True intraocular pressure (IOP) vs measured IOP. For every 1–mmHg increase in measured IOP, the true IOP increases by 0.931 mm Hg in whitesubjects and by 0.994 mm Hg in black subjects. The correction for true IOPwas made based on the additive correction table by Ehlers et al.4
Herndon LW, Weizer JS, Stinnett SS. Central Corneal Thickness as a Risk Factor for Advanced Glaucoma Damage. Arch Ophthalmol. 2004;122(1):17-21. doi:10.1001/archopht.122.1.17
To determine if central corneal thickness (CCT) is related to the levelof glaucoma severity at the initial examination.
The initial visit to a glaucoma specialist by consecutive patients withprimary open-angle glaucoma from 1997 to 2002 was reviewed retrospectively.Each patient's age, sex, race, family history of glaucoma, number of glaucomamedications, visual acuity, spherical equivalent, intraocular pressure, CCT,visual field data, and vertical and horizontal cup-disc ratios were analyzed.
Three hundred fifty eyes of 190 patients met the inclusion and exclusioncriteria. Central corneal thickness was significantly lower in blacks (mean,537 µm) than in whites (mean, 556 µm). Lower CCT was significantlyassociated with worsened Advanced Glaucoma Intervention Study score, worsenedmean deviation of visual field, increased vertical and horizontal cup-discratios, and increased number of glaucoma medications. In multivariate analysis,lower CCT was significantly associated with worsened Advanced Glaucoma InterventionStudy score, worsened mean deviation of visual field, and increased verticaland horizontal cup-disc ratios.
Central corneal thickness is a powerful clinical factor in determiningglaucoma severity at the initial examination by a specialist. Measuring CCTmay aid the ophthalmologist in identification of glaucoma patients at highrisk for progression.
Applanation tonometry has been considered to be the gold standard fordetermining intraocular pressure (IOP) for almost 50 years. It is less likelyto be influenced by variables such as scleral rigidity, which can significantlyaffect measured values obtained by Schiotz tonometry.1 Goldmannand Schmidt2 discussed the influence of variationsof central corneal thickness (CCT) on IOP measured by applanation in theirlandmark article. However, they believed that significant variations in CCToccurred only rarely. After a commercially available optical pachymeter becameavailable, a positive correlation between CCT and IOP measured by applanationwas found.3 This relationship has been confirmedin animal and human studies where intraocular cannulation has been performedto experimentally control IOP.4,5 Centralcorneal thickness has recently been recognized as a significant risk factorfor progression of ocular hypertension to primary open-angle glaucoma (POAG)in the Ocular Hypertension Treatment Study.6 Thisstudy was the first to prospectively demonstrate that a thinner CCT predictsthe development of POAG. Several other studies have shown that eyes with normal-tensionglaucoma have a mean CCT less than that of normal eyes, while eyes with ocularhypertension have a mean CCT greater than that of normal eyes.7- 14 Inthe current study, we examined the effect of CCT on the degree of glaucomadamage seen in POAG patients at the initial examination by a glaucoma specialist.
All patients were evaluated at the Duke University Eye Center (Durham,NC) and retrospectively selected from our research database. Approval of thestudy was granted by the Duke University institutional review board. We reviewedthe initial visit of each consecutive new POAG patient seen by a glaucomaspecialist (L.W.H.) from 1997 to 2002. The diagnosis of POAG was based onpatients having an IOP of 22 mm Hg or higher at the initial visit, or historically,characteristic glaucomatous optic neuropathy with diffuse or focal optic rimthinning, hemorrhage, cupping, or nerve fiber layer defects indicative ofglaucoma, and corresponding visual field loss. Each of the patients had tohave pachymetry performed within 1 month of the initial visit. Stereo discphotographs or a detailed optic nerve drawing, and automated visual fieldtesting (if appropriate) were also required within 1 month of the initialvisit. Exclusion criteria were corneal or retinal disease or a secondary causeof glaucoma, including pseudoexfoliation and pigment dispersion syndromes.Patients who had undergone any corneal surgery or retinal laser procedures,including panretinal photocoagulation, were not included in this study. Therewere 2 patients with normal-tension glaucoma who were included in the study.
