## Introduction

Glaucoma is a progressive optic neuropathy involving characteristic structural pathological changes in the optic nerve head.1, 2, 3 Estimates of the number of patients bilaterally blind because of glaucoma ranges between 7 and 8 million people.4, 5, 6 Glaucoma is an increasingly important public health concern due to our ageing population demographics, and has been identified as the second leading cause of blindness world-wide and the first cause of irreversible blindness.

It is irreversible because it results from the degeneration of retinal ganglion cells, and to date there is no cure for neuronal central nervous system degeneration, or means of regenerating lost neurones.

After more than a century of active research in the management of glaucoma, the ‘Holy Grail’ of glaucoma treatment is still elusive, and we have to be pragmatically content with our attempts to prevent further progression of glaucomatous damage, rather than curing the disease.

Glaucoma progression is strongly associated with a number of risk factors.7, 8, 9, 10, 11, 12, 13, 14 Some of these factors are unmanageable like ethnicity and age, while others can be manipulated, with varying degrees of success, in an attempt to slow or arrest progression, like intraocular pressure (IOP) and possibly vascular dysregulation.

Reducing IOP is presently the evidence-based, most accepted, and most practised therapeutical approach of glaucoma patients.12, 15, 16

Currently, topical ocular hypotensive medications, with its different classes, as well as filtering surgery (trabeculectomy and non-penetrating surgery) are in the forefront of therapeutic modalities to reduce IOP.15, 17, 18, 19, 20, 21, 22, 23

This review article looks at the potential advantages and disadvantages of topical medications vs filtering surgery and vice versa. It does not directly address the question of initial treatment of glaucoma,20, 21, 24, 25, 26, 27, 28, 29 or what is the better treatment30 of glaucoma, as other review articles had, but rather looks in a more specific fashion on the pros and cons of each in relation to IOP reduction. In other words, this review article deals with the situation once the decision has been made to reduce IOP.

## What is the ideal treatment to reduce IOP?

Based on available knowledge,12, 15, 18, 23, 24, 28, 30 an ideal treatment to reduce IOP should achieve different objectives. It should offer sufficient reduction in IOP, possibly in the low teens. It should provide this reduction on a long-term bases and not just momentarily or sporadically. It should be associated with minimal IOP fluctuation, IOP fluctuation being identified as a significant and independent risk factor.

It should, if possible, encourage patient compliance or, better still, be totally independent from the compliance factor. On top of all that the ideal treatment should offer tolerable systemic and local side effects, or again better still be devoid of side effects.

Last but by any means not least, in a world exceedingly aware of the heavy medical care expenses,31 an ideal treatment to reduce IOP should be economically sound.

### Sufficient reduction of IOP

Although it has been suggested that IOP reduction should be individualised to specific target pressure32 for each specific patient, in the majority of our patients, we are mostly aiming at pressures in the low teens.33

In. the Advanced Glaucoma Intervention Study (AGIS),33 eyes were randomised to laser trabeculoplasty or filtering surgery when medical therapy failed. In the roughly one quarter of eyes in the study in which the IOP was always lower than 18 mmHg (mean IOP, 12.3 mmHg) during 6 years of follow-up, there was no net change in the mean visual field score. While 15% of such eyes were said to have worsened by four AGIS visual field units, an equal percentage ‘improved’ by the same criterion, perhaps providing an estimate of the false-positive rate of progression. In contrast, in those eyes in which the IOP was greater than 17 mmHg more than half the time (average IOP, 20.2 mmHg), eyes lost an average of approximately three AGIS units during 8 years of follow-up. A clear dose–response relationship between IOP and risk of progressive field loss was evidenced for intermediate categories.

With surgery, levels of IOP in the low teens, or even lower, are technically achievable,20, 21, 34, 35, 36, 37 and that is independent of the IOP values prior to surgery. What determines IOP levels postoperatively are mainly the degree of surgical precision intraoperatively, the resistance to outflow posed by the scleral flap in trabeculectomy, at least initially. In the long run, the degree of wound scarring30 postoperatively and whether this has been manipulated by antimetabolites,21, 35 plays a role of paramount importance.

