Matching-adjusted indirect comparison between garadacimab and lanadelumab for the long-term prophylactic treatment of patients with hereditary angioedema

Garadacimab and lanadelumab, two monoclonal antibodies targeting respectively activated FXII [6] and kallikrein [10] in the pathophysiology of HAE, have demonstrated favorable outcomes as long-acting prophylactic treatments for HAE [6,10,23], but have not been compared in any head-to-head trials. To fill this knowledge gap, the present study conducted MAICs to estimate the comparative efficacy of garadacimab 200 QM (using IPD from VANGUARD and the phase II trial) and lanadelumab 300 Q2W/Q4W (using SLD from HELP).

In comparison to lanadelumab 300 Q2W, garadacimab 200 QM reduced the rate of time-normalized number of moderate and/or severe HAE attacks by 75% and improved AE-QoL scores from baseline to day 182 by an average of 17 points; both of these findings were statistically significant. Furthermore, in the comparison to lanadelumab 300 Q4W, garadacimab 200 QM reduced both the rate of time-normalized number of HAE attacks and HAE attacks requiring on-demand treatment by 71%, reduced the rate of time-normalized number of moderate and/or severe HAE attacks by 85%, and improved AE-QoL scores from baseline by an average of 21 points; all of these findings were statistically significant. Additionally, patients receiving garadacimab 200 QM were three-times more likely to be attack-free than patients receiving lanadelumab 300 Q4W, and this finding was statistically significant. Garadacimab also showed improvements of the remaining outcomes in comparison to lanadelumab 300 Q2W, but these results were not statistically significant.

Both garadacimab and lanadelumab inhibit bradykinin production, but they have different molecular targets in the signaling cascade. While garadacimab targets activated factor XIIa, lanadelumab inhibits a downstream target kallikrein, both of which reduce the bradykinin formation characteristic of HAE [7]. Given that the present indirect treatment comparisons showed garadacimab improved outcomes or provided at least similar benefit compared with two regimens of lanadelumab, it is possible that the inhibition of a mediator earlier in the pathway may more effectively improve outcomes for LTP in patients with HAE.

Another possible reason for the observed results is the difference in treatment schedules between trials. In the VANGUARD and phase II trials evaluating garadacimab, patients were given a loading dose (400 mg in VANGUARD, and 100 mg in phase II) once in the first month of treatment with subsequent doses of 200 mg every month for the remainder of the study duration. The HELP trial evaluating lanadelumab did not have a loading dose schedule. Although it is unclear what the impact of this loading dose has on these results, given the relatively short duration of the trials – both 26 weeks, we can speculate that the impact of the garadacimab loading dose may have influenced the beneficial outcomes observed in our analysis via increased bioavailability or through some other biological interaction of pharmacodynamic/pharmacokinetic properties leading patients to reach the therapeutic window sooner, which was then maintained by subsequent doses. While MAICs can adjust for population-level differences, inherent residual limitations to study designs may persist.

The impact of HAE on health-related QoL and patient burden should be considered when a patient is presented with LTP treatment options. Validated patient-reported outcome measures, such as the AE-QoL, are recommended to monitor disease activity, impact, and control for LTP in patients with HAE [29]. In both VANGUARD and HELP trials, AE-QoL was assessed prior to treatment administration. However, AE-QoL was assessed more frequently in VANGUARD (every month for 6 months) than in HELP (only at baseline, day 98, day 182 and follow-up). These differences in assessment frequency cannot be addressed by the present analysis, but monthly garadacimab showed statistically significant favorable change in AE-QoL score when compared with both Q2W and Q4W doses of lanadelumab. Therefore, these consistent findings suggest that the impact of assessment frequency is minimal on QoL in patients with HAE. Additionally, a minimal clinically important difference in AE-QoL score of at least 6 points for patients to notice a change has been established [30]. In this analysis, garadacimab was shown to improve AE-QoL by 17 points when compared with lanadelumab Q2W and by 21 points when compared with Q4W, suggesting that the QoL improvement with garadacimab is both statistically significant and clinically meaningful. In terms of treatment burden, patient compliance in LTP is paramount in achieving treatment success, and this can be affected by dosing schedules. While patient burden was not assessed in this analysis, in cases where the clinical efficacy between treatments is similar, a patient may prefer infrequent dosing to improve compliance and lessen the treatment burden. For this reason, once monthly dosing may be preferrable to biweekly dosing.