For each patient, we recorded age, sex, race, family history of glaucomain a first-degree relative, and presence of diabetes mellitus or systemichypertension as reported by the patient. For each eye, we recorded Snellenvisual acuity, number of glaucoma medications prescribed, spherical equivalent,IOP by Goldmann applanation tonometry, average CCT, visual field data whereavailable, and vertical and horizontal cup-disc ratios. Combination eyedrops,such as timolol/dorzolamide, were counted as 2 glaucoma medications; oralpressure-lowering medications, such as acetazolamide, were counted as 1 glaucomamedication. Goldmann applanation IOP was measured before dilation; when 2or more predilation measurements were charted, the average of those pressureswas recorded. Visual field data included type of visual field analysis performed,Advanced Glaucoma Intervention Study (AGIS) score, mean deviation of visualfield, fixation losses, false-negative responses, and false-positive responses.Right and left eyes were analyzed separately. The AGIS score has been describedin detail previously.15 In brief, the visualfields are graded on a scale of 0 to 20 based on the degree of damage on theTotal Deviation printout. A score of 0 represents a normal visual field; 1to 5 represents mild disease; 6 to 11, moderate disease; 12 to 17, severedisease; and 18 to 20, end-stage glaucoma.
Each patient's CCT was measured using an ultrasonic pachymeter (StorzCompuscan Ultrasonic Pachymeter System; Storz, St Louis, Mo [used 1997 to2000]; and DGH 550 Pachette 2; DGH Technology, Exton, Pa [used 2000 to 2002]).The average of 5 CCT readings was recorded. Patients also underwent visualfield testing by Humphrey automated 24-2 (Humphrey Systems, Dublin, Calif)or 30-2 full-threshold or Swedish Interactive Thresholding Algorithm (SITA)standard perimetry protocols if their visual function allowed, or by Humphreyautomated 10-2 SITA standard perimetry if their glaucomatous damage was deemedvery severe. Patients underwent Goldmann manual perimetry if they were notable to complete Humphrey automated visual field testing. Patients whose visualacuity was too poor for automated or manual visual field testing in that eyedid not have data for that eye entered into the visual field categories. Onlypatients who had reliable Humphrey automated 24-2 or 30-2 SITA standard orfull-threshold perimetry within 1 month of their initial visit had their visualfield data included in the statistical analysis. Reliable Humphrey automatedperimetry was defined by having fewer than 2 of the following characteristics:fixation losses less than 20%, false-positive responses less than 33%, orfalse-negative responses less than 33%. This reliability criteria was adaptedfrom the AGIS reliability ratings.15 Each Humphrey24-2 or 30-2 visual field was scored by one masked grader (J.S.W.) accordingto the AGIS scoring system.15 In the case of30-2 Humphrey visual fields, the outermost circumference of testing pointswas not included in the scoring so that the remaining testing points fit theAGIS scoring template for 24-2 visual fields. One masked grader (L.W.H.) determinedvertical and horizontal cup-disc ratios for each eye. Cup-disc ratios werejudged using stereoscopic optic disc photographs from each patient's initialvisit when available, or by evaluating detailed chart drawings from the initialvisit when photographs were not available.
Initially, descriptive statistics (number and percentage for categoricalvariables, and number, mean, and SD for continuous variables) were obtained.Subsequently, each variable was assessed individually for its relationshipto the degree of damage from glaucoma (AGIS score, mean deviation, and verticaland horizontal cup-disc ratios) and number of medications used for glaucoma.For categorical variables (sex, race, and family history of glaucoma), analysisof variance was used to assess whether differences in the outcome variablesexist between categories of these predictor variables. The relationship betweencontinuous predictor variables (corneal thickness, age, spherical equivalent,and IOP) and the outcome variables was assessed using linear regression. Finally,all predictor variables were combined in a single regression model to assesstheir joint effects on the outcome variables. The model for each outcome variablewas reduced using backward elimination until it contained only significantpredictors. In all statistical models, the correlation between the 2 eyeswas accounted for by using a mixed model with a random subject effect.
