There is insufficient evidence to recommend lens extraction as a treatment for primary open-angle glaucoma: an evidence-based perspective
Conflict/competing interest: None declared.
Dr Mark J Walland, 55 Victoria Parade, Fitzroy, Vic. 3065, Australia. Email: email@example.com
Cataract extraction in primary open-angle glaucoma has not been thought to provide a clinically useful or predictable decrease in IOP. This concept has now been challenged, with the opposite belief being promulgated: namely, that lens exchange should be considered as treatment for glaucoma. This revelation could bring a significant change in the glaucoma treatment paradigm.
There are no randomised controlled trials to guide the role of lens extraction in primary open-angle glaucoma. The available evidence suggests at most a modest reduction in IOP from cataract extraction – greater in the presence of pseudoexfoliation – which is likely to be of marginal benefit, and only in milder forms of open-angle glaucoma. There is currently no evidence of any quality to suggest that lens extraction routinely represents a clinically useful treatment for primary open-angle glaucoma.
Cataract surgery is a vision-restoring and sometimes life-transforming operation. The surgery is performed by the majority of general ophthalmologists, so the skill is not concentrated in the hands of just a few specialists.
There is evidence from randomized clinical trials on the role of cataract extraction in the management of primary angle closure disease;1,2 recent reviews have provided guidelines for their application and possible extrapolation to the role of clear lens extraction in this condition.3,4
In the treatment of primary open angle glaucoma (POAG), cataract extraction has been considered to provide a modest and variable decrease in intraocular pressure (IOP) that cannot be relied upon to predictably and adequately lower the IOP for the objectives of glaucoma management. An evidence-based review in 2002 concluded that there was only weak evidence for a 2- to 4-mmHg reduction in IOP after extracapsular cataract extraction or phaco-emulsification cataract surgery, and that sustained IOP control was only achieved in those who underwent combined cataract and trabeculectomy surgery.5 Another recent review concluded that ‘cataract surgery alone may be of limited clinical benefit in lowering IOP (in POAG)’.6
A recent retrospective and unmasked study has, however, concluded otherwise and suggested that the IOP lowering from phaco-emulsification is effective and safe enough to consider lens exchange as treatment for glaucoma and possibly for prevention of progression in ocular hypertension.7 Postulates have also been provided to suggest that the lens could be a ‘phacomorphic’ cause for the glaucoma even with open-angles. Wide exposure of this work – particularly in the trade press – has meant that an increasing trend is developing which considers lens extraction as routinely providing a benefit in lowering IOP, especially where the preoperative IOP is high,a–d even extending to a claim that ‘in patients with IOP greater than 25 mm Hg, one can achieve 10 mm Hg drops in IOP (with cataract surgery alone) vs. 13 mm Hg for tube and trabeculectomy’.e
Given the level of ambient ‘chatter’ on this issue, and the putative causal role of the lens in this disease, we believe it possible that a change in glaucoma practice patterns may be occurring. Especially for any treatment with potential for widespread application, it is important that any such change be not only evidence-based, but also based on evidence appropriate for the question being asked. We therefore wish to critically evaluate the peer-reviewed literature and formulate preliminary guidelines for situations where the IOP-lowering effects of lens extraction alone may be appropriate as treatment for POAG.