The comparative efficacy of lanadelumab has been previously assessed in HAE via two network meta-analyses (NMAs). One of which investigated lanadelumab versus berotralstat versus intravenous C1-esterase inhibitor [31], and another that investigated lanadelumab versus intravenous C1-esterase inhibitor [32]. Both NMAs found lanadelumab to be favored over the comparator in terms of efficacy (e.g., reduction in HAE attack rate, achieving ≥90% reduction in monthly HAE attacks, and extending attack-free intervals) [31,32]. Since garadacimab was not included in the aforementioned published NMAs, the present study is the first indirect treatment comparison analysis utilizing IPD in the form of MAICs to indirectly compare garadacimab and lanadelumab for the LTP treatment of HAE, and it is the first indirect treatment comparison to evaluate lanadelumab's comparative impact on QoL. Additionally, a prospective study was conducted in Hong Kong that investigated the efficacy, safety, and QoL of LTP in HAE using garadacimab or lanadelumab, but only aggregate outcomes were reported and the two treatments were not compared [33].

Several strengths of this study should be noted. First, analyses were performed according to best practice for conducting and reporting MAICs following the NICE DSU TSD 18 to ensure transparency and reproducibility [15]. Further, a rigorous feasibility assessment was performed to highlight any sources of inter-trial heterogeneity between the included studies (Supplementary Appendix B). One such source of heterogeneity was variation in trial duration/follow-up time across the included studies. While VANGUARD and HELP had a similar follow-up time of 26 weeks, the phase II trial for garadacimab had a notably shorter follow-up time of 12 weeks. Since patients are more likely to experience fewer events for the binary endpoint included (proportion of attack-free patients) simulated IPD was generated for HELP to account for this difference.

Additionally, since placebo was a common comparator in VANGUARD, the phase II trial for garadacimab, and HELP, anchored analyses were conducted for all outcomes, except proportion of attack-free patients due to zero events which necessitated unanchored analysis. As such, the MAICs only needed to adjust for differences in TEMs across trials. Scenario analyses were also conducted to explore how the results and ESS were impacted by incrementally adjusting for all TEMs in order of importance, as indicated by clinical experts. The base case MAIC analyses for each outcome adjusted for all feasible clinically important TEMs to ensure trial populations were aligned as closely as possible. The results across unadjusted and adjusted scenarios analyses showed similar relative efficacy of garadacimab 200 QM to lanadelumab 300 Q2W or lanadelumab 300 Q4W. This demonstrates the robustness of results and reinforces the findings from the base case analyses in this study.

This study was not without limitations. The language of the SLR was restricted to include English-only articles at the study selection stage, which may have led to the exclusion of relevant documents in other languages. The literature review results were validated by clinical experts in an attempt to mitigate the language bias. Given that most of the key studies identified were published in English journals, it is likely that this was a minor limitation. However, it should be noted that this restriction was not applied to the search strategy.

Regarding the analysis, one limitation was the inability to fully match patient populations according to eligibility criteria across trials. Despite efforts made to align populations, VANGUARD and, especially the phase II trial for garadacimab, were narrower for some eligibility criteria compared with HELP resulting in unresolved heterogeneity between trials. Key sources of heterogeneity included trial phase (the phase II trial for garadacimab differed from VANGUARD and HELP which were phase III trials), treatment period duration (the phase II trial had a shorter treatment period at 12 weeks than VANGUARD and HELP at 26 weeks), dosing schedules (every 4 weeks in the phase II trial for garadacimab, once monthly in VANGUARD, and every 2 or 4 weeks in HELP), and trial eligibility criteria (e.g., age and baseline HAE attack rate; the phase II trial for garadacimab enrolled patients ≥18 years old whereas VANGUARD and HELP enrolled patients aged ≥12 years old, and minimum baseline HAE attack rates were ≥2 attacks per month in the phase II trial for garadacimab and ≥1 attack per month in VANGUARD and HELP). Additionally, there is inherent heterogeneity in trial designs that was unable to be accommodated by the MAIC. These differences include trial phases (VANGUARD and HELP were phase III), study sponsors (VANGUARD and phase II garadacimab trials were sponsored by CSL Behring; HELP was sponsored by Shire), study locations (all trials enrolled patients from Canada, Germany and the US; VANGUARD enrolled patients from Hungary, Israel and the Netherlands; the phase II garadacimab trial enrolled patients from Israel; and HELP enrolled patients from Italy, Jordan, Puerto Rico and the UK), and the use of loading doses (present in the VANGUARD and phase II garadacimab trials; absent in the HELP trial). Although MAICs can adjust for differences in TEMs, the uncertainty in the analysis estimates must be considered in the context of these remaining differences.

