Jackson et al. use a multi-institutional longitudinal dataset of treated glaucoma patients to study the prevalence of rapid progressors as defined by global and central visual field (VF) loss. They find that the prevalence of rapid progression defined by global VF loss (rate of mean deviation [MD] change < -1 dB/year) is approximately 12%, in line with previously published studies. They define rapid central VF loss by computing the slope of the total deviation values of the 12 test locations in the central 10 degrees (MTD10 < -1 dB/year) and find a similar rate of rapid progression (11.7%). Additionally, rather than solely relying on pre-specified cutoffs, they define rapid progression based on normal distribution cutoffs. This approach found poor overlap between progressing eyes defined by global (MD) and central (MTD10) definitions of rapid progression. For instance, when defining rapid progression as the bottom 0.5% of the normal distribution of slopes, they found only one in four eyes progressing by central change (MTD10) progressed by global change (MD).
Based on these results, one may ask whether following global VF, central VF, or some other measure of VF worsening is more valuable. Some areas of the VF will progress more dramatically in specific patients (due to disease progression or statistical noise). For this reason, focusing on additional regions of the VF may pick up progression that analysis of the full VF does not. But can one also make a functional argument for the central visual field being most critical? While some papers have suggested central loss is more associated with quality of life, better peripheral vision is more associated with less frequent/ notable events (falls, motor vehicle accidents, difficulty searching) that are nonetheless quite important and impactful.
These results highlight the prevalence of treatment failures in routine clinical practice
Additionally, these results highlight the prevalence of treatment failures in routine clinical practice. Future work must seek to reduce the rate of vision loss in the approximately 10% of patients undergoing treatment who progress at an unacceptably fast pace. Recent advances in artificial intelligence models, which forecast eyes at higher risk for rapid progression (PMID: 36944385) with early data, are likely a step in the right direction. Still, much work remains to make these models accessible to the clinician. Further, we need a better understanding of how various treatment decisions impact the rate of visual field progression. While large clinical trials (UKGTS, LIGHT, HORIZON, etc.) allow us to understand the effect of a limited number of treatments on VF change in narrow subpopulations of patients, we often lack evidence they need to guide clinical decisions in the face of unique combinations of patient characteristics and an evergrowing array of therapeutic options encountered in real-world clinical settings. Again, interpretable and causal models developed using sizeable real-world datasets and trained to assess the impact of various treatments on VF trajectory may help us better understand the impact of our decisions on a patient's risk for rapid progression.