SGEM#279: Do You Really Want to Hurt Me and Use a Placebo Control for a Migraine Trial?

Subscribe: RSS

Date: January 10th, 2020

Reference: Dodick DW et al. Ubrogepant for the Treatment of Migraine. NEJM 2019

Guest Skeptic: Dr. Anand Swaminathan is an Assistant Professor of Emergency Medicine at St. Joseph’s Hospital in Paterson, NJ. He is also the managing editor of EM:RAP and associate editor at REBEL EM.

Case: A 23-year-old man with a history of migraines presents with two days of headache, nausea and photo-photophobia typical of his prior migraines. He’s tried a number of medications at home including ibuprofen, acetaminophen, aspirin and sumatriptan without any considerable improvement in symptoms. You start to offer him your standard medications like metoclopramide and haloperidol when he asks about a new drug he heard about called ubrogepant.

Background: Migraine headaches are a chronic neurologic disease characterized by throbbing, often unilateral headaches that are often associated with nausea, vomiting, photophobia and phonophobia. It is a common disease and can be severe enough to impede on people’s lives.

Headaches themselves are not only a common emergency department presentation but one that is filled with potential dangers. There are a number of causes of headache that are life and limb threatening – subarachnoid hemorrhage (SGEM#201), meningitis, encephalitis, cerebral venous thrombosis, vertebral artery dissection among other things but, most headaches are benign in nature.

There is an international classification system of headaches (IHS 2018). The current system classifies them into primary and secondary headaches. An important part of our job as emergency physicians is to differentiate the lethal headache from the benign headache.

Though we rarely make a de novo diagnose of migraines in the emergency department, many patients with migraines present to us for symptom management. The pathophysiology of migraines is both complicated and poorly understood but there are a number of potential treatments including NSAIDs, acetaminophen, aspirin, neuroleptics, triptans and even propofol.

More recently, calcitonin gene-related peptide antagonists (CGRPs) have emerged as a new potential treatment. The first big study that came out on these drugs was published in the NEJM in 2019 and was entitled Rimegepant, an Oral Calcitonin Gene-Related Peptide Receptor Antagonist for Migraine (Lipton et al).

Now, we have a second study published in the NEJM on a related drug, ubrogepant.

Clinical Question: Does ubrogepant increase the percentage of patients who were free from pain and absent of the most bothersome migraine-associated symptom at two hours from initial dose in comparison to placebo?

Reference: Dodick DW et al. Ubrogepant for the Treatment of Migraine. NEJM 2019

Population: Adult patients (18-75 years of age) with at least a one-year history of migraine with or without aura that met criteria from the International classification of headache disorders and had migraine onset before the age of 50. Patients had to have a history of migraines between 4-72 hours and a history of migraine attacks separated by at least 48 hours of freedom from headache. Additionally, they had to have suffered from two to eight migraines per month over the last three months.
- Exclusions: Patients with 15 or more headaches/month on average in the previous six months. Hard to distinguish the type of headache. Use of acute migraine treatment on ten or more days in the previous three months. Participated in a trial involving CGRP. Had clinically significant cardiovascular or cerebrovascular disease. History of hepatitis in the last six months or laboratory findings of liver disease (elevated AST, AST, Bilirubin or low serum albumin).
- Additional Exclusions from ClinicalTrials.gov
  - Has a history of migraine aura with diplopia or impairment of level of consciousness, hemiplegic migraine, or retinal migraine
  - Has a current diagnosis of new persistent daily headache, trigeminal autonomic cephalgia (eg, cluster headache), or painful cranial neuropathy
  - Required hospital treatment of a migraine attack 3 or more times in the previous 6 months
  - Has a chronic non-headache pain condition requiring daily pain medication
  - Has a history of malignancy in the prior 5 years, except for adequately treated basal cell or squamous cell skin cancer, or in situ cervical cancer
  - Has a history of any prior gastrointestinal conditions (eg, diarrhea syndromes, inflammatory bowel disease) that may affect the absorption or metabolism of investigational product; participants with prior gastric bariatric interventions which have been reversed are not excluded
- Intervention: Ubrogepant 50 mg or 100 mg
- Comparison: Placebo
- Outcomes:
  - Co-Primary Outcome: Freedom from pain at two hours from initial dose of medication. Absence of the most bothersome symptom associated with migraine two hours from initial dose of medication.
  - Secondary Outcomes: Change in severity of headache at two hours, sustained pain relief, sustained freedom from pain, absence of photophobia, absence of photophobia and absence of nausea at two hours from initial dose. Adverse events were also collected.

