Systemic Therapies for Inflammatory Eye Disease
In most 'inflammologies' (rheumatology, renal medicine and the rest) the last decade has been characterized by major multicentre studies that have established the efficacy and safety of established or novel therapies in a randomized controlled format (vs. placebo or standard care). In inflammatory eye disease we have performed very few such studies, but encouragingly there has been a major effort to address the blocks that have hitherto limited our progress. A striking characteristic of this new phase is the extent to which the whole inflammatory eye disease community is working together with increasing international and cross-specialty collaboration. As we consider these issues below we have focused on uveitis, but the arguments apply equally well to most other forms of inflammatory eye disease (such as ocular mucous membrane pemphigoid, autoimmune keratitis, scleritis, myositis, inflammatory orbitopathies and inflammatory neuropathies):
Uveitis has an annual incidence of 14–50 per 100 000, with a prevalence of around 38–115 per 100 000 in the general population. Since the majority of these cases are acute anterior uveitis responsive to topical therapy, the population of patients who need systemic therapy (and are eligible for clinical trials of immunosuppressants) is small. Other inflammatory diseases such as scleritis have a higher proportion requiring systemic therapy, but are overall much less common. It is partly in response to this, that increasing collaboration has emerged in recent years. One example is the SITE consortium, a collaboration of five academic ocular inflammation practices in the United States who undertook a retrospective cohort study primarily to investigate long-term adverse events occurring in patients on systemic therapies for ocular inflammatory disease; although not the primary aim, this study also provides interesting data on outcomes in the use of these drugs. The study group comprised 7957 (of whom 2340 were treated with systemic immunosuppression) patients observed over 68751 visits, spanning 14.910 person years. Whilst accepting the limitations inherent in a retrospective study, this is an important milestone in being the largest dataset of immunosuppression in ocular inflammatory disease.
Ocular inflammatory disease is very heterogeneous group. Within uveitis, classification is generally by anatomical grouping and etiology, including the presence or absence of systemic disease. 'Splitting' i.e. defining a very pure cohort (e.g. Posterior uveitis of Birdshot pattern) leads to a very small target population, whereas 'lumping' enables easier recruitment but may be clinically meaningless due to the range of disease included. This is true both for routine clinical practice and for clinical trials. In clinical trials maximization of the signal: noise ratio is critical. An intervention may be highly effective for a disease but will fail to return a statistically significant benefit if it is trialed on too broad a group of patients (e.g. patients who have conditions that look superficially similar but differ fundamentally in etiology).
In addition to its heterogeneity as a group, the uveitis subtypes are imperfectly defined with no diagnostic criteria with high enough sensitivity and specificity to reliably separate all cases of one uveitis entity from another. This leads to considerable variation in classification and diagnosis. To some extent such markers may only become apparent as we better understand disease etiology. This has occurred in the field of retinal dystrophies where there has been a shift from grouping by somewhat arbitrary pictorial descriptions (e.g. butterfly-shaped macular dystrophy) to defining disease by the gene responsible (e.g. a peripherin/RDS retinal dystrophy). This is important not only because it leads to a more clear-cut and objective classification of disease, but also because a classification based on real differences in etiology is more likely to translate into an appropriately targeted therapeutic approach.
Within uveitis the definition of syndromes by pattern and the lack of clear diagnostic markers leads to significant variation between clinicians in the classification of uveitis entities. Currently a major international collaboration - the Standardization of Uveitis Nomenclature consortium - is seeking to define classification criteria for 28 different uveitis syndromes. The consortium have identified 194 uveitic terms and 'dimensions' including 87 unique terms that are specifically used by uveitis specialists to describe signs and symptoms. These were mapped to the 28 uveitic syndromes under consideration. Currently 250 cases per uveitis syndrome are being gathered which will be used to validate the mappings and to form the basis of a classification criteria and a proposed ontology for uveitis. Although this system is based on pattern rather pathogenesis, this project does have the potential to significantly improve standardization of uveitis classification with benefits for both clinical practice and trials.