Of the 434 medical records reviewed, 350 eyes of 190 patients met theinclusion and exclusion criteria for this study. Descriptive statistics fordemographic and background variables are presented in Table 1. The average age of the 190 patients was 65.6 years (range,22-97 years). Patients were 51% male and were predominantly black (55%) andwhite (42%). Ninety-nine percent of patients had a diagnosis of POAG, and39% had a self-reported family history of glaucoma. Nineteen percent had ahistory of diabetes, and 53% had a history of hypertension. Eighty-one percentof right eyes and 79% of left eyes had Humphrey visual field tests using theSITA standard method.
Descriptive statistics for ophthalmic measures are presented separatelyfor right and left eyes in Table 2.Of the 350 eyes, 178 were right eyes and 172 were left eyes. The average visualacuity in the right eyes was 20/38 (0.284 logMAR units) and in the left eyeswas 20/42 (0.321 logMAR units). The average spherical equivalent was –0.82diopters OD and –0.69 diopters OS. Average IOP was about 19 mm Hg OU.The mean CCT was 544 µm OD and 546 µm OS. The average AGIS scorewas about 6.2 OU. The average mean deviation of visual field was –8.3dB OD and −8.1 dB OS. The mean vertical cup-disc ratio was about 0.80OU, and the mean horizontal cup-disc ratio was 0.76 OU. The average numberof medications was 1.6 OD and 1.7 OS.
In white patients, the mean CCT was 556 ± 41 µm, whilein black patients, the mean CCT was 537 ± 38 µm. This was a statisticallysignificant difference (P<.001). There was nostatistically significant difference in CCT between men and women, with menhaving a mean CCT of 544 µm and women having a mean CCT of 547 µm(P = .64). There was also no statistically significantdifference in CCT between patients with and without a family history of glaucoma;those with a positive family history had a mean CCT of 547 µm, whilethose without had a mean CCT of 544 µm (P =.83).
Results of univariate modeling are presented in Table 3. When considered alone, CCT was a significant predictorof every outcome variable. An increase in CCT was associated with an improvedAGIS score (P = .001), an improved mean deviationof visual field (P<.001), a decrease in vertical(P<.001) and horizontal (P =.003) cup-disc ratios, and a decrease in the number of medications used forglaucoma (P = .04).
Age was a significant predictor of AGIS score (P =.01) and mean deviation of visual field (P = .002).An increase in age was associated with a worsened AGIS score and a worsenedmean deviation of visual field. Intraocular pressure was a significant predictorof mean deviation of visual field (P = .01). An increasein IOP was associated with an improved mean deviation of visual field.
There was a significant difference among races in the number of medicationsused (P = .01). For both eyes combined, Hispanicsused the smallest average number of medications (1.3), followed by whites(1.8), blacks (2.1), Asians (2.7), and those of unknown race (4.5). The numberof patients of races other than black or white was very small, however.
There was a significant difference between those with and without afamily history of glaucoma in terms of vertical cup-disc ratio (P = .02), horizontal cup-disc ratio (P = .07),and number of medications used (P = .009). Thosewith a positive family history had a slightly greater vertical cup-disc ratio(0.87) than those without (0.83) and a slightly greater horizontal cup-discratio (0.84) than those without (0.80). Those with a family history used moremedications (2.3) than those without (1.9). Spherical equivalent was not asignificant predictor of any outcome variable. Also, there was no differencebetween men and women for any outcome variable.
Results of multivariate modeling are presented in Table 4. When all variables were considered simultaneously, theonly significant predictor of AGIS score was CCT (P =.001). For an increase of 10 µm of corneal thickness, AGIS score improvedby 0.31 points.
