Gepants for abortive treatment of migraine: A network meta‐analysis

Abstract Objectives To evaluate and compare the efficacy and safety of gepants for abortive treatment of migraine by network meta‐analysis. Materials & Methods Publications, which were randomized controlled trials (RCTs) about gepants for abortive treatment of migraine, were acquired from Pubmed and Cochrane Library. The literatures screening and quality assessment followed the Cochrane handbook. Review manager 5.3 and Addis v1.16.8 were utilized for data analyzing. Results Totally, 15 RCTs were included in the network meta‐analysis. The trials enrolled were with high quality. There are 7 treatments were analyzed: BI 44370 TA, MK‐3207, olcegepant, rimegepant, telcagepant, ubrogepant, and placebo. Of these trials, 11,118 patients and 10,917 patients were assigned to one of 7 treatments randomly for efficacy assessment and safety assessment, respectively. In meta‐analysis of direct comparisons, all gepants were superior to placebo in achieving pain freedom 2 hr postdose and only rimegepant and telcagepant were higher than placebo in incidence of any adverse events. In network meta‐analysis, the rank best 3 drugs were olcegepant, BI 44370 TA, and MK‐3207 for efficacy outcomes. And the rank best 3 drugs were BI 44370 TA, placebo, and ubrogepant for safety outcomes. Conclusion Gepants were effective for abortive treatment of migraine. The most effective treatment of gepants for migraine might be olcegepant which were administrated transvenously. And all of gepants were safe for migraine treatment with single dose.

abortive treatment and preventive treatment (Lambru, Andreou, Guglielmetti, & Martelletti, 2018). The most widely prescribed abortive treatment of migraine is triptans, which are the serotonin 5-HT receptor agonists (Leroux & Rothrock, 2019). But triptans are not always effective for abortive treatment of migraine and with a high incidence of adverse events (Leroux and Rothrock, 2019). And the most serious adverse events are cardiovascular effects (Leroux & Rothrock, 2019).

| Data selection
Database including Pubmed and Cochrane Library were queried using the following terms: migraine disorders, migraine without aura, migraine with aura, calcitonin gene-related peptide, receptors, calcitonin gene-related peptide, and calcitonin gene-related peptide receptor antagonists. The searching results were filtered by a clinical trial. The cutoff date was December 15, 2019.
According to the PICO principle, the publications of RCTs published in English and matching the following criteria were enrolled: (a) the participants are diagnosed with migraine, (b) the interventions were gepants for an acute attack of migraine, (c) the comparisons were other gepants or placebo.

| Data extraction and analysis
The procedure of data extraction and analysis was published in our previous publications (Hong & Liu, 2016). In brief, the assessing of risk of bias was followed with Cochrane collaboration' tool for evaluating risk of bias. The primary outcomes were incidence of pain freedom 2 hr postdose and any adverse events. The secondary outcomes were incidence of nausea freedom 2 hr postdose, phonophobia freedom 2 hr postdose, photophobia freedom 2 hr postdose, treatment-related adverse events, abnormal liver function, and chest discomfort.

| Statistical analysis
The direct comparisons between different gepants or placebo were analyzed by Review manager 5.3 (Cochrane Collaboration). α less than 0.05 was set as the significant level. The network meta- Chain Monte Carlo (MCMC) method. The consistency of the network meta-analysis was assessed by node-splitting analysis (Dias et al., 2010). When p value was more than .05, the consistency model was chosen for drawing conclusions and ranking the included treatments. Otherwise, inconsistency model was utilized to analyze the data. Odds ratio (OR) and 95% confidence interval (CI) was selected as the effect magnitude.

| Ethical statement
All of data analyzed in this article were from articles published, so the ethical approvement was not required.

| RE SULTS
Totally, we included 15 RCTs in the network meta-analysis, after the removal of repetitions and unmatched publications. Six of 15 RCTs were phase 2 trials, the rest were phase 3 trials. Of these trials enrolled, 7 treatments were analyzed: BI 44370 TA, MK-3207, olcegepant, rimegepant, telcagepant, ubrogepant, and placebo. All  Table 1. All of trials were two-grouped studies. Of these trials, 11,118 patients and 10,917 patients were assigned to one of seven treatments randomly for efficacy assessment and safety assessment, respectively. The mean sample size was 1589 per group (range from 73 to 4,250) for efficacy assessment and 1,560 per group (range from 73 to 4,114) for safety assessment. Only one trial had high risk in incomplete outcome data (Ho et al., 2012).
So, the quality of overall trails enrolled was good and their designs were similar. The risk of bias of trials enrolled was shown in Figure 1. Figure 2 showed the network of comparisons for efficacy/safety.
All gepants had one placebo-controlled randomized trial at least, but there no existed head to head comparisons between gepants.

| Efficacy
Regarding primary efficacy outcome (pain freedom 2 hr postdose), the heterogeneity was no exist between 6 pair-wise comparisons.
And all gepants were superior to placebo in meta-analysis of direct comparisons (Table 2). In the network meta-analysis, olcegepant was the rank 1 gepants to achieve pain freedom. And the next two were BI 44370 TA and MK-3207 (Table 3).
Regarding the secondary outcomes, the comparison between olcegepant and placebo was missing. In nausea freedom 2 hr postdose, all gepants were superior to placebo except MK-3207. And the rank best drug was BI 44370 TA, the next two were rimegepant and ubrogepant. All of gepants were superior to placebo in achieving phonophobia freedom 2 hr postdose and photophobia freedom 2 hr postdose. And in the network meta-analysis of phonophobia freedom 2 hr postdose, the rank best 3 were BI 44370 TA, rimegepant, and telcagepant. Meanwhile, in the photophobia freedom 2 hr postdose, the rank best 3 were BI 44370 TA, rimegepant, and telcagepant also. The detail information was showed in Tables 2 and 3.