One of the advantages of managing inflammation in the eye (as opposed to elsewhere in the body) is that the transparent nature of the cornea and visual axis enables us to see and score inflammatory activity and titrate treatment accordingly. The flip-side is that we have accepted these same subjective activity indices in clinical trials. Non-invasive technologies which can objectively quantify the key parameters of cellular activity, flare and vitreous haze are needed. With regard to the assessment of AC flare, laser flare meters have potential but current models are relatively time-consuming and cumbersome, with some concerns over variability especially in the non-ideal patient (e.g. posterior synechiae, patient mobility). Some models have sought to also quantify cellular activity but with limited success. With regard to the posterior segment the NEI vitreous haze score described by Nussenblatt et al. (which depends on the clinician scoring the clarity of the optic disc which can be compared to a standard set of photographs) is the key outcome recognized by the FDA. Although it is the gold-standard, it has a number of limitations including that it is (1) subjective; (2) non-continuous, leading to very large steps in disease activity between categories; (3) poorly discriminatory at lower levels of vitreous haze, with most cases of active uveitis being scored at 0.5+ or 1+; and (4) limiting of recruitment and sensitivity in a clinical trial context (where a 2 point change is usually required to be counted as a significant change).
An adaptation of this technique proposed by Davis and colleagues is to score clarity on photographs (rather than on the live biomicroscopic view of the NEI technique) which can then be compared to a more extensive set of reference images leading to a potentially more discriminatory 0–9 score (vs. the 0, 0.5, 1, 2, 3 and 4 scale of the SUN modification of the Nussenblatt). Overall however this is still a subjective technique and shares most of the key limitations of the standard vitreous haze score. A truly objective measure is likely to be based on quantification of vitreous density or reflectivity using current or future imaging technologies.
Whilst it is absolutely right that our key aim is to retain and restore vision, we must recognize that visual acuity is a very poor marker of the efficacy of a drug in inflammatory eye disease. The impact of uveitis on visual acuity will depend on both the activity of the disease and the damage caused by the disease. Thus whilst vision may improve due to treatment-induced improvement in cystoid macular edema, vitritis, keratic precipitates and aqueous clarity, this benefit may be obscured by loss of vision due to cataract, band keratopathy, macular scarring or even glaucoma. A further complexity arises in that some of these types of damage are reversible (notably cataract and band keratopathy) leading to the slightly bizarre scenario that their correction has to be specifically prohibited within most clinical trials. Additionally it is impossible to accurately quantify the extent to which each factor is contributing to the reduced visual acuity. Although the clinician is used to making these judgment calls every day in clinic - for example when deciding the extent to which advancing cataract or persistent cystoid macular edema is responsible for a fall in vision – these estimates are very approximate and frequently shown to be incorrect. It should also be recognized that visual acuity is a subjective parameter which despite every effort towards standardization may be affected by factors such as patient mood, general health and compliance. Visual outcome remains the core purpose for clinical trials in inflammatory eye disease, but it is less clear how it should be utilized as a trial endpoint.
The increasing and important focus on patient reported outcome measures (PROMs) as a component of clinical trials may start to capture some of these aspects of the overall benefit experienced by the patient. It is encouraging that more recent clinical trials in uveitis have often incorporated measures both of quality of life and health utility. These include specific vision-related quality of life measures (such as the 25-item NEI-Visual Function Questionnaire, NEI-VFQ25 or the Vision-related Quality of Life Core Measure, VCM-1) general health-related quality of life measure (such as the short form-36, SF-36) and health utility measures (such as the EuroQol 5-dimension, EQ-5D, and Visual Analogue Scores).
In the past pharmaceutical companies have shown little interest in ocular inflammatory disease, partly due to the relatively small market. Encouragingly the last decade has seen a surge of interest with industry-sponsored studies of both novel agents – such as the calcineurin antagonist voclosporin and the anti-IL17 agent secukinumab – and of novel routes for older drugs (e.g. intravitreal corticosteroid implants, intravitreal sirolimus) and providing ultimately robust evidence of anti-TNF blockade.