The studies investigating the IOP lowering associated with cataract surgery have been well summarized.8 We confine our attention to the effects in POAG (Table 1), having considered the effects of lens extraction in angle closure disease in detail elsewhere.3,4
Table 1. Summary of publications which included details on the effect of phaco-emulsification cataract extraction on IOP in primary open-angle glaucoma
|Kim 19999||R||31||16.4 months||18.1 ± 3.1||15.2 ± 2.9||−2.9 ± na||P < 0.001|
|Shingleton 199910||R||61||12 months||17.0 ± 4.9||15.9 ± 4.0||−1.1 ± na||NS|
|Hayashi 200111||P||68||24.1 months||20.7 ± 5.4||15.2 ± 3.8||−5.3 ± 4.8||P < 0.0001|
|Mathalone 200512||R||24||24 months||17.0 ± 4.6||15.1 ± 3.2||−1.9 ± 4.9||NS|
|Shingleton 200613||R||55||5 years||18.4 ± 3.4||16.6 ± 3.5||−1.8 ± 3.5||P = 0.005|
|Poley 20097||R||124||0−10 years||17.8 ± 0.9||15.1 ± 2.9||−2.7 ± 4.3||P < 0.001|
|Shoji 200714||R||35||36 months||16.7 ± 1.1||15.6 ± 3.4||−1.1 ± na||P = 0.145|
|Fea 201015||P||21||15 months||17.3 ± 3.0||15.7 ± 1.1||−1.6 ± 3.2||na|
|Radcliffe 201016||P||32||3−60 months||na||na||−0.45 ± na||P = 0.4597|
|Samuelson 201117||P||117||12 months||18.4 ± 3.2||17.4 ± na||−1.0 ± 3.3||na|
The literature concerning the effect of cataract extraction on IOP in POAG is muddied by frequent uncertainty about whether a diagnosis of POAG was actually established. If this statement seems facile, we emphasize that the term POAG implies that the angle is open. Accordingly ‘POAG is a diagnosis of exclusion that can be made only once gonioscopy has been performed’ and secondary causes excluded.18 The inadequacy of angle and other findings to fully characterize the type of glaucoma in some reports may reasonably cast doubt on the conclusions reached. Despite the recognized role of cataract surgery in primary angle closure disease, the use of gonioscopy is rare in most publications that primarily report IOP lowering following cataract surgery for POAG.
Cataract surgery and intraocular pressure
A variable decrease in IOP has been noted with extracapsular cataract extraction.19 Early reports with phaco-emulsification showed a modest but inconstant IOP decrease of around 1.0–3.5 mmHg.9,20 In 1999, a retrospective study of 143 patients (164 eyes) undergoing clear corneal phaco-emulsification stratified cases into no glaucoma, glaucoma suspect or glaucoma eyes.10 Angle findings or classification of glaucoma type was not mentioned, but eyes with Pigment Dispersion Syndrome or pseudoexfoliation (PXF) were included. A statistically significant IOP reduction was seen in the non-glaucoma group at 12 months' follow up. In both the glaucoma suspect and glaucoma groups, there was no significant IOP difference in the glaucoma cases at 12 months, although a reduction in glaucoma medication usage was seen. The authors felt that ‘clear corneal phaco-emulsification should not be performed with the intent of achieving better IOP control; but the IOP would likely remain stable and possibly slightly reduced at 1 year.’ In a subsequent retrospective report 3- and 5-year follow up was provided and IOP decreases of 1–2 mmHg were noted. The wide confidence intervals prevented valid predictions and there was no change in use of glaucoma medications.13
Retrospective findings of clear corneal phaco-emulsification in 345 eyes which included 58 eyes with medically controlled open-angle glaucoma – but specifically excluded narrow angle eyes – showed a statistically significant reduction in IOP at 12 and 24 months' follow up in the control but not in the glaucoma group. Results are interpreted cautiously given losses to follow up of around 60% in both groups at 2 years. The authors speculated that the IOP decrease only in the non-diseased eyes might be explained by a healthier trabecular outflow mechanism.12 A further confounding factor in retrospective studies such as these is that eyes with a higher postoperative IOP (perhaps glaucoma eyes) may be less likely to become lost to follow up because of their high IOP.
In a retrospective review in 2003, the effects on the IOP of cataract extraction in 297 PXF patients and 427 non-PXF patients were studied.21 A decrease in mean IOP was seen in each group, and was greater in the PXF group. The authors speculated that trabecular lavage during phaco-emulsification surgery may have produced the larger decrease in the PXF group; the possibility of preoperative angle compromise secondary to lax zonules was not mentioned.22 Gonioscopy findings were not reported for these studies, but we imagine the angle would have been accurately documented for the glaucoma patients.
A prospective study found modest (−1.85 mmHg) IOP reduction post phaco-emulsification in 71 PXF eyes compared to 112 non-PXF eyes (−0.62 mmHg; P = 0.0037). Both groups had 29 glaucoma patients.23 The authors mention that decrease in IOP was associated with increased volume of irrigation, but the data suggest the finding may merely be a surrogate marker of PXF, indicating a need for more irrigation in PXF cataract cases. Another prospective, but non-randomized study of the effects of phaco-emulsification with clear corneal incision in patients with PXF syndrome demonstrated a 3.5-mmHg reduction in IOP in pseudoexfoliatives versus a 0.48-mmHg reduction in controls at 1 year postoperatively.24
Data from these publications indicate what is already accepted: a small reduction in IOP – that is more marked in the presence of PXF – may occur with phaco-emulsification in an open-angle. Most authors and reviews conclude that the quantum of IOP reduction following phaco-emulsification is neither sufficient nor adequately predictable for most established POAG cases.
