The times they are a‐changin’: time to change glaucoma management

Abstract

We live in a time with vastly increased and clinically important knowledge of glaucoma. For a long time, glaucoma was somewhat of a stepchild in ophthalmology. Critics stated that glaucoma experts could not agree on how to define the disease, how to treat it, how to follow it and – by the way – maybe treatment made no difference. That led to a climate, where it was hard to give solid information to colleagues and healthcare providers on how glaucoma care could be organized and implemented, and to give evidence-based advice to patients. This in turn sometimes made it difficult to secure enough resources for glaucoma care. Clearly, fundamental clinical research was needed to fill in important knowledge gaps to provide a solid foundation for good glaucoma care. The large randomized glaucoma trials – often referred to as the ‘alphabet soup’ [The Advanced Glaucoma Intervention Study (AGIS 1994), The Collaborative Initial Glaucoma Treatment Study (Musch et al. 1999). The Collaborative Normal Tension Glaucoma Study (CNTGS 1998), The Early Manifest Glaucoma Trial (EMGT) (Leske et al. 1999), The Ocular Hypertension Treatment Study (OHTS) (Gordon & Kass 1999) and The European Glaucoma Prevention Study (EGPS) (Miglior et al. 2002)] – were designed to provide evidence for glaucoma care, and through them, the most important research needs have now been met. Important knowledge has also come from other clinical studies. Many of the important knowledge gaps that existed 10 years ago have thus been filled. This means that it is time to consider whether clinical glaucoma care has taken advantage of the new knowledge, or whether changes are now motivated. What is it that we have learned in the last decade? The most fundamental issue is that we now know that reducing intraocular pressure helps. The Early Manifest Glaucoma Trial and CNTGS have shown that lowering intraocular pressure (IOP) reduces the risk of progression, that is, treatment prolongs time to progression and, therefore, rate of disease progression (Collaborative Normal-Tension Glaucoma Study Group 1998; Heijl et al. 2002). Patients whose eyes pressures are always quite low show only little progression (AGIS Investigators 2000). The Ocular Hypertension Treatment Study clearly showed that pressure reduction in patients with ocular hypertension reduces the incidence of glaucoma damage (Kass et al. 2002). The randomized trials have also shown that treatment effects are surprisingly large. In EMGT, a mean IOP reduction of 5 mmHg reduced the risk of progressing by half (Heijl et al. 2002; Leske et al. 2003). Also in OHTS, the same IOP reduction reduced glaucoma incidence by half (Kass et al. 2002). Actually risk reductions were 10–14% per mmHg in EMGT (Leske et al. 2003), OHTS (Gordon et al. 2002) and EGPS (Miglior et al. 2007). In the more recent Canadian glaucoma Study, which included treated patients with lower IOP levels than the other studies, risk reduction was as high as 19% per mmHg (Chauhan et al. 2008a). These findings are clinically very important, because they show that over a long-time perspective an extra pressure reduction of just a few mmHg might make a meaningful or sometimes even a great difference (cf. below). As the percentages of risk reduction seem to be larger in studies with lower IOP levels, the results also suggest that, for example, 2 mmHg additional pressure reduction might be more important in a patient progressing at lower IOP levels than in a patient with higher pressure. A fundamental piece of knowledge is that IOP reduction leads to decreased risk of progression even in eyes with IOP within the statistically normal range (Collaborative Normal-Tension Glaucoma Study Group 1998; Heijl et al. 2002). Glaucoma is a multifactorial disease, and we must remain hopeful that we in the future may successfully exploit treatment modalities other than IOP lowering, that is, treatment that addresses vascular insufficiency or offers neuroprotection. Yet, even today, we must remember that further reducing IOP may be very helpful in treated eyes that do not do well despite the fact that measured IOP is always ‘normal’. Another important observation is that most patients with glaucoma do progress if monitored for long enough with at least moderately sensitive tools. This is so even if IOP is always measured within the statistically normal range. Thus, for example, in EMGT, 59% of patients randomized to the treatment arm had shown definite progression, a large proportion with all IOP readings within ‘normal’ limits (Leske et al. 2007). Therefore, progression paradigms in glaucoma have changed. It used to be considered that any progression was a reason to step up treatment or at least to consider such a step. Now, it is realized that a change of treatment depends on the magnitude of the progression, and whether the progression rate is high enough to constitute a threat to the quality of life (QoL) of the patient during his remaining lifetime. The trials have also shown that also early progression can be identified with great statistical power using standard automated perimetry, if only field testing is performed often enough, and event analyses are used to identify progression. Thus, definite visual field progression in EMGT was associated with a worsening of mean deviation (MD) by <2 dB (Heijl et al. 2003) with a specificity of at least 99% per test (Heijl et al. 2008). This indicates that at least 15 steps of progression can be