In topical medications, IOP in the low teens is less likely to be achieved,38 but nevertheless possible specially with newer classes of medications19 and with the use of combination topical medical therapies, while the level of IOP is largely dependent on IOP values before commencing therapy. Topical medications depending on their mechanisms of action, whether reducing production or increasing outflow, tend to knock down percentages of prior levels.

It is fair though to acknowledge that not every patient requires an IOP in the low teens in order to halt his glaucoma progression. In the collaborative initial glaucoma treatment study,24 it would seem that patients with mild, initial damage can do well at pressures in the mid-to-upper teens, while those with more advanced damage indeed do better when pressure is reduced to the lower teens.

## Long-term IOP reduction

In spite of initial IOP reduction of some medications, the effect seems to ware off in many cases. Watson et al39 examined the long-term efficacy of monotherapy with topically applied beta-blocking agents. Analysis showed that less than half the eyes initially treated with topical beta blockers might be expected to still be being treated with their original medication after 5 years. The rest required either additional medication or trabeculectomy.

Long-term IOP reduction capabilities of other newer classes of medications are still not fully determined.40, 41 One study19 with a 2-year follow-up found that Latanoprost significantly reduced IOP from pre-treatment values, and this reduction was maintained over the 24-month treatment period with no sign of upward drift.

Surgery also shows deterioration of its results with time. Chen et al42 studied the long-term outcomes of primary trabeculectomies that were successful at 1 year. This was a retrospective study of patients with various types of glaucoma who had trabeculectomies that were successful at 1 year and who had a follow-up of at least 10 years. In all, 40 patients (40 eyes) were enrolled, who had primary trabeculectomies that were successful at 1 year and who had a follow-up range of 10–21 years. Control of IOP was evaluated at 5, 10, and 15 years and at the last obtainable follow-up. Successful control of IOP was defined as IOP less than 21 mmHg or a reduction of 33% if preoperative IOP was less than 21 mmHg. Their results show that, if an eye was considered successful by IOP at 1 year, the probability of successful control of IOP was 82% at 5 years and 67% at 10 and 15 years. If an eye did not require further glaucoma surgery at 1 year, the probability that it would not still need further surgery at 5 years was 90%, at 10 years 75%, and at 15 years 67%. They concluded that loss of IOP control and progression of glaucomatous damage occurs over time despite initial success at 1 year.

Another study43 examined the long-term results (1–14 years) of trabeculectomies with 5-fluorouracil injections that were successful at 1 year. In a retrospective noncomparative case series. The authors identified 87 patients (87 eyes) who had trabeculectomies with 5-fluorouracil injections that were successful at 1 year and had a follow-up range of 1.0–14.7 years (mean 8.1, standard deviation of 4.4 years). All patients had previously failed glaucoma surgery (66.7%), cataract surgery (47.1%), or other diagnoses making them at high risk for failure. Successful control of IOP was defined as IOP less than 21 mmHg or a reduction of 33% if preoperative pressure was less than 21 mmHg. Statistical analysis was performed using Kaplan–Meier life table analysis. If an eye is considered successful by IOP at 1 year, the probability of successful control is 61% at 5 years, 44% at 10 years, and 41% at 14 years. They concluded that, despite successful IOP control at 1 year, trabeculectomies with 5-fluorouracil injections show a continual loss of IOP control over time.

## Minimal IOP diurnal fluctuations

Large diurnal fluctuations in IOP have been identified as an independent risk factor in patients with glaucoma. In a retrospective study of 114 patients under treatment for primary open-angle glaucoma (POAG) and ocular hypertension OHT over an 11-year period of observation, Niesel and Flammer44 described, more than a quarter of a century ago, a highly significant correlation between IOP and progression of visual field defects. This correlation could be shown for the visual field outer boundary in 81 eyes with OHT and for typical visual field defects in 33 eyes with chronic glaucoma. The relationship was, however, only significant when both the standard deviation of the annual IOP and the influence of cataract development upon visual acuity were considered. If only the mean IOP, not considering the standard deviation, is considered, this correlation is rendered insignificant. In another retrospective study45 by the same group they described clearly a significant correlation between concentric constriction during 11 years of observation and the IOP fluctuation.

