Corneal thickness in screening, diagnosis, and management James D. Brandt

Purpose of review “I do not object to having a patient of mine subjected Variability in central corneal thickness (CCT) is a potent to examination with a mechanical tonometer, but ex- confounder of most tonometry techniques, especially pect very little from this test since digital tonometry Goldmann applanation tonometry. The Ocular Hypertension by an expert is a much more accurate test…” Treatment Study (OHTS) provided important information regarding predictive factors for the eventual development of Isador Schnabel (1842–1908) before the Vienna Ophthal- glaucoma in patients at risk of the disease. Among the most mological Society, 1908 striking of the OHTS’ findings was that CCT was a powerful predictor of glaucoma risk. In this review, studies subsequent Introduction to the OHTS report will be reviewed and placed in the context Ever since the recognition by Bannister in the 16th cen- of what is known about CCT and its relation with tonometry tury that certain forms of blindness were associated with and glaucoma risk. a firm eye, ophthalmologists have been trying to measure Recent findings intraocular pressure (IOP). The simultaneous explosion Several well-designed studies have since expanded on this of ophthalmic knowledge and medical instrument- hypothesis, confirming that CCT bears an inverse relation with making in the 19th century led to mechanical tonometers the risk of developing glaucoma damage. Other investigators of varied underlying principals and designs. These de- have confirmed the presence of racial differences in CCT and vices were met with almost uniform suspicion by the pilot studies suggest that CCT may vary systematically in leaders of the day. More than a century later, it seems different forms of glaucoma. that perhaps their skepticism was well founded. Summary The absence of a widely-accepted algorithm for the correction Goldmann applanation tonometry (GAT) gained wide- of IOP measurements should not prevent the widespread spread and rapid acceptance after its introduction in the adoption of pachymetry as part of the comprehensive eye 1950s. It was reasonably priced, based on easily under- exam, as knowledge of an individual’s CCT provides valuable stood physical principles, fit seamlessly into the work- information about their glaucoma risk. flow of the slit-lamp examination, and appeared to pro- vide accurate, reproducible measurements. GAT’s status Keywords as a gold standard went largely unchallenged for 50 years, glaucoma, tonometry, pachymetry, ocular hypertension, LASIK even though Professor Goldmann himself recognized various potential sources of error for the device in his first Curr Opin Ophthalmol 15:85–89. © 2004 Lippincott Williams & Wilkins. description of his tonometer. In particular, Goldmann and Schmidt acknowledged that their design assump- tions were based on a central corneal thickness (CCT) of 500 µm and that the accuracy of their device would vary if CCT deviated from this value [1]. Given the paucity of Glaucoma Service, Department of Ophthalmology, University of California, Davis, published data at the time, 500 µm seemed a reasonable Sacramento, California, USA assumption for the average patient. We now know that Correspondence to James D. Brandt, MD, Glaucoma Service, Department of CCT varies greatly among the general population, to a Ophthalmology, University of California, Davis, 4860 Y Street, Suite 2400 Sacramento, CA 95817-2307, USA degree that impacts the accuracy of GAT in daily prac- Tel: 916 734 6969; fax: 916 734 6992; e-mail: [email protected] tice. Current Opinion in Ophthalmology 2004, 15:85–89

Abbreviations In 1975, Ehlers [2] cannulated 29 otherwise normal eyes undergoing cataract surgery and correlated corneal thick- CCT central corneal thickness EMGT Early Manifest Glaucoma Trial ness with errors in GAT. He found that GAT most ac- GAT Goldmann applanation tonometry curately reflected true intracameral IOP when CCT was IOP intraocular pressure LASIK laser in situ keratomileusis 520 µm, and that deviations from this value resulted in an OHTS Ocular Hypertension Treatment Study over- or underestimation of IOP by as much as 7 mm Hg © 2004 Lippincott Williams & Wilkins per 100 µm. Numerous investigators have since demon- 1040-8738 strated that CCT varies far more among otherwise nor- mal individuals than Goldmann and Schmidt ever 85 86 Glaucoma dreamed; differences in CCT are seen among different glaucoma, and pseudoexfoliation glaucoma. The study racial and ethnic groups [3–5], and may lead to misclas- outcome is glaucoma progression as measured by quan- sification of patients with normal tension glaucoma and titative