Another limitation was related to data availability of all TEMs. Anchored MAIC methodology assumes that all TEMs are balanced across trials; however, these analyses could only account for characteristics reported in the comparator publications. Therefore, it is possible that covariates not considered in this analysis could have impacted the outcomes of interest. For example, history of anxiety and history of depression were respectively ranked as the third and fourth most important covariates to adjust for by clinical experts; however, these characteristics were not reported by HELP and could not be adjusted for in these analyses. Another limitation was related to low event numbers in the outcome of proportion of attack-free patients. Since zero subjects achieved attack-free status in the placebo arms of VANGUARD and the phase II trial for garadacimab, it was inappropriate to conduct an anchored MAIC for this outcome. Instead, an unanchored MAIC was performed. Unanchored MAICs rely on the assumption that both prognostic factors and TEMs are balanced across trials; however, only one subject achieved an attack-free status in the placebo arm of HELP which suggests that prognostic differences across trials may be minimal. Lastly, it should be acknowledged that some TEMs may not be accounted for in the MAICs of QoL due to reporting limitations. However, given the importance of improving QoL in aiding clinical decision-making in HAE, it was of interest for these analyses.

This study has practical implications in the treatment of patients with HAE using LTP. While indirect treatment comparisons are not replacements for direct head-to-head evidence of comparative efficacy, these analyses serve as a tool to support clinical, regulatory and reimbursement decision-making. Future directions in the LTP for HAE disease space should involve indirect analyses of safety outcomes to provide a holistic understanding of treatment benefit. Patients with HAE vary in demographics and disease-related factors, and thus population-adjusted analyses like MAICs provide valuable insight in the absence of direct evidence. This is particularly valuable in regions where modern diagnostics and therapies are limited; as international guidelines for HAE are regularly revised and updated, this analysis may be used to aid in developing a global standard for LTP in patients with HAE [29].

The results of these MAICs show improved efficacy and QoL with garadacimab compared with lanadelumab in the long-term prophylactic treatment of patients with HAE. Compared with lanadelumab Q2W, garadacimab statistically significantly reduced the rate of time-normalized number of moderate and/or severe HAE attacks by at least 75% and significantly improved QoL as assessed by a change from baseline in AE-QoL total score compared with lanadelumab Q2W. Compared with lanadelumab Q4W, treatment with garadacimab provided statistically significant improvements in all clinical outcomes assessed, which includes number of HAE attacks, proportion of attack-free patients, number of HAE attacks requiring on-demand treatment, number of moderate and/or severe HAE attacks and change from baseline in total AE-QoL score. Overall, these results suggest that garadacimab may provide improved therapeutic benefit compared with lanadelumab in the long-term prophylactic treatment of patients with HAE.

All authors participated in the conception and design of the study. S Walsh, A Haltner, M Bartlett and IA Samjoo contributed to the collection and analysis of the data. All authors contributed to the interpretation of the data and critically reviewed for importance of intellectual content for the work. All authors were responsible for drafting or reviewing the manuscript and for providing final approval. All named authors meet the International Committee of Medical Journal Editors (ICMJE) criteria for authorship for this article, take responsibility for the integrity of the work and have given their approval for this version to be published.

The authors acknowledge Joanna Bielecki who developed, conducted and documented the database searches; and Elizabeth Halloran, Kerise Clarke and Amrita Debnath for their assistance with the literature review. All are employed by EVERSANA, Canada.

This work was supported by CSL Behring. The authors have received no other financial and/or material support for this research or the creation of this work apart from that disclosed.

J Sears, Y Li, I Jacobs and N Ray are employees of CSL Behring, USA. M Kelly is an employee of CSL Behring Ltd. S Gavata-Steiger and C Nenci are employees of CSL Behring AG. I Pragst is an employee of CSL Innovation GmbH, Germany. S Walsh, A Haltner, M Bartlett and IA Samjoo are employees of EVERSANA, Canada and USA, which was a paid consultant to CSL Behring in connection with the development of this manuscript. The authors have no other competing interests or relevant affiliations with any organization or entity with the subject matter or materials discussed in the manuscript apart from those disclosed.

Medical writing and editorial assistance were provided by S Walsh, A Haltner, M Bartlett and IA Samjoo from EVERSANA, Canada and USA. Medical writing and editorial assistance were funded by CSL Behring Ltd.

This work is licensed under the Attribution-NonCommercial-NoDerivatives 4.0 Unported License. To view a copy of this license, visit https://creativecommons.org/licenses/by-nc-nd/4.0/