Neurologists should consider access (e.g., cost and insurance coverage), efficacy, side effects, and comorbidities and contraindications when choosing a preventive therapy, she added. Verapamil and memantine are well tolerated and appropriate choices if the goal is to avoid side effects in general. If weight gain or fatigue is a concern, then topiramate and venlafaxine should be considered. Neurologists should avoid prescribing antiepileptic drugs if cognitive symptoms are a concern, said Dr. Burch. Beta blockers and venlafaxine would be better options in this case.
In clinical trials of CGRP therapies, the rates of adverse events were similar between the active and control arms. “But it’s become fairly clear that the clinical trials did not fully capture the side-effect profile that we are seeing in clinical practice,” said Dr. Burch. In a paper currently in review, she and her colleagues retrospectively studied 241 patients that they had treated with CGRP monoclonal antibodies at their headache center. The most common adverse events were constipation (43%), injection-site reaction (24%), muscle or joint pain (17%), and fatigue (15%). Furthermore, CGRP antagonists were associated with maternal hypertension, fetal growth restriction, and fetal mortality in animal studies. The current recommendation is to avoid CGRP monoclonal antibodies during pregnancy or in any patient who is at risk of becoming pregnant, said Dr. Burch.
How should neurologists assess preventive efficacy?
The assessment of a medication’s preventive efficacy “is a moving target in the headache world,” said Dr. Burch. “Historically, we have used headache days per month, and that is still, according to the International Headache Society clinical trials guidelines, how we should be judging whether a medication is working or not. But that doesn’t necessarily tell us what’s going to happen to an individual patient in front of us.”
In 2017, the Institute for Clinical Effectiveness Research compared data for old and new migraine treatments in a network meta-analysis. They all tended to reduce the number of monthly migraine days by one to two, compared with placebo. When one analyzes clinical trials of the drugs using this criterion, “most of these treatments come out about the same,” said Dr. Burch.
More recently, investigators have examined responder rates. They commonly report the proportions of patients who had a reduction in headache days of 50%, 75%, or 100%, for example. To extrapolate responder rates from the trial participants to the general population, a neurologist must know which groups of patients got worse on treatment, said Dr. Burch. Furthermore, the responder rates for older medications are unknown, because they were not examined. This situation makes comparisons of newer and older therapies more complicated.
Phase 3 trials of the CGRP drugs included analyses of the therapies’ 50% responder rates. This rate was about 42% for the 70-mg dose of erenumab and 50% for the 140-mg dose. The 50% responder rates for fremanezumab were 47.7% for the 225-mg dose and 44.4% for the 675-mg dose. In two trials of galcanezumab, the 50% responder rate for the 120-mg dose was approximately 60%, and the rate for the 240-mg dose was about 59%. The 50% responder rates for eptinezumab were 50% for the 100-mg dose and 56% for the 300-mg dose. The 50% responder rate across all trials was around 50%-60% in the active group, which is roughly 25% over the placebo group, said Dr. Burch.
Another measurement of efficacy is the efficacy-to-harm ratio, which is derived from the number needed to treat and the number needed to harm. To calculate this ratio, however, harm needs to be assessed adequately during a clinical trial. Although the ratio can provide a clinically relevant overview of a drug’s effects, patients may differ from each other in the way they evaluate efficacy and harm.
In addition, many questions about preventive treatment of migraine have no clear answers yet. It is uncertain, for example, how long a patient should receive preventive treatment and when treatment should be withdrawn, said Dr. Burch. “Can we expect that a lot of people are going to need to be on it for life, or is there a subpopulation who will get better and [for whom] we can withdraw [treatment]?” she asked. “How do we identify them?” Also, more data are needed before neurologists can understand why a given patient responds to one treatment, but not to another. It is difficult to predict which patients will respond to which treatments. Finally, it remains unclear how much of patients’ improvement can be attributed to regression to the mean, rather than preventive treatment.