The Ocular Hypertension Treatment Study: Topical Medication Delays or Prevents Primary Open-angle Glaucomain African American Individuals

Background  The prevalence of glaucoma is higher in African American individualsthan in white individuals.

Objective  To report the safety and efficacy of topical ocular hypotensive medicationin delaying or preventing the onset of primary open-angle glaucoma (POAG)among African American participants in the Ocular Hypertension Treatment Study.

Methods  Eligibility criteria included age between 40 and 80 years, intraocularpressure between 24 and 32 mm Hg in one eye and between 21 and 32 mm Hg inthe other eye, and no evidence of glaucomatous structural or functional damageby standard clinical measures. Participants were randomized to either theobservation group or medication group. Of the 1636 participants randomized,408 were self-identified as African American.

Main Outcome Measure  The primary outcome was the development of reproducible visual fieldabnormality and/or reproducible optic disc deterioration attributed to POAG.

Results  Among African American participants, 17 (8.4%) of 203 in the medicationgroup developed POAG during the study (median follow-up, 78 months) comparedwith 33 (16.1%) of 205 participants in the observation group (hazard ratio,0.50; 95% confidence interval, 0.28-0.90; P = .02).

Conclusion  Topical ocular hypotensive therapy is effective in delaying or preventingthe onset of POAG in African American individuals who have ocular hypertension.

The Ocular Hypertension Treatment Study (OHTS) reported that reducingintraocular pressure (IOP) with topical antiglaucoma medication delayed orprevented the onset of primary open-angle glaucoma (POAG) in individuals withocular hypertension; at 60 months, the cumulative probability of developingPOAG was 4.4% in the medication group and 9.5% in the observation group.¹ Prior studies comparing treated and untreated eyeswith ocular hypertension generally had small sample sizes, used various definitionsof glaucoma, used single agents to reduce pressure, and included fairly homogenouspopulations of patients.^2-5 Thesefactors contributed to a lack of consensus regarding the benefits of earlytherapy. The OHTS is the first randomized treatment study of ocular hypertensionto recruit sufficient numbers of African American participants to examinethe therapeutic benefit of ocular hypotensive medication in this group.

Glaucoma is the leading cause of blindness among African American individuals.In the Baltimore Eye Survey, the age-adjusted prevalence rates of POAG were4 to 5 times higher in African American individuals than among white individuals.⁶ At the time of publication of the OHTS primary outcomearticle, there was a trend for treatment to be protective in African Americanindividuals, but the results did not attain statistical significance.¹ There may be several reasons why the protective trenddid not achieve statistical significance, including the shorter length offollow-up of the African American participants. The purpose of this articleis to present longer follow-up data on the cohort of African American participantsenrolled in the OHTS, which, in fact, demonstrate that topical ocular hypotensivemedication delays or prevents the onset of POAG in this group as well.

Eligibility criteria included: age, 40 to 80 years; an IOP of 24 mmHg or higher and 32 mm Hg or lower in one eye and 21 mm Hg or higher and 32mm Hg or lower in the fellow eye; normal and reliable visual fields as determinedby the University of California Davis Visual Field Reading Center, Sacramento;and normal optic discs on clinical examination and on stereoscopic photographsas determined by the Bascom Palmer Eye Institute Optic Disc Reading Center,Miami, Fla. The qualifying IOP was the mean of 4 to 6 IOP measurements pereye taken on 2 separate visits. Individuals were excluded if they had a visualacuity worse than 20/40 in either eye, previous intraocular surgery otherthan uncomplicated cataract extraction, or background diabetic retinopathyor other diseases capable of causing visual field loss or optic disc abnormalities.Individuals signed an informed consent form approved by the institutionalreview board of each clinic.