On the other hand, two recent papers7,25 are prominent in current discussions around the role of cataract extraction as treatment for POAGa–d and their high exposure merits a more exhaustive review.
In the first, a retrospective review was undertaken of 588 eyes which were either normotensive or ocular hypertensive undergoing cataract extraction.25 The eyes were stratified into five groups based on preoperative IOP; the greatest decrease in IOP after surgery was seen in those eyes with the highest preoperative IOP. This finding was independent of age, and was maintained at 1 and 10 years after surgery. There was no mention of gonioscopy findings.
The second paper retrospectively reviewed the long-term effects of phaco-emulsification and intraocular lens implantation in glaucomatous eyes.7 A total of 124 eyes were again stratified into five groups depending on preoperative IOP. It is not clear when the IOP categories were selected, but from the retrospective nature of the study it is likely this was during the data analysis phase. The authors state that only eyes with glaucoma were reviewed, but also indicate that gonioscopy was not recorded in many patient charts and was thus not reported. The number of IOP measurements on which the analysis was based is also not clear. The patients were a heterogeneous mix with ‘all eyes (preoperatively having) … treatment with glaucoma drops, surgery (trabeculectomy or iridectomy), or both or had deterioration in the visual field, optic nerve, or both’. Eyes with the highest mean preoperative IOP (24.7 mmHg) had greatest decrease in postoperative IOP (34%), whilst the group with the lowest preoperative IOP had a mean 16% increase in IOP (37% had a decrease; 56% had an increase). In addition, the majority of eyes used the same or more drops after surgery (65 same; 17 increased eye-drop usage).
In reviewing the relevant literature on the effect of cataract extraction on postoperative IOP, the authors acknowledge the greater IOP decreases reported in the angle closure as compared to the open-angle studies.26,27 We remain uncertain – given this foreknowledge – why the authors did not seek to correlate any angle metrics with the preoperative IOP. We are also uncomfortable with the notion that a published cohort of glaucoma cases could lack mention of gonioscopy findings or exclusion of secondary glaucoma. One might speculate that the IOP reductions of greater magnitude following lens extraction represent the effect of lens removal in undiagnosed angle closure or perhaps PXF. It also remains to be explained why many glaucoma eyes on treatment should have such elevated preoperative IOP if truly POAG. With the possible exception of one cohort of the myocilin family of glaucoma patients,28 extreme elevation of IOP is more commonly seen in either primary angle closure glaucoma (PACG), juvenile or secondary forms of glaucoma (such as steroid-related or uveitic). The authors hypothesize a role for the lens in the causation of open-angle glaucoma. Their sophistry on the putative effect of cataract surgery in decompressing trabecular beams – quoting historical work of Mann29 and Tripathi and Tripathi30 in both papers – certainly cannot be entertained until the preoperative status of the angles is clarified.
More importantly, and basic to any critical analysis is the fact that the retrospective study design makes it difficult to attribute causality. The lack of masking and absence of consideration of regression to the mean detract from the authors' logic. Furthermore, the subsequent speculation does not fulfil any modern concept of causality. And even if all this were not so, their explanation that the expansion of the crystalline lens with age compresses trabecular beams still remains speculative: it fails to account for why IOP is not routinely elevated in all older patients (even given a greater prevalence of glaucoma with increasing age), and discounts the uncertain association even between angle closure and lens thickness.31–33 If such ‘phakomorphic’ glaucoma were causative or even part of a ‘sufficient component’ cause model, one might also expect a higher prevalence of POAG in regions with a high prevalence of cataract. That is clearly not the case. The authors also do not address why patients whose IOP is low preoperatively tended to have an increase in IOP despite the same morphological changes presumably being brought by removal of crystalline lens bulk. This result hints at the possibility of regression to the mean; those with high baseline IOPs could naturally also be subject to this effect. The authors' efforts to redefine ‘phakomorphic glaucoma’ are therefore premature.