identified with perimetry between a normal field and a perimetrically blind field (Fig. 1). Perhaps surprisingly, this is a considerably larger number of steps of progression than with present-day structural imaging techniques. Thus, sensitivity to change, which depend on long-term fluctuation, and dynamic range is still somewhat smaller with optical coherence tomorgraphy (OCT), laser polarimetry and confocal laser tomography where the number of steps between a normal and a glaucoma blind eye, is approximately 6–10 (Jampel et al. 2006; Leung et al. 2008, 2011) at specificities of approximately 95%. One likely reason for this is that event analyses for perimetry have had considerably longer time to develop than those for structure, and available perimetric event analyses for the detection of change exploit data from individual test point locations and can detect local, focal progression (Fitzke et al. 1996; Bengtsson et al. 1997). With frequent perimetry, progression can be ascertained early with high statistical power. In Early Manifest Glaucoma Trial, definite progression was associated with a mean deterioration of mean deviation of <2 dB. The diagram depicts the number of independent progression events that could theoretically be detected with frequent perimetric testing. Presently, structural tests can diagnose glaucoma with high sensitivity and specificity (Bengtsson et al. 2010), but for follow-up, structural tests may best be performed as ancillary tests, but not at the expense of visual field testing (Chauhan et al. 2008b). The trials and other studies have also shown that rates of progression vary tremendously among patients, in clinical glaucoma care, and also in those few studies that have followed glaucoma patients untreated. Thus, some patients progress extremely slowly, while others are fast progressors – often defined as those who show a worsening of MD by 1.5 or 2 dB or more per year. We therefore cannot guess the progression rate of individual patients very well. Today, there is at least some knowledge on commonly encountered rates of progression. In treated glaucoma, mean rates often seem to be perhaps 0.4–0.6 dB/year, but with the very large interindividual variability. Natural history data are available form CNTGS and EMGT. In the latter study, the average rate was a little over 1 dB/year in the whole patient cohort and similar in patients with high-tension glaucoma (IOP ≥21 mmHg) (Heijl et al. 2009) (Fig. 2). This corresponds to going from a normal field to perimetric blindness in approximately 30 years. Progression rates are considerably slower in normal-tension glaucoma (IOP ≤ 20 mmHg) with mean rates of approximately 0.4 dB/year both in CNTGS and in EMGT patients (Anderson et al. 2001; Heijl et al. 2009). Patients with untreated exfoliation glaucoma in EMGT progressed at disturbingly high rates; the mean was over 3 dB/year, which corresponds to going from a normal field to perimetric blindness in approximately 10 years. Progression rates in untreated glaucoma have also been calculated from cross-sectional epidemiological data (Quigley et al. 1996). Rates of progression vary greatly among patients with glaucoma, but also among groups of patients. This graph shows mean rates of progression in untreated glaucoma patients from Early Manifest Glaucoma Trial (Heijl, Bengtsson et al. 2009). The fact that rates of progression cannot be guessed has resulted in new advice on management of patients with newly detected glaucoma. Frequent perimetry is required the first years after diagnosis, if we want to be able to detect rapidly progressing eyes before too much additional damage has occurred. Thus, to detect an eye that progresses by 2 dB/year in 2 years with 80% statistical power, we need to perform three visual field per year, during those first 2 years (Chauhan et al. 2008b). Such frequent perimetry will not only help detect rapid progressors; it will also help to give us measured individual rates of progression. Establishing such rates of progression as a part of standard glaucoma care is now also part of modern management recommendations, for example, by the European Glaucoma Society (2008) and by the Swedish Ophthalmological Society (Heijl et al. 2010). Perimetric progression rates in treated glaucoma tend to be linear if the measurement unit is logarithmic, for example, MD values or mean sensitivites of visual field index (VFI) values (Bengtsson et al. 2009). This means that we can assume with at least some level of confidence that if a glaucoma patient continues with the same treatment as during the period when we measured his rate of progression, or with similar IOP values, it is likely that he will continue to progress at approximately the same rate. This, of course, is of considerable help, when we need to consider future treatment. Glaucoma is still the second most common cause of blindness. Many years ago, Hattenhauer et al. (1998) found in a retrospective chart review in Minnesota that after 20 years of follow-up after diagnosis 22% of patients who were initially diagnosed with manifest glaucoma with field loss were blind in both eyes. In a recent study by Forsman et al. (2007) in a well-defined geographical catchment area in Finland, 15% of patients with glaucoma were blind in both eyes at the last visit before death. One reason is that glaucoma is often diagnosed late with considerable damage (Grødum et al. 2002), increasing the risk for blindness. This is a problem in itself. In ophthalmic care, we can help early detection by