Corneal Scarring and Vision in Keratoconus - Vision Research ...

5 downloads 435 Views 235KB Size Report
and by photography, and measurement of corneal curvature. The ... (K.Z., J.T.B., J.J.N.); Southern California College of Optometry, Fullerton,. California (T.B.E.); ...
Cornea 19(6): 804–812, 2000.

© 2000 Lippincott Williams & Wilkins, Inc., Philadelphia

Corneal Scarring and Vision in Keratoconus A Baseline Report from the Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study Karla Zadnik, O.D., Ph.D., Joseph T. Barr, O.D., M.S., Timothy B. Edrington, O.D., M.S., Jason J. Nichols, O.D., M.S., Brad S. Wilson, M.A., Kimberly Siegmund, Ph.D., Mae O. Gordon, Ph.D., and the CLEK Study Group

Purpose. The multicenter Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study is a prospective, observational study of 1,209 keratoconus patients. We report on the correlation of corneal scarring with clinical and patient-reported variables at the baseline visit. Methods. Patients completed a questionnaire on their vision, effect of glare, contact lens wear, and work-related issues. Clinical examination included high- and low-contrast visual acuity, refraction, assessment of corneal scarring by the clinician and by photography, and measurement of corneal curvature. The correlation of central corneal scarring with visual acuity and patient-reported variables was analyzed using multiple regression analysis and generalized estimating equations. Results. High- and low-contrast visual acuity with habitual and optimal correction is reduced in scarred eyes. Multiple regression analyses controlling for age, contact lens wear, and disease severity show that central scarring is associated with poorer visual acuity and increased patient-reported symptoms of glare. Restrictions on day-to-day activities do not appear to be associated with corneal scarring above and beyond the effects of keratoconus alone. Conclusions. Corneal scarring in keratoconus is significantly associated with decreased high- and low-contrast visual acuity. Key Words: Corneal scarring—Keratoconus—Visual acuity— Low-contrast visual acuity.

Although the ectatic area of the cornea is usually displaced inferiorly, the corneal irregularity disrupts vision, even in the earliest stages. Keratoconus patients are usually managed initially with spectacles; rigid contact lenses are prescribed when spectacles no longer provide adequate vision. Typically, 10–20% of keratoconus patients undergo penetrating keratoplasty, usually because of contact lens intolerance, decreased visual acuity, and/or central corneal scarring.2–5 Previous large-scale studies of keratoconus focused on disease incidence, prevalence,4,6 etiologies,7 or trends in the clinical management of keratoconus.2,3,5,8–12 Few characterized the course of the disease and its associated factors in large samples of keratoconus patients.13,14 All previous studies relied on retrospective evaluations of keratoconus patients’ records. Information on apical corneal scarring in keratoconus is scarce. Only two previous retrospective studies reported the prevalence of corneal scarring, reporting a range of prevalence for corneal scarring in keratoconus from 24 to 29%.14 One study specifically did not assess corneal scarring because of a “lack of . . . a consistent description for each eye regarding . . . apical scarring”3 and another did not quantify the frequency of scarring other than to note its appearance in patients with advanced disease.12 Information on the effect of corneal scarring on patient outcomes is elusive. We know that elevated corneal scars and associated abrasions can prohibit the comfortable, long-term wear of rigid contact lenses.15 We know that corneal scars that either limit visual acuity or decrease contact lens comfort are indications for surgical intervention.2,4,5,16,17 There is a suggestion that oval, “sagging,” inferiorly displaced “cones” are associated with more corneal scarring.1,8 The relation between central corneal scarring and visual acuity has not been described. Although several smallscale studies investigated visual function in keratoconus,18–24 only two preliminary investigations characterized vision and scarring in the same patients25,26 Burger and co-workers25 reported that the presence of a corneal scar (as reported by a clinician on slit-lamp biomicroscopic examination) reduced both high- and low-contrast visual acuity by two lines on average. In addition, disability glare was abnormal in scarred eyes, indicating that light scatter from central corneal scars may reduce vision in moderate keratoconus. Barr and Yackels26 found that corneal scarring was associated with a greater than normal difference between

