consented with 100% agreement, as were standardized recommendations for laboratory testing and a consensus-based algorithm for management of liver transaminase elevations. CONCLUSIONS:Delphi-based expert recommendations, developed in co-creation with patient representatives, provide a framework to minimize complications associated with gene therapy and establish the basis for standardized post-marketing data collections. The methodology used in this Delphi-consensus-group can serve as a blueprint for future gene therapy approvals.

Background and objectivesThe severity of the phenotype of spinal muscular atrophy (SMA) is highly variable, yet little is known about the phenotypic variation among siblings. We systematically investigated the phenotypic variability of therapy-naïve 5q-SMA siblings leveraging a large multicentre cohort from the SMArtCARE registry.ResultsClinical information was available from 132 siblings of 65 families. There were 24 (18.2%) type 1, 38 (28.7%) type 2, 54 (40.9%) type 3 patients, and 16 (12.1%) presymptomatic individuals. In 17 families (32.1%), there was discordance in the type of SMA among symptomatic siblings. We found no influence of gender on discordance in SMA type among siblings (p = 0.528). The median age at disease onset within all sibships varied by 6 months (interquartile range (IQR) = 1-30). There was no correlation in age of onset among siblings (r = 0.405; p = 0.052). Among siblings who lost ambulation, the median interval between the start of wheelchair use was 12 months, but the maximal interval was 18 years. In one pair of siblings, one sibling lost the ability to walk at the age of 13, whereas the other sibling was still ambulatory at the age of 54. In 6 sibling pairs (9.5%), only one of both siblings had a history of scoliosis surgery. Analysing SMN2 copy numbers, in one sibling pair (1.8%) 1 SMN2 gene copy was detected, while 10 (17.5%) had 2 copies, 23 (40.4%) had 3 copies, and 17 (29.8%) had 4 copies. Concordance in SMN2 copy numbers across siblings was observed in 90% of families. With increasing SMN2 copy number, the median differences in age of onset among siblings increased without reaching statistical significance.ConclusionThis study reports considerable phenotypic variability in therapy-naïve SMA sibships that cannot solely be explained by differences in SMN2 copy numbers.

BACKGROUND:Approved spinal muscular atrophy therapies greatly improve clinical outcomes; however, substantial motor function deficits persist. Apitegromab, a fully human monoclonal antibody, selectively inhibits myostatin activation, improving muscle function. We aimed to assess the safety and efficacy of apitegromab in patients with nonambulatory type 2 or type 3 spinal muscular atrophy receiving nusinersen or risdiplam. METHODS:SAPPHIRE, a double-blind, placebo-controlled, phase 3 trial, was done in 48 hospitals in Belgium, France, Germany, Italy, Poland, Spain, the Netherlands, the UK, and the USA. Eligible participants were aged 2-21 years, had genetically documented SMN-deficient nonambulatory type 2 or type 3 spinal muscular atrophy, an estimated life expectancy greater than 2 years, Hammersmith Functional Motor Scale-Expanded (HFMSE) scores 10-45, and had received at least 10 months' nusinersen or at least 6 months' risdiplam therapy at screening. Participants aged 2-12 years were randomly assigned 1:1:1 to receive apitegromab 20 mg/kg, apitegromab 10 mg/kg, or placebo every 4 weeks; participants aged 13-21 years were randomly assigned 2:1 to receive apitegromab 20 mg/kg or placebo every 4 weeks. All participants, parents or caregivers, investigators, and site personnel were unaware of the treatment assignment. The primary endpoint, change from baseline in HFMSE at 12 months, was assessed in participants aged 2-12 years who received at least one dose of apitegromab or placebo and had at least one post-baseline evaluable HFMSE assessment (modified intention-to-treat set). Comparisons of the combined apitegromab dose (20 mg/kg and 10 mg/kg) versus placebo and the 20 mg/kg dose versus placebo were done with a mixed-effects model with repeated measurement. Safety was analysed in all participants who received at least one dose of apitegromab or placebo through evaluation of adverse events, physical examinations, vital signs and cardiac assessments, laboratory evaluations, and concomitant medications. SAPPHIRE is registered with ClinicalTrials.gov, NCT05156320, and is completed. FINDINGS:From March 28, 2022, to Sept 4, 2024, we enrolled 188 patients (156 in the population aged 2-12 years and 32 in the population aged 13-21 years); of whom 128 participants received apitegromab and 60 participants received placebo. At 12 months, least squares mean difference in HFMSE change from baseline was 1·8 (95% CI 0·30 to 3·32, p=0·019) points for participants aged 2-12 years receiving apitegromab versus placebo (least squares mean 0·6 vs -1·2). Least squares mean difference in HFMSE change from baseline was 1·4 (95% CI -0·34 to 3·13; p=0·11) for apitegromab 20 mg/kg versus placebo (least squares mean 0·2 vs -1·2). The incidence and severity of adverse events were similar between apitegromab and placebo, and consistent with spinal muscular atrophy and background spinal muscular atrophy therapy. The most frequently reported adverse events were pyrexia (apitegromab, 33 [26%] of 128 vs placebo, 17 [28%] of 60), nasopharyngitis (32 [25%] vs 14 [23%]), cough (30 [23%] vs 12 [20%]), vomiting (29 [23%] vs ten [17%]), upper respiratory tract infection (28 [22%] vs 18 [30%]), and headache (27 [21%] vs 12 [20%]). No patients discontinued due to adverse events. INTERPRETATION:Participants in the apitegromab treatment groups (combined 20 mg/kg and 10 mg/kg dose) achieved statistically significant improvements in motor function compared with placebo; however, the least squares mean difference was not significant between apitegromab 20 mg/kg and placebo. Overall, SAPPHIRE results build on findings from the phase 2 TOPAZ trial, showing improved motor function with a generally well tolerated safety profile, supporting the use of muscle-targeting therapy for spinal muscular atrophy. FUNDING:Scholar Rock.