visual field criteria and optic disc changes as de- ocular hypertension [6,7]. termined by a disc reading center. The identification of subjects took place in Malmö and Helsingborg, Sweden. The Ocular Hypertension Treatment Study A total of 255 patients were enrolled with randomization The importance of CCT in the management of glaucoma concluded in April 1997. patients, particularly those with ocular hypertension, was recently driven home by findings from the Ocular Hy- The primary outcome result, with median follow-up of 6 pertension Treatment Study (OHTS) [8,9]. In the years, was published in October 2002 [11]. The treat- OHTS, patients were recruited who had untreated GAT ment group maintained a mean IOP reduction from measurements in one eye between 24 and 32 mm Hg on baseline of 25%, whereas the observation group main- two separate occasions (the other had to be between 21 tained its baseline IOP through the course of follow-up. and 32 mm Hg), with no secondary cause of elevated Progression was less frequent in the treatment group IOP. The patients all had normal visual fields and optic (45%) compared with the observation group (62%) and nerves. CCT was measured approximately 2 years after occurred later. The median time to progression was 66 enrollment was completed. Among the OHTS partici- months in the treatment and 48 months in the observa- pants, 25% had CCT values greater than 600 µm [5]. If tion groups. The EMGT provided unequivocal evidence one uses Ehler’s correction of approximately 7 mm that reducing the IOP makes a difference in slowing Hg/100 µm deviation from the nominal value of 520 µm, glaucomatous damage. It thus corroborates the major then as many as 50% of OHTS subjects had corrected conclusion of OHTS, but at a more advanced stage of the IOP values upon entry less than 21mm Hg! In a multi- disease process, and it reaffirms the primary role of IOP variate model of baseline characteristics predictive of in the pathogenesis of glaucomatous damage. which subjects would develop glaucoma, CCT proved to be the most potent [9]. In a publication analyzing baseline risk factors for pro- gression of glaucoma among EMGT participants, Leske The OHTS results demonstrate that many patients are et al. reported that CCT was not a significant predictor of being misclassified in terms of glaucoma risk on the basis progression [12••]. This seeming contradiction of the of erroneous IOP estimates by GAT. Clearly, many in- OHTS findings are best understood by recognizing the dividuals with elevated GAT measurements but no other following. findings suggestive of glaucoma probably have normal true IOPs and do not need treatment or even increased 1. The OHTS used IOP as its primary entry criterion glaucoma surveillance. (at least from the standpoint of recruitment and screening), with normality of visual fields and optic The Early Manifest Glaucoma Trial nerves confirmed afterwards. If the influence of In contrast to the OHTS, which studied patients initially CCT as a predictive factor is primarily as a con- enrolled without glaucoma damage, the Early Manifest founder of IOP measurement, then the entry criteria Glaucoma Trial (EMGT) studied subjects with glauco- for OHTS and its study design is perfectly set up to matous damage documented at enrollment [10]. Like demonstrate that CCT causes a misclassification of the OHTS, the EMGT measured CCT in all partici- risk—thus the powerful effect CCT had in predict- pants shortly after enrollment was completed. The ing who went on to get glaucoma. EMGT randomized patients with primary open-angle 2. In contrast, the EMGT was a population-based glaucoma, normal tension glaucoma, and pseudoexfolia- study, with patients recruited based on the presence tion glaucoma to either treatment with topical betaxolol, of damage regardless of IOP. This study design ␤ a 1 selective adrenergic antagonist, combined with ar- eliminated the recruitment bias an IOP cutoff can gon laser trabeculoplasty, or to observation. As a safety cause. Indeed, a large portion of the subjects were measure for eyes with sustained IOP elevation greater those whom many would consider as having normal than or equal to 35 mm Hg, delayed treatment was per- tension glaucoma. Thus, at the outset, the EMGT mitted for the observation group, as was the addition of started with patients who had demonstrated the pro- latanoprost, a topical prostaglandin analog, after it be- pensity to sustain damage at whatever their true IOP came commercially available in 1996. might be; errors in IOP measurement become less important in such a situation. It may be that if and Eligible subjects were from 50 to 80 years of age, with when we have a correction algorithm and apply it to newly diagnosed, previously untreated chronic open- the EMGT data, a dose–response relation between angle glaucoma. The diagnosis of glaucoma required re- true IOP and progression would be demonstrated. peatable visual field defects in at least one eye and in- Finally, the EMGT was a relatively small sample (at cluded primary open-angle glaucoma, normal-tension least compared with the OHTS) and may not have Corneal thickness and glaucoma Brandt 87