Eligible individuals were randomized in equal proportions to eitherthe medication group or the observation group. Participants who were randomizedto the medication group began using topical ocular hypotensive medicationto achieve a target IOP reduction of (1) an IOP 24 mm Hg or lower and (2)a 20% reduction in IOP from the average of the qualifying IOP and baselineIOP but not necessarily lower than 18 mm Hg. All commercially available topicalocular hypotensive medications were included. Participants completed follow-upvisits every 6 months. Semiannual visits included ocular and medical history;measurements of refraction, best-corrected visual acuity, and Humphrey 30-2visual fields; slitlamp examination; direct ophthalmoscopy; and IOP measurement.In addition, dilated fundus examination and stereoscopic optic disc photographywere performed at annual visits. Information on adverse effects was collectedat each visit using a patient-completed checklist of ocular and systemic symptoms(Glaucoma Symptom Checklist and the Medical Outcomes Study Short Form with36 questions [SF-36]) and ocular and medical history as elicited by clinicpersonnel. The study design has been described elsewhere.^7,8

The primary outcome was development of POAG in one or both eyes. Thiswas defined as a reproducible visual field abnormality or as a clinicallysignificant reproducible optic disc deterioration attributed to POAG by themasked endpoint committee. Development of a visual field abnormality was determinedby masked certified readers at the Visual Field Reading Center. A technicallyacceptable visual field was considered abnormal if the corrected pattern standarddeviation had a P<.05 or if the glaucoma hemifieldtest results were outside normal limits by STATPAC 2 criteria.⁸ Becausemost abnormal visual fields were found to be normal on retest, the protocolwas changed effective June 1, 1997, so that confirmation of abnormality required3 consecutive abnormal visual fields with the same type, location, and indexof abnormality.⁹ If 3 consecutive visual fieldsmet criteria for abnormality, the Visual Field Reading Center initiated theendpoint review process. Additional details about the process of reviewingvisual fields were given in a previously published report.^7,8

Optic disc deterioration was determined by masked certified readersat the Optic Disc Reading Center. Optic disc deterioration was defined asa generalized or localized thinning of the optic disc neuroretinal rim comparedwith baseline stereoscopic optic disc photographs in side-by-side comparisons.If optic disc deterioration was confirmed by 2 consecutive sets of stereoscopicoptic disc photographs in independent masked reviews, the Optic Disc ReadingCenter initiated the endpoint review process. Additional details about theprocess of reviewing optic disc photographs were given in a previously publishedreport.^7,8,10

The Data and Safety Monitoring Committee approved the termination ofthe overall trial when the last randomized participant reached 5 years offollow-up, as specified in the original protocol. The primary outcome article,which was published in June 2002, included data through November 8, 2001,on 125 participants who had developed POAG of the 1636 participants randomized.¹ This article reports on 148 participants who developedPOAG whose first abnormality was detected by June 1, 2002, with additionaldata collected through September 11, 2003, to confirm reproducibility of abnormalities.The protocol as described in the baseline article and the primary outcomearticle continued through June 1, 2002.

The overall target sample size of 1500 participants (750 per randomizationgroup) was selected assuming a 40% reduction in the 5-year incidence of POAGin the medication group, a 2-sided α error of .05, and statistical powerof 0.90. Our goal at study onset was to recruit 400 African American participants,approximately 25% of the overall sample, to provide a valid estimate of thetreatment effect in this group.⁸

All comparisons of randomization groups were made on an intention-to-treatbasis. For the purposes of the primary efficacy analysis, the time to developPOAG was determined by the date of the first abnormal finding that was subsequentlyconfirmed and attributed to POAG. The primary hypothesis was tested usingthe Mantel log rank test to compare the cumulative risk of developing POAGamong African American participants randomized to the medication or to theobservation group. Cox proportional hazards analysis was used to estimatethe hazard ratio of POAG in the medication group compared with the observationgroup, adjusting for the influence of baseline factors.

The potential for differential treatment protection on the risk of POAGby race was estimated using Cox proportional hazards models that adjustedfor baseline IOP and heart disease as well as those baseline factors thatdiffered by race, age, sex, vertical cup-disc ratio, corneal thickness, meandeviation, systemic hypertension, and diabetes mellitus. The final Cox proportionalhazards model that we report included pattern standard deviation rather thanmean deviation to reduce potential confounding due to lens opacification.Analyses were performed with SAS software (version 8.2; SAS Institute, Cary,NC). P values were 2 sided.