Asrani et al46 have demonstrated that the diurnal IOP range and the IOP range over multiple days were significant risk factors for glaucomatous progression, even after adjusting for office IOP, age, race, gender, and visual field damage at baseline. This implies that we can no longer rely on an IOP in the statistically normal levels under treatment in office visits. Patients could still suffer from glaucomatous progression because of high fluctuations. What remains to be identified is the risk of this progression in large ‘within statistically normal’ fluctuations compared to fluctuations outside of what is statistically normal.

Migdal et al38 compared the long-term functional outcome in POAG in medically treated patients vs surgically treated patients. Among many results from this study, they observed that patients in the surgery-treated group had the lowest mean IOPs and with fewer peaks and troughs. The maximum mean IOP was 15.5 mmHg and the minimum mean IOP was 13.1 mmHg for surgery, compared with 22.1 and 15.9 mmHg for medicine.

Another study47 was designed specifically to compare the IOP fluctuations in glaucoma patients under ocular hypotensive therapy with those of patients previously submitted to trabeculectomy. IOP peaks and fluctuations for the same patients in response to the water-drinking test (WDT) were also examined. The study included 30 POAG patients using ocular hypotensive medications and with no history of previous intraocular surgery (medical group), and 30 POAG patients previously submitted to one or more trabeculectomies though taking no medication at the time of the study (surgical group). All patients were submitted to a diurnal tension curve-DTC followed by the WDT. The IOP peak and IOP fluctuation during the diurnal tension curve were significantly greater in the medical group than in the surgical group. The same was observed following the WDT. From an overall baseline IOP of 10.6 mmHg, the mean IOP change following the WDT was 13% in the surgical group and 40% in the medical group. The study concluded that patients submitted to trabeculectomy have less IOP fluctuations during the diurnal tension curve and following a water-drinking provocative test.

This observation does constitute a definite advantage of surgery over medical treatment in that respect, thus potentially offering better potential chance of stabilisation or retardation of the glaucomatous disease process. One criticism to these two studies, though, was the inclusion of patients under different classes of ocular hypotensive medications under the medical group. Different classes have different effects on IOP diurnal curves, as has been demonstrated by Orzalesi et al48 They compared the around-the-clock IOP reduction induced by timolol 0.5%, latanoprost 0.005%, and dorzolamide in patients with POAG or OHT. This was a crossover trial, 20 patients with POAG or OHT were treated with timolol, latanoprost, and dorzolamide for 1 month. The treatment sequence was randomised. All patients underwent measurements for four 24-h tonometric curves: at baseline and after each 1-month period of treatment. The between-group differences were tested for significance by means of parametric analysis of variance. To compare the circadian IOP rhythms in the POAG–OHT and control groups, the acrophases for each subject were calculated. All the drugs significantly reduced IOP in comparison with baseline at all times, except for timolol at 0300 h. Latanoprost was more effective in lowering IOP than timolol at 0300, 0600, and 0900 h (P=0.03), noon (P=0.01), 2100 h and midnight (P=0.05), and was more effective than dorzolamide at 0900 h, noon (P=0.03), and 1500 and 1800 h (P=0.04). Timolol was more effective than dorzolamide at 1500 h (P=0.05), whereas dorzolamide performed better than timolol at midnight and 0300 h (P=0.05). In this study, Latanoprost seemed to lead to a fairly uniform circadian reduction in IOP, whereas timolol seemed to be less effective during the night-time hours. Dorzolamide was less effective than latanoprost, but led to a significant reduction in nocturnal IOP. Although the study compares the IOP-lowering effect of the medications over a diurnal curve, the study lacks, or does not report, information on the differences in fluctuations in the diurnal curve between the different medications. In fact, if one compares different diurnal curves of different medications, latanaprost seems to have its curve in the lowest level, but if one examines the IOP diurnal curves presented in this study, it appears that latanaprost provides a similar range of fluctuations to timolol. Another concern is that we are judging the mean IOPs and not individual curves, theoretically single individuals might be at a higher risk of glaucomatous progression due to high fluctuations, which would not be apparent from mean IOP curves, this is of special concern in an era where individualised therapeutic decisions are stressed.