the statistical power to find such a relation. The tion. In an intriguing small study of 26 eyes in Japanese EMGT investigators have published few details on patients with pseudoexfoliation syndrome, Inoue et al. the range and distribution of CCTs they measured in [18] recently found corneas thinner than age-matched their racially homogeneous population; it is quite controls without pseudoexfoliation, regardless of wheth- possible it was far narrower than what was found in er glaucoma was present. the OHTS. If there is an effect of CCT on progres- sion rates in established disease, the EMGT might The laser in situ keratomileusis–thinned cornea be too small, and the range of IOPs and CCTs too A million laser in situ keratomileusis (LASIK) procedures narrow, to detect it. are performed each year among mostly young to middle- aged myopes. Myopia is a strong risk factor for the de- Central corneal thickness as a glaucoma risk factor velopment of glaucoma [19], and many of the patients The OHTS convincingly demonstrated that CCT plays undergoing LASIK today are destined genetically to de- an important role in risk stratification among patients in velop glaucoma in the coming decades. Although we do whom glaucoma was of concern. Medeiros et al. have not yet know how to correct a GAT measurement made provided additional support of a role for CCT in risk on a LASIK-thinned cornea, it is clear that in many cases determination. They assessed a large cohort of ocular GAT will grossly underestimate IOP. The problem will hypertensive individuals with a battery of structural and arise 10 or 15 years from now when patients neglect to psychophysical tests and followed them longitudinally, inform their ophthalmologist that they had LASIK years looking for reproducible signs of early glaucoma. Patients ago in 2003, and a GAT measurement of 18 mm Hg is with frequency doubling technology perimetry defects regarded as normal despite a 425-µm cornea. Unless had thinner corneas than those with normal results pachymetry becomes a part of the routine eye examina- [13••]. Similarly, patients with reproducible short wave- tion (or tonometry technologies independent of CCT are length automated perimetry defects, an indicator of early developed and widely disseminated), patients like this glaucomatous damage, had thinner corneas than subjects will fall through the cracks. with normal short wavelength automated perimetry test- One promising new technology that may prove useful is ing [14••]. the dynamic contour tonometer. This device consists of an electronic strain gauge embedded in a contoured plas- Central corneal thickness differences among tic tip. When in contact with the cornea, the tonometer racial groups tip creates a tight-fitting shell on the corneal surface Several investigators recently provided further evidence without applanation of corneal tissue. The assumption is that African-American subjects, as a group, tend to have that the tonometer compensates for all forces exerted on thinner corneas than their white counterparts. Nemesure the cornea, allowing the strain gauge to measure IOP et al. [15], following a CCT survey of participants in the largely independently of corneal properties. Preliminary Barbados Eye Survey, reported that black participants data with the device suggest that it is more accurate in had thinner corneas (mean thickness 529.8 µm) than LASIK-thinned corneas than GAT [20•]. white participants (545 µm). No relation between IOP and CCT was found in this population-based survey. Implications for clinical practice Shimmyo et al. [16] performed a retrospective biometric Confronted with the expanding evidence that CCT is an review of patients at a large center, also important ocular parameter that should be measured in finding that African-American patients had thinner cor- clinical practice, ophthalmologists understandably won- neas than white patients seeking refractive surgery; they der how to obtain the measurements and what to do with found no difference in CCT among white, Asian, and the information. Newer technologies such as optical co- Hispanic patients in their population. This is in contrast herence tomography [21] and partial coherence interfer- to the findings of the population-based Los Angeles