Keratoconus is an ectasia of the axial cornea characterized by corneal distortion and irregular astigmatism. Keratoconus affects both eyes in most cases but is classically described as a markedly asymmetric disease. It is characterized by symptoms of visual distortion, observable corneal irregularity that worsens with time, classic slit-lamp biomicroscopic signs (Vogt’s striae and Fleischer’s ring), stromal thinning, and progressive corneal scarring.1 Submitted December 15, 1999. Revision received March 21, 2000. Accepted March 27, 2000. From The Ohio State University College of Optometry, Columbus, Ohio (K.Z., J.T.B., J.J.N.); Southern California College of Optometry, Fullerton, California (T.B.E.); Department of Ophthalmology and Visual Sciences, Washington University Medical School, St. Louis, Missouri (B.S.W., M.O.G.); and the Department of Preventive Medicine, University of Southern California, School of Medicine, Los Angeles, CA (K.S.), U.S.A. Address correspondence and reprint requests to Dr. K. Zadnik, The Ohio State University College of Optometry, 338 West Tenth Avenue, Columbus, OH 43210-1240, U.S.A. E-mail: [email protected]

804

CORNEAL SCARRING AND VISION IN KERATOCONUS high- and low-contrast visual acuity. Again, corneal scarring was determined by slit-lamp biomicroscopic examination. The Collaborative Longitudinal Evaluation of Keratoconus (CLEK) Study developed a protocol for the unbiased determination of corneal scarring. We use standardized photography and centralized masked reading to document corneal scarring. This system has proven to be repeatable and valid.27,28 The purpose of this report from the CLEK Study is to document the correlation of baseline corneal scarring with baseline patient outcomes, including visual acuity, contact lens wear, glare disability, and employment.

METHODS CLEK Study Design The CLEK Study is a 16-center observational study of patients with keratoconus. In total, 1,209 eligible patients were enrolled between May 31, 1995, and June 29, 1996. Patients were eligible if they satisfied all three of these entry criteria: (i) were at least 12 years old; (ii) had an irregular corneal surface in at least one eye by distorted keratometric mires, scissoring of the retinoscopic reflex, or irregularity in the red reflex with the direct ophthalmoscope; (iii) had either Vogt’s striae, a Fleischer’s ring of at least 2 mm arc, or corneal scarring characteristic of keratoconus in at least one eye. Patients were excluded if they had undergone bilateral corneal transplants or if they had nonkeratoconic ocular disease in both eyes, e.g., cataract, intraocular lens implants, macular disease, and/or optic nerve disease other than glaucoma. The protocol used for the study was described in detail elsewhere.29 The procedures outlined below are those relevant to the present report. Each patient completed the National Eye Institute-Visual Function Questionnaire (NEI-VFQ)30,31 and a Patient History Form. Questions specific to this report from the Patient History Form are as follows. (i) Patients estimated the effect of glare on their activities by answering the question, “How much are you hindered, limited, or disabled by glare (dazzling light) in each of the following activities? (Your usual daily activities; reading shiny paper (such as magazine pages); driving toward the sun; driving toward oncoming headlights at night; walking outside on a sunny day).” Responses included “Not at all,” “A little bit,” “Some,” “Quite a lot,” “Can’t do because of vision,” and “Don’t do for reasons other than vision.” (ii) Patients estimated their visual limitations by answering the question, “How much does your vision hinder, limit, or disable you in each of the following activities? (Usual daily activities; recognizing people or objects across the street; reading the labels in stores and supermarkets; reading a magazine, newspaper, telephone book; or sewing; daytime driving; nighttime driving; kind or amount of work you can do; your education; your recreation and hobbies)” on a scale with responses “Not at all,” “A little bit,” “Some,” “Quite a lot,” “Can’t do because of vision,” “Don’t do for reasons other than vision.” (iii) Contact lens– wearing patients assessed their contact lens wear by responding “Yes” or “No” to the question, “Can you wear your contact lenses for enough hours to permit leisure activities?” (iv) Patients reported their employment and work status by responding “Yes” or “No” to the questions “Are you currently employed or have you been employed (part-time or full-time) in the last 12 months?”; “Have you changed jobs in the last 12 months?”; or “In the last 12 months did you miss any days of work due to your keratoconus?”

805

(v) Patients rated their own vision by responding separately for each eye to the question, “How would you rate your vision with your usual correction? (Excellent, very good, good, fair, or poor).” Additionally, a series of questions about the patient’s ocular history, general health, and contact lens history were completed in an interview format. The interviewer collected data on the patient’s ocular and medical history, family history, and history of contact lens wear, as well as on the patient’s health-related and visionrelated quality of life. Questions specific to this report are as follows: (i) To report their eye-rubbing habits, patients answered, separately for each eye, “Yes,” “No,” or “Unsure” to the question, “Do you rub your eyes vigorously?” (ii) To report contact lens comfort, contact lens–wearing patients answered, separately for each eye, “Very comfortable,” “Quite comfortable,” “Comfortable,” “Somewhat irritating,” or “Irritating” to the question, “In general, how comfortable are your contact lenses?”