Medication class-specific or even medication-specific studies comparing IOP fluctuations in surgically vs medically controlled glaucomatous patients are in dire need, if this point is to be resolved. Though, the bulk of evidence, for the time being, seems to point to a relative advantage of surgery over medications in that respect.

## The issue of compliance and persistence

Though compliance is not a real issue in surgically treated patients, it does pose a serious challenge to the efficiency of medical treatment. 49, 50, 51, 52, 53, 54, 55

Clinically significant noncompliance with glaucoma medications has been well documented. One study56 documented the prevalence of noncompliance in a Greek cohort. Clinically significant noncompliance (more than two doses missed per week) was established in 44% of patients examined. Men and those using eye drops more than four times a day were more likely to default. Noncompliant patients exhibited higher mean IOP (22.9 vs 18.5 mmHg; P>0.001) and worse visual field loss (10.8 vs 7.0 dB; P=0.008) compared with compliant patients. Involuntary non-compliance was also common in this group, with only 53% instilling their eye drops accurately.

Another study52 was designed to assess the levels of compliance in elderly glaucoma patients on timolol eye drops. In all, 24% of patients were admitted to omitting eye drops either occasionally or frequently. A total of 51% were found to have had insufficient drops dispensed to comply with treatment as prescribed. In noncomplaint patients, the mean period without drops was 85 days of the year, with a maximum of 165 days.

Compliance could theoretically, but safely, be assumed to be much worse in developing countries compared to developed countries, for obvious reasons. It could be assumed to improve with fewer medications and fewer doses per day, more easily tolerated side effects, as well as with better understanding of the nature and gravity of the disease process.

One study57 examined the causes of noncompliance with drug regimens in glaucoma patients. The results showed that forgetfulness was the number one reported reason for noncompliance.

In the literature,49, 50, 51, 52, 53, 54, 55, 56, 57 rates of noncompliance range between 23 and 51%; whatever the rate may be in different communities and age groups, surgery has a clear advantage over medications in this respect.

One useful way in assessing compliance with eyedrop medications is the miniature compliance monitor, as purposed by Kass et al.58 The medication monitor resembles commercially available 30-ml eyedrop bottles in size, shape, and weight. It electronically records the date and time of each medication administration over a 6-week period.

Glaucoma patients’ persistence with long-term pharmacotherapy is also an issue of concern when discussing ocular hypotensive medications. Two studies59, 60 have identified persistency as a significant factor that may influence not only health outcomes, but also long-term costs and health planning. One study compared the persistency (time on initial therapy) of latanoprost vs beta-blocker monotherapy. The authors reported that patients receiving a beta-blocker as initial therapy were 3.8 times more likely to change therapy than those initially treated with latanoprost.

## Ocular side effects and complications

Surgery carries with it a set of serious ocular side effects and complications, most notably endophthalmitis. 61, 62, 63, 64 The use of antimetabolites35, 36, 43, 64, 65, 66 with trabeculectomy ups the stakes, on the one hand, better success rates are achieved, on the other higher complication rates are usually reported. Incidence of endophthalmitis is 0.2–1.5%67, 68 in trabeculectomies without antimetabolites (this should be compared to the incidence of endophthalmitis after cataract extraction, which ranges between 0.07 and 0.12%69); in trabeculectomies with 5-FU it is 3.0%70 and in trabeculectomies with MMC it is 2.1%.71

Another much dreaded, but fortunately a rare complication is expulsive haemorrhage. One study72 reported an incidence of 0.57% after trabeculectomy. Another study73 examined delayed suprachoroidal haemorrhage (DSCH) after glaucoma filtration procedures. Of a total of 1863 trabeculectomy procedures, DSCH developed in nine of 615 (1.5%) trabeculectomies without antimetabolite, 30 of 1248 (2.4%) trabeculectomies with antimetabolite.