Anterior segment changes from cataract surgery
Changes in angle contour and anterior chamber depth (ACD) would perhaps be less likely to be significant in eyes that truly had open-angles. An increase in ACD and chamber angle has been noted angle closure glaucoma with no significant difference seen between these parameters in open-angle glaucoma versus controls.26 It has also been reported that while IOP was significantly reduced in glaucoma patients following cataract surgery, a larger percentage of patients with PACG (40.0%) did not require glaucoma medications postoperatively compared to POAG (19.1%).11 The IOP decrease after phaco-emulsification is reported to be more in eyes with a shorter ACD; a large ACD led to short-term increase in IOP (which the authors speculated might have been due to less comprehensive clearance of viscoelastic).34 Angle closure eyes are known to occasionally produce a dramatic spike in immediate postoperative IOP:35 one study demonstrated a similar increase in open-angle glaucoma patients as well.36
A study of Anterior Segment Optical Coherence Tomography changes following cataract surgery in 53 eyes (eight with ‘glaucoma’) found no predictive value of preoperative Anterior Segment Optical Coherence Tomography for postoperative IOP, although the majority of the angles were ‘open’ so the result is perhaps unsurprising. Gonioscopy findings were not provided.37 In contrast, a novel index for predicting the degree of IOP reduction based on a ratio of the preoperative IOP to ACD (pressure to depth ratio) has been reported.38 There was a consistently greater than 4-mmHg IOP reduction in patients with a pressure to depth ratio of more than 7 (IOP measured in mmHg/ACD in mm). The ACD was found to change (the authors state ‘decrease’), on average, by 1.10 mm postoperatively. The authors believe that any role of ACD is likely related to the effect of lens thickness on this parameter. Gonioscopy was regrettably not part of the protocol for the volunteers recruited for the study. It should be noted that unlike regression, correlation cannot be used to predict the IOP drop: correlation is used to establish the closeness of the association between two variables rather than predict the effect of one on the other. Also, both correlation and regression require assumptions not necessarily fulfilled in the study.39
Phaco-emulsification with previous trabeculectomy surgery
There are several reports of the postoperative IOP following phaco-emulsification performed in the setting of a prior functioning trabeculectomy filter (and most often, therefore, also an iridectomy), and this has been generally associated with a slight postoperative increase in IOP40–43 or no change.44 These results are surprising if cataract extraction alone is purported to lower IOP (but are also potentially confounded in each case by the possible negative effects of cataract surgery on bleb function).
Ancillary data from glaucoma surgical studies
The control arm of studies comparing new techniques in glaucoma surgery combined with cataract surgery to cataract surgery alone yield valuable insights. In an iStent (Glaukos Corp., Laguna Hills, CA, USA) investigation, 36 patients with open-angle glaucoma (confirmed by gonioscopy), and IOP >18 mmHg on one or more medications were randomized.15 As compared with enrolment IOP on medication, the IOP at 16 months' follow up after 1-month medication washout showed the mean IOP in the cataract-only group to have risenfrom 17.3 + 3.0 to 19.2 + 3.5 mmHg (however, 24% required no postoperative hypotensive treatment). A prospective trial randomized 240 patients to iStent combined with cataract surgery versus cataract surgery alone.17 The report was confounded by the free use of glaucoma medications postoperatively – and in this case by a protocol requirement to keep the postoperative IOP ≤ 21 mmHg with medication – but the control arm of cataract surgery alone showed a decrease of 1.0 ± 3.3 mmHg from baseline medicated IOP, and with a small reduction in mean requirement for glaucoma medication postoperatively (1.4 vs. 1.0). In a normal Normal Tension Glaucoma population, a retrospective study reported IOP lowering of phacoviscocanalostomy and cataract surgery compared to cataract surgery alone: the decrease in IOP in the cataract-only group was significant only for the first 6 months; mean IOP decrease at 36 months was 1.0 ± 3.0 mmHg; 9.5% and 2.9% of cases respectively achieved a 20% and 30% decrease in IOP without postoperative glaucoma medication.14
A recent presentation derived from the Ocular Hypertension Treatment Study reported that cataract extraction in non-glaucomatous, untreated ocular hypertensive eyes (33 patients, 51 eyes) produced a decrease in IOP from 23.9 ± 3.7 preoperatively to 19.6 ± 3.6 mmHg postoperatively (P < 0.001).f It could be speculated that these data suggest that there may be an effect of glaucoma medications in modifying the IOP response from cataract extraction, or even that a ‘law of diminishing returns’ also applies to cataract surgery when utilized as a component of glaucoma management.d
In contrast, a recent report from the Collaborative Initial Glaucoma Treatment Study provides a pertinent insight.16 In a study designed to evaluate the possibility of trabeculectomy in one eye having an effect on the IOP of the untreated fellow eye, the report details IOP outcomes in the fellow eyes of 300 prospectively evaluated, open-angle glaucoma patients who underwent trabeculectomy in the first eye. In order to avoid any extraneous influences, fellow eyes undergoing any surgery (cataract extraction, Argon Laser Trabeculoplasty or trabeculectomy) were censored from the data. The data on this censored group are useful for our purposes: there were 32 fellow eyes undergoing cataract extraction. There was no significant reduction in mean postoperative IOP in this censored group (P = 0.4597). The lack of effect in this group is telling, in particular because all cases had confirmed POAG (which might be considered usually to be bilateral: 175 of the fellow eyes underwent trabeculectomy); no fellow eyes in the surgical arm were treated with medical therapy, and the investigation was also a well-designed, prospective trial.
