A terse but pointed letter published in the journal Advances in Therapy is challenging one of the more striking claims to emerge recently in the treatment of hereditary angioedema: that intravenous recombinant human C1 esterase inhibitor is roughly four and a half times more likely than the oral drug sebetralstat to resolve an acute attack completely. The letter, written by Noemi Hummel and Agnieszka Kopiec, does not dispute the underlying data, but argues that the statistical machinery used to generate that headline number rests on assumptions so fragile that the result should be read with considerable caution.
Because no head-to-head trial has ever pitted the two treatments against each other, the researchers turned to indirect comparison, a technique that borrows strength from separate clinical trials.
Indirect comparisons come in two broad flavors, and the distinction matters enormously here. Anchored comparisons preserve the randomization of the original trials by linking them through a shared comparator, typically placebo, which protects against hidden biases. Unanchored comparisons, by contrast, simply match patients from one trial against patients from another, and their validity depends entirely on the assumption that every variable capable of modifying treatment effect, and every variable that influences outcomes regardless of treatment, has been identified, measured and adjusted for. That assumption, the letter’s authors note, is inherently difficult to verify and becomes increasingly strained when the trials being compared differ substantially from one another.
And differ they did. In the recombinant C1 inhibitor trials, every single patient experienced a severe or very severe attack, compared with just 17.5 percent of patients in KONFIDENT. The timing of treatment also diverged sharply: patients in the C1 inhibitor trials waited a median of 242 to 267 minutes from attack onset before receiving therapy, largely because they had to travel to a clinic within five hours for an intravenous infusion, whereas participants in KONFIDENT were instructed to self-administer sebetralstat immediately, resulting in median times of just 41 to 51 minutes. Both attack severity and time to treatment are recognized as factors that influence treatment outcomes, yet the primary analysis adjusted only for attack location and the use of long-term prophylaxis, leaving these two powerful variables unmatched.
The letter also raises a subtler problem with the primary endpoint itself. Time to complete attack resolution within 24 hours was measured directly in KONFIDENT using a patient-reported severity instrument, but that instrument was never collected in the recombinant C1 inhibitor trials. To bridge the gap, the analysts reconstructed the endpoint from visual analogue scale scores, mapping them onto severity categories using an algorithm informed by patient and clinician input. Such harmonization may be unavoidable when trials measure outcomes differently, but it means the pivotal endpoint was never directly observed in one of the two treatment programs, and the result depends wholly on the validity and transportability of the mapping algorithm. Intriguingly, the letter points out that both trial programs did use comparable seven-point scales for their respective primary endpoints of time to symptom relief, which could have facilitated a more straightforward translation between studies.
Perhaps the most striking evidence of instability comes from the spread of results across the analytical approaches the original authors themselves employed. The anchored matching-adjusted comparison of time to complete resolution produced a hazard ratio of 1.41, with a confidence interval stretching from 0.63 to 3.18, a range that includes the possibility of no difference at all. The primary unanchored analysis yielded a hazard ratio of 4.52, and a Bucher indirect comparison pushed the estimate to 9.76, with a confidence interval of 4.94 to 19.29. Three defensible-looking analyses of the same data thus produced estimates ranging from a modest and statistically uncertain advantage to a nearly tenfold effect, a variability the letter argues should itself temper any confident conclusion.
The redosing analysis, the letter contends, has problems of its own. It was limited to the C1 1310 trial and KONFIDENT, and while an anchored version used the placebo arms as a common comparator, those placebo arms were not equivalent. In C1 1310, patients randomized to placebo who needed redosing actually received the active C1 inhibitor, whereas placebo patients in KONFIDENT who required redosing simply received more placebo. The supposedly untreated anchor arm was therefore not composed exclusively of untreated patients. Beyond that, redosing behavior reflects more than clinical response: the willingness to take a second dose is likely lower when it requires an intravenous infusion than when it means swallowing another tablet, so administration route and treatment burden can shape the outcome independently of efficacy.
The Bucher analysis of redosing drew further criticism. By replacing the placebo data from C1 1310 with placebo data from KONFIDENT, the analysis broke the internal randomization logic of the original trial and potentially introduced confounders. The Bucher method assumes that treatment effects are exchangeable across the trials connected through the common comparator, but attack severity and treatment timing differed between the two programs, and those variables are believed to modify treatment effects. Moreover, because KONFIDENT placebo patients who needed redosing continued receiving placebo, they fared worse than their counterparts in C1 1310, a difference in protocol design that the letter argues may have artificially inflated the apparent advantage of recombinant C1 inhibitor.
The subgroup analysis of patients with severe or very severe attacks attracted a separate set of objections. Despite the substantial differences between the trials, the authors drew their conclusions from an unanchored comparison within this subgroup, and for that analysis no survival curve was reported; the calculation relied on the reported median time to complete resolution and assumed a constant event rate over time, an assumption that could not be verified for sebetralstat and that allows results to deviate in any direction. The letter adds that the reported confidence interval captures only the uncertainty arising from sample size, not the uncertainty introduced by the methodological assumptions themselves, giving the result a false appearance of precision.
Hummel and Kopiec are careful not to dismiss indirect comparisons altogether. In the absence of head-to-head trials, they write, such analyses can provide genuinely useful information for clinicians, payers and health technology assessment bodies, particularly in a rare disease where direct comparative evidence may never be generated. Their objection is narrower and more pointed: the primary conclusion that complete symptom resolution is 4.5 times more likely with recombinant C1 inhibitor than with sebetralstat 600 mg rests on an unanchored analysis that matched only two variables, ignored attack severity and treatment timing despite their recognized influence on outcomes, and depended on a derived endpoint never directly measured in one of the treatment programs. In their opinion, the methodological limitations of the comparison merit contextualizing its findings, a reminder that in rare-disease research, the elegance of a statistical number can outrun the strength of the evidence beneath it.
Subject of Research: Methodological critique of a matching-adjusted indirect treatment comparison of recombinant C1 inhibitor versus sebetralstat for on-demand treatment of hereditary angioedema attacks
Article Title: Letter to the Editor Regarding “Comparative Efficacy of Recombinant C1 Inhibitor Versus Sebetralstat for On-Demand Treatment of Hereditary Angioedema Attacks: A Matching-Adjusted Indirect Treatment Comparison”
Article References: Hummel, N., & Kopiec, A. (2026). Letter to the Editor Regarding “Comparative Efficacy of Recombinant C1 Inhibitor Versus Sebetralstat for On-Demand Treatment of Hereditary Angioedema Attacks: A Matching-Adjusted Indirect Treatment Comparison”. Advances in Therapy. https://doi.org/10.1007/s12325-026-03836-5
Image Credits: AI Generated
DOI: 10.1007/s12325-026-03836-5
Keywords: hereditary angioedema, recombinant C1 inhibitor, sebetralstat, indirect treatment comparison, matching-adjusted indirect comparison, KONFIDENT trial, methodology, on-demand treatment, hazard ratio, redosing, clinical trials, Advances in Therapy
News Source: Ophelia Keating. (October 28, 2026). Researchers Challenge Claim That C1 Inhibitor Beats Sebetralstat in Hereditary Angioedema. Scienmag.