Significant predictors of mean deviation of visual field were CCT (P = .006), age (P = .02), andIOP (P = .02). For an increase of 10 µm ofCCT, the mean deviation of visual field improved by 0.34 dB; for an increaseof 10 years of age, the mean deviation of visual field worsened by 0.88 dB;for an increase of 1 mm Hg of IOP, the mean deviation improved by 0.21 dB.
Significant predictors of vertical cup-disc ratio were CCT (P<.001) and family history of glaucoma (P =.01). For an increase of 10 µm of CCT, the vertical cup-disc ratio decreasedby 0.008. At an average value of CCT (ie, 545 µm), patients with a familyhistory of glaucoma had a vertical cup-disc ratio that was 0.043 greater thanthose without a family history of glaucoma.
Significant predictors of horizontal cup-disc ratio were CCT (P = .003) and family history of glaucoma (0.042). For anincrease of 10 µm of CCT, horizontal cup-disc ratio decreased by 0.007.At an average value of corneal thickness (ie, 545 µm), patients witha family history of glaucoma had a horizontal cup-disc ratio that was 0.038greater than those without a family history.
Significant predictors of number of medications used for glaucoma wereage (P<.001), race (P<.001),spherical equivalent (P = .04), and family historyof glaucoma (P = .03). For an increase of 10 yearsof age, the number of medications increased by 0.23; for an increase of 1diopter of spherical equivalent, the number of medications decreased by 0.06.Adjusting for all other variables, patients with a family history of glaucomaused 0.31 more medications that those without a family history. The patternamong races was similar to that in the univariate model.
We mentioned previously that black patients had significantly thinnercentral corneas compared with white patients. The measured IOP in black patients(19.4 ± 5.9 mm Hg) was not different compared with that in white patients(18.4 ± 6.0 mm Hg). Ehlers and associates4 createdan additive correction table for CCT measurements greater or less than themean CCT in their study. They extrapolated that applanation tonometry is overestimatedor underestimated by approximately 5 mm Hg for every 70-µm differencein measured CCT from normal CCT. If we adjust the measured IOPs by this correctionfactor to determine the true IOPs, using 545 µm as a normal CCT, thereis a significant difference (P<.001) between blacksand whites (20.0 ± 6.4 mm Hg vs 17.6 ± 6.3 mm Hg, respectively).This difference between measured and true IOPs is shown in Figure 1.
To our knowledge, we are the first investigators to report the associationof CCT with POAG severity. The effect of CCT on ocular hypertension and normal-tensionglaucoma has already been well documented.7- 14 Inour study, patients with POAG who had a thinner CCT tended to have more severeglaucomatous damage on initial examination by a glaucoma specialist. Centralcorneal thickness was the most consistent predictor of degree of glaucomatousdamage as measured by the outcome variables.
It has been suggested that a thicker CCT may be protective against glaucomatousdamage, since CCT in ocular hypertensive patients tends to be thicker thanin POAG patients.7,9,10,14 Itis well known that IOP measured by applanation should be adjusted to correctfor CCT measurements that are higher or lower than the mean CCT of approximately545 µm in the general population.12 Inour univariate and multivariate analyses, increased IOP was associated withan improved mean deviation of visual field and was not significantly associatedwith any other outcome variable. Thinner CCT was associated with worseningin all outcome variables except number of glaucoma medications. It is unclearwhy having a higher IOP should be associated with an improved mean deviationof visual field in our study. Perhaps the association of higher IOP with increasedCCT explains this phenomenon, although after correcting for the true IOP,the difference still persisted.
Our study confirms the observation that mean CCT in black patients islower than mean CCT in white patients. La Rosa et al16 foundthat the mean difference in CCT between all black and white subjects was 27.3µm, while the mean difference between black and white glaucoma patientswas 32.2 µm. Nemesure et al17 measuredpachymetry on 1142 participants in the Barbados Eye Study and found that blackparticipants tended to have thinner corneas (mean thickness, 529.8 µm)than mixed (black and white) (537.8 µm) and white (545.2 µm) participants,respectively. In our study, the mean difference in CCT between black and whiteglaucoma patients was 19 µm. This may help to explain why blacks are4 to 5 times more likely than whites to be diagnosed as having glaucoma.18 In the Ocular Hypertension Treatment Study, whenrace is entered into a multivariate analysis, it is no longer a statisticallysignificant predictor of progression to glaucoma; CCT and vertical cup-discratio remain significant predictors, however.6 Inour univariate and multivariate analyses, black race significantly predictsan increase in the number of glaucoma medications, although race is not significantlyassociated with any other outcome variables.