The history of medicine is littered with unproven therapies, some of which have caused more harm than good. Even in recent years, treatments that were widely accepted on the basis of weak evidence founded on retrospective data and vocal advocacy have subsequently been discredited by higher quality data and evidence.45–48
The effectiveness of lens or cataract extraction in targeted cases of PACG seems plausible, quality evidence has been provided, and recommendations for the place of this surgery in management have been made.1–4
On the other hand, and to state what should be obvious: one cannot conclude that cataract extraction lowers IOP in POAG when, in the absence of gonioscopy, it has not been determined that the patient has POAG. A large number of the cases included in what are largely retrospective studies to date have potential for diagnoses other than POAG: especially PACG, but also PXF, Pigment Dispersion Syndrome and others. This inevitably confounds the interpretation and precludes any meaningful meta-analysis. The Collaborative Initial Glaucoma Treatment Study data on the effect of cataract extraction on IOP in confirmed POAG indicates no benefit. Uncritical repetition of the results of some other studies has lent a veneer of credibility and validity to their data which detailed analysis indicates may be undeserved. Basic critical appraisal skills suggest that conclusions about the causal role of the lens and about lens exchange as treatment for ocular hypertension and glaucoma cannot be based, for example, on a retrospective, unmasked study that did not use gonioscopy to exclude closure and did not account for phenomena like regression to the mean.8 The extension of the results of many of these studies to suggest that lens extraction may routinely be expected to yield a therapeutically useful IOP-lowering benefit in POAG is highly dubious. Arguments that this operation produces fewer complications than alternative glaucoma surgical therapies are hardly relevant if it is shown that the operation is not therapeutically effective in the first place.7 The results of many retrospective – and even some prospective – studies on the effects of cataract extraction on IOP are compromised by an ‘as-needed’ protocol for use of postoperative topical hypotensive medications. A glaucoma primary treatment trial of cataract extraction versus standard treatment (or observation) would ultimately be needed for final clarification: proceeding to such a randomised controlled trial would depend on approval by an ethics committee.
It may be argued that reports of the effect on IOP of lens extraction without gonioscopy findings echoes the ‘real world’ scenario of cataract extraction, where gonioscopy is not usually performed. We have recently advocated strongly that gonioscopy should form a part of the routine assessment for all patients;18 even if this is not accepted, it is beyond our comprehension that gonioscopy could go unreported in any series of glaucoma cases, and is not thought to be mandatory when IOP is of sufficient concern that therapeutic benefit to IOP from cataract surgery is being considered. This neglect is regrettably not unique: indeed, we do not consider it impossible that a body of research concerning apparent POAG (or Normal Tension Glaucoma) remains suspect because – despite the best intentions of the researchers – the absence of or deficiency in technique of gonioscopy49 has failed to detect potentially occludable angles.