Authors’ Conclusions:“A higher percentage of participants who received ubrogepant than of those who received placebo had freedom from pain and absence of the most bothersome symptom at 2 hours after the dose. The most commonly reported adverse events were nausea, somnolence, and dry mouth. Further trials are needed to determine the durability and safety of ubrogepant for acute migraine treatment and to compare it with other drugs for migraine.”

Quality Checklist for Randomized Clinical Trials:

The study population included or focused on those in the emergency department. No
The patients were adequately randomized. Yes
The randomization process was concealed. Yes
The patients were analyzed in the groups to which they were randomized. No
The study patients were recruited consecutively (i.e. no selection bias). Unsure
The patients in both groups were similar with respect to prognostic factors. Unsure
All participants (patients, clinicians, outcome assessors) were unaware of group allocation. Yes
All groups were treated equally except for the intervention. Yes
Follow-up was complete (i.e. at least 80% for both groups). No
All patient-important outcomes were considered. No
The treatment effect was large enough and precise enough to be clinically significant. Unsure

Key Results: They enrolled 1,672 patients with roughly equal numbers allocated to each of the three groups. The mean age was around 40 years and almost 90% were female. The modified ITT analysis excluded 345 (21%) of participants.

Ubrogepant was superior to placebo in treating migraine headaches.

Primary Outcomes: (100mg/50mg/placebo)
- Freedom from pain at two hours: 21%/19%/12% (both doses statistically better than placebo but, not better than the other). That gives an absolute difference of about 8% and Number Needed to Treat for Benefit (NNTB) of 13
- Absence of most bothersome symptom at two hours: 38%/39%/28%. This is an absolute difference of 10% with a NNTB of 10.
Secondary Outcomes:
- Pain relief at two hours (61%/61%/49%) and sustained pain relief (38%/36%/21%) was better with ubrogepant compared to placebo.
- Serious Adverse Events:There were five SAE with all of them being in the intervention group (two appendicitis, pericardial effusion, spontaneous abortion and seizure). Only the seizure was considered related to the trial drug. Six patients had ALT levels three times the upper limit of normal (one in the placebo group and five in the treatment group). Only one of the treatment group was considered possibly related to the trial regimen. Details are in the supplemental appendix.

1. Patients: We had a few issues with the patients included in this study. First, these were not emergency department patients but rather those recruited from outpatient clinic. Whether or not these are the same patients that present to the emergency department is unknown.

We are also unsure if the patients were recruited consecutively. This is an important aspect to avoid potential selection bias. Remember that when we use the term “bias” we are not talking about random noise in the data but something that systematically moves us away from the “truth”.

The third question we had about the included patients was whether or not both groups were similar with respect to prognostic factors. Baseline demographics are reported in Table 1. However, things like number of headaches/month, refractory headaches in the past, and other things are not reported. This could impact the results and therefore the conclusions.

2. Comparison to Placebo: Randomized control trials (RCTs) are considered an ideal study design to establish causality and effect of a medication. Drug intervention RCT design requires that the intervention be compared to something (active drug, standard treatment, no treatment or placebo).

It is widely agreed upon that comparison to placebo is acceptable when no proven intervention exists (Millum and Grady 2013). In contrast, placebo comparison is not considered acceptable in life-threatening conditions if there is an available treatment that is known to prolong life. The use of placebo for comparison in non-life-threatening conditions has been hotly debated for decades, particularly when an accepted treatment exists.