always inspecting the optic disc and measure intraocular pressure in patients over 50, who undergo ophthalmic examinations for any reason. This would only be a partial help; however; a screening programme must probably be developed and subjected to large-scale testing in the field to solve the detection problem. But clearly most patients who develop visual impairment or blindness from glaucoma do so while undergoing treatment. Many clinicians have noted that more drastic measures to reduce IOP, for example, surgery, are often not performed until most of the visual function is already lost in the treated eye. The question is why this is so. One reason can be that glaucoma management is based far too much on tonometry alone. This was obvious, for example, in a national questionnaire, reported in a systematic literature review of open-angle glaucoma by the Swedish Board of Health Technology Assessment, which showed that patients with glaucoma were subjected to perimetric testing only approximately every second year (Lindén et al. 2011). There are also data from the US which indicate that in ordinary clinical practice perimetry is performed too rarely to detect rapidly progressing patients or establish rates of progression within just a few years after diagnosis (Quigley et al. 2007). The concept target pressure is certainly valuable, and it makes a lot of sense to base target pressure on the risk of future impairment or loss of QoL. In this way, patients with more damage and younger age should have lower target pressures. We must realize, however, that the initial target pressure is just a guess based on risk factors. After measuring a patient’s individual rate of progression, target pressure should be redefined (cf. below) (European Glaucoma Society 2008; Heijl et al. 2010). Today patients with glaucoma should be followed not only with tonometry – to ensure that the target pressure has been met – but also with repeated measurement of glaucoma damage. It is only by measuring damage that we can assess rate of progression to see whether the initial target pressure was adequate or whether target pressure has to be lowered further and treatment be stepped up. Currently, it is preferable to measure damage with perimetry for two reasons: One is that perimetric results are results of visual function testing. They show how much visual reserve that is available, and we know pretty well where on the visual field scale that QoL is affected in a clear way. Certainly, structural parameters frequently show progression in patients with glaucoma, but the agreement with perimetry is small (Chauhan et al. 2001; Leung et al. 2011). With structural testing, this relationship is presently quite unclear. Further, we still do not have tools to translate structural measurements, like optic disc rim area or retinal nerve fibre layer thickness, to the functional domain. Another reason to presently prefer field testing is the earlier recognition of progression with perimetry than with structural testing (cf. above) and probably a smaller test-retest variability of global visual field measures (MD or VFI) than of global structural measurements (Jampel et al. 2006). Currently, structural testing, particularly with OCT, is very helpful for diagnosis (Bengtsson et al. 2010), but probably considerably less valuable for follow-up of patients with established glaucoma with field loss. This may change in the future with the present fast development of imaging technologies and techniques for statistical interpretation of imaging methods. The age-function diagram is very useful in glaucoma management (Fig. 3). This diagram has age on the x-axis and a global index of visual function on the y-axis. Usually, the index is MD that stands for the mean deviation from the age-corrected normal threshold value. Mean deviation is expressed in dBs. A normal visual field has an MD of around 0 dB, while a perimetrically blind eye has an MD value of approximately −30 dB or even less, in the Humphrey perimeter. The MD value in the Octopus perimeter is very similar but has a positive sign in defective fields, since MD in the Octopus means mean defect. Plotting perimetric test results over time in the age-function diagram gives an immediate and intuitive picture of how a glaucoma eye is developing. The age-function diagram is a powerful and intuitive tool in modern glaucoma management. Age is plotted on the x-axis and visual function on the y-axis, usually with diminishing function downwards. The unit on the y-axis is usually mean deviation in the Humphrey system or mean defect in the Octopus system. An arbitrary line can be drawn, for example, at the level of MD = −15 dB or a VFI value of 50%. Vision-related quality of life is clearly and measurably reduced in patients who have field loss at this level or worse in their better eye. A goal of glaucoma management is to prevent loss of QoL [European Glaucoma Society (EGS) Guidelines]. Reduced QoL is often not obvious in patients with glaucoma. Patients with glaucoma as a group often show no differences in QoL compared with age-matched controls when tested with general QoL tools (Wändell et al. 1997) or with vision-specific tools when analyses are age-adjusted (Wu et al. 2008). Similarly, patients with glaucoma in CIGTS demonstrated excellent quality of life even when tested with vision-specific QoL instruments (Wren et al. 2009). Quality of life is certainly related to the stage of disease and with MD and VFI scores in patients with glaucoma (Hyman et al. 2005; van