The Baltimore Eye Survey (BES) showed that the average IOP among blackand white subjects in the general population was similar.19 Inboth races, the screening IOP was higher in glaucomatous eyes than in eyesin the general population. In white subjects, eyes that received glaucomatreatment had an IOP 2 to 3 mm Hg higher than eyes without glaucoma, on average.Glaucomatous eyes that were not yet receiving treatment had an IOP 6 to 7mm Hg higher than normal eyes. In black subjects, treatment status made littledifference; the IOP in glaucomatous eyes was 4 to 5 mm Hg higher than in therest of the population whether they received treatment or not. Untreated blacksubjects with POAG had lower IOP than untreated white subjects in the BES,which would suggest that the optic nerve in black subjects may be more sensitiveto IOP than that in white subjects. It is difficult to draw comparisons betweenour study and the BES, as most of our patients were receiving treatment; however,perhaps the true IOP of the untreated black subjects in the BES (as correctedfor variations in CCT) would be higher than that of the untreated white subjects,which would account for the seemingly increased susceptibility to glaucomatousoptic nerve damage among blacks.
We found that having a family history of glaucoma was associated withhaving worse disease, as defined by increased horizontal and vertical cup-discratios. Also, patients with a family history of glaucoma used more glaucomamedications. Having a family history of glaucoma, however, had no effect onthe degree of visual field loss as defined by AGIS scores and mean deviationindices. These findings are in agreement with a recent article by Landerset al,20 who found that a family history ofPOAG had no influence on the severity of visual field at diagnosis. They didfind, however, that having a family history of glaucoma was associated witha better visual field at diagnosis in patients younger than 50 years but notin patients 50 years or older.
There are some minor limitations in our study. We did not intend tofollow patients' progression of glaucomatous damage. Rather, our data reflectthe findings at the initial examination by a glaucoma specialist. Also, weare aware that there may be some referral bias in our patient population.Patients referred to a glaucoma specialist at a tertiary care institutionmay have more advanced, intractable glaucoma than those in the general populationand therefore may not represent the majority of POAG patients. Because ofa lack of standardized grading methods, we were not able to grade the fewGoldmann manual perimetry studies and 10-2 Humphrey visual fields that wereperformed, so patients with the most severe glaucomatous visual field damagecould not be included in the visual field analysis because they were unableto perform SITA standard or full-threshold Humphrey visual field testing.Finally, the AGIS grading system was designed for full-threshold strategies,but most of our patients had SITA analysis. It is not clear if one can directlyapply AGIS criteria to SITA in a seamless manner; however, we were consistentwith all of our interpretation, which should quantify our results in a uniformfashion.
In conclusion, CCT is a significant predictor of glaucomatous damageas measured by AGIS score, mean deviation of visual field, and vertical andhorizontal cup-disc ratios in patients at the initial examination by a glaucomaspecialist. Measuring CCT in glaucoma patients may help identify those patientswho are at higher risk for developing severe glaucomatous sequelae, thus enablingthe ophthalmologist to treat their disease more aggressively.
Corresponding author and reprints: Leon W. Herndon, MD, Duke UniversityEye Center, Box 3802, Duke University Medical Center, Durham, NC 27710-3802.
Submitted for publication April 28, 2003; final revision received August17, 2003; accepted August 25, 2003.
This study was supported in part by Clinical Vision Research DevelopmentAward EY11725 from the National Eye Institute, National Institutes of Health,Bethesda, Md (Dr Stinnett).