Other complications that do occur more commonly include hypotonic maculopathy (8.9%), bleb leaks (8–14.6%), hyphaema (24.6%), and choroidal detachment (14.1%).74

Of special interest is cataract incidence after trabeculectomy. This complication, aside from its deleterious effect on vision, often necessitates another intraoperative surgery, which could adversely affect the initial trabeculectomy results. The cataractogenic effect of surgery has been well documented.24, 75 Cataract incidence post trabeculectomy has been reported to be as high as 20.2% in a follow-up of 12 months.74

It is worth mentioning that such complication rates are not reported with nonpenetrating surgery.76, 77, 78, 79, 80, 81, 82, 83 The vast majority of studies in the literature report significantly lower early postoperative complications compared to trabeculectomy. It also reports lower incidence of postoperative cataract formation. It is necessary though to acknowledge that the randomised controlled trials reporting on cataract incidence after nonpenetrating surgery do not have the same long-term follow-ups as in the case of trabeculectomy.

Ocular hypotensive medications are not devoid of ocular side effects and complications. For one, the early manifest glaucoma trial15 reported increases in clinical nuclear lens opacity gradings with medical treatment (P=0.002). Similar observation, but not statistically significant, was observed in the ocular hypertensive treatment study,84 where 6.4% of the treatment group had cataract surgeries compared to 4.3% of the observation group (P=0.06).

It is worth mentioning that the last two studies’ medication groups were under different medications. We have no information about which ocular hypotensive medication would put a patient under a higher risk of cataract occurrence.

The noxious effect of ocular hypotensive medications on the ocular surface has been long identified. One study85 reported that administration of a single topical medication preserved with benzalkonium chloride, irrespective of type, for 3 months or more induced a significant degree of subclinical inflammation, detected as increased expression of HLA-DR on conjunctival epithelial cells. Another study86 reported that latanoprost treatment induces ocular surface changes which are more evident in POAG patients who are also affected by allergic conjunctivitis. The authors hypothesised that these findings are probably related to the very high latanoprost concentration of benzalkonium chloride and to its bedtime administration, which further amplifies the toxicity.

An increasing number of studies,24, 38, 87, 88 both experimental and epidemiological, have provided evidence that filtering glaucoma surgery may be less effective than initially described. Of a number of risk factors for failure, duration and number of antiglaucoma drugs prior to surgery seem to play a critical role and highly accumulated antiglaucoma topical treatments significantly reduce the success rates.89 Inversely, trabeculectomies performed as a primary procedure do offer higher success rates than trabeculectomies performed after a history of ocular hypotensive medications.

Again, very little is known about which medication, in which patient is associated with which adverse effects, and if that is clinically relevant.

## Systemic side effects and complications

Aside from the systemic side effects of general anaesthesia, which is rarely resorted to in glaucoma surgery90, 91, 92 nowadays, surgery appears to be at a clear advantage in this respect. Ocular hypotensive medications have a long list of potential systemic side effects93, 94, 95, 96 (Table 1).

Among others, pulmonary effects of beta-blockers are exceptionally worrisome. Two drops of 0.5% timolol equates to a 10 mg oral dose. This is not enough to cause symptoms in many patients, but unfortunately glaucoma and airways disease frequently coexist. Glaucoma affects 5% of people over 65 years97 and incidence98 of asthma in elderly patients (above 65 years old) is 4% in males and 7% in females. This should not be understood in the sense that beta-blockers are contraindicated in patients above 65 years, only patients with a clinically relevant bronchial asthma does run a risk with beta-blockers.