Reflecting on Now: The Age of Enlightenment
In most 'inflammologies' (rheumatology, renal medicine and the rest) the last decade has been characterized by major multicentre studies that have established the efficacy and safety of established or novel therapies in a randomized controlled format (vs. placebo or standard care). In inflammatory eye disease we have performed very few such studies, but encouragingly there has been a major effort to address the blocks that have hitherto limited our progress. A striking characteristic of this new phase is the extent to which the whole inflammatory eye disease community is working together with increasing international and cross-specialty collaboration. As we consider these issues below we have focused on uveitis, but the arguments apply equally well to most other forms of inflammatory eye disease (such as ocular mucous membrane pemphigoid, autoimmune keratitis, scleritis, myositis, inflammatory orbitopathies and inflammatory neuropathies):
Problem 1: Small Target Population
Uveitis has an annual incidence of 14–50 per 100 000, with a prevalence of around 38–115 per 100 000 in the general population. Since the majority of these cases are acute anterior uveitis responsive to topical therapy, the population of patients who need systemic therapy (and are eligible for clinical trials of immunosuppressants) is small. Other inflammatory diseases such as scleritis have a higher proportion requiring systemic therapy, but are overall much less common. It is partly in response to this, that increasing collaboration has emerged in recent years. One example is the SITE consortium, a collaboration of five academic ocular inflammation practices in the United States who undertook a retrospective cohort study primarily to investigate long-term adverse events occurring in patients on systemic therapies for ocular inflammatory disease; although not the primary aim, this study also provides interesting data on outcomes in the use of these drugs. The study group comprised 7957 (of whom 2340 were treated with systemic immunosuppression) patients observed over 68751 visits, spanning 14.910 person years. Whilst accepting the limitations inherent in a retrospective study, this is an important milestone in being the largest dataset of immunosuppression in ocular inflammatory disease.
Problem 2: Heterogeneous Disease Groups
Ocular inflammatory disease is very heterogeneous group. Within uveitis, classification is generally by anatomical grouping and etiology, including the presence or absence of systemic disease. 'Splitting' i.e. defining a very pure cohort (e.g. Posterior uveitis of Birdshot pattern) leads to a very small target population, whereas 'lumping' enables easier recruitment but may be clinically meaningless due to the range of disease included. This is true both for routine clinical practice and for clinical trials. In clinical trials maximization of the signal: noise ratio is critical. An intervention may be highly effective for a disease but will fail to return a statistically significant benefit if it is trialed on too broad a group of patients (e.g. patients who have conditions that look superficially similar but differ fundamentally in etiology).
Problem 3: Poorly Defined Phenotypes
In addition to its heterogeneity as a group, the uveitis subtypes are imperfectly defined with no diagnostic criteria with high enough sensitivity and specificity to reliably separate all cases of one uveitis entity from another. This leads to considerable variation in classification and diagnosis. To some extent such markers may only become apparent as we better understand disease etiology. This has occurred in the field of retinal dystrophies where there has been a shift from grouping by somewhat arbitrary pictorial descriptions (e.g. butterfly-shaped macular dystrophy) to defining disease by the gene responsible (e.g. a peripherin/RDS retinal dystrophy). This is important not only because it leads to a more clear-cut and objective classification of disease, but also because a classification based on real differences in etiology is more likely to translate into an appropriately targeted therapeutic approach.
Problem 4: Diagnostic Inconsistency
Within uveitis the definition of syndromes by pattern and the lack of clear diagnostic markers leads to significant variation between clinicians in the classification of uveitis entities. Currently a major international collaboration - the Standardization of Uveitis Nomenclature consortium - is seeking to define classification criteria for 28 different uveitis syndromes. The consortium have identified 194 uveitic terms and 'dimensions' including 87 unique terms that are specifically used by uveitis specialists to describe signs and symptoms. These were mapped to the 28 uveitic syndromes under consideration. Currently 250 cases per uveitis syndrome are being gathered which will be used to validate the mappings and to form the basis of a classification criteria and a proposed ontology for uveitis. Although this system is based on pattern rather pathogenesis, this project does have the potential to significantly improve standardization of uveitis classification with benefits for both clinical practice and trials.