Recruitment was extended by 6 months, from 24 months to 30 months, toachieve enrollment of 400 African American participants. Between February28, 1994, and October 31, 1996, 1636 individuals with documented informedconsent were randomized. Two hundred three (25%) of the 817 participants randomizedto receive topical medication were self-identified as African American. Twohundred five (25%) of the 819 participants randomized to the observation groupwere self-identified as African American.

For purposes of this article, OHTS participants were classified as eitherself-identified African American (not of Hispanic origin) or "other," whichincludes white (n = 1137), Hispanic (n = 59), Asian (n = 14), American Indianor Alaskan Native (n = 4), and unknown (n = 14). Detailed description of clinicaland demographic characteristics of participants have been reported.⁸

Among African American participants enrolled in the OHTS, no statisticallysignificant differences in baseline demographic or clinical factors were foundbetween groups (all comparisons, P>.05) (Table 1).

The median follow-up was 78 months for African American participantsand 84 months for other participants. The visit completion rate was 76.3%for African American participants and 81.2% for other participants (P<.001). Among African American participants, the visitcompletion rate was essentially identical for the observation and medicationgroups (P = .98). Technically acceptable visual fieldsand stereoscopic optic disc photographs were obtained at 99% and 96%, respectively,of the specified completed follow-up visits for African American participants.

In the medication group, 3 (1.5%) of 203 African American participantsand 34 (5.5%) of 614 other participants were withdrawn from medication orchose to stop using medication for 6 months or more during the study (P = .02). In the observation group, 11 (5.4%) of 205 AfricanAmerican participants and 30 (4.9%) of 614 other participants received topicalocular hypotensive medication for 6 months or longer during the study (P = .78). In most cases, treatment was initiated by theOHTS clinician out of concern for the participant's high IOP.

The baseline and follow-up IOP for the medication and observation groupsare reported by race in Table 2.The distribution of IOP at baseline and across the course of follow-up visitsfor the African American participants is plotted in Figure 1. Among African American participants, the IOP goal wasmet in both eyes in 1807 (87.6%) of 2063 scheduled follow-up visits and inonly 1 eye in 137 (6.6%) of 2063 scheduled follow-up visits. Among other participants,the IOP goal was met in both eyes in 6152 (84.8%) of 7251 scheduled follow-upvisits and in 1 eye in 594 (8.2%) of 7251 scheduled follow-up visits. Figure 2 shows the percentage of AfricanAmerican participants who were prescribed each class of topical ocular hypotensivemedication during the follow-up period. At 60 months, 64 (42.7%) of 150 AfricanAmerican participants in the medication group were prescribed 2 or more topicalocular hypotensive medications compared with 195 (38.3%) of 509 other participantsin the group (P = .34) At the 60-month visit, fewerAfrican American participants were prescribed β-adrenergic antagonistscompared with other participants (85 [57%] of 150 African American participantsvs 351 [69%] of 509 other participants; P = .005)and were more likely to be prescribed prostaglandin agonists (78 [52.0%] of150 African American participants vs 203 [39.9%] of 509 other participants; P = .008).

Table 3 reports the progressand outcome of randomized participants by race unadjusted for follow-up time. Table 4 reports whether the first POAGendpoint was ascertained by visual field abnormality, by optic disc deterioration,or both. Among African American participants, the percentage developing POAGin the medication group during the entire follow-up period (median follow-up,78 months) was significantly lower (17 [8.4%] of 203) compared with the observationgroup (33 [16.1%] of 205) (hazard ratio, 0.50; 95% confidence interval [CI],0.28-0.90; Mantel log rank P = .02) (Figure 3). Among other participants, the percentage developing POAGduring the entire follow-up period (median follow-up, 84 months) was significantlylower in the medication group (27 [4.4%] of 614) compared with the observationgroup (71 [11.6%] of 614) (hazard ratio, 0.36; 95% CI, 0.23-0.57; P<.001). The protective effect of medication among African Americanparticipants (hazard ratio, 0.50) was not statistically different from itsprotective effect among other participants (hazard ratio, 0.36; P = .40 for race interaction).