| Safety
Regarding primary safety outcomes, only rimegepant and telcagepant were higher than placebo in incidence of any adverse events in pair-wise meta-analysis (Table 2). And in the network meta-analysis, the rank best 3 drugs were BI 44370 TA, placebo, and ubrogepant (Table 4).
Regarding secondary safety outcomes, the comparison of olcegepant and placebo was missing in treatment-related adverse events. And there were no differences between all gepants and placebo. And in the network meta-analysis, the rank best 3 drugs were BI 44370 TA, placebo, and ubrogepant. In the incidence of abnormal liver function, the comparisons between BI 44370 TA and placebo, MK-3207 and placebo or olcegepant and placebo were not estimable, because the number of patients suffered from abnormal liver function was zero. And there were no differences between the rest TA B L E 2 Summary estimates for efficacy and acceptability in meta-analysis of direct comparisons between CGRP antagonists or placebo gepants and placebo. In the incidence of chest discomfort, the trials about BI 44370 TA, MK-3207, and olcegepant had not reported this event. And the comparisons between rimegepant and placebo were not estimable, because the number of patients suffered from chest discomfort was zero. So, the network meta-analysis of abnormal liver function and chest discomfort was hard to draw a conclusion.
The detail information was showed in Tables 2 and 4. Figure 3 showed the estimate effect values of different comparisons.

| D ISCUSS I ON
Our study showed that all of gepants with different dosages and administrated once were effective and safe for abortive treatment for migraine based on the results of 15 RCTs with high quality. And our results might help clinicians to choose the type and dosage of gepants. In terms of primary efficacy, all gepants were superior to placebo, and the most effective of gepants was olcegepant according to network meta-analysis. Regarding to primary safety outcome, only rimegepant and telcagepant were inferior to placebo, but there are no existed fatal adverse events in gepants group.
In the meta-analysis, we assessed the efficacy of gepants for improving associated symptoms of migraine. It was a pity that the data of olcegepant about associated symptoms of migraine was missing.
And we found that most gegpants could improve nausea except MK-3207, which nausea was its drug-related adverse event (Hewitt, Aurora, et al., 2011). And the best gepants to achieve nausea freedom was BI 44370 TA. All of gepants could improve phonophobia and photophobia, and the best gepants to achieve phonophobia freedom and photophobia freedom 2 hr postdose were BI 44370 TA.

TA B L E 3 Rank probability of efficacy of gepants
So, the best gepants to improve associated symptoms of migraine was BI 44370 TA.
Concerning the safety of gepants, although the incidence of any adverse events of rimegepant and telcagepant was higher than placebo, but there was no difference between gepants and placebo in treatment-related adverse events. There are concerns about potential cardiovascular risk after CGRP blockade. Olcegepant, a gepants administrated intravenously, had not reported the incidence of cardiovascular events (Olesen et al., 2004). And so on BI 44370 TA and MK-3207 (Diener et al., 2011;Hewitt, Aurora, et al., 2011). These three gepants were discontinued because of different reasons.
In two phase 3 clinical trials had not reported the cardiovascular events when rimegepant were administrated with 75 mg orally Lipton, Croop, et al., 2019). But one patient in rimegepant group experienced transaminase concentration greater than 3 fold of the upper limit of normal (ULN) .
And 13 patients suffered from transient transaminase concentration F I G U R E 3 Network meta-analysis of efficacy and acceptability of CGRP antagonists for migraine. The odds ratios (ORs) of comparisons of drugs are between the column-defining drug and the row-defining drug. Regarding efficacy, ORs higher than 1 favors the column-defining drug. Regarding acceptability, ORs lower than 1 favors the column-defining drug increasing in rimegepant group, but there were no difference between placebo and rimegepant group .
Ubrogepant, which was distinct from MK-3207 and telcagepant, were evaluated in migraine for acute treatment. And there are three patients suffered from chest discomfort and 1 patient experienced chest pain after ubrogepant treatment Lipton, Dodick, et al., 2019;Voss et al., 2016). And there were 11 patients experienced liver function lesion which transaminase concentration greater than 3 fold of ULN Lipton, Dodick, et al., 2019;Voss et al., 2016). And in our meta-analysis, the incidence of abnormal liver function and chest discomfort were no differences between gepants and placebo, which were consistence with original trials. So, gepants with single dose were safety for an abortive treatment of migraine.
The limitations of present study were as follow. Firstly, the follow-up period of trials enrolled was short, and the results of safety might be underestimated. Secondly, the sample sizes of different gepants were varied widely, which might affect the rank of gepants.
Finally, our results apply only to abortive treatment of migraine and have not offered the preventive treatment of migraine.
In conclusion, gepants were effective for abortive treatment of migraine. The most effective treatment of gepants for migraine might be olcegepant which were administrated transvenously. And all of gepants were safe for migraine treatment with single dose.

CO N FLI C T O F I NTE R E S T
None.

AUTH O R S CO NTR I B UTI O N S
Peiwei Hong and Jing Xiao put forward the idea. Peiwei Hong, Tianlin Tan, and Yao Liu acquired the data. Peiwei Hong and Jing Xiao analyzed the data and wrote the first draft. Tianlin Tan and Yao Liu revised the draft.

DATA AVA I L A B I L I T Y S TAT E M E N T
All of data were extracted from previously published data.