Contact lens examination results stratified by optical coherencetomography (OCT) thickness measurements.
Brown JC, Solomon SD, Bressler SB, Schachat AP, DiBernardo C, Bressler NM. Detection of Diabetic Foveal EdemaContact Lens Biomicroscopy Compared With Optical Coherence Tomography. Arch Ophthalmol. 2004;122(3):330-335. doi:10.1001/archopht.122.3.330
To compare contact lens biomicroscopy with optical coherence tomography(OCT) for the detection of diabetic foveal edema.
Study participants consisted of a convenient cohort of consecutive patientswith diabetes mellitus seen at the Wilmer Eye Institute's Retinal VascularCenter, Baltimore, Md. Case characteristics were recorded and eyes were examinedby 1 of 4 retina specialists by means of contact lens biomicroscopy. Edemainvolving the center of the macula was assessed as definitely present, questionablypresent, or definitely not present. The OCT testing was performed and interpretedby trained technicians, masked to the physicians' assessment of foveal edema.Agreement between OCT and contact lens examination for the absence or presenceof foveal edema was evaluated.
One hundred seventy-two eyes of 95 patients with diabetes were enrolledin August and September 2002. Foveal thickness was objectively measured byOCT in 170 (99%) of 172 cases. We found excellent agreement between OCT andcontact lens examination for the absence or presence of foveal edema whenOCT thickness was normal (≤200 µm) or moderately to severely increased(>300 µm). However, agreement was poor when foveal thickness was mildlyincreased on OCT (201-300 µm).
Agreement between contact lens examination and OCT for the detectionof diabetic foveal edema is poor when OCT thickening is mild. This suggeststhat contact lens biomicroscopy is relatively insensitive for the detectionof mild foveal thickening apparent on OCT. Additional studies are needed toinvestigate the natural course of cases with mildly increased foveal thicknesson OCT that do not appear thickened clinically.
The clinical gold standard for the detection of macular edema is viewingthe fundus with a contact lens at the slitlamp through a pharmacologicallydilated pupil. This is a complex psychomotor process that is highly dependenton observer skill and experience, patient cooperation, the degree of pupillarydilation, the amount of media opacity, and the pattern and extent of retinaledema. Recently, devices that can objectively measure retinal morphologiccharacteristics, such as the optical coherence tomography (OCT) scanner1 and the retinal thickness analyzer,2 haveentered the commercial market. Many studies have reported that OCT3- 12 andthe retinal thickness analyzer6,13- 16 canaccurately and reliably quantify macular retinal thickening in diabetic patients.
Strom et al7 recently carried out a studythat demonstrated good agreement between stereoscopic color fundus photographyand OCT for the detection of retinal edema. However, a comparison of OCT withthe current clinical standard of contact lens biomicroscopy for the detectionof macular edema has not been reported. The goal of this study was to determinethe level of agreement between OCT and contact lens biomicroscopy for thedetection of diabetic foveal edema and to search for characteristics thatmight predict disagreement between these modalities. Because of the tremendouspublic health impact of diabetic retinopathy and the skill and equipment neededfor biomicroscopic examination, detection of macular edema by an objectivetechnique such as OCT followed by the prompt initiation of treatment couldimprove visual outcomes for many diabetic patients. The use of OCT was chosenfor the objective measurement of retinal thickness in this study because ofour perception that this instrument was more widely available in ophthalmicpractices than other objective measurement devices.
Study interventions were performed in addition to, and not as a substitutefor, routine ophthalmologic care as part of new-patient and follow-up clinicalencounters. All aspects of this investigation were approved by The Johns HopkinsInstitutional Review Board, Baltimore, Md, and all subjects gave informedconsent before enrollment in the study.
Study participants consisted of a convenient cohort of consecutive diabeticpatients with varying levels of retinopathy, including the absence of retinopathy,examined in the Retinal Vascular Center at the Wilmer Eye Institute, Baltimore,during a 6-week period in August and September 2002. "Convenient" impliedthat every patient seen during that period in the Retinal Vascular Centerwith diabetic retinopathy who might meet the study's inclusion criteria wasinvited to participate by one of the study investigators (J.C.B.) if thatinvestigator was present in the clinic and not interacting with another studyparticipant at the time. Each eye of a study participant was eligible forenrollment provided that the eye underwent ophthalmologic examination andOCT testing during the study visit. Exclusion criteria included the presenceof any retinal or choroidal disease, other than diabetes, that could affectretinal thickness or preclude identification of edema involving the centerof the macula. Media opacity was not an exclusion criterion provided thatthe investigator could assess the presence or absence of retinal edema onbiomicroscopic examination.