La- ometry [22] seem likely to become part of a future bio- tino Eye Study [17], which found CCTs among their metric workstation. In the meantime, ultrasonic Hispanic patients intermediate between values reported pachymeters are reasonably priced (less than US$3000), for African-American and white populations. extremely accurate, and easy to use: the acquisition of CCT data can safely be delegated to technical personnel Central corneal thickness differences among in a busy clinical practice. How to use CCT measure- glaucoma syndromes ments in daily practice, however, is not as straightfor- Are different glaucoma syndromes associated with dif- ward; there is wide disagreement among investigators as ferences in CCT? Normal-tension glaucoma is associated to whether there is an adequately validated correction with thinner CCTs than primary open-angle glaucoma, algorithm; without a validated algorithm, the argument but this may simply represent a misclassification of pa- goes, clinicians cannot use the data. tients with primary open-angle glaucoma as having nor- mal-tension glaucoma on the basis of tonometry error The question of whether such an algorithm exists is, in and therefore may not represent a true biologic associa- my opinion, not particularly important in daily practice. 88 Glaucoma

Indeed, the clinician should be cautious in rigidly ex- ers—patients who did not seem to fit the mold of a pres- trapolating any algorithm into practice. For example, sure-sensitive disease. As new technologies are devel- Ehlers’ oft-quoted study [2] was based on a small num- oped to measure IOP with greater precision (and on a ber of eyes (29) that included a relatively narrow range of continuous basis), it seems likely that a much tighter CCTs (450 to 590 µm). The interested reader is referred dose–response relation between glaucoma damage and to a detailed exploration of the mechanical characteristics IOP will be found. Incorporating corneal thickness into of the cornea and the role of CCT in GAT error by our thinking is but the beginning of this transformation. Orssengo and Pye [23] in which a mathematical and en- gineering model closely approximates Ehlers’ and other References and recommended reading published cannulation data. Papers of particular interest, published within the annual period of the review, have been highlighted as: A general recommendation supported by the data so far • Of special interest is that one can take far better care of patients simply by •• Of outstanding interest 1 Goldmann H, Schmidt T: U¨ ber applanationstonometrie. Ophthalmologica categorizing corneas as thin, average, or thick, just as it is 1957, 134:221–242. important to recognize that optic discs come in small, 2 Ehlers N, Bramsen T, Sperling S: Applanation tonometry and central corneal medium, and large, allowing the clinician to interpret thickness. Acta Ophthalmol (Copenh) 1975, 53:34–43. disc configurations accordingly. 3 La Rosa FA, Gross RL, Orengo-Nania S: Central corneal thickness of Cau- casians and African Americans in glaucomatous and nonglaucomatous popu- Conclusion lations. Arch Ophthalmol 2001, 119:23–27. 4 Foster PJ, Baasanhu J, Alsbirk PH, et al.: Central corneal thickness and intra- The story of corneal thickness is but the tip of the ice- ocular pressure in a Mongolian population. Ophthalmology 1998, 105:969– berg. The Goldmann tonometer simply measures the 973. force needed to deform the cornea in a standardized 5 Brandt JD, Beiser JA, Kass MA, et al.: Central corneal thickness in the Ocular manner. IOP is derived from the force measurement in- Hypertension Treatment Study (OHTS). Ophthalmology 2001, 108:1779– 1788. directly, based on a number of assumptions about cor- 6 Ehlers N, Hansen FK: Central corneal thickness in low-tension glaucoma. neal deformability. Corneal deformability in turn repre- Acta Ophthalmol (Copenh) 1974, 52:740–746. sents a summation of the cornea’s curvature, elastic 7 Copt RP, Thomas R, Mermoud A: Corneal thickness in ocular hypertension, properties, surface tension, and the IOP. Although CCT primary open-angle glaucoma, and normal tension glaucoma. Arch Ophthal- is probably the major component of corneal elasticity, it mol 1999, 117:14–16. [see comments] is likely not the only component. The mix of collagen 8 Kass MA, Heuer DK, Higginbotham EJ, et al.: The Ocular Hypertension Treat- ment Study: a randomized trial determines that topical ocular hypotensive types, corneal hydration, packing density of collagen medication delays or prevents the onset of primary open-angle glaucoma. fibrils, the extracellular matrix, and other factors un- Arch Ophthalmol 2002, 120:701–713. doubtedly vary among individuals; in some patients 9 Gordon