Examination A detailed examination was performed at baseline on each eligible patient. Bailey-Lovie32 (School of Optometry, University of California, Berkeley, CA) distance visual acuities were measured in three ways. (i) Entrance visual acuity was measured with the patient’s visual correction in place that he or she wore to the baseline visit. Both high- and low-contrast visual acuities were measured with habitual correction, for each eye separately and with both eyes together. (ii) Best corrected visual acuity was measured by assessing high- and low-contrast visual acuities with best correction (for rigid contact lens wearers: their rigid contact lenses with optimal spherocylindrical overrefraction, for those patients who do not wear rigid contact lenses: a CLEK Study trial contact lens with base curve radius equal to the steep keratometric reading plus optimal overrefraction), for each eye separately. (iii) Finally, manifest refraction visual acuity was measured by using highcontrast Bailey-Lovie visual acuity with manifest refraction, for each eye separately (see Appendix). Keratometry was performed by taking two readings in the flattest and steepest meridians. The keratometric range was extended, if needed, as outlined previously.29,33 Slit-lamp biomicroscopy was performed according to a standardized protocol. Corneal scarring was scored as “definitely not scarred,” “probably not scarred,” “probably scarred,” and “definitely scarred” and was graded for size, density, and location. A protocol for determining the first definite apical clearance lens (FDACL) to provide a measure of corneal curvature was developed specifically for the CLEK Study.34 A rigid contact lens from the CLEK Study trial lens set with a base curve radius equal to the steep keratometric reading was applied. If the initial trial lens was judged to be flat centrally, steeper trial lenses were applied to the eye for fluorescein pattern evaluation. This procedure was repeated until an apical clearance pattern was achieved. If the initial trial lens was judged to be steep centrally, flatter trial lenses were applied to the eye until an apical touch fluorescein pattern was observed. Four parallelepiped photographs of the central cornea were taken, and the CLEK standardized corneal photography protocol was used to determine the presence or absence of corneal scarring.28 After pupillary dilation, two oblique photographs were taken of the entire cornea. Exposed but undeveloped film was mailed to the CLEK Photography Reading Center at The Ohio State University College of Optometry for centralized develop-

Cornea, Vol. 19, No. 6, 2000

806

K. ZADNIK ET AL.

ment, labeling, and grading. Photography readers are masked to information regarding the patient, the eye, and the clinic.

Definition of Scarring For the purposes of this report, corneal scarring is defined as a corneal scar within a 6-mm diameter central region (as reported by the examining clinician), which is classified as a “definite” corneal scar by both the clinician and by the CLEK Photography Reading Center (95% of the eyes judged by the clinician only as having a “definite” scar have those scars within the central region). These criteria define a sample of 727 patients with no scars, 226 patients with unilateral scars, and 133 patients with bilateral scars at baseline. (The remaining 123 patients include 118 patients who had undergone corneal surgery in either eye and five patients with missing data on scarring status for one or both eyes.)

Statistical Methods Data for this report are limited to baseline data from the clinical examination, questionnaires completed by the patient, and corneal scarring assessments by readers at the CLEK Photography Reading Center. The analysis dataset excludes patients with a corneal transplant (n ⳱ 116) or epikeratoplasty (n ⳱ 2) in either eye at baseline. The outcome variables selected for analysis for this report were selected for their relevance to patients. Univariate and bivariate descriptive statistics were computed using categorical, rank, and interval analyses. The glare tolerance scale score, “glare,” is the mean of responses to five questions on the effect of glare on vision. The “hindrance from glare” scale score is the mean of nine questions about the effect of keratoconus on visual activities of daily living. Questions for each scale were moderately to highly intercorrelated, and the missing data rate was very low. The sample for descriptive statistics for outcome variables that are eye specific (i.e., measured for each eye) includes both eyes from each patient in the analysis dataset. Eye-specific outcome variables include monocular visual acuities, the sphere value from the manifest refraction, contact lens comfort, vision rating, and eye rubbing. The sample for descriptive statistics for patient-specific variables is all patients in the analysis dataset. Patient-specific variables that were studied include binocular visual acuities and patient-reported “glare,” “hindrance from glare,” “vision,” “use of contact lenses for leisure activities,” and “employment status.” For the purposes of analysis, patient scarring status is coded as scarring in neither eye, one eye, or both eyes. All visual acuities are reported as the number of letters correctly read by the patient. The association between baseline demographic (age, marital status, gender, and education) or clinical factors (scarring, steep kera-