Patient characteristics, including age, race, gender, duration of diabetes,type of diabetes, history of previous laser photocoagulation or intraocularsurgery, lens status, and visual acuity, were recorded. Age was defined asthat at the time of examination. Race and gender were self-reported by thepatient. Duration of diabetes was defined as time between the self-reportedage at initial diabetes diagnosis and the age at the time of examination.Type of diabetes was defined as early onset (age <30 years) or late onset(age ≥30 years) on the basis of age at initial diagnosis. Past ocular treatmentand phakic status were determined by review of the patient's medical records,including those of referring physicians, and the ophthalmologic examination.Visual acuity was determined by means of the patients' habitual refractivecorrection with a standard Early Treatment Diabetic Retinopathy Study chartand was recorded as logarithm of the minimum angle of resolution.
Each subject was examined by 1 of 4 retina specialists (S.D.S., S.B.B.,A.P.S., or N.M.B.), who performed contact lens biomicroscopy after pharmacologicpupillary dilation. The investigators assessed retinal thickness at the centerof the macula in terms of edema being definitely present, questionably present,or definitely not present. The level of retinopathy was recorded as none;mild, moderate, or severe nonproliferative; or proliferative. Immediatelyafter clinical examination, the chart note from that day was reviewed by thestudy coordinator (J.C.B.). The presence or absence of a vitreomacular interfaceabnormality was recorded. For cases without foveal thickening or with questionablefoveal thickening on clinical examination, the presence or absence of definiteextrafoveal macular edema was also recorded.
After clinical examination and review of the chart note were completed,OCT (OCT3; Zeiss-Humphrey Systems, Dublin, Calif) and measurement of pupillarydiameter were carried out by a trained OCT technician, masked to the physician'sassessment of foveal edema. The OCT images were generated with the use ofsix 6-mm radial scans in a spokelike pattern according to manufacturer protocolas described in the user's manual.17 The OCToperator closely monitored patient fixation under direct visualization, andscanning was repeated until all reasonable attempts had been made to obtainexcellent fixation maintained over the entire 1.92 seconds. Each of the 6line scans then was reviewed individually to determine whether some or allof the scans imaged the center of the fovea. If none of the line scans imagedthe foveal center, then the scan was repeated. Once all reasonable attemptshad been made to obtain an adequate line scan, the foveal thickness was manuallymeasured by placing calipers at the vitreous-retina and retina-RPE interfaces.Each scan then was interpreted by a second masked observer (C.D.), who assessedthe quality of the OCT image as either adequate or inadequate and recordedthe retinal thickness in microns at the center of the macula.
The clarity of media was assessed indirectly by means of stereoscopicfundus photographs obtained as part of routine care when available. Photographswere used only if obtained on the same day as the clinical evaluation andwere assessed by ophthalmologists who train graders at the Wilmer PhotographicReading Center, Baltimore. The quality of media was recorded as poor, fair,or good.
Descriptive statistics such as frequency, range, mean, and standarddeviation were calculated for case characteristics. Percentage agreement,weighted κ, and Pearson correlation coefficient were calculated to comparecontact lens examination results with the OCT data. Logistic regression analysisusing backward elimination was carried out to search for selected case characteristicsthat might be associated with disagreement between OCT and clinical examinationor the presence of questionable foveal edema by contact lens examination.Characteristics included in the analyses were quality of media, pupillarydiameter, vitreomacular interface abnormality, extrafoveal macular edema,level of retinopathy, and severity of OCT thickening.
The design of this study required a finite cutoff for the upper levelof normal foveal thickness in patients without diabetes. Unfortunately, normalizeddata on foveal thickness in healthy control subjects obtained with the OCT3were not available at the time of this investigation. Therefore, we reviewedthe existing literature in which foveal thickness was measured by OCT in healthycontrols without diabetes3,6,8,10- 12 (summarizedin Table 1). Our analysis suggeststhat 200 µm represents a reasonable and convenient cutoff for the upperlevel of normal foveal thickness in healthy nondiabetic adults. We definedan OCT foveal thickness of 200 µm or less as normal, 201 to 300 µmas mild thickening, 301 to 400 µm as moderate thickening, and greaterthan 400 µm as severe thickening.
