4.2.1. Efficacy
In Study 079, a statistically significant difference was seen between onabotulinumtoxinA and placebo for the following secondary outcomes: frequency of headache days and frequency of migraine/probable migraine days. The effect sizes for Study 080 were greater than those for Study 079, and statistical significance was obtained for more outcomes. In Study 080, statistically significant differences between groups were obtained for the primary outcome (headache days) and the following secondary outcomes: severe HIT-6 category score, moderate/severe headache days, total cumulative hours of headache occurring on headache days, migraine/probable migraine days, and headache episodes. In both treatment groups, the main improvement for all efficacy end points was noticed at week 12, with a small additional improvement noticed at week 24, indicating that the main response was achieved after the first treatment with a small incremental benefit after the second onabotulinumtoxinA treatment. Statistical significance was also reached for other outcomes including all three domains of MSQ in both trials, and acute headache pain medication days and acute headache pain medication overuse in Study 080. However, it is unclear whether statistical adjustments were made to account for multiple testing. Hence, despite similar study protocols, there were important differences in the findings between Studies 079 and 080.
The reasons for the better results obtained in Study 080 are unclear, but were perhaps due to differences in baseline patient characteristics between onabotulinumtoxinA and placebo patients in Study 079. At baseline in Study 079, patients on onabotulinumtoxinA had fewer headache episodes and migraine episodes. Furthermore, in Study 080, the primary outcome was changed to headache days prior to unblinding when it became evident that there was no statistically significant difference between treatment groups in headache episodes in Study 079. This change in primary end point may limit the interpretation of results of Study 080.
HRQoL is an important outcome according to the patient group inputs provided for this review and following discussion with the clinical expert involved in the review; as such, it was chosen as a key efficacy outcome for the review. In the included trials, HRQoL was measured using MSQ. The mean changes from baseline for the three HRQoL domains measured by MSQ demonstrated clinically important and consistent results across both trials at week 12 and at the end of the DB phase in patients who received onabotulinumtoxinA treatment. In both studies, patients who received onabotulinumtoxinA treatment exceeded the established within-group MCIDs. The clinical expert consulted for this review concurred with this finding, as he noted that the improvement in change from baseline obtained with onabotulinumtoxinA was clinically important. Patients receiving placebo had an improvement in all three domains of MSQ; however, the changes did not exceed MCIDs. The between-group differences were statistically significant at weeks 12 and 24; however, the MCIDs for between-group differences in MSQ score are not available, and it is unclear whether the differences in MSQ between onabotulinumtoxinA and placebo were clinically important. Furthermore, from the range of domain scores reported, it would seem that some patients did not understand the MSQ or did not answer the questionnaire truthfully, because some patients reported scores of 100 points (highest function) at baseline, yet at week 12 and week 24, some reported a difference in score of 100 points.
Other patient-reported outcomes, which included HIT-6 scores, were also chosen as key efficacy outcomes. The between-group difference in HIT-6 at week 24 favoured the onabotulinumtoxinA group, and the difference was clinically important. The clinical expert involved in the review stated that HIT-6 is self-reported by patients and can be completed over the Internet. Some patients may have exaggerated their symptoms in order to be included in the clinical trial. This would have been reflected in a large proportion of patients (approximately 93%) reporting “severe” HIT-6 scores at baseline.
At baseline, there were 20 days of headache, of which, 19 days were migraine. The clinical expert involved in the review indicated that some patients may find it difficult to differentiate between headaches and migraines. The clinical expert also indicated that a decrease in 2 or more headache days per week is likely clinically important. In the included trials, headache days decreased by approximately eight to nine days (from 20 days) per 28-day period for onabotulinumtoxinA at the end of the DB phase. Placebo patients achieved a decrease of six to seven headache days per 28 days. The between-group difference was approximately one to two headache days, which is unlikely to be clinically important according to the clinical expert. The clinical expert further indicated that a treatment failure is considered a lack of return to EM, but treatment failure was not specifically measured in the trials. Potentially, patients could go back and forth between CM and EM with onabotulinumtoxinA. There are no data on how often a patient could stop/start onabotulinumtoxinA treatment if he or she fluctuates between the two subtypes of migraine because of onabotulinumtoxinA treatment.
In both studies, migraine-free status was evaluated in each of the trials as the proportion of patients with a 100% decrease in migraine days from baseline, with only a small proportion of patients (fewer than 5%) achieving migraine-free status.
The change in cumulative hours of headache favoured the onabotulinumtoxinA group. Compared with placebo, onabotulinumtoxinA patients had an additional 40 hours of headache-free time (and thus approximately one work week). However, the clinical expert believed that this outcome is not a useful measure and that it is unreliable, as it is difficult for patients to accurately report the number of hours with headache.
The greater reduction in headache and migraine days seen with onabotulinumtoxinA compared with placebo was not reflected in a greater reduction in acute headache pain medication intake. However, the analysis of acute headache pain medication intake was limited, because patients were not required to report medication dose. Nonetheless, both treatment groups decreased their medications by eight to 10 intakes (five to six days) without completely eliminating their need for medications. The clinical expert confirmed that, even with onabotulinumtoxinA treatment, the patients’ need for acute pain headache medications would not likely be eliminated.