The estimate of the effect of treatment among African American participantswas not substantially altered after adjusting for baseline age, visual fieldpattern standard deviation, vertical cup-disc ratio, IOP, and corneal thickness,which was measured after randomization (hazard ratio, 0.41; 95% CI, 0.22-0.75).Among African American participants, treatment appeared to be protective againstboth reproducible visual field abnormality attributed to POAG (hazard ratio,0.51; 95% CI, 0.25-1.06; P = .07) and reproducibleoptic disc deterioration attributed to POAG (hazard ratio, 0.36; 95% CI, 0.17-0.78; P = .007).

Adjusting for time of follow-up, the risk of developing POAG was higheramong African American participants compared with others in both groups. Inthe medication group, the risk among African American participants was 2 timeshigher during the course of the study compared with other participants (hazardratio, 2.03; 95% CI, 1.10-3.73; Mantel log rank P =.02) (Table 3) (Figure 4). In the observation group, the risk among African Americanparticipants was 58% higher compared with other participants (hazard ratio,1.58; 95% CI, 1.04-2.39; Mantel log rank P = .03)(Table 3) (Figure 5).

To ascertain the safety of treatment among African American participants,the medication and observation groups were compared for participant self-reportof symptoms (Glaucoma Symptom Checklist and SF-36) and for medical and ocularhistory as collected by clinic staff during the course of the study. All P values reported in the safety section are unadjustedfor multiple comparisons between groups.

On the Glaucoma Symptom Checklist, there was no evidence that the medicationgroup had more ocular or systemic symptoms compared with the observation groupduring follow-up (P>.05), except for symptoms associatedwith administration of prostaglandin analogues. Changes in iris color, darkeningof eyelids, and growth of eyelashes were reported by 19 (18.8%) of 101 AfricanAmerican participants prescribed a prostaglandin analogue for 6 months orlonger compared with 20 (13.8%) of 145 African American participants in theobservation group (P = .29). On the SF-36, therewere no differences between randomization groups on the physical componentor mental health component at any annual visit (P>.05).

In medical and ocular histories collected by clinic staff, there wereno differences among African American participants in either group in totalhospitalizations (P = .24), worsening of preexistingconditions (P = .55), mortality (P = .24), percentage reporting 1 or more adverse events (P = .79), and percentage reporting 1 or more serious adverse events(P = .19). There was a trend for a higher percentageof any cancer in the medication group (medication group, 9% vs observationgroup, 4.5%; P = .08) and a trend for prolonged hospitalizations(medication group, 2% vs observation group, 0%; P =.06). There were differences in some adverse events between groups. AmongAfrican American participants, clinic personnel classified a higher percentageof adverse events in the medication group as life threatening (medicationgroup, 3.5% vs observation group, 0.5%; P = .04).Among African American participants, a higher percentage in the medicationgroup compared with the observation group was reported to have psychiatricadverse events (18.4% vs 8.3%, respectively; P<.001).Clinic staff classified 4 (2.0%) of 201 psychiatric adverse events in themedication group as serious and 1 (0.5%) of 204 in the observation group asserious (P = .21). None of the 5 serious psychiatricadverse events in the medication group were judged to be "probably" or "definitely"related to study medication by the clinicians. Among African American participants,a higher percentage in the medication group reported genitourinary symptomscompared with the observation group (16.9% vs 10.8%, respectively; P = .08). Clinic staff classified 12 (6.0%) of 201 genitourinary adverseevents in the medication group as serious and 5 (2.5%) of 204 in the observationgroup as serious (P = .09). None of the 12 seriousgenitourinary adverse events in the medication group were judged to be "probably"or "definitely" related to study medication.

No difference was found between randomization groups by race in thepercentage of participants reporting 1 or more adverse events (interactiontest for race and randomization group, P = .58) orin the percentage of participants reporting 1 or more serious adverse events(interaction test for race and randomization group, P =.16).

Among African American participants, no differences in Early TreatmentDiabetic Retinopathy Study visual acuity were found between randomizationgroups throughout the study (P>.05 at all follow-upperiods except at 66 months when visual acuity in the medication group averaged1 letter lower than in the observation group; P =.04). There was a slight excess in the rate of cataract surgery for AfricanAmerican participants in the medication group (10 [5.1%] of 198) comparedwith the observation group (5 [2.5%] of 202) (P =.17).