Gestel et al. 2010; Sawada et al. 2011), while measurable utility seems to remain unaffected until very late disease stages (van Gestel et al. 2010) at least in some studies. Yet, some studies have identified measurable reduction in vision-related QoL even with early-stage glaucomatous visual field loss (Wu et al. 2008). It is therefore difficult to define a maximum lifetime maximum target for glaucomatous visual field loss. There is pretty good agreement, however, that QoL has certainly been reduced when at least half the visual field has been lost in a patient’s best eye, that is, at an MD value of approximately −15 dB or worse. This, of course, does not mean that progression at better levels is innocuous, but at least keeping visual function above the −15 dB level during the patient’s entire remaining lifetime is an important goal, which is not always easy to meet. There are similar diagrams in most perimeters (Fig. 4), but despite the fact that they have been available in all Octopus and Humphrey perimeters for more than 20 years, these printouts are unknown to most users and much underutilized. Mean deviation plots have been available in the Humphrey and Octopus perimeters for over 20 years. Even if they do not have age on the x-axis and even if the time axis of the Humphrey plot is not proportional to the time between tests, they are very valuable tools to assess disease development. Despite their clear value for clinical management, these tools have been much underutilized, and many perimeter users do not even know that they have access to these printouts. A modern development of the age-function diagram exists in the Humphrey perimeter, in the glaucoma progression analysis (GPA®) program (Fig. 5). The new diagram has been developed to be as easily comprehensible as possible. It has also been placed on the same sheet of paper as the result of the most recent (plus two baseline) fields with the hope that clinicians will choose the printout mode as a routine for all patients with glaucoma who are followed up over time, so that the trend is always visible in routine glaucoma care. In this newer diagram, MD has been replaced with VFI – the Visual Field Index, which is expressed in % instead of in dBs (Bengtsson & Heijl 2008). The time axis is correct and shows the patient’s age now and also for the first fields and for all follow-up fields. The modern VFI plot of the GPA program of the Humphrey perimeter is an age-function diagram, showing rate of progression. If data are sufficient, the current trend will be extrapolated up to 5 years, representing an educated guess of how the patient will progress if treatment is not changed, or if IOP remains approximately the same. Usage of age-function diagrams is quite intuitive. A few years after diagnosis with enough data points in the diagram, it is quite easy to see the general trend. It is important to look at the slope of the regression line rather than at the p-value. The slope is often of considerable importance even if it is statistically nonsignificant. Three different scenarios are depicted in Fig. 6 and commented in the legends of the figure. As mentioned earlier, one should extrapolate the regression line for the patient’s expected lifetime plus some more years, to see whether the current trend is associated with great risk of loss of QoL. If so, intensified treatment should be considered. Naturally, the functional status of both eyes is of importance, but in ophthalmological treatment, decisions are often made for each eye separately. Therefore, while the binocular field, or the field in the better eye, is of greatest importance, we should look at the development of each eye separately and adjust treatment accordingly. After 2 or 3 years of repeated visual field testing, rate of progression can usually be assessed with some certainty. The slope of the line is intuitively comprehensible. Projecting the current slope for the patient’s remaining lifetime helps determine the risk of serious loss of quality of life. With a slow and ‘safe’ slope, treatment can be maintained unchanged (A), while with a very steep slope (C), radical change of target pressure/treatment must be considered. In the common ‘in between’ situation (B), one should consider lowering IOP further particularly if this can be achieved with little risk of side-effects of complications. The age-function diagram may also be useful at diagnosis or during the first few years after diagnosis, when rate of progression has not yet been determined. The functional status of each eye may be plotted in the diagram. Initial treatment decisions can then be made depending on the risk of future loss of QoL (Fig. 7). Present level of functional damage and patient age, and life expectancy are of great importance. Even if we cannot guess the rate of progression of the individual patient or eye, a general slope, as shown in the safe and unsafe zone diagram, may be of some help. It is of course not difficult to understand that a young glaucoma patient, for example, 65 years of age, has a much larger risk of suffering loss of vision-related QoL than a considerably older patient with the same level of damage. Neither is it difficult to understand that between two patients of the same age but with considerably different levels of damage, the patient with more damage is at larger risk. Nevertheless, plotting or looking at age and level of damage in an age-function diagram is so simple and immediately comprehensible, that it might remind us to treat different patients