Exploratory research in immune TTP and other ultra‐rare diseases—Balancing statistical idealism with pragmatic discovery

We appreciate the thoughtful comments by Dr. Liu highlighting the methodological challenges that face rare disease research such as our prospective study in 39 immune thrombotic thrombocytopenic purpura (TTP) survivors investigating risk factors for silent cerebral infarction (SCI),1 which is associated with cognitive dysfunction in this population.2 Dr. Liu highlights that with such a small sample size, the study may be underpowered to detect clinically meaningful associations and that models including multiple covariates with few events are likely to be overfitted. Small sample sizes are often unavoidable in ultra-rare disease research (immune TTP has a prevalence of only 13 per million individuals).3, 4 In this setting, there is a need to balance statistical idealism with the practical necessity of generating evidence for underserved populations. Smaller cohorts are scientifically and ethically appropriate, as they enable timely investigation of disease characteristics, mechanisms and potential therapies while minimizing the burden on a highly vulnerable patient population.5 Nevertheless, we agree that these statistical challenges must be acknowledged and appropriately managed as far as possible. We have done our best to rationally and transparently select variables while investigating risk factors for SCI. The requested univariate analysis for all candidate variables is already shown in supplemental table S1. We additionally included prespecified variables chosen for biological plausibility, most notably the ‘total days of thrombocytopenia’, which served as a surrogate for the duration of active, acute TTP activity. The reason to pick this over caplacizumab use (which also reduces duration of thrombocytopenia) alone was that even patients who had received caplacizumab had prior TTP episodes treated without caplacizumab. Given the small sample size that may limit the ability to show statistical significance even in univariate analysis, we believe it is critical not to arbitrarily discard variables that may have significant biological relevance and are potentially modifiable (duration of thrombocytopenia is modifiable with the addition of caplacizumab to standard regimens). We have previously shown that lower levels of ADAMTS13 (A Disintegrin And Metalloproteinase with Thrombospondin Type 1 Motif, Member 13) in clinical remission are associated with ischaemic stroke.6 We could not evaluate this variable in our model since we did not have remission ADAMTS13 levels available for the entire period since diagnosis for most of these patients. Only six patients had complete availability of every 3-month ADAMTS13 measurement activity during their entire period of clinical remission; however, this is being remedied by continued enrolment of additional patients starting shortly after their recovery from their first episode of TTP. Larger studies are needed to address the limitations of our exploratory analysis. Rather than additional statistical manipulation of a small dataset, we have initiated prospective multicentre studies through IMPACT (Initiation of a cohort to define pathogenic Mechanisms, Precision diagnosis And Complications of Thrombotic Microangiopathies) consortium, a part of the US Rare Diseases Clinical Research Network,7 which will enrol a much larger cohort of immune TTP patients to longitudinally examine cerebrovascular health and allow rigorous investigation of risk factors, pathological mechanisms and biomarkers. We also welcome collaboration with the international TTP research community, which is essential to generate robust and generalizable real-world data to support trials of interventions to improve outcomes for TTP survivors worldwide.8 Shruti Chaturvedi: Conceptualization; writing – original draft; funding acquisition. Data sharing not applicable to this article as no datasets were generated or analysed during the current study.