Retinitis pigmentosa (RP) is a group of inherited retinal degenerative diseases with a prevalence of approximately 1:4000, affecting 2 million persons globally. RP is genetically heterogeneous (associated with >3000 mutations in ~100 genes). Currently, there are more than 50 retinal gene therapy, clinical trials, most of which are for relatively rare diseases such as RP. Two problems confronting RP clinical trials are: 1) the target patient population is small and 2) patient heterogeneity is significant, even among patients with similar phenotypes.
N-of-1 trials are randomized, prospective, controlled, multiple crossover trials in a single patient. The effects of one or more treatments are studied by following individual patients who receive alternative treatments for pre-specified time intervals, termed “periods”. The various treatments alternate in a randomized order through several crossover periods, termed “blocks”. Typically, a “run-in” period is included in which one can assess patient tolerance and adherence as well as permit washout of any previous treatments. Ideally, the patients, physicians, and data analysts are masked. One can have randomized allocation to treatment cycles if the cycles differ in their structure.
Some notable features of N-of-1 trials include the following. Aggregated N-of-1 trials constitute level 1 evidence. By contrast, parallel group, randomized, clinical trials (RCTs), provide level 2 evidence. The N-of-1 trial design is just a subgroup of RCTs and resembles a crossover trial, but it is just for one person at a time. The experimental protocols of parallel group RCTs can be used, such as allocation concealment and double-masking. N-of-1 trials allow assessment of efficacy in an individual subject since each subject participates in the treatment and control group at different times. In contrast, parallel group RCTs cannot assess treatment benefit in a given patient since subjects are enrolled either in the treatment or in the control group, but not both.
N-of-1 trials have four notable advantages:
1. Each patient is assured exposure to the experimental treatment, which may facilitate recruitment. This design also might be attractive to the parents of children with RP.
2. Each patient serves as their own control, which reduces variance and minimizes confounding.
3. There is substantial patient input on efficacy and safety, which could simplify clinical decision-making if the aggregated data do not demonstrate a treatment effect for the overall population. This outcome would not be surprising, given the heterogeneity of patients with similar RP phenotypes. It is important to note, though, that the aggregated trial data are essential, as they guide initial clinical recommendations for patients who have not been enrolled in the trial.
4. Enrollment of patients with comorbidities, generally discouraged in parallel group RCTs, is more easily managed with an N-of-1 trial design since each patient serves as their own control.
Features of conditions that favor the use of an N-of-1 trial design include:
1. The condition is rare (small patient population).
2. The condition is chronic.
3. The condition is slowly progressive during the time of the trial.
4. The primary outcome is clinically important, repeatedly measurable, and treatable.
Features of therapies that favor the use of an N-of-1 trial design include:
1. The treatment ameliorates but does not cure the disease.
2. The treatment has a reversible effect on the primary outcome.
3. The treatment has a relatively rapid onset of measurable effect after exposure and rapid cessation after withdrawal.
4. The treatment induces changes that are measurable repeatedly and clinically relevant.
Limitations of N-of-1 trials include the following. First, patient dropout in N-of-1 studies has a disproportionate impact compared to parallel group RCTs because each participant contributes at least twice the information (subjects contribute both to treatment and control arms of the study). Second, if the effect of a treatment period affects subsequent periods (excluding the washout period), then a “carryover effect” is present. Persistence of a treatment effect into the subsequent period of a block is likely to invalidate the measurement of the primary outcome during the second period. This limitation can be avoided by including a washout between treatment periods or by randomizing block duration as well as treatment assignment. Third, if the cycles of an N-of-1 trial are long, then a confounder known as the “period effect” may become important. A period effect occurs when different outcomes are attributable to the calendar time in which the treatment is received, which could occur if a outcome under treatment is exacerbated in the winter and ameliorated in the summer, and the trial extends through both the summer and winter.
While it is relatively easy to envision the application of an N-of-1 trial design to pharmacotherapy, it may not be obvious that this trial design could be applied to gene therapy. For example, if gene therapy has a measurable effect only if it is reversibly enhanced by an exogenous device/agent, then the effect of the gene therapy in the presence and absence of the device/agent might be assessed in an N-of-1 trial. Sahel and coworkers reported results of optogenetic therapy for RP patients. This therapy, which induces light sensitivity in retinal ganglion cells, was effective only if light amplifying goggles were worn because the light sensitivity of the chromophore is relatively low. So, the trial design might have the patient, who has been genetically altered, wearing the light amplifying goggles during the first period, and in the second period, visual acuity, the primary outcome, would be measured in the absence of the goggles.
In summary, N-of-1 trials are randomized, prospective, controlled, multiple crossover trials in a single patient. The effects of one or more treatments are studied by following individual patients, who receive alternative treatments (e.g., therapeutic intervention versus placebo). N-of-1 trials may provide a path to assess treatments for rare diseases with rigor equal to or greater than that of parallel group RCTs provided that: 1) the disease is reasonably stable during the trial; and 2) the disease has a sign or symptom that responds reversibly to the therapy; and 3) the primary endpoint can be measured repeatedly. N-of-1 trials may improve the feasibility and affordability of clinical trials for patients with rare inherited retinal diseases.
Dr. Zarbin is employed by Rutgers University. He has recieved consulting fees or honorarium from Genentech/Roche, Novartis Pharma AG, Life Biosciences, Perfuse Therapeutics, Selphagy, NVasc, Iduna, and Boehringer Ingelheim. He has received payment for lectures from Novartis Pharma AG, and has stock options in NVasc. Finally, he has patents with Rugers University.
Image by z_wei / Getty