MO, Beiser JA, Brandt JD, et al.: The Ocular Hypertension Treatment Study: baseline factors that predict the onset of primary open-angle glau- these other factors may dwarf the effect of CCT on the coma. Arch Ophthalmol 2002, 120:714–720. accuracy of IOP estimation. A LASIK flap can be lifted 10 Leske MC, Heijl A, Hyman L, et al.: Early Manifest Glaucoma Trial: design and easily a year or more after the procedure. Does this flap baseline data. Ophthalmology 1999, 106:2144–2153. slide over its bed during applanation, altering corneal 11 Heijl A, Leske MC, Bengtsson B, et al.: Reduction of intraocular pressure and rigidity in ways we can’t imagine? Might the potent IOP- glaucoma progression: results from the Early Manifest Glaucoma Trial. Arch Ophthalmol 2002, 120:1268–1279. lowering effect of the prostaglandin analogues result not 12 Leske MC, Heijl A, Hussein M, et al.: Factors for glaucoma progression and only from their impressive effects on outflow but also •• the effect of treatment: the early manifest glaucoma trial. Arch Ophthalmol through an effect on the corneal extracellular matrix (and 2003, 121:48–56. This study provides a detailed analysis of factors predictive of glaucoma progres- thus the cornea’s deformability and elastic properties)? sion among participants in the Early Manifest Glaucoma Trial. Central corneal thick- No one knows. Finally, some have proposed that the risk ness was not found to be a predictive factor, in contrast to the Ocular Hypertension Treatment Study. See the text for a possible explanation of this seeming paradox. relation of CCT and glaucoma represents not just an artifact of GAT but also a biomechanical relation be- 13 Medeiros FA, Sample PA, Weinreb RN: Corneal thickness measurements •• and frequency doubling technology perimetry abnormalities in ocular hyper- tween CCT and the support structures of the optic tensive eyes. Ophthalmology 2003, 110:1903–1908. This study demonstrated that CCT is thinner in ocular hypertensive individuals with nerve. These hypotheses are pure speculation at this abnormalities in frequency doubling technology perimetry, a technique commonly point but seem worthy of serious investigation in coming used in screening that can detect early glaucomatous field loss. years. 14 Medeiros FA, Sample PA, Weinreb RN: Corneal thickness measurements •• and visual function abnormalities in ocular hypertensive patients. Am J Oph- thalmol 2003, 135:131–137. We now grudgingly recognize that our ability to accu- This study demonstrates that CCT is thinner in ocular hypertensive individuals with rately measure IOP is far weaker than we’ve ever imag- abnormalities in short wavelength automated perimetry, a technique that can de- tect the earliest glaucomatous field loss. ined; we have spent half a century believing that a 15 Nemesure B, Wu SY, Hennis A, et al.: Corneal thickness and intraocular flawed one-time measurement told us enough about our pressure in the Barbados eye studies. Arch Ophthalmol 2003, 121:240– patients on which to base clinical decision-making. A 244. failure to question whether our measurement techniques 16 Shimmyo M, Ross AJ, Moy A, et al.: Intraocular pressure, Goldmann appla- nation tension, corneal thickness, and corneal curvature in Caucasians, were sufficiently accurate to guide patient care has led us Asians, Hispanics, and African Americans. Am J Ophthalmol 2003, to propose a variety of hypotheses to explain the outli- 136:603–613. Corneal thickness and glaucoma Brandt 89

17 Hahn S, Azen S, Ying-Lai M, et al.: Central corneal thickness in Latinos. Invest An initial evaluation report of a promising new tonometry technique. Ophthalmol Vis Sci 2003, 44:1508–1512. 21 Wirbelauer C, Scholz C, Hoerauf H, et al.: Noncontact 18 Inoue K, Okugawa K, Oshika T, et al.: Morphological study of corneal endo- with slit lamp-adapted optical coherence tomography. Am J Ophthalmol thelium and corneal thickness in pseudoexfoliation syndrome. Jpn J Ophthal- 2002, 133:444–450. mol 2003, 47:235–239. 19 Mitchell P, Hourihan F, Sandbach J, et al.: The relationship between glaucoma 22 Rainer G, Petternel V, Findl O, et al.: Comparison of ultrasound pachymetry and myopia: the Blue Mountains Eye Study. Ophthalmology 1999, and partial coherence interferometry in the measurement of central corneal 106:2010–2015. thickness. J Cataract Refract Surg 2002, 28:2142–2145. 20 Kaufmann C, Bachmann LM, Thiel MA: Intraocular pressure measurements 23 Orssengo GJ, Pye DC: Determination of the true intraocular pressure and • using dynamic contour tonometry after laser in situ keratomileusis. Invest modulus of elasticity of the human cornea in vivo. Bull Mathematical Biol Ophthalmol Vis Sci 2003, 44:3790–3794. 1999, 61:551–572.