Would You Let Sleeping Dogmas Lie?

Rising obesity levels in the general population have meant that obstructive sleep apnea (OSA) – a condition where the upper airway repeatedly collapses during sleep – is now a common disorder of the adult population. People with OSA suffer disrupted sleep, as repeated interruptions to their breathing results in neurological arousal and reduced blood oxygen saturation. Unsurprisingly, recurrent episodes of hypoxia over time can have an impact on health, including increased risk of cardiovascular disease, metabolic impairment and stroke (1).

Over recent years, many have considered that there may be a link between OSA and primary open-angle glaucoma (POAG), rationalized by the following hypotheses:

• Recurrent episodes of hypoxia are associated with reduced optic nerve head perfusion
• OSA-induced hypoxia leads to an increased risk or progression in glaucomatous optic neuropathy, plus
• Mechanical factors affectingintraocular pressure (IOP) may also be involved.

If a link was established, this would provide important insights into POAG pathophysiology, particularly the contribution of vascular dysfunction, and a strong rationale for screening individuals with one condition for the other. The problem is that it isn’t clear whether OSA and POAG are linked, as previous studies have reported conflicting results: smaller prospective studies tend to report an association, whereas larger retrospective cohort studies tend to find no such association (2). When you look closer, many of these studies had significant methodological limitations in terms of sample size and study design.

Isn’t it time that someone provided some strong data? That is exactly what Michael Goldacre and his collaborators (Tiarnan Keenan and Raph Goldacre) have delivered in their recently published paper – although the results may not be entirely as anticipated by some (3).

In a cohort of over 67,000 patients with OSA – collated from linked English hospital episode statistics (1999–2011) – they observed that there was no significant increase in the risk of POAG when compared with a reference cohort (containing over 2,600,000 people; Figure 1).

For completeness, there was also no significant risk of OSA in a cohort of over 87,000 patients with POAG. Not only did they observe that OSA and POAG were not positively associated in these cohorts, the closeness of the rate ratios to one suggested little, if any, association, even when subgroups such as sex and age groups were analyzed. Interestingly, they also observed that OSA was significantly associated with POAG in the first year of admission but not in subsequent years, suggesting that additional cases of sleep apnea may have been diagnosed in the short-term at the point of POAG diagnosis (presumably from referral by ophthalmologists). Some would argue that Goldacre and his collaborators have presented strong evidence that it may be time for ophthalmologists to consider putting the hypothesis of a link between POAG and OSA to bed.

Moving on from POAG, Goldacre and his collaborators also identified a novel association between OSA and age-related macular degeneration (AMD; Figure 1), and this significantly increased risk was also observed when the OSA cohort was analyzed by sex and age. They also analyzed an obesity cohort – aware of the knowledge that obesity has been identified as a risk factor for both OSA and AMD – and revealed that the relationship between obesity and AMD was not as strong as the observed association between OSA and AMD, indicating that OSA had a significantly independent effect.

But what does this all mean? We spoke with Tiarnan Keenan – the corresponding author for the study – for his thoughts on the findings and where the field may be headed.

What were the driving factors for this research?

My original motivation was based more around examining the potential link between OSA and glaucoma. The idea of an association between these two conditions has been generating a lot of interest in recent years – and has important implications – but existing studies in this area have been relatively small and inconsistent. By using a ‘big data’ approach through a collaboration with Michael Goldacre, Professor of Public Health at the University of Oxford, we were able to perform the most highly powered single study to date to address this question.

But I also have a particular interest in AMD. So while we were planning the study, I wanted to see if sleep apnea might be associated with an increased risk of AMD. At that time, only one small study (4) had reported a related finding, that patients with retinal diseases responding poorly to anti-VEGF therapy had higher risk of sleep apnea. We were extremely well placed to perform this analysis; we had previously undertaken some of the largest studies in similar areas, for example,  demonstrating the absence of any association between Alzheimer’s disease and AMD (5), and a small association between arthritis 
and AMD (6).

Did the result that OSA was not associated with POAG surprise you?

I was not particularly surprised by this. I think that the rhetoric in this area had rather outstripped the evidence. While there were plausible biological reasons to consider a link between sleep apnea and glaucoma, the published data so far had been inconsistent. In particular, previous studies had generally been relatively small in size and had methodological limitations.

In fact, I was slightly surprised that the rate ratios in our analysis were so precisely and consistently close to 1 (i.e. no increased risk in either temporal direction). This was despite substantial interest in this area in recent years, which could have led to an artefactual finding of a positive association (through ophthalmologists referring many additional glaucoma patients for sleep studies).

Indeed, our analysis of the association by time interval provided some interesting insights in this respect, showing that additional diagnoses of sleep apnea were generated in the short-term following glaucoma diagnoses, but that these diagnoses were just brought forward in time rather than representing a genuinely positive association.

Were you surprised that OSA had a significantly independent effect on AMD when compared with obesity?