The sample for predictive analyses of POAG consisted of 148 randomizedparticipants who developed POAG (50 African American participants and 98 otherparticipants) and 1471 participants who did not develop POAG (350 AfricanAmerican participants and 1121 other participants). Inclusion in the predictiveanalyses required the completion of at least 1 follow-up visit (17 of 1636participants did not complete any follow-up visits). In the univariate predictivemodel, self-identified African American race was associated with a 71% increasein risk of developing POAG compared with other participants (hazard ratio,1.71; 95% CI, 1.2-2.40). However, African American participants had largerbaseline vertical and horizontal cup-disc ratios (0.45 ± 0.18 SD and0.42 ± 0.17 SD, respectively) compared with other participants (0.37± 0.19 SD and 0.34 ± 0.18 SD, respectively) and thinner centralcorneal measurements (554.9 µm ± 38.7 µm SD) compared withother participants (578.3 µm ± 36.5 µm SD). In a multivariateCox proportional hazards model that was stratified by randomization groupand adjusted for these factors as well as age, sex, history of diabetes, systematichypertension and heart disease, IOP, and pattern standard deviation, AfricanAmerican race was no longer statistically significantly associated with anincreased risk of developing POAG (hazard ratio, 1.1; 95% CI, 0.76-1.66).

The OHTS is the first randomized trial on the prevention of POAG toenroll a large number of African American participants. This is importantgiven the high prevalence of glaucoma and glaucoma-related blindness in AfricanAmerican individuals.^6,11 At thetime of the publication of the primary outcome article in 2002, there wasa trend for treatment to be protective in African American participants, butthe results did not attain statistical significance. With additional follow-up,the OHTS demonstrates that topical ocular hypotensive medication reduces theincidence of POAG in individuals of African derivation with ocular hypertension.This finding is consistent with the overall findings of the OHTS publishedpreviously.¹ Among African American participantsin the OHTS, 17 (8.4%) of 203 in the medication group and 33 (16.1%) of 205in the observation group developed POAG. The median follow-up among AfricanAmerican participants was 78 months in this article compared with 72 monthsin the previous article.¹ There was a slighttrend for treatment to be less protective in African American participantsthan in other participants, but this difference was small and not statisticallysignificant.

Despite substantial treatment benefit, the incidence of POAG was stillsignificantly higher among African American participants compared with otherparticipants in both the medication and observation groups. In the medicationgroup, African American participants had twice the hazard of developing POAGcompared with other participants, despite similar baseline and treated follow-upIOPs. In the observation group, African American participants had a 58% higherhazard of developing POAG, despite similar baseline and follow-up IOPs. Theresults of the OHTS are consistent with reports of higher prevalence of glaucomaand glaucoma-related blindness in African American individuals.^6,11-14 Amongthe factors that may contribute to the increased prevalence of glaucoma andglaucoma-related blindness in individuals of African origin are the following:genetic susceptibility to POAG;^15-19;higher prevalence of comorbidity such as cardiovascular disease²⁰;earlier onset of POAG^15-17,19;later detection of POAG; and economic and social barriers to treatment.^21,22 Data from the OHTS address some ofthese possible explanations. In the OHTS, African American participants hada higher prevalence of risk factors for POAG than other participants.²³ In addition, a higher proportion of African Americanparticipants compared with other participants reported a history of hypertensionand diabetes, factors that have been associated with an increased risk ofPOAG by some investigators.^18,24-29 Inthe OHTS, we did not find that diabetes and hypertension were associated withincreased risk of POAG.²³ However, since informationon systemic conditions was collected by self-report without confirmation bymedical records or testing, it is difficult to determine the true extent thathypertension or diabetes may have contributed to the risk of developing glaucoma.Furthermore, individuals with moderate to severe diabetes were excluded fromthe trial because of the possible effect of retinopathy on visual fields.The OHTS does not directly address the questions of age of onset of POAG ordelay in detection among African American individuals. Nor does the OHTS addressthe question of whether a lower IOP goal would have been more protective againstPOAG. In the OHTS, there was no charge for medication and copayments wereminimized, which should diminish but not eliminate potential barriers to treatment.In the OHTS, the study protocol was standardized so that differences in managementamong participants were minimized. However, adherence to medical therapy mustalways be considered as a potential confounding factor.