We approached 107 patients and requested that they consider participationin the study. Ninety-seven patients (91%) agreed, completed the informed consentprocess, and were enrolled. Two patients were excluded after enrollment becauseone was unable to complete OCT testing during the clinic visit as a resultof time constraints and another left before OCT testing without offering anexplanation. One hundred seventy-two eyes of 95 patients completed the study.Case characteristics are summarized in Table 2 and Table 3.The OCT scans were of sufficient quality for interpretation in 170 (99%) of172 cases. In both cases of uninterpretable scans, the OCT operator attributedthe poor image quality to media opacity.
Of the 172 eyes, edema involving the center of the macula by biomicroscopywas definitely present in 33 (19%), questionably present in 14 (8%), and definitelynot present in 125 (73%) of the cases. Objective foveal thickness measurementswere obtained by OCT in 14 (100%) of 14 cases with questionable foveal edemaby contact lens examination. The clinical presence of foveal thickening demonstratedgood positive correlation with increasing OCT thickness (Pearson coefficient= 0.634; P<.001).
Of 100 eyes with OCT thickness no greater than 200 µm (no OCTthickening), definite edema by biomicroscopy was detected in only 3 eyes,questionable edema in 11 eyes, and no edema in 86 eyes. Of 44 eyes with OCTthickness greater than 200 µm but no greater than 300 µm (mildOCT thickening), definite edema by biomicroscopy was detected in 8 eyes (18%),questionable edema in 2 eyes (5%), and no edema in 34 eyes (77%). Of 16 eyeswith OCT thickness greater than 300 µm but no greater than 400 µm(moderate OCT thickening), definite edema by biomicroscopy was detected in12 eyes (75%) and no edema in 4 (25%). Of 12 eyes with OCT thickness greaterthan 400 µm (severe OCT thickening), definite edema by biomicroscopywas detected in 10 eyes (83%), questionable edema in 1 (8%), and no edemain 1 (8%). Results organized by OCT thickness stratification are summarizedin Figure 1.
Overall agreement between results of contact lens examination and OCTwas only 119 (69%) of 172 eyes (weighted κ = 0.378; P<.001). However, the majority of disagreement occurred for caseswith mild OCT thickening (>200 µm but ≤300 µm), where agreementwas present in only 10 (23%) of 44 eyes. When cases of mild thickening wereexcluded, overall agreement was good and improved to 109 (85%) of 128 eyes(weighted κ = 0.697; P<.001).
Logistic regression analysis was carried out to search for associationsbetween selected case characteristics and disagreement between OCT and contactlens biomicroscopy data or the clinical determination of questionable fovealedema. False-negative clinical assessments (the absence of thickening by contactlens examination for cases with OCT thickness >200 µm) were associatedwith mild OCT thickening (P<.001) and milder retinopathy(P = .054). False-positive clinical assessments (thepresence of thickening by contact lens examination for cases with OCT thickness≤200 µm) were not associated with any of the characteristics evaluated.The presence of questionable foveal edema by contact lens examination wasassociated with the presence of definite extrafoveal macular edema (P<.001). Logistic regression associations are summarizedin Table 4.
This prospective, masked, clinical case series is, to our knowledge,the first reported systematic comparison of contact lens biomicroscopy byskilled retina specialists with OCT for the detection of diabetic foveal edemain a large consecutive series of diabetic patients from one clinical center.By enrolling consecutive patients without visual acuity or media opacity restrictions,we studied a representative cohort of diabetic patients receiving care atour center. The nearly equal numbers of men and women, inclusion of patientswith both early- and late-onset diabetes, and the relatively balanced distributionof retinopathy levels supports the external validity of our findings. Whitesand African Americans were well represented, but Hispanic Americans were notablyabsent from the study cohort.
