Inherited Retinal Disorders: Using Evidence as a Driver for Implementation

Susceptibility to human diseases is typically influenced by the complex interplay between genetic, environmental/lifestyle and stochastic factors. The relative contribution of genetic alterations to disease predisposition is variable, and genetic disorders form a continuum: on the one end of the spectrum are common conditions (e.g. age-related macular degeneration and diabetic retinopathy) that are influenced by multiple factors to the degree that the direct effect of individual genetic changes is blurred; on the other end of the spectrum, there are conditions like rod-cone dystrophy, for which the identification of defects in a single gene can predict disease development with relatively high accuracy (Fig. 1) [1-3]. The term monogenic or Mendelian is used to describe the latter group of disorders. Although each individual condition in this group is rare (defined in the EU as affecting less than 1 in 2,000 individuals [4]), collectively they are common, and their cumulative impact on affected families and healthcare systems is substantial [5, 6]. This article focuses on monogenic retinal disorders and discusses the importance of building a strong evidentiary foundation for genetic ophthalmology.

The modern era of retinal genetics began in 1986 with the identification and cloning of RB1, the gene associated with retinoblastoma [7, 8]. Two years later, mutations in the OAT gene were found to cause gyrate atrophy [9] and, in 1990, the rhodopsin (RHO) gene was linked to dominant rod-cone dystrophy [10] and the CHM gene was linked to choroideremia [11, 12]. In the three decades that followed these early breakthroughs, over 270 genes have been associated with retinal disorders (Fig. 2) [13]. These discoveries have ushered a new era for the field of retinal genetics and catalysed the development of powerful genomic tests that have revolutionised diagnostics for monogenic retinal disease [14, 15]. It has also been possible to translate this growing genomic knowledge into targeted interventions. Notable examples include gene augmentation therapies [16], anti-sense oligonucleotide treatments [17] and pre-implantation genetic screening [18].

Contemplating such progress is gratifying, but the challenges lying ahead remain considerable. A central problem is that of integrating research findings into healthcare practices and policies [19-21]. Important considerations include the cost, the benefit and the evidentiary basis of each proposed intervention. Genetic testing is discussed here as an example: although the importance and value of genetic testing for monogenic retinal disorders has been repeatedly highlighted over the past decades [22-24], variation in the current provision of testing remains significant [25, 26]. Considering economic factors is necessary but insufficient. Two important questions that need to be answered are how beneficial are these tests and how strong is the evidence that supports their routine use?

In 2017, the US National Academies of Sciences, Engineering, and Medicine released a report titled “An Evidence Framework for Genetic Testing” [27]. This document discusses the clinical applications and usefulness of genetic tests, and examines how relevant evidence is generated, evaluated and synthesised. A key concept in this report is that of clinical utility defined as “the ability of a test to improve clinical outcomes measurably and to add value for patient management decision making compared with current management without genetic testing” [27, 28]. This term has been used in the context of genetic testing for over 20 years and, over this period, it has been construed both narrowly and broadly [29-31]. Scholars who refer to clinical utility by its narrowest definition focus on the ability of a screening or a diagnostic test to lead to an improved health outcome (impact on mortality, morbidity, disability; e.g. excising a Wilms tumour detected following a genetic test for aniridia). Conversely, broader definitions may include any change in management (e.g. preventing additional investigations or introducing personalised surveillance measures) or any outcome that is important to the affected individual or family (e.g. early resolution of uncertainty, better understanding of condition, effect on reproductive or life planning) [31]. It is noteworthy that other test parameters such as analytical validity (i.e. the ability to accurately identify variants of interest) and clinical validity (i.e. the diagnostic accuracy) are related but not overlapping with clinical utility (Table 1). Also, clinical utility is not tied to cost-effectiveness as the latter can be only evaluated at the site level. At present, rigorous evidence on the clinical utility of genetic testing for monogenic retinal disorders is lacking – as indeed it is lacking for most other monogenic disorders (i.e.it is unclear who should be tested, with what test and when). Notably, there is no consensus on what constitutes sufficient evidence to justify implementation of genetic testing – are randomised trials required, or can observational studies aligned to mecha-nistic reasoning suffice? Clearly, randomised trials are expensive and hard to justify if there is no genuine uncertainty among experts (principle of equipoise) [32]. In any case, evidence gaps need to be identified and addressed.

Conducting clinical research/trials in rare disorders like monogenic retinal disease poses unique challenges. A key issue is how to avoid conducting underpowered studies. Two evident obstacles are disease heterogeneity and geographic dispersion of affected individuals. The paucity of natural history data poses another important barrier; without a firm understanding of disease progression, choosing meaningful outcome measures and designing and powering clinical trials is challenging [33-36]. Furthermore, many monogenic retinal disorders primarily affect children adding further complexity to potential study design [34]. It is apparent that to overcome these hurdles, stakeholder engagement at an unprecedented level will be required [37].

In recent years, several initiatives have focused on improving our ability to conduct research on rare diseases in general and on monogenic retinal disorders in particular. A notable example is the European Reference Network on Rare Eye Diseases (ERN-EYE), an initiative with a leading role in the development of key infrastructure such as international patient registries and computational tools to standardise terminology [38, 39]. Another example is a forum/workshop organised by the Association for Research in Vision (ARVO) and Foundation Fighting Blindness in 2016 with the goal of identifying appropriate trial endpoints for inherited retinal disorders [40]. Outside ophthalmology, substantial efforts have been put on rethinking clinical trial design for small populations, defining core sets of patient-centred outcome measures, optimising error rates when there are constraints on the available sample size and identifying models of “real-world” evidence collection [33, 35-37, 41-43]. However, a standardised framework integrating all these elements is still missing.

In December 2017, the US Food and Drug Administration (FDA) in a landmark decision approved the first gene augmentation therapy for an inherited disorder. The drug was voretigene neparvovec, and it can be used to treat RPE65-associated retinal dystrophy in adults and children [44]. In November 2018, approval by the European Medicines Agency (EMA) was also obtained, and many European countries are presently conducting Health Technology Assessments to decide whether the therapies should be approved for reimbursement [45, 46]. Numerous other gene therapies are presently under development or undergoing benefit-risk assessment. There is little doubt that some of these ongoing trials will demonstrate safety and efficacy. However, how many will meet the required standards of evidence for reimbursement bodies remains to be determined [47]. Nonetheless, it only is by conducting high-quality trials and by gathering as robust evidence as possible that we can resolve clinical agnosticism and ensure a significant impact [48].

To obtain sufficient sample sizes for research, we need to engage geographically, socially and economically diverse patient populations. To reach as many affected individuals as possible, the effort of all ophthalmologists who see patients with monogenic retinal disorders will be required. A few steps that could accelerate the translation of our growing genetic knowledge into effective interventions include: (i) ensuring that all individuals with monogenic retinal disorders that come to medical attention receive an accurate diagnosis [49]; (ii) capturing clinical information efficiently [14, 50] and collecting natural history data in a rigorous and systematic way (e.g. in registries [51]); (iii) appreciating the statistical challenges and focusing on optimising research designs [33-36, 52, 53] (Table 2).

I have been most fortunate to work with so many generous and gifted people over the years. I am especially grateful to my mentors Graeme Black, Andrew Webster and Tony Moore. This work is partly supported by the UK National Institute of Health Research (NIHR) Clinical Lecturer Programme (CL-2017-06-001).

The author has no conflicts of interest to declare.