What makes a trial outcome clinically relevant? 

When no treatment exists for a blinding disease, any improvement compared to natural history may be clinically relevant. For example, while the role of photodynamic therapy in the widespread management of neovascular age-related macular degeneration was short lived, it did modify the natural trajectory of visual acuity decline, at least in some populations. Similarly, a first-in-class therapy that slows the progression of geographic atrophy, even slightly, will likely be highly clinically relevant. 

When a validated therapy already exists, noninferiority trials can help define the clinical relevance of a new treatment beyond visual acuity; e.g., by decreasing treatment burden or offering an alternative, potentially surgical, approach to management. 

Sometimes clinically relevant outcomes only become apparent with long-term follow-up. Despite monthly dosing, nearly 18 percent of patients in RIDE/RISE developed proliferative diabetic retinopathy within three years, a finding both clinically relevant and valuable in highlighting the shortcomings of pulsatile
anti-VEGF monotherapy. 

Traditionally, Food and Drug Administration approval of a pharmaceutical for a new retinal indication hinged directly on vision. More recently, the FDA has accepted anatomic biomarkers that have a valid correlation with vision to serve as the primary endpoint. Examples include optical coherence tomography-based vitreomacular traction, autofluorescence-based GA area and, most recently, fundus photography-based changes on the Diabetic Retinopathy Severity Scale. The association of improvements in DRSS scores with the clinically relevant outcomes of PDR and development of diabetic macular edema within PANORAMA are currently being debated.

Two-year results of the Diabetic Retinopathy Clinical Research Network Protocol V give us the opportunity to reconsider what is clinically relevant for DME patients with good vision. On page 38, John Pitcher, MD, and Namrata Saroj, OD, explore these outcomes. At baseline, mean central subfield thickness was 311 µm, arguably very mild DME. Between 25 and 34 percent of laser and observation eyes lost VA and received aflibercept rescue. Ultimately at two years, 16 to 19 percent of each arm experienced a ≥5-letter VA loss. 

Based on this data, if we agree that many patients with very early DME don’t warrant pulsatile treatment with current-generation anti-VEGF agents, we must also keep in mind that early DME continues to represent a non-optimal, pathologic state. In the context of DME being a major cause of global visual impairment, then we must consider that mild DME is not acceptable and remains highly clinically relevant. For the next generation, ideally more durable and less invasive treatments may be needed to shift the threshold for initiating treatment. RS