In a longitudinal clinical registry, different measurement instruments might have been used for assessing individuals at different time points. To combine them, we investigate deep learning techniques for obtaining a joint latent representation, to which the items of different measurement instruments are mapped. This corresponds to domain adaptation, an established concept in computer science for image data. Using the proposed approach as an example, we evaluate the potential of domain adaptation in a longitudinal cohort setting with a rather small number of time points, motivated by an application with different motor function measurement instruments in a registry of spinal muscular atrophy (SMA) patients. There, we model trajectories in the latent representation by ordinary differential equations (ODEs), where person-specific ODE parameters are inferred from baseline characteristics. The goodness of fit and complexity of the ODE solutions then allows to judge the measurement instrument mappings. We subsequently explore how alignment can be improved by incorporating corresponding penalty terms into model fitting. To systematically investigate the effect of differences between measurement instruments, we consider several scenarios based on modified SMA data, including scenarios where a mapping should be feasible in principle and scenarios where no perfect mapping is available. While misalignment increases in more complex scenarios, some structure is still recovered, even if the availability of measurement instruments depends on patient state. A reasonable mapping is feasible also in the more complex real SMA dataset. These results indicate that domain adaptation might be more generally useful in statistical modeling for longitudinal registry data.

Hintergrund: In der physiotherapeutischen Behandlung von Patienten mit spinaler Muskelatrophie (SMA) sind dringend Anpassungen erforderlich, da durch die krankheitsmodifzierende Therapien neue Phänotypen entstehen. Ziel dieser Studie war es, das aktuelle physiotherapeutische Management von Patienten mit SMA innerhalb des SMArtCARE-Netzwerks zu erfassen. SMArtCARE ist ein krankheitsspezifisches Register für Patienten mit 5q-SMA mit 74 teilnehmenden Zentren in Deutschland, der Schweiz und Österreich.

Hintergrund: Durch die zeitliche Abfolge der Zulassungen von Nusinersen, Risdiplam und Onasemnogene-Abeparvovec (OA) hat sich die Therapielandschaft der SMA in den letzten Jahren weiter verändert. Ausgenommen der eingeschränkten Zulassung von OA, gibt es bislang wenig publizierte Daten dazu, welche Kriterien für die Therapiewahl oder einen Therapiewechsel herangezogen werden. SMArtCARE ist ein krankheitsspezifisches Register, das seit 2017 den Krankheitsverlauf von SMA Patienten erfasst.

Background and objectives: Disease-modifying treatments (DMT) have dramatically changed phenotypes in patients with spinal muscular atrophy (SMA). Because publications regarding standards of care were published before DMTs emerged, detailed recommendations and guidelines for physiotherapeutic management are still lacking. The objective of this study was to map the physiotherapeutic management of patients with SMA within the SMArtCARE network, a disease-specific registry for patients with 5q-SMA with 83 participating centers in Germany, Switzerland, and Austria.Methods: An online survey using a modified Delphi approach was conducted among physiotherapists with two questionnaire rounds between June 2022 and June 2023. Seven physiotherapeutic experts developed and revised the questionnaires focusing on the main topics of stretching, positioning, mobility and exercise, and chest physiotherapy. The second questionnaire was based on eight different case studies.Results: The second questionnaire was sent to 148 participants with a response rate of 28%. Most of the physiotherapists were well experienced in treating SMA patients. There was a strong consensus that home-based stretching should be used in pediatric patients with contractures regardless of their motor function. Muscle strengthening training was considered to be essential for all sitters and for walkers with moderate motor function restriction by a strong consensus. For all patients with respiratory involvement there was a consensus for prophylactic respiratory therapy.Conclusion: Our results describe the current physiotherapeutic management and recommendations within the SMArtCARE network. These findings highlight the need for an individualized approach, and the necessity of developing and adjusting existing guidelines.

