Applied Clinical Trials
Tony Fiorino, Chief Medical Officer of electroCore, discusses challenges he has faced with device studies, and will elaborate on his experiences about the differences between device and drug trials.
Device and drug trials pose a variety of challenges; they are, after all, completely different in study design, have varying regulatory pathways, as well as differing executional challenges. In this interview, Tony Fiorino, Chief Medical Officer of electroCore, will discuss challenges he has faced with device studies, and will elaborate on his experiences about the differences between device and drug trials.
Moe Alsumidaie: As a non-invasive vagus nerve stimulation (nVNS) device, what challenges did you experience in your migraine and cluster headaches studies?
Tony Fiorino: We have completed five studies targeting cluster headache and migraine and are currently running a sixth study in migraine prevention. The biggest hurdle in running these studies is ensuring patient compliance. I have observed studies in which we were able to identify 10% -20% of subjects who did not have adequate compliance. So, that is an area that we are increasingly focused on in our current studies. Our compliance initiatives include implementing appropriate site and subject training and providing subjects with tools to better understand the study’s requirements. We are also considering the incorporation of digital notification systems embedded in the device to better engage the patient; essentially building digital health in the device itself.
MA: How did you address noncompliance in your previous studies?
TF: With our prior study, we observed during data analysis that there was a subgroup of subjects who were not compliant. We, subsequently, conducted post hoc analyses of the study looking specifically at the compliant patient population.
MA: Could you discuss the endpoints that you used in your study to validate safety and efficacy?
TF: In all of the studies, certainly with migraines in the field where there are lots of clinical trials in both the acute and preventive setting, we did not look to reinvent the wheel. We wanted data that would be interpretable by neurologists in the context of the other kinds of migraine studies they see. Our efficacy endpoints included pain-free scales at various time points after the initiation of therapy in the acute setting, percentage of responders, percentage of patients with improvement in pain, and duration of the episode. On the prevention side we looked at the reduction in migraine days, reduction in headache days, and reduction in migraine medications (in the migraine studies), and reduction in the number of attacks per week and use of acute treatment medication during the study (in the cluster headache studies).
MA: What are the differences and similarities between medical device and pharmaceutical studies? Moreover, how are device studies different than pharmaceutical studies from a study design standpoint?
TF: With devices, we have the ability to obtain analytical data on compliance by querying the device, and diaries are also another way of assessing compliance; we tend to rely more on patient diaries, as we have determined that to be an appropriate compliance measure. With a drug, you have pharmacokinetics, so if you happen to be doing a PK study, you can determine whether or not the patient is compliant. In the future I think it will be important to validate diaries with device data to improve the quality of our compliance assessments, and to ensure that doses were delivered when they were supposed to have been delivered. Another challenge is operational and study design in nature; the control. We have a device that produces a physical effect when you use it. It's a marketed device, so patients, or subjects in the study can go to our website and see the patient education modules; there are FDA-approved instructions that specifies device effect during stimulation. It's hard to create a sham device that mimics the experimental device or arm of the study without unblinding subjects on the sham arm. We've played around with different kinds of sham devices, from totally inoperable shams, to using the same device but placing it in a different location, to shams that deliver a low-grade stimulation that we believed didn't activate the vagus nerve. With pharmaceuticals, you can easily create a pill or injectible agent that mimics the investigational product without unblinding the subject.
MA: Do you think it is feasible to conduct a head-to-head study with the standard of care, like a drug, for example, versus your device?
TF: In the PREVA (PREVention and Acute treatment of chronic cluster headache) study, it was standard of care plus or minus gammaCore, and we did see a decrease in use of acute intervention medications in those cluster patients. However, in general, head-to-head trials are a risky venture, which is why device and biopharma industry sponsors do so few of them. Additionally, drugs tend to focus on different endpoints than devices, and the comprehensive therapeutic effect and mechanisms of action are very different. Further, safety profiles differ between drugs and devices; nVNS has a fantastic safety profile, and I am blown away by it. It is hard to compare the safety profile of nVNS therapy versus a pharmaceutical therapy; they are completely different, like comparing apples and oranges. That approach does not support good study design and valid scientific results.
MA: Outsourcing, especially with smaller companies, is a critical area. However, some companies decide to in-source their operations or vertically integrate their clinical operations. What strategy have you taken for clinical trial execution, and why did you take that strategy?
TF: Historically, I have run small biotech companies where we did not have much manpower, and we had a limited number of programs. So, building out an in-house ClinOps team to run studies did not make sense; outsourcing made much more sense. At electroCore, initially, studies were in-sourced, however, as the studies got larger, operations moved towards a more outsourced model. In order to maximize and scale our resources, we're now looking at whether or not it is feasible to bring certain functions in-house and leverage them across multiple studies.