The baseline numbers of ER visits and hospitalizations were too small to draw conclusions about the impact of onabotulinumtoxinA on the use of these resources. There was no improvement in work productivity despite improvements in key outcomes for the review, such as HRQoL.
The outcomes were subject to variations in interpretation. For example, there may have been instances when a single episode of headache may have been counted as two episodes, as it may be difficult to distinguish between episodes especially if acute medications had been used with some temporary benefit. Similarly, it may have been difficult to distinguish between two or three shorter headaches occurring back to back, compared with one headache occurring over several days. This is underscored at baseline by the fact that some patients experienced more than one headache per day. The clinical expert indicated that this was surprising, and that it was more likely that the headaches experienced on the same day are in fact the same headache.
Blinding was difficult to maintain in light of the AEs expected with onabotulinumtoxinA; as such, placebo-treated patients who guessed that they had been randomized to placebo would have had no expectation of improvement. The other issue with improper blinding is that the effect of onabotulinumtoxinA could have been overestimated.
More than 60% of patients had a history of prophylactic medication use; conversely, less than 40% of patients had not used prior prophylactic treatment. This is surprising given, that the average duration of illness was 17 to 20 years and that the average age of patients was 40 years. The manufacturer’s listing request is for use in patients who have failed (due to lack of efficacy, intolerance, or clinical contraindication) three or more prior oral prophylactic medications. The trials showed that patients who had a history of prophylactic use of three or more medications had a response similar to the overall patient population. However, this subgroup analysis was post hoc and the manufacture did not conduct a test for interaction. Furthermore, “failure of prophylactic treatment” was not clearly defined by the manufacturer. Hence, it is challenging to draw a conclusion for this subpopulation. Moreover, the manufacture did not conduct an analysis on the subpopulation of patients who did not use three or more prophylactic medication, and as such, we do not know if prophylactic treatment-naive patients or patients who used less than three prophylactics would benefit from onabotulinumtoxinA treatment.
Except for males, results from the subgroup analyses (females and patients with or without overuse of acute headache pain medications) were consistent with the overall patient population results for almost all of the outcomes. Results for males were not different between-treatment groups, possibly due to the small sample of males included in the trial. A subgroup analysis according to duration of illness was not conducted.
It is possible that some of the included patients suffered from MOH instead of CM, a type of migraine recognized as being distinct from CM according to ICHD-3.1 These patients would likely have benefited from stopping all acute pain medications as a first step in the treatment of their CM, and by doing so, would have reverted to EM. Hence the response seen would have been due to acute headache pain medication discontinuation and not due to onabotulinumtoxinA treatment. However, the clinical expert involved in the review indicated that there is considerable overlap between the two types of migraine and in clinical practice it is difficult to distinguish between MOH and CM.
Other generalizability issues are worth noting. Patients had suffered from CM for an average of 17 to 20 years; stratifying the randomization of patients to treatment groups and/or pre-specified subgroups analysis according to duration of illness would have been informative. Results may only be applicable to patients with long-standing disease, as it is unknown if the duration of illness would affect a patient’s response to treatment. Furthermore, the results may not be generalizable to patients with comorbid illnesses who require the use of drugs that are used in the prophylactic treatment of CM and other conditions, for example propranolol used in hypertension or amitriptyline used in depression. In the included trials, more than 90% of patients reported a HIT-6 score greater than 65. The trials results may not be applicable to patients with less severe headaches.
Very few patients required additional “follow-the pain” doses. It should be noted that the administration of onabotulinumtoxinA is a complicated procedure which requires proper training of clinicians. Hence, patients may be required to access specialized treatment centres.
In the OLE phase, patients received onabotulinumtoxinA treatment for 32 weeks (APPENDIX 6: SUMMARY OF OPEN-LABEL EXTENSION PHASE OF STUDY 079 AND STUDY 080). Irrespective of group assignment in the DB phase, there were no between-group differences at the end of the study for the various outcomes measured. This is not surprising because, in the DB phase, when an improvement in an outcome was achieved with onabotulinumtoxinA, it occurred after the first dose. With the second dose, the incremental benefit was small. Therefore, patients who had been on placebo in the DB phase would have obtained relief immediately with onabotulinumtoxinA in the OLE phase.
Irrespective of group assignment in the DB phase, there were improvements in MSQ and HIT-6 scores at the end of the study compared with baseline in both Studies 079 and 080. Acute headache pain medications could not be completely stopped, with more than 70% of patients still requiring acute pain medications at week 56. However, fewer than 20% of patients overused acute pain medications at week 56. The frequency of headache days and migraine/probable migraine days decreased by 10 to 11 days per month at week 56, from approximately 19 to 20 days per month at baseline. This means that patients continued to experience on average eight to nine migraines per month, reverting back to a diagnosis of EM.
Results from the OLE phase should be interpreted with caution, given that all patients received open-label treatment and the outcomes measured were subjective. In addition, no large increase in efficacy was noticed at week 56 when compared with week 24. Finally, in the OLE phase, the rate of discontinuation was high, with more than 25% of patients discontinuing treatment before week 56. However, few patients (2% to 4%) discontinued the study due to a lack of treatment efficacy.