Ordinary differential equations (ODEs) can provide mechanistic models of temporally local changes of processes, where parameters are often informed by external knowledge. While ODEs are popular in systems modeling, they are less established for statistical modeling of longitudinal cohort data, e.g., in a clinical setting. Yet, modeling of local changes could also be attractive for assessing the trajectory of an individual in a cohort in the immediate future given its current status, where ODE parameters could be informed by further characteristics of the individual. However, several hurdles so far limit such use of ODEs, as compared to regression-based function fitting approaches. The potentially higher level of noise in cohort data might be detrimental to ODEs, as the shape of the ODE solution heavily depends on the initial value. In addition, larger numbers of variables multiply such problems and might be difficult to handle for ODEs. To address this, we propose to use each observation in the course of time as the initial value to obtain multiple local ODE solutions and build a combined estimator of the underlying dynamics. Neural networks are used for obtaining a low-dimensional latent space for dynamic modeling from a potentially large number of variables, and for obtaining patient-specific ODE parameters from baseline variables. Simultaneous identification of dynamic models and of a latent space is enabled by recently developed differentiable programming techniques. We illustrate the proposed approach in an application with spinal muscular atrophy patients and a corresponding simulation study. In particular, modeling of local changes in health status at any point in time is contrasted to the interpretation of functions obtained from a global regression. This more generally highlights how different application settings might demand different modeling strategies.

Background Since the approval of onasemnogen abeparvovec (OA) for gene addition therapy in children with spinal muscular atrophy (SMA), there has been a considerable increase of evidence regarding its effectiveness and safety. Consequently, the previous recommendations needed to be revised. Objective The primary objective was to develop an evidence- and expert-based best practice protocol ensuring optimal patient safety and comprehensive support for affected families. The harmonization of treatment algorithms is expected to facilitate the collection of standardized real-world data, laying the foundation for future evidence-based adjustments. Methods A modified, two-part Delphi process was selected as a standardized methodology. Experts specializing in SMA from all 31 neuromuscular treatment centers within Germany, Austria and Switzerland, and patient advocacy groups participated in an industry-independent Delphi panel. Existing evidence concerning effectiveness, safety, and guidelines of OA was analyzed in a systematic literature followed by development of consensus statements regarding its effectiveness. Results Strong consensus was reached regarding the following statements on effectiveness: (1) OA gene addition therapy for SMA demonstrates a clear advantage compared to the natural progression of the disease. (2) Superiority of any of the three approved disease-modifying therapies has not been proven. (3) Earlier initiation of therapy with fewer symptoms and shorter disease duration leads to better outcomes. (4) There is no clinical evidence supporting the superiority of combining two treatments over monotherapy. Conclusions: The systematic literature analysis constitutes the basis for the subsequent part 2, which involves the generation of expert-based recommendations for the surveillance of SMA gene addition therapy.

Duchenne muscular dystrophy (DMD) is a rare genetic disorder that primarily affects males. Beginning in childhood, patients experience ambulatory loss, heart failure and need ventilation. Disease management has improved, however, DMD remains debilitating, and has no cure. The rarity of the disease makes research difficult, and German prevalence data are lacking. Cost and resource utilization estimations are based on small sample sizes or self-reported data, limiting generalizability and adds the potential for recall bias. With a retrospective study on a healthcare claims database, we adapted algorithms to identify DMD patients and categorized them by disease stages 1–4 (early ambulatory, late ambulatory, early non-ambulatory, late non-ambulatory) with increasing disease progression. We analyzed annual prevalence, burden of disease, healthcare resource utilization and direct medical care costs, by time under observation (patient year). From 2016 to 2021, we identified 134 patients for which we could determine a disease stage and determined an extrapolated prevalence of DMD in Germany between 14.85 (95

