Abstract
Purpose :
Clinical trials for emerging glaucoma therapies require demonstrating a significant treatment effect on visual field (VF) based outcomes. Detecting a significant effect may require large sample sizes due to the inherent variability of VF tests and the slow rate of glaucoma worsening in most treated patients. Using a large dataset of treated patients, we study the effect of various levels of VF variability and rates of worsening on clinical trial sample size.
Methods :
We included 15,495 eyes from 8,921 patients followed for glaucoma or suspect glaucoma with ≥ 5 reliable longitudinal VFs. We defined VF worsening and variability using mean deviation (MD) slope from ≥ 5 VFs and the standard deviation percentile of MD slope residuals (SD-MD-Res), respectively. We performed a grid search using 6 thresholds for VF worsening (-0.5 to -1.0 dB/year) and variability (bottom 10th to 50th SD-MD-Res percentiles). The percentage of eyes in our dataset for each unique VF worsening and variability threshold combination was calculated. Clinical trials were simulated with 1 visit at baseline and a follow-up visit every 3 months after that (i.e., 10 visits over 2.25 years). The sample size for a trial with 80% power and 50% treatment effect was estimated by randomly selecting N eyes from the worsening set with low variability.
Results :
Figure 1 shows smaller proportions of patients qualify for trial enrollment with faster VF worsening and lower variability rates. Figure 2 shows that the required sample size decreases with faster worsening and lower variability rates. Without a variability cutoff, the sample size decreases from 256 eyes to 66 eyes for a worsening rate of −0.50 dB/year to −1.00 dB/year. With a 50% variability cutoff, the sample size decreases from 88 eyes to 22 eyes for a worsening rate of −0.50 dB/year to −1.00 dB/year. Lower variability provides modest improvement in the sample size for a −0.50 dB/year worsening rate, while lower variability cutoffs offer minimal improvement for a −1.00 dB/year worsening rate.
Conclusions :
Using eyes with faster VF worsening and low variability can reduce the sample size needed for clinical trial enrollment. However, the advantages of stricter variability cutoffs are limited when the worsening rate is high. The reduced sample size benefit must be weighed against the drawback of a smaller pool of eligible clinic patients for trial enrollment.
This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.