Attempts have been made though to offer better safety profile ocular hypotensive medications. Betaxolol (Betoptic) is a cardioselective beta-blocker, but is associated with poorer IOP control.99 Carbonic anhydrase inhibitors31, 48, 100, 101, 102, 103, 104, 105, 106 applied topically are associated with less serious systemic side effects, but are also less potent than if orally administered. Prostaglandins19, 41, 48, 96 are, however, very effective at lowering the IOP with, as far as we know, minimal systemic side effects, and have become the treatment of first choice in many cases.

The advantage of beta blockers, still, is that it has been tried and tested, we know a lot about beta-blockers and we probably do not know the same amount of information about newer classes of ocular hypotensive medications. This makes beta-blockers still a very valid option.

## Economical burden of IOP-lowering strategies

The cost of surgical reduction of IOP decreases with time, where the cost of surgery can be divided by the number of years of life expectancy. The opposite is true for ocular hypotensive medications where the cost increases with time and could be multiplied by the number of years of life expectancy.

The cost of medication107, 108 for a latanaprost-treated patient per year is 337$, while it is 336$ and 288$for betaxolol and dorzolamide, respectively. The daily cost of latanaprost is 0.87$; this should be put in context with the fact that, according to the United Nations and the World Bank, more than one billion to 1.3 billion people live on a daily income of less than 1\$ (one dollar) a day. This makes glaucoma a surgical disease in most of the developing countries. Developing countries have the majority of glaucoma patents, and thus surgery is the treatment of choice to the majority of glaucoma patients. Unfortunately, very little industry research funding is being allocated for research in glaucoma surgery, which the majority of glaucoma patients are poised to benefit from.

There is evident lack of studies related to economic evaluation in glaucoma. Kobelt109 states that the genuine lack of a useful outcome measure, and the impossibility to calculate the absolute annual risk of vision loss at given levels of the one parameter that is being treated, IOP, has essentially limited the research to resource utilisation. Another limitation is that economical studies usually take in consideration data from industrialised countries, which could be misleading if applied to developing countries’ circumstances.

## Conclusions

In essence, surgery has the potential to fulfil many features of an ideal approach to reduce IOP over drugs. It can lower the IOP to low teens, achieve long-term IOP reduction, minimise IOP fluctuations, lower the cost, and minimise systemic side effects. The major drawback, though, is the potentially devastating, but rare, ocular side effects.

Although surgery is usually the first-line treatment in developing countries, it is still resorted to as a final attempt to reduce IOP in developed countries. The possibility of employing surgery as a first-line treatment is limited by the high incidence of potential ocular complications.

Beta-blockers are effective and relatively safe ocular hypotensive medications, but have a well-established list of side effects; it has been tried and tested over long-term follow-ups. The real advantage lies in the fact that the amount of information that we possess about beta-blockers is relatively large. Prostaglandins may offer certain advantages of limited side effects and effective IOP reduction, but longer follow-ups are in dire need to provide evidence of efficacy and safety.

There is evidence that nonpenetrating surgery could offer a safer option to trabeculectomy, but the widespread practice is hindered by its surgical complexity, which results in long learning curves. Being technically demanding, it is very difficult to employ in mass surgical treatment, specially in developing countries. The use of implants with nonpenetrating surgery is an extra considerable cost. Another disadvantage is that nonpenetrating surgery does not seem to achieve its previously defined target IOP, in a significant percentage, without the postoperative use of goniopuncturing, which necessitates the access to laser equipment.110

There is an urgent need to improve our surgical options in order to reduce the related ocular complications. If possible, safer and simpler surgical procedures should be developed to tackle the bulk of our glaucoma problem in developing countries.

Reducing IOP, is surgery better than drugs? As is the case in many aspects of glaucoma, indeed as in life itself, there are no easy answers to such questions. What could be at a clear advantage for one patient could be an absolute contraindication for another. In fact, many patients, specially in the developing world, do not have the luxury of an option.

We look forward to the day when effectively reducing IOP would not be such an important matter, when we can manipulate other risk factors, such as vascular autoregulation and neuronal damage, to the advantage of our glaucoma patients.