Hereditary Angioedema Prophylaxis: Lanadelumab and Garadacimab Outcomes Look Nearly Identical, Letter Argues

A newly published letter to the editor in the journal Drugs in R&D is challenging the conclusions of a recent network meta-analysis that ranked the monoclonal antibody garadacimab as the most probable effective long-term prophylactic treatment for hereditary angioedema. Writing in the open-access letter, Irmgard Andresen, Natalie Khutoryansky, Sarah L. Feeny and Alissa Dangel argue that treatment outcomes with lanadelumab given every two weeks and garadacimab are in fact very similar, and that several methodological problems undermine the meta-analysis, particularly its comparison involving lanadelumab dosed every four weeks.

The letter responds to a publication titled “Network meta-analysis of pharmacological therapies for long-term prophylactic treatment of patients with hereditary angioedema” by Walsh and colleagues. That analysis used a Bayesian network meta-analysis, a statistical technique that allows indirect comparisons between treatments that have never been tested against each other in a head-to-head clinical trial. The treatments compared were garadacimab at 200 mg monthly, lanadelumab at 300 mg every two weeks and every four weeks, subcutaneous C1-inhibitor at 60 IU/kg twice weekly, and berotralstat at 150 mg daily. In the absence of direct head-to-head trials, such indirect comparisons are often used to compare treatment options, but the letter’s authors emphasize that inherent limitations of network meta-analyses must be considered before drawing clinical conclusions from them.

The most important limitation, according to the letter, concerns the source of the lanadelumab efficacy and safety data. These were drawn from the HELP Study, a phase III trial in which participants were randomly assigned at initiation to receive placebo, lanadelumab every two weeks, or lanadelumab every four weeks. However, the summary of product characteristics and label for lanadelumab requires all patients to begin treatment on the every-two-week regimen, with extension of the dosing interval to every four weeks considered appropriate only for patients who are stable or well controlled, for example attack-free, on the more frequent schedule. Because participants in the HELP Study did not receive this clinical-practice regimen of starting every two weeks and then extending the interval once stable, the letter argues that the trial’s data on every-four-week lanadelumab do not reflect the effectiveness of that regimen in real-world practice, and that the comparison between garadacimab and every-four-week lanadelumab in the meta-analysis should therefore not be incorporated into treatment decision-making.

A second concern relates to how the garadacimab evidence base was assembled. Data from a 12-week phase II study were combined with data from the six-month phase III VANGUARD study, and this combined garadacimab dataset was then compared with data from the six-month phase III HELP Study of lanadelumab. The letter’s authors point out that combining 12-week and six-month garadacimab data and comparing the result with six-month lanadelumab data favors garadacimab for non-time-normalized endpoints, such as the proportion of patients who remain attack-free and the occurrence of adverse events. The reasoning is straightforward: the longer a patient remains in a study, the more likely they are to experience an event, so pooling a shorter trial with a longer one and comparing against a longer trial alone distorts such endpoints.

The letter also raises a fundamental statistical objection to the modeling choices made by Walsh and colleagues. The meta-analysis authors conducted both random-effects and fixed-effects models but based their conclusions about relative treatment effects solely on the fixed-effects model. In the random-effects model results, which were presented in the supplementary materials of the original analysis, there were no significant differences for any endpoints between garadacimab and lanadelumab at either the every-two-week or every-four-week dosing interval. The fixed-effects model, which the original authors selected given the sparsity of the treatment network, assumes a single true effect across all studies. The letter argues this assumption is implausible in studies of patients with hereditary angioedema, given the heterogeneity of study designs, endpoint definitions and dosing regimens. Notably, the letter states that the original authors did not discuss their own model-fit comparison, which showed that the random-effects model better fitted the data, with meaningfully lower deviance information criteria and residual deviance. The letter’s authors describe this finding as unsurprising, citing the widely referenced work of Borenstein and colleagues, who explain why in the vast majority of meta-analyses the random-effects model would be the more appropriate choice.

