The Problem With Dry Eye Trials

In this issue of JAMA Ophthalmology, McCann et al¹ are to be commended for undertaking a critical review of the literature on the efficacy and adverse effects of various dry eye therapies. Overall, 71 systematic reviews evaluating interventions for dry eye were included, and of the 26 systemic reviews deemed reliable, no conclusive evidence was identified for the effectiveness of any dry eye intervention. These interventions included environmental modifications, polyunsaturated fatty acid supplementation, artificial tears and lubricants, punctal occlusion, intense pulsed light therapy, vectored thermal pulsation therapy (Lipiflow [Johnson & Johnson]), topical corticosteroids, topical cyclosporine A, topical secretagogues, and autologous serum. Of these interventions, the authors suggest that the strongest, yet still inconclusive, evidence existed for the use of air humidification, punctal plugs, and topical corticosteroids for dry eye symptoms. When examining these findings, questions arise as to why, despite having so many studies examining dry eye therapeutics (>8000 PubMed hits with this term), data regarding efficacy are lacking.

The first possible explanation is that most dry eye studies are recruiting patients with a very heterogenous group of syndromes, conditions, and diseases. As commonly used, the term dry eye encompasses diseases with different pathologic mechanisms (eg, inflammation, stem cell depletion, neurosensory dysfunction) that cause ocular surface pain (frequently characterized as dryness, grittiness, burning, irritation) and/or visual disturbances and/or present with some collection of tear instability, corneal epithelial disruption, decreased tear production, and/or alterations of the eyelid margin.² This means that distinct syndromes, conditions, and diseases, including Sjögren disease, ocular rosacea, nocturnal lagophthalmos, and neuropathic pain, are frequently lumped into the single diagnostic category termed dry eye. It should come as no surprise in the current era of personalized medicine that a single therapy is not likely to be effective for a set of diverse conditions with different pathologic causes.

A serious limitation in the field of ocular surface disease is our relative inability to accurately subtype an individual’s specific problem based on the underlying disease mechanisms. As such, our ability to deliver precision therapy for their specific condition is limited. This translates into enrolling individuals into clinical trials whose therapeutic interventions do not target the underlying problem. For example, not all individuals with dry eye symptoms have detectable inflammation on the ocular surface.³ However, these individuals may still meet the inclusion criteria for enrollment into a clinical trial that targets ocular surface inflammation, thereby reducing the power of the study. Furthermore, as we have learned from the field of uveitis, there are many different pathologic patterns of ocular inflammation, each requiring a different optimal therapy. Not surprisingly, studies have similarly demonstrated different tear inflammatory profiles among patients with symptoms and signs of dry eye.⁴ This likely explains why symptomatically, some individuals prefer cyclosporine-based products whereas others prefer lifitegrast.⁵ As a community, we need to better subtype dry eye into categories based on treatable mechanisms. Current diagnostic tests are largely insufficient for this, highlighting the trouble in recruiting a relatively homogeneous population of patients with dry eye to assess a new therapy.

A second issue hampering these clinical trials is their design. Most dry eye studies are designed to meet a prespecified symptom and sign end point, despite our clear understanding that patients are focused on relief of their symptoms, which do not always correlate with physical signs.⁶ Instead, it is understood that nerve activation drives symptoms,⁷ but current diagnostic tests do not robustly evaluate nerve function. Another important issue is that selected end points are often not well aligned with therapeutic mechanisms. For example, epithelial disruption is often used as an end point, even though this physical sign is influenced by a variety of factors outside the scope of study (eg, air pollution, abnormal corneal sensation, anatomic abnormalities, iatrogenic insult in the examiner’s office). A rethinking about biological relevance and appropriate end points is needed to improve clinical trial design and interpretation.

A third issue is that current therapeutic trial interpretation is limited by the fact that comparisons are often with a control and not with another active therapy. This reality limits translatability of clinical trial results into real-world application.

A fourth issue that is critical and rarely discussed is the importance of having well-trained experts examine the patients entered into these trials, ensuring that the inclusion and exclusion criteria are truly being met. If a thorough eye examination is not performed, including everting the upper eyelid and examining the posterior eyelid margin and conjunctiva, to evaluate for other ocular surface disorders, study outcomes may be put in jeopardy. Enrolling individuals with lagophthalmos, superior limbic keratoconjunctivitis, anterior basement membrane dystrophy, and Salzmann nodules as a cause of dry eye symptoms and signs into clinical trials designed to address other aspects of tear health is problematic, reducing the power of a study to detect the effectiveness of a new intervention. In fact, we frequently see individuals diagnosed with dry eye in the setting of these other entities, some of whom are participants in the latest multicentered study for a new therapeutic.

A final possible explanation, albeit cynical, is that there are simply no effective dry eye therapies currently available. This must be considered given the many published studies evaluating the multiple different forms of therapy developed for dry eye with the conclusions noted in the recent article.

Beyond issues with the trials themselves, it is important to also consider the limitations of systematic review designs and meta-analysis strategies. Only 37% of the systematic reviews included in the study by McCann et al¹ were deemed reliable, with incomprehensive literature search and inappropriate meta-analysis being the key methodologic limitations. The spectrum of quality among systematic reviews on ocular surface disease corroborate with findings from the recent Tear Film and Ocular Surface Society Evidence Quality Report⁸ and emphasizes the importance of considering potential bias in the evidence synthesis process and whether or not the methods are appropriate for evaluating the research question. Inappropriate use of meta-analyses, eg, by not properly accounting for included study designs or incorrectly computing the weight of included studies, has the potential to mislead findings, particularly if clinical or methodologic diversity across studies is not carefully considered.

In the end, “absence of evidence is not evidence of absence,” as attributed to Martin Rees and/or Carl Sagan. Although it is important to recognize the limitations of our current knowledge base and to design thoughtful studies to target these gaps, we still need to try to effectively treat the millions of individuals worldwide who have dry eye. At the same time, we need to understand that, in the US alone, billions of dollars per year are being spent on dry eye therapies, with a lack of strong evidence to support the effectiveness of many of these treatments. Taking into account findings highlighted by McCann et al¹ in the current issue of JAMA Ophthalmology, clinicians can combine a balanced consideration of findings from systematic reviews while at the same time blending the science and art of medicine to treat our patients to the best of our abilities.