tometry, flat keratometry, FDACL, and contact lens correction) and outcomes are analyzed using multiple regression and generalized estimating equations. Multiple regression is used to describe the percentage of the total variance in outcome variables (visual acuity, the sphere value from the manifest refraction, contact lens comfort, self-reported glare and vision rating, eye rubbing, and employment status) explained by scarring and other important covariates and to identify factors that are associated with these outcome variables. The percentage of variance explained in each outcome variable is estimated for the joint effect of all variables and for each variable separately. Multiple regression models assume that each observed response is independent of the others, so for patient-specific outcomes, e.g., binocular visual acuity, all measurements can be used. For eye-specific variables, e.g., monocular visual acuity, we create two samples, the first consisting of one eye selected randomly from each patient in the analysis dataset and the second consisting of the eyes not selected. Multiple regression analyses are run separately on each sample. The results in both samples are similar, and we report variance estimates from multiple regression analyses from the sample of selected eyes only. Generalized estimating equations (GEE), which permit the use of data from both eyes, are used to confirm the selection of covariates from multiple regression analyses and for significance testing. GEE provide a method to account for the correlation structure between the eyes and adjusts the confidence intervals appropriately. P values for the association of scarring with outcome variables are estimated using GEE models adjusting for age, contact lens correction, and corneal curvature (FDACL). For eyespecific outcome variables, e.g., monocular visual acuity, the GEE model uses the observed value for both eyes. For patient-specific outcome variables, e.g., binocular visual acuity, the GEE model uses a “recoded” value for the following variables. The scarring status of each eye was recoded as the number of scarred eyes (neither, one, or both eyes). Contact lens correction of each eye was recoded as the number of eyes corrected by contact lenses (neither, one, or both eyes). FDACL was recoded as the steeper FDACL of the two eyes and the flatter FDACL of the two eyes.

RESULTS Table 1 presents the distribution of monocular entrance visual acuity according to whether an eye is scarred or unscarred. In general, scarring shifts the distribution of visual acuity toward worse visual acuity. For example, 21.5% of scarred eyes have monocular entrance high-contrast visual acuity from 20/41 to 20/ 69, whereas only 11.2% of unscarred eyes have monocular entrance high-contrast visual acuity from 20/41 to 20/69. Likewise,

TABLE 1. Monocular entrance high- and low-contrast Bailey-Lovie visual acuity by corneal scarring status 20/20 or better

20/21 to 20/40

20/41 to 20/69

20/70 to 20/199

20/200 or worse

Total

392 (23.3%) 37 (7.5%)

992 (59.0%) 267 (54.1%)

188 (11.2%) 106 (21.5%)

74 (4.4%) 45 (9.1%)

35 (2.1%) 39 (7.9%)

1,681 (77.3%) 494 (22.7%)

29 (1.7%) 0 (0.0%)

569 (33.9%) 61 (12.4%)

664 (39.6%) 161 (32.7%)

345 (20.6%) 209 (42.4%)

72 (4.3%) 62 (12.6%)

1,679 (77.3%) 493 (22.7%)

High-contrast Unscarred eyes Scarred eyes Low-contrast Unscarred eyes Scarred eyes

Both eyes of each patient are included, but patients with a corneal transplant or epikeratoplasty in either eye are excluded. Each column contains the number of eyes and the corresponding percentage of the total number of unscarred or scarred eyes in parentheses.

Cornea, Vol. 19, No. 6, 2000

CORNEAL SCARRING AND VISION IN KERATOCONUS

807

TABLE 2. Monocular best corrected high- and low-contrast Bailey-Lovie visual acuity by corneal scarring status

High-contrast Unscarred eyes Scarred eyes Low-contrast Unscarred eyes Scarred eyes

20/20 or better

20/21 to 20/40

20/41 to 20/69

20/70 to 20/199

20/200 or worse

Total

578 (34.6%) 57 (11.8%)

959 (57.4%) 280 (58.1%)

105 (6.3%) 90 (18.7%)

24 (1.4%) 46 (9.5%)

5 (0.3%) 9 (1.9%)

1,671 (77.6%) 482 (22.4%)

28 (1.7%) 0 (0.0%)

726 (43.6%) 68 (14.4%)

661 (39.7%) 173 (36.7%)

232 (13.9%) 200 (42.4%)

20 (1.2%) 31 (6.6%)

1,667 (77.9%) 472 (22.1%)