The argument against the use of placebos in these circumstances is guided by the Declaration of Helsinki. This documents state:

“In any medical study, every patient — including those of a control group, if any — should be assured of the best proven diagnostic and therapeutic methods.”

Thus, if an effective treatment exists, it should be prescribed to patients (Simon 2000).

Those researchers arguing for use of a placebo comparator counter that even in the presence of effective treatment, placebo control may be necessary when:

“there are compelling and scientifically sound methodological reasons for its use and the participants in the study will not face additional risks of serious or irreversible harm from exposure to placebo” (Keranen et al 2015).

Even when a widely accepted treatment exists, some researchers argue that informed consent can be used to justify the use of placebo.

However, research participants are likely to believe that participation in a trial will lead to benefit and this therapeutic misconception may not be resolved simply by informed consent (Chiodo et al 2000). Patients randomized to the placebo group, when accepted active treatment exists, must not be subjected to additional risks or harms but in the absence of harm, placebo-control would be reasonable (Temple and Ellenberg 2000). The problem becomes what definition of harm to use.

Despite the ethical issues surrounding placebo-control studies, there are numerous prior studies that violate the tenets of the Declaration of Helsinki as well as the ideal of “prima no nocere.” Examples include research into the treatment of onchocerciasis with ivermectin, ondansetron in chemotherapy-induced emesis, ACE inhibitors in congestive heart failure and antihypertensive agents (Rothman and Michels 1994). If a placebo-control approach is used, rationale for this design is necessary. However, Keranen et al found that only about one-third (35%) of RCTs actually do this while the risk of placebo is often (83%) and not explained to patients (Keranen et al 2015).

With regards to abortive migraine treatment, there are a number of options with established efficacy. Three options that are considered first line include: antidopaminergics, triptans and nonsteroidal anti-inflammatory drugs (NSAIDs) (Friedman 2016, AHS Statement 2019). Lipton et al state that up to 66% of patients respond to triptans based on prior work (Lipton et al 2019). Lipton et al do not provide justification for the placebo-control methodology in the manuscript which is particularly important in the setting of effective alternative treatment.

While an argument can be made that treating migraine headaches with placebo does not lead to long-term harm, patients suffering from migraines often experience severe, debilitating pain often resulting in an inability to work or perform typical daily activities. I have suffered from migraines in the past and would be very upset to have received a placebo when effective treatments were available. Thus, a placebo-control study exhibits, at best, questionable ethical standards and subjects patients to possibly unnecessary harm.

So why would researchers choose to compare their drug to placebo as opposed to comparing to standard therapy? There may be a number of possible explanations for this choice. Comparing a new drug to placebo is easier to establish efficacy than to a known effective therapy. However, a non-inferiority study design could be used to compare the new drug to established treatment if the new drug had benefits over the prior treatment i.e. ease of use, cost, reduced side effects.

Some argue that historically, the Food and Drug Administration (FDA) has withheld drug approval when placebo-controls were not used in establishing efficacy. However, these concerns are likely exaggerated (Chiodo et al 2000, Orentlicher 2001).

The most obvious reason for using a placebo-control methodology is that pharmaceutical companies may believe:

“it is their interest to compare new drugs with placebo rather than existing therapy, even when better information for patients and physicians would be provided by an active control” (Orentlicher 2001).

Time to Talk a Little Nerdy

Demonstrating superiority to placebo is easier than demonstrating superiority to an effective therapy and is more likely to result in positive findings for the drug and the pharmaceutical company. This comes at the harm of patients and, adds little to our understanding of treatment for the disease.

If you want to dive into the ethics of using placebos in clinical trials you can check out Time to Talk a Little Nerdy (TTLN) as part of the EMRAP family of shows.

3. Co-Primary Outcomes: How many times will I have to use this great quote from the movie Highlander 1986…”there can be only one”, primary outcome? I might have to start using the best 80’s movie every (The Princess Bride 1987) quote: “You keep using that word (primary), I do not think it means what you think it means”.