It is not surprising that the increased risk of POAG associated withparticipants who self-identified as African American in the OHTS appears largelyattributable to differences in specific clinical factors and not race itself.Since the completion of the mapping of the human genome, the validity of raceas a construct for classifying individuals has been questioned.^30,31 Moreover,data from the 2000 US Census demonstrates that self-identification of raceis problematic, given that more than 800 000 individuals self-identifiedas both African American and white.³² The OHTSpredictive analyses underscore the importance of measuring clinical risk factors,specifically corneal thickness and cup-disc ratio, in the management of individualpatients rather than relying on the perceived race of that individual.

Because the safety of topical ocular hypotensive medications was systematicallyassessed on all participants, OHTS data allow us to compare the safety ofmedication in African American participants with other participants as wellas safety in African American participants in the medication and observationgroups. There were no differences by race in the overall rate of adverse eventsor the rate of serious adverse events in the OHTS. Among African Americanparticipants, there were no overall differences by randomization group ineither the ocular or systemic self-reported complaints except changes in iriscolor, darkening of eyelids, and growth of eyelashes, which were more frequentlyreported with the use of the prostaglandin analogues in the medication group.Among African American participants, there was a trend for cataract surgeryto be more frequently performed in the medication group than in the observationgroup. A similar trend was noted among other participants as well. These trendsare consistent with the higher incidence of lens opacities reported in patientsreceiving topical ocular hypotensive medication in the Barbados Eye Studyand the Early Manifest Glaucoma Trial and require further study.^33,34 Amongboth African American participants and other participants, clinical staffclassified a higher percentage of psychiatric events and genitourinary adverseevents as serious in the medication group compared with the observation group.With the large number of statistical comparisons conducted and with classificationof these data by unmasked clinic personnel, it is difficult to interpret thesedifferences. However, these findings warrant continued evaluation.

In summary, topical ocular hypotensive therapy is effective in delayingor preventing the onset of glaucoma in African American individuals with ocularhypertension. However, the therapeutic benefit of medication does not implythat every patient of African descent with ocular hypertension requires treatment.Clinicians should consider factors such as corneal thickness, cup-disc ratio,IOP, age, general health, and life expectancy in determining who should betreated rather than relying on the race of the individual as an indicationfor treatment. Adverse effects should be monitored closely, and patients shouldbe encouraged to adhere to their prescribed regimens. Finally, given the protectivebenefit of topical therapy among African American individuals, it is importantto identify those at moderate to high risk of developing POAG so they canbe evaluated for possible medical treatment.

Corresponding author and reprints: Mae O. Gordon, PhD,Ocular Hypertension Treatment Study Coordinating Center, Department of Ophthalmologyand Visual Sciences, Washington University School of Medicine, Box 8203, 660S Euclid, St Louis, MO 63110 (e-mail: mae@vrcc.wustl.edu).

Submitted for publication November 11, 2003; final revision receivedJanuary 29, 2004; accepted February 3, 2004.

This study was supported by grants EY09307and EY09341 from the NationalEye Institute and the National Center on Minority Health and Health Disparities,National Institutes of Health, Bethesda, Md; Merck Research Laboratories,White House Station, NJ; and by an unrestricted grant from Research to PreventBlindness, New York, NY.

Drugs were donated by the following pharmaceutical companies: AlconLaboratories Inc, Fort Worth, Tex; Allergan Therapeutics Group, Irvine, Calif;Bausch & Lomb Pharmaceutical Division, Tampa, Fla; CIBA Vision Corporation,Duluth, Ga; Merck Research Laboratories, White House Station; Novartis OphthalmicsInc, Duluth, Minn; Otsuka America Pharmaceutical Inc, Rockville, Md; and Pharmacia& Upjohn, Peapack, NY. Pachymeters were loaned to clinical centers byDGH Technology, Exton, Pa.