Background:Evidence for the efficacy of nusinersen in adults with 5q-associated spinal muscular atrophy (SMA) has been demonstrated up to a period of 16 months in relatively large cohorts but whereas patients reach a plateau over time is still to be demonstrated. We investigated the efficacy and safety of nusinersen in adults with SMA over 38 months, the longest time period to date in a large cohort of patients from multiple clinical sites. Methods:Our prospective, observational study included adult patients with SMA from Germany, Switzerland, and Austria (July 2017 to May 2022). All participants had genetically-confirmed, 5q-associated SMA and were treated with nusinersen according to the label. The total Hammersmith Functional Motor Scale Expanded (HFMSE) and Revised Upper Limb Module (RULM) scores, and 6-min walk test (6 MWT; metres), were recorded at baseline and 14, 26, and 38 months after treatment initiation, and pre and post values were compared. Adverse events were also recorded. Findings:Overall, 389 patients were screened for eligibility and 237 were included. There were significant increases in all outcome measures compared with baseline, including mean HFMSE scores at 14 months (mean difference 1.72 [95% CI 1.19-2.25]), 26 months (1.20 [95% CI 0.48-1.91]), and 38 months (1.52 [95% CI 0.74-2.30]); mean RULM scores at 14 months (mean difference 0.75 [95% CI 0.43-1.07]), 26 months (mean difference 0.65 [95% CI 0.27-1.03]), and 38 months (mean difference 0.72 [95% CI 0.25-1.18]), and 6 MWT at 14 months (mean difference 30.86 m [95% CI 18.34-43.38]), 26 months (mean difference 29.26 m [95% CI 14.87-43.65]), and 38 months (mean difference 32.20 m [95% CI 10.32-54.09]). No new safety signals were identified. Interpretation:Our prospective, observational, long-term (38 months) data provides further real-world evidence for the continuous efficacy and safety of nusinersen in a large proportion of adult patients with SMA. Funding:Financial support for the registry from Biogen, Novartis and Roche.

Giant axonal neuropathy (GAN) is a progressive neurodegenerative disease affecting the peripheral and central nervous system and is caused by bi-allelic variants in the GAN gene, leading to loss of functional gigaxonin protein. A treatment does not exist, but a first clinical trial using a gene therapy approach has recently been completed. Here, we conducted the first systematic study of GAN patients treated by German-speaking child neurologists. We collected clinical, genetic, and epidemiological data from a total of 15 patients representing one of the largest cohorts described thus far. Average age of patients was 11.7 years at inclusion. The most frequently reported symptoms (HPO coded) were gait disturbance and muscle weakness, abnormality of muscle size, and abnormal reflexes. In line with the frequency of homozygous variants, in five families, parents reported being at least distantly related. In 14 patients, diagnosis was confirmed by molecular genetic testing, revealing eight different GAN variants, four being reported as pathogenic in the literature. Proteomics of white blood cells derived from four patients was conducted to obtain unbiased insights into the underlying pathophysiology and revealed dysregulation of 111 proteins implicated in diverse biological processes. Of note, diverse of these proteins is known to be crucial for proper synaptic function and transmission and affection of intermediate filament organisation and proteolysis, which is in line with the known functions of gigaxonin.

Dystrophinopathies are the most common muscle diseases, especially in men. In women, on the other hand, a manifestation of Duchenne muscular dystrophy is rare due to X-chromosomal inheritance. We present two young girls with severe muscle weakness, muscular dystrophies, and creatine kinase (CK) levels exceeding 10,000 U/L. In the skeletal muscle tissues, dystrophin staining reaction showed mosaicism. The almost entirely skewed X-inactivation in both cases supported the possibility of a dystrophinopathy. Despite standard molecular diagnostics (including multiplex ligation-dependent probe amplification (MLPA) and next generation sequencing (NGS) gene panel sequencing), the genetic cause of the girls' conditions remained unknown. However, whole-genome sequencing revealed two reciprocal translocations between their X chromosomes and chromosome 5 and chromosome 19, respectively. In both cases, the breakpoints on the X chromosomes were located directly within the DMD gene (in introns 54 and 7, respectively) and were responsible for the patients' phenotypes. Additional techniques such as Sanger sequencing, conventional karyotyping and fluorescence in situ hybridization (FISH) confirmed the disruption of DMD gene in both patients through translocations. These findings underscore the importance of accurate clinical data combined with histopathological analysis in pinpointing the suspected underlying genetic disorder. Moreover, our study illustrates the viability of whole-genome sequencing as a time-saving and highly effective method for identifying genetic factors responsible for complex genetic constellations in Duchenne muscular dystrophy (DMD).