Even within the fixed-effects model, the only significant difference between garadacimab and lanadelumab every two weeks was in the time-normalized number of moderate or severe attacks. Yet the letter highlights that definitions of attack severity were not the same across the two pivotal trials. In the HELP Study of lanadelumab, a mild attack was defined as one that required no medical intervention, whereas in the VANGUARD study of garadacimab, attacks that required treatment with medications such as analgesics could still be categorized as mild. The same attack profile could therefore have been classified as moderate severity in HELP but mild in VANGUARD. According to the letter, this inconsistency undermines the core network meta-analysis assumption of consistency and precludes meaningful comparison between the studies.

The letter further questions the clinical relevance of the treatment rankings reported in the meta-analysis. The original authors concluded that garadacimab ranked as the most probable effective treatment among all comparators assessed, with lanadelumab every two weeks or subcutaneous C1-inhibitor ranking second across most outcomes, based on the probability of being best and on Surface Under the Cumulative Ranking curves. The letter counters that such rankings do not consider the absolute magnitude or clinical relevance of the probable difference between treatments. As an example, the original authors calculated a probability of being best of 73 percent for garadacimab on the primary endpoint of the time-normalized number of attacks, but the 95 percent credible interval for the effect of garadacimab relative to lanadelumab, spanning 0.58 to 1.32, shows that the difference between the two treatments is not statistically significant. Moreover, the mean monthly attack rates in the two pivotal trials were very similar: 0.27, with a 95 percent confidence interval of 0.05 to 0.49, for garadacimab in VANGUARD, and 0.26, with a 95 percent confidence interval of 0.14 to 0.46, for lanadelumab every two weeks in HELP. The letter argues that these nearly identical rates call into question the clinical relevance of any differences in the ranking probabilities.

Reporting standards form a final layer of criticism. The letter states that while the original authors claimed adherence to the PRISMA-NMA standards, several requirements were not met. The analysis did not provide simple summary data for each intervention group and treatment outcome, data that the letter says would have demonstrated the similarity of clinical outcomes between lanadelumab every two weeks in HELP and garadacimab in VANGUARD. Nor did the authors provide information on the public availability of datasets, statistical code or model specifications required by the PRISMA 2020 statement, preventing independent validation of the reported findings. The letter also notes that Bayesian methods, as applied in the analysis, have been criticized for allowing subjectivity, because results can depend on the choice of prior distributions. Best practice is to pre-specify the priors used for key model parameters such as treatment effects, especially with small sample sizes where priors can exert a stronger influence, but the original authors did not report their choice of prior distributions, preventing any assessment of how the results might have been affected by this aspect of the modeling.

The letter closes by acknowledging the original authors’ own observation that multiple sources of evidence, including trial data, patient preferences and clinician expertise, must be considered alongside comparative efficacy estimates from indirect treatment comparisons. It adds that lanadelumab has been available since 2018 and that a wealth of longer-term effectiveness data exists from single-arm real-world studies, data that could not be included using the methodology and narrow parameters selected by Walsh and colleagues. Taken together, the letter’s authors conclude that the limitations they identified raise concerns about the appropriateness, validity and credibility of the conclusions drawn regarding different long-term prophylaxis treatments for hereditary angioedema, specifically the comparison between lanadelumab and garadacimab, and particularly the comparison involving lanadelumab dosed every four weeks. As a letter to the editor, the publication presents a critical re-examination of a published analysis rather than new clinical trial findings, and readers should weigh both the original meta-analysis and this critique when considering the comparative evidence for these prophylactic therapies.

Subject of Research: Methodological critique of a network meta-analysis comparing lanadelumab and garadacimab for hereditary angioedema prophylaxis

Article Title: Treatment Outcomes with Lanadelumab Every 2 Weeks and Garadacimab are Very Similar in Patients with Hereditary Angioedema

Article References: Andresen, I., Khutoryansky, N., Feeny, S. L., & Dangel, A. (2026). Treatment Outcomes with Lanadelumab Every 2 Weeks and Garadacimab are Very Similar in Patients with Hereditary Angioedema. Drugs in R&D, 26(1), 31-33. https://doi.org/10.1007/s40268-026-00539-8

Image Credits: AI Generated

DOI: 10.1007/s40268-026-00539-8

Keywords: hereditary angioedema, lanadelumab, garadacimab, network meta-analysis, prophylaxis, HELP Study, VANGUARD, fixed-effects model, random-effects model, PRISMA-NMA, Bayesian methods, Drugs in R&D