Both eyes of each patient are included, but patients with a corneal transplant or epikeratoplasty in either eye are excluded. Each column contains the number of eyes and the corresponding percentage of the total number of unscarred or scarred eyes in parentheses.

scarred eyes is statistically significant for both high- and lowcontrast best corrected acuity, but the difference is more marked for low- than for high-contrast monocular best corrected acuity. The mean low-contrast, best corrected visual acuity is 25.2 ± 14.2 (SD) letters correct for scarred eyes compared to 35.9 ± 10.8 (SD) letters correct for unscarred eyes. Mean differences, adjusting for age, contact lens correction, and FDACL, were statistically significant (p < 0.0001). Scarring is associated with a more minus manifest refraction sphere (Table 1). Table 4 gives the means, standard deviations, and p values from multivariate GEE analyses for binocular visual acuity and selfreported, patient-specific outcomes. Patients with bilateral scarring have the poorest binocular visual acuity compared to patients with unilateral scarring or patients with no scarring. Patients with both eyes scarred at baseline have binocular entrance high-contrast visual acuity of 45.7 ± 11.0 (SD) letters correct compared to 50.1 ± 7.6 (SD) letters correct for patients with unilateral scarring and 52.7 ± 6.7 (SD) letters correct for patients with no scarring. The association between the number of eyes scarred and reduced binocular entrance high-contrast visual acuity was statistically significant (p ⳱ 0.007) after adjusting for age, contact lens correction, and FDACL. The association of scarring with poorer binocular visual acuity is more marked for binocular low-contrast visual acuity than for binocular high-contrast visual acuity. Patients with both eyes scarred have binocular low contrast visual acuity of 30.9 ± 12.9 (SD) letters correct compared to 36.5 ± 10.3 (SD) letters correct for patients with unilateral scarring and 41.3 ± 9.2 (SD) for patients with neither eye scarred. The association of the number of eyes scarred and reduced binocular low-contrast visual acuity was statistically significant (p < 0.001) after adjusting for covariates.

42.4% of scarred eyes have monocular entrance low contrast visual acuity from 20/70 to 20/199, whereas only 20.6% of unscarred eyes fall into this same category. The same relationships emerge in Table 2, which represents the distribution of monocular best corrected visual acuity as a function of whether an eye is scarred or unscarred. Again, the distribution of scarred eyes is shifted toward worse visual acuity. Further, comparing Tables 1 and 2, it is evident that monocular best corrected acuity is better than monocular entrance visual acuity for both high- and low-contrast in the unscarred eyes. In the case of the scarred eyes, there is little improvement over monocular entrance levels with best correction. For example, 12.4% of the scarred eyes have monocular low contrast entrance visual acuity of 20/40 or better, whereas 14.4% of the scarred eyes have low contrast visual acuity of 20/40 or better with best correction in place. Table 3 presents means, standard deviations, and p values from multivariate GEE analyses for monocular visual acuities and selfreported, eye-specific outcomes. Monocular high-contrast visual acuity of scarred eyes at baseline is consistently worse by one to two lines (5 to 10 letters) compared to unscarred eyes. This differential is observed for monocular entrance visual acuity, both high and low contrast. However, the difference between scarred and nonscarred eyes is greater for entrance low-contrast acuity than for high-contrast acuity. The mean acuity for monocular lowcontrast entrance visual acuity is 20.8 ± 18.8 (SD) letters correct for scarred eyes compared to 32.1 ± 14.5 (SD) letters correct for unscarred eyes. Mean differences, adjusting for age, contact lens correction, and FDACL, were statistically significant (p ⳱ 0.006). Differences between scarred and unscarred eyes at baseline are also observed for monocular best corrected visual acuity at both high and low contrast. The difference between scarred and un-

TABLE 3. Eye-specific outcomes for scarred and unscarred eyes at baseline Nonscarred eyes† Mean Entrance monocular VA High-contrast Low-contrast (letters correct) Best corrected, monocular VA High-contrast Low-contrast (letters correct) Manifest refraction sphere (D) Contact lens comfort Vision rating Eye rubbing? (% yes)

Scarred eyes‡ SD

Mean

SD

p value*

45.5 32.1

13.1 14.5

36.6 20.8

19.0 18.8

0.17 0.006

49.5 35.9 −4.80 2.5 3.0 48.6%

8.2 10.8 5.10 1.2 1.2

41.6 25.2 −8.80 2.5 3.3 47.2%

13.1 14.2 6.00 1.1 1.1

0.0003