The irregular presence of PXF in a number of studies – excluding those where it was the specific subject of investigation21,23,24– also confounds the interpretation of the effects of cataract surgery alone on IOP. It seems probable that cataract surgery in PXF may confer an exaggerated benefit in IOP lowering, possibly via trabecular lavage as has been suggested, but also potentially as a consequence of chamber deepening where zonular laxity has caused some shallowing preoperatively.22
In summary: it seems likely that lens extraction in an open-angle may decrease the IOP by a mean of 1–3 mmHg. This lowering is mild, unpredictable and frequently not sustained. The available body of work suggest that the non-glaucomatous eyes may be more likely to achieve this benefit. Despite exhortations to the effect that ‘every 1 mmHg counts’, the new approaches in glaucoma care involve the setting of an IOP target rather than incremental tweaking of the pressure. One might thus question whether unpredictable IOP-lowering equivalent to a second- or third-line medical therapy represents sufficient justification for surgical intervention on the lens – the more so if this was to be considered in the absence of visually significant cataract – and whether in fact an incremental gain in IOP control of this amount represents any meaningful benefit except in the mildest forms of glaucoma. Lens extraction is also unproven with respect to removing diurnal variation in the way trabeculectomy does;50,51 risks a postoperative IOP spike – particularly in occult angle closure eyes; and, given the regular persistence of postoperative flare, may compromise the timely performance of a trabeculectomy if subsequently required.52
The existing evidence does not support lens extraction as an incremental step in open-angle glaucoma management. In mild or moderate POAG that is well controlled at target on one to two medications where all we need is status quo and would be happy with an additional 1–2 mmHg reduction or a decrease in medication dependence, and where cataract surgery was indicated in its own right, then phaco-emulsification with intraocular lens implantation alone is appropriate treatment. In uncontrolled or moderate to severe glaucoma controlled on multiple medications, consideration should be given to the use of glaucoma surgery in the management plan. We believe there is insufficient evidence indicating reliable benefit from lens removal in open-angle glaucoma to allow consideration of clear lens extraction. Similarly, whilst one is clearly happy to accept any IOP lowering that is achieved when cataract surgery is undertaken in an ocular hypertensive, clear lens extraction as a treatment for open-angle Ocular Hypertension is not supported by evidence and remains conceptually challenged. Concern remains that in the ‘selection of treatment based on authoritarian advice or (on) our conviction that therapy seems to work or ought to work – as opposed to hard evidence that it does work – not only may worthless treatment be applied; sometimes it is downright harmful’.53
Unlike in PACG, there are no randomized controlled trials to guide the role of lens extraction in POAG. The prospective studies, those including requisite glaucoma clinical assessment, as well as evidence-based reviews, seem to suggest at most a modest reduction in IOP from cataract extraction, although more in the presence of PXF. Secondary data from a well-conducted randomized controlled trial confirm no significant decrease in IOP in established POAG. The mild and variable nature of any IOP decrease from lens extraction suggests marginal benefit only in milder forms of POAG. There is currently no evidence of any quality to suggest that lens extraction routinely represents a clinically useful treatment for POAG. The present evidence base clearly cannot countenance consideration of clear lens extraction as treatment for or prevention of POAG. The contribution of the lens to the development of POAG is best treated currently as an unproven hypothesis. We believe that medical practice should be evidence-based rather than hypothesis-based.54 The possible role of phaco-emulsification as treatment for POAG should be based on evidence of the type that has accumulated for primary angle closure disease and remains to be established.
‘… the intensity of the conviction that a hypothesis is true has no bearing on whether it is true or not’.
Sir Peter Medawar
Non-peer reviewed sources
- a. Boyle EL. Phacoemulsification could play key role in glaucoma treatment paradigm. Ocular Surgery News U.S. Edition 2009; April 25.
- b. Laber D. Reports of phaco lowering IOP. Eye World Asia Pac 2010; 6(3). Available from: http://aspx.apacrs.org/apacrs-publication/Default2.aspx?id=699
- c. Larkin H. Can cataract surgery control glaucoma? Eurotimes 2009; July, pp. 6–7.
- d. McDonald JE II. Physicians talk solutions for patients with cataract and glaucoma. Available from: http://www.eyeworld.org/article-pondering-alternatives-to-trabeculectomy
- e. Lindstrom R. Results show cataract surgery alone a good option for combined glaucoma and cataract cases. Ocular Surgery News U.S. Edition 2010; February 25.
- f. Mansberger SL, Gordon MO, Jampel H, et al. Abstract. Change in intraocular pressure after cataract extraction in the observation group of the Ocular Hypertension Treatment Study (OHTS). American Glaucoma Society 21st Annual Meeting 2011.