The term primary according to Merriam-Webster’s dictionary means first rank, importance, or value. So, decide on what is the most important outcome, design your study accordioning, and report this finding. Everything else can be secondary. By having more than one primary outcome (co-primary or composite outcome) you make the target larger (statistical significance) and therefore easier to claim a positive trial and get published.

4. Modified Intention-To-Treat (mITT) Analysis: A pure ITT is to take all the patients immediately after randomization and analyses them in the groups to which they were allocated. This is the conservative way to look at the data and is the preferred method for superiority trials.

A per-protocol approach analyses participants not by group allocation but by whether or not they received the intervention. This is a quality indicator for non-inferiority studies.

Between these dichotomous extremes is the mITT. These authors used a mITT analysis to report their data. Rather than analyzing patients allocated to each group they excluded a number of patients.

Patients were only included if they:

“took an initial dose of ubrogepant or placebo, recorded a baseline rating for the severity of the migraine headache, and recorded at least one rating for the severity of the migraine headache after the initial dose or recorded the presence or absence of at least one migraine- associated symptom at or before the 2-hour time point after the initial dose.”

If this did not happen, patients were excluded from the efficacy analysis.

A mITT analysis can nudge the results (bias) towards finding efficacy. This may over-estimates the effect size for efficacy of ubrogepant and makes us less confident in the results.

5. Industry Funding: This was an industry supported trial. The publication says:

“Confidentiality agreements were in place between the sponsor (Allergan) and the authors. The sponsor developed the trial protocol in collaboration with external consultants, provided the trial drug and placebo, and gathered and analyzed the data. The manuscript was prepared by the sponsor, with contributions from all authors and with assistance from a professional medical writer employed by the sponsor.”

This does not make the data wrong, but it should make us more skeptical.

Doing research is expensive and the funding needs to come from somewhere. Our current model includes industry involvement in studies that can take many forms. This introduces potential biases and it would be better if our system had no industry involvement.

If and until that ever happens, SGEMers need to keep their skeptical radar turned on when they see that industry has played a role in the design, analysis, writing, publication and dissemination of studies.

Comment on Authors’ Conclusion Compared to SGEM Conclusion: While it appears that ubrogepant is superior to placebo, this isn’t the question we should be asking. Rather, we should be asking if ubrogepant is better than standard therapy for migraine headaches.

SGEM Bottom Line: Ubrogepant and the other CGRP receptor antagonists are not ready for widespread use in the emergency department for patients who present with a migraine headache.

Case Resolution: You discuss with the patient that the novel CGRP receptor antagonists have potential to benefit patients with moderate to severe migraines but, that there’s limited evidence for their use in the emergency department and that we have no idea if these drugs are more effective or safer than our standard medications .

You decide instead to treat the patient with 10 mg of metoclopramide which has a modest reduction in pain. You follow this with 2.5 mg of haloperidol which results in the patient’s pain reaching a level of 2 out of 10 and resolution of his nausea and photo-phonophobia. He is improved enough to go home and states he will follow up with his neurologist in the next week or so.

Dr. Anand Swaminathan

Clinical Application: We do not know if ubrogepant is better, worse or the same efficacy as existing treatments for migraine headaches. Until data on patients presenting to the ED with migraine is reported that includes a comparison to standard active treatment and safety data, this drug doesn’t belong in our armamentarium.

What Do I Tell My Patient? There is a new class of drug that have been approved to treat migraine headaches. The research was not done on patients like you in the emergency department. The studies also did not compare the new drugs to our existing treatments. We do not know if it is more or less effective than what we already use. The good news is the treatments we do have is very effective. We can also offer you a dose of steroids (SGEM#28) to prevent you suffering a rebound headache (NNT 9).

Keener Kontest: Last episode’s winner was a repeat win for Dr. Steven Stelts from New Zealand. He knew Dr. Anton von Troltsch is credited for developing the otoscope?

Listen to the podcast to hear this weeks’ trivia question. If you know the answer, send an email to TheSGEM@gmail.com with “keener” in the subject line. The first correct answer will receive a cool skeptical prize.

Other FOAMed: