Phenotypic clustering in tuberous sclerosis complex reveals four distinct disease trajectories

Tuberous sclerosis complex (TSC) is a phenotypically heterogeneous autosomal dominant epilepsy, neuropsychiatric and tumoural predisposition disease, occurring because of germline variants in the TSC1 or TSC2 genes. Despite an improving understanding of the varied phenotypes with which TSC can present, there remains an incomplete understanding of the disease trajectory and the genotype-phenotype relationship in this disorder. We sought to examine whether an unbiased clustering approach could uncover subgroups of disease trajectories in TSC and enhance understanding of the genotype-phenotype correlation. In this observational, prospective, multicentre natural history cohort of patients with confirmed diagnosis of TSC (TSC Alliance natural history database), data collected from 2006-2022 were used to identify groups of co-occurring phenotypes. This was a multicentre study involving 18 TSC clinical network centres in the USA. Nine hundred and forty-seven individuals were included, all of whom had a clinical diagnosis of tuberous sclerosis complex. Each patient was required to have complete characterization of 29 phenotype features associated with TSC. The primary outcomes were consensus clusters of clinical features defining subgroups of patients with TSC and their association with genotype. Nine hundred and forty-seven individuals (50% male) across the TSC Alliance natural history database were included in this study, and 29 clinical features were used to define clusters of phenotypes to define disease trajectories. Four reproducible and distinct disease subgroups were identified: angiomyolipoma-predominant TSC (cluster 1), TSC with infantile spasms (cluster 2), neuropsychiatric TSC (cluster 3) and a milder phenotype of TSC (cluster 4). Variants in the Rho domain of hamartin and the TSC1 binding domain of tuberin preferentially associated with cluster 1, with increased likelihood of angiomyolipomas, dermatological findings and subependymal giant cell astrocytoma. Four distinct disease subgroups exist in TSC and associate differentially with variant location, informing deep genotype-phenotype correlation in TSC with potential impact for personalizing disease surveillance, treatment and clinical trial end-point choice. Additional prospective data are needed to confirm these findings.