differently, with both lower target pressures and increased attention and closer follow-up in patients at higher risk. Conversely, the age-function diagram can help us not overdoing glaucoma management in old patients with very little loss, who therefore have a very small risk of ever suffering measurable losses of QoL. Before rate of progression has been measured, this age-function diagram with ‘safe’ and ‘unsafe’ zones could be of some value. The slanted white line between the zones represents the median rate of progression in a large study of clinically treated patients. The two eyes in (A), that have the same amount of damage, obviously, have very different risks to progress to the red zone with more advanced damage. The eye marked with red has a much higher risk, because the younger patient has a longer life expectancy. Similarly in (B) with two eyes of patients of the same age, the red symbol eye carries a much higher risk of severe field loss and impairment. In situations where the measured rate of progression is unsatisfactory, one may use the age-function diagram to consider other treatment options. If the measured rate is much higher than acceptable (e.g. Fig 6C), one may immediately conclude that IOP must be reduced quite drastically. If measured IOP levels have been reasonable (e.g. 16–20 mmHg), the new target levels must be very low, probably 10–12 mmHg, which will require rather profound changes of management, including surgery. In situations where progression rates and risks are not so high, but still unacceptable, one may use the age-function diagram to try to estimate the effects of small and moderate reductions of the progression rate. Considering the results on the relationship between risk of progression and IOP from the large trials (cf. above), a reasonable guess is that the rate of progression may be diminished by 10–15% per mmHg of further IOP reduction. Also rather modest reductions in pressure, which may be achieved by adding one more drug, going to combination drops or performing laser treatment, may then result in considerable gains over a long-time perspective (Fig. 8). The patient may gain several years without large losses of vision-related QoL. Even a very modest extra reduction of IOP may make a big difference in the long run, saving many years of good sight and retained quality of life. Knowledge on glaucoma has increased considerably during the last decade, and glaucoma management can now stand on a much firmer evidence-based foundation than before. Treatment works and treatment effects are large. Nevertheless, entirely ‘safe’ IOP levels cannot be reached without great cost in side-effects, risks of surgical complications, comfort and money. Instead treatment must be tailored to the individual needs of each patient. Treatment aims at preventing vision-related losses of QoL, and thus assessing this risk must be part of all management decisions. We have learned that progression is the rule even in patients who always have IOP readings within the statistically normal range. The individual rate of progression is very variable and impossible to predict. Target pressure therefore is an educated guess only and must be reassessed depending on disease development, that is, progression of glaucoma damage. Tonometry alone is not enough. Therefore, a new standard in modern glaucoma management is the need to establish individual rates of progression in the great majority of patients with manifest glaucoma. This will require more frequent perimetric examinations during the first few years after diagnosis, than what has been common in the past. Quality of life is closely related to visual field status. The age-function diagram therefore is a powerful tool in modern glaucoma management. At diagnosis, this diagram with the safe and unsafe zone concept can be used to help set the initial target pressure and decide initial treatment. After a few years when glaucoma damage has been assessed five or six times and rate of progression has been established, the age-function diagram can help us determine whether the measured rate is safe or unsafe. At that stage, it can also be used to help estimate the long-term effects of changes of treatment strategy and to identify those patients who need radical changes to treatment before too much further visual damage has occurred. The changes needed to improve standard glaucoma care to a level where we can always answer a patient’s question on how he is changing over time, and on the risks of future problems are not great. Tradition and faulty notions about the difficulties of perimetry are larger threats to progress than lack of economical resources. First of all, glaucoma is not an expensive disease for society and healthcare providers. Secondly, perimetry is not very expensive. As an example, increasing the frequency of visual field testing in Sweden from an average of one field test every second year to one every year has a lower total cost for the whole country, than what a single university department spends on anti-vascular endothelial growth factor drugs every year. If reimbursement does not cover that cost, it is up to us as a profession and to patient organizations, to strive to change the rules. Glaucoma management needs to be individualized. Treatment and follow-up should be tailored to the needs of each patient, based on the risk of future losses of vision-related QoL. In this way, we should be able to limit the impact of glaucoma damage and lessen the burden of impairment caused by glaucoma.