I was really interested (but not overly surprised) by this finding. Evidence has been accumulating that OSA is linked to a whole host of other conditions, including cardiovascular disease, metabolic impairment and stroke. In addition, we have some evidence that sleep apnea leads to increased systemic oxidative stress, inflammation and VEGF up-regulation, all of which are strongly implicated in AMD pathogenesis. It therefore seemed highly plausible to consider a link between OSA and AMD.

Previous well-known epidemiological studies did examine potential associations between AMD and various systemic conditions. However OSA was generally overlooked in these studies. With the advent of ‘big data’ approaches such as ours, and increased interest in the field of OSA, I believe that the time was ripe for this discovery.

What impact do you expect your findings to have?

Our findings have important implications. For a start, it’s important to emphasize how common OSA is. Around one in five adults has OSA, but over 85 percent of individuals with clinically significant OSA have never been diagnosed. This includes the elderly population, where undiagnosed OSA is very common but tends to
be overlooked.

In clinical practice, we have often had a tendency to treat neovascular AMD in relative isolation from the patient and any other medical conditions. At its most extreme, this has meant treating the optical coherence tomography (OCT) rather than the individual. With the knowledge that sleep apnea increases the risk of neovascular AMD and interferes with its treatment, we may now need to probe deeper into our patients’ medical history to reduce their risk of visual loss. If more conditions follow, we may be facing a paradigm shift where our patients’ sleep and general health are just as important as their OCT and fundal findings. 

As retinal specialists, we certainly have strong precedents for this in other conditions that we treat, particularly diabetic retinopathy and retinal vascular occlusion.

Do you feel that a questionnaire on sleep symptoms is going to be enough?

The use of a validated sleep questionnaire (e.g. the Berlin Questionnaire, with just 10 questions) is definitely a good starting place – this is ideal as a screening tool. I believe that we should definitely now be using these on neovascular AMD patients labeled as ‘poor responders’ to anti-VEGF therapy – ideally we should actually be using them on all neovascular AMD patients at the point of diagnosis. Treating sleep apnea at this point would provide a great opportunity to improve the visual outcomes with anti-VEGF therapy, as well as improving patients’ sleep and general health. But our results in this paper suggest we could be using questionnaires even earlier. If treating sleep apnea turns out to reduce the excess risk of neovascular AMD, we should be giving sleep questionnaires to all AMD patients, not just those who have progressed to neovascular disease.

Clearly, while validated sleep questionnaires are ideal as screening tools, the gold standard for diagnosis is polysomnography. So patients could be referred for sleep studies on the basis of the questionnaire results, in order to improve detection rates and use services most efficiently. In fact, this has already been happening to some extent – with local variation – in the glaucoma field, particularly by glaucoma specialists who have taken a special interest in sleep apnea. So the precedent exists. It would seem a good time to enter into talks with sleep physicians to discuss these ideas and refine referral pathways.

What will be the key benefits for patients?

Those with neovascular AMD will benefit in two ways. First, those with undiagnosed sleep apnea should have better visual outcomes and reduced anti-VEGF treatment burden with prompt therapy for sleep apnea. In addition, they should have improved sleep and general health if their sleep condition is picked up following AMD diagnosis.

How will your findings benefit ophthalmologists?

Ophthalmologists should benefit in turn. With anti-VEGF therapy as the current mainstay for neovascular AMD treatment, the label of ‘poor responder’ can mean a frustrating barrier in the ophthalmologist’s ability to offer sight-saving care. The recognition of any condition with implications for neovascular AMD risk or treatment is of great benefit to the ophthalmologist. Apart from anything else, picking up and treating sleep apnea may mean fewer injections and more money saved for macular services (which are stretched in the UK).

Do you think your findings will benefit other clinicians?

Sleep physicians and scientists are also likely to be interested in our findings. We have discovered another important condition where OSA increases risk, which underlines the importance of their work and research. The advice given to individuals diagnosed with sleep apnea should therefore expand, adding to the motivation for improving their sleep and using continuous positive airway pressure where applicable. I would suggest they are also asked to consult an optometrist at the point of OSA diagnosis, and made aware of the link with AMD.

What will be key to determine in future research?

First, I would be very interested to see our novel findings replicated in another country. Beyond that, an important question is whether OSA increases the risk of early AMD and/or progression to geographic atrophy (GA), as well as risk of neovascular AMD. Another vital question is whether there is any interaction with the main genetic variants for AMD – i.e. CFH-CFHR at chromosome 1 and ARMS2/HTRA1 at chromosome 10. For example, sleep apnea might contribute to both GA and neovascular AMD but only in those with CFH risk variants (i.e. complement-driven AMD). Alternatively, it might increase risk of progression from early AMD to neovascular disease in all individuals, irrespective of genotype. Clearly we will need to make these important distinctions in the future, particularly in the context of increasingly personalized medicine. Ultimately, the most important question we need to answer is whether treatment of sleep apnea reduces the excess risk of neovascular AMD that we have discovered.

Where to next?

We are conducting further research into other systemic conditions of hypoxia. We are very interested to see whether there is something unique about OSA, or whether other hypoxic conditions such as pulmonary diseases also carry increased AMD risk. Any expansion to our list of associated conditions has important implications for clinical practice.