Problem 5: Subjective Outcome Measures
One of the advantages of managing inflammation in the eye (as opposed to elsewhere in the body) is that the transparent nature of the cornea and visual axis enables us to see and score inflammatory activity and titrate treatment accordingly. The flip-side is that we have accepted these same subjective activity indices in clinical trials. Non-invasive technologies which can objectively quantify the key parameters of cellular activity, flare and vitreous haze are needed. With regard to the assessment of AC flare, laser flare meters have potential but current models are relatively time-consuming and cumbersome, with some concerns over variability especially in the non-ideal patient (e.g. posterior synechiae, patient mobility). Some models have sought to also quantify cellular activity but with limited success. With regard to the posterior segment the NEI vitreous haze score described by Nussenblatt et al. (which depends on the clinician scoring the clarity of the optic disc which can be compared to a standard set of photographs) is the key outcome recognized by the FDA. Although it is the gold-standard, it has a number of limitations including that it is (1) subjective; (2) non-continuous, leading to very large steps in disease activity between categories; (3) poorly discriminatory at lower levels of vitreous haze, with most cases of active uveitis being scored at 0.5+ or 1+; and (4) limiting of recruitment and sensitivity in a clinical trial context (where a 2 point change is usually required to be counted as a significant change).
An adaptation of this technique proposed by Davis and colleagues is to score clarity on photographs (rather than on the live biomicroscopic view of the NEI technique) which can then be compared to a more extensive set of reference images leading to a potentially more discriminatory 0–9 score (vs. the 0, 0.5, 1, 2, 3 and 4 scale of the SUN modification of the Nussenblatt). Overall however this is still a subjective technique and shares most of the key limitations of the standard vitreous haze score. A truly objective measure is likely to be based on quantification of vitreous density or reflectivity using current or future imaging technologies.
Problem 6: The 'Distraction' of Visual Acuity
Whilst it is absolutely right that our key aim is to retain and restore vision, we must recognize that visual acuity is a very poor marker of the efficacy of a drug in inflammatory eye disease. The impact of uveitis on visual acuity will depend on both the activity of the disease and the damage caused by the disease. Thus whilst vision may improve due to treatment-induced improvement in cystoid macular edema, vitritis, keratic precipitates and aqueous clarity, this benefit may be obscured by loss of vision due to cataract, band keratopathy, macular scarring or even glaucoma. A further complexity arises in that some of these types of damage are reversible (notably cataract and band keratopathy) leading to the slightly bizarre scenario that their correction has to be specifically prohibited within most clinical trials. Additionally it is impossible to accurately quantify the extent to which each factor is contributing to the reduced visual acuity. Although the clinician is used to making these judgment calls every day in clinic - for example when deciding the extent to which advancing cataract or persistent cystoid macular edema is responsible for a fall in vision – these estimates are very approximate and frequently shown to be incorrect. It should also be recognized that visual acuity is a subjective parameter which despite every effort towards standardization may be affected by factors such as patient mood, general health and compliance. Visual outcome remains the core purpose for clinical trials in inflammatory eye disease, but it is less clear how it should be utilized as a trial endpoint.
The increasing and important focus on patient reported outcome measures (PROMs) as a component of clinical trials may start to capture some of these aspects of the overall benefit experienced by the patient. It is encouraging that more recent clinical trials in uveitis have often incorporated measures both of quality of life and health utility. These include specific vision-related quality of life measures (such as the 25-item NEI-Visual Function Questionnaire, NEI-VFQ25 or the Vision-related Quality of Life Core Measure, VCM-1) general health-related quality of life measure (such as the short form-36, SF-36) and health utility measures (such as the EuroQol 5-dimension, EQ-5D, and Visual Analogue Scores).
Problem 7: Low Commercial Interest
In the past pharmaceutical companies have shown little interest in ocular inflammatory disease, partly due to the relatively small market. Encouragingly the last decade has seen a surge of interest with industry-sponsored studies of both novel agents – such as the calcineurin antagonist voclosporin and the anti-IL17 agent secukinumab – and of novel routes for older drugs (e.g. intravitreal corticosteroid implants, intravitreal sirolimus) and providing ultimately robust evidence of anti-TNF blockade.