Contact lens biomicroscopic examination was chosen over a comparisonwith a handheld 90- or 78-diopter lens, even though evaluation of diabeticretinopathy in clinical practice may be performed more often with noncontactlenses. There are at least 3 reasons for using contact lens biomicroscopyrather than a noncontact lens examination for assessing the presence or absenceof edema clinically. First, a previous study at our clinical center demonstratedthat noncontact lens biomicroscopy failed to detect questionable or definiteclinically significant macular edema that was detected by contact lens biomicroscopyin approximately 10% of cases.18 Therefore,use of a contact lens examination is more likely to identify edema. Second,contact lens examinations facilitate the ability to avoid blinking and eyemovement that might interfere with clinical assessment of macular edema comparedwith noncontact lens biomicroscopy. Third, the optics of the contact lensprovide increased axial magnification, which facilitates the detection ofsubtle areas of edema that might not be detected with less axial magnification.These findings have led us to use contact lens examination routinely for evaluatingthe presence or absence of macular edema. (The routine use includes instillationof a drop of topical anesthetic followed by use of a contact lens with a hardcontact lens wetting solution. This solution provides enough viscosity toallow coupling of the lens to the cornea, especially if the patient looksdown into the lens at the time of application, with little likelihood of anair bubble to limit one's view. Also, the hard contact lens wetting solutiondoes not interfere with subsequent examinations by indirect ophthalmoscopyor retinal photography.) In summary, the best clinical examination to assessthe presence or absence of macular edema was judged to be contact lens examination.We estimate that, if noncontact lens biomicroscopy of the macula had beenperformed, approximately 10% more cases of macular edema detected on OCT mightnot have been detected by clinical examination.
We found good agreement between OCT and contact lens examination forthe presence or absence of foveal edema when OCT thickness was considerednormal (≤200 µm) and when it was judged to be moderately to severelyincreased (>300 µm). However, agreement was poor when OCT foveal thicknesswas mildly increased (201-300 µm). It is not known whether results wouldhave been similar if macular edema not involving the fovea had been evaluated.However, recent studies by Hee et al3 as wellas Goebel and Kretzchmar-Gross11 have demonstrateda strong correlation between extrafoveal and foveal thickness measurements,suggesting that results would be similar.
The failure of contact lens biomicroscopy to correlate with mild edemadetected by OCT in this study is indirectly corroborated by several previousinvestigations. Shahidi et al19 showed thatnoncontact slitlamp biomicroscopy failed to detect retinal edema at locationswith an average thickness of 1.5 times normal by examination with a retinalthickness analyzer. Yasukawa et al13 reportedthat OCT demonstrated foveal thickening in 9 (31%) of 29 eyes that appearednormal by noncontact biomicroscopic examination and 6 of 6 eyes with fovealthickening by slitlamp biomicroscopy. Oshima et al15 foundthat eyes with diabetic retinopathy and no macular edema on noncontact biomicroscopicexamination had a mean foveal thickness on OCT of 283 ± 116 µmcompared with 564 ± 168 µm for cases with clinically significantmacular edema. Lattanzio et al12 reported that,among eyes with diabetic retinopathy, those with no macular edema on noncontactbiomicroscopic examination had a mean foveal thickness on OCT of 228 ±53 µm, those with non–clinically significant macular edema hada mean foveal thickness on OCT of 322 ± 124 µm, and those withclinically significant macular edema had a mean foveal thickness on OCT of476 ± 146 µm. All of these investigations suggest that clinicalexamination is relatively insensitive for the detection of mild macular edemaevident by objective imaging methods.
We found a strong non–statistically significant trend toward anassociation between false-negative clinical assessments (the absence of thickeningby contact lens examination for cases with thickness greater than 200 µmon OCT) and milder retinopathy. This was caused by 8 (35%) of 23 cases withmild nonproliferative retinopathy being labeled as having false-negative responsescompared with 30 (21%) of 145 cases with more advanced retinopathy and wasindependent of OCT thickness measurements. The cause and significance of thisassociation is not known. It is possible that investigators were more vigilantwith biomicroscopic examinations when retinopathy was more advanced.
Pupillary diameter and media opacity were not associated with the clinician'sassessment of questionable foveal edema or disagreement between OCT and clinicalexamination results. However, the average pupillary diameter was 7.8 mm witha minimum of 5 mm. Furthermore, media opacity data were available only for98 (57%) of 172 cases, and the quality was graded as fair or good in 92 (94%)of 98 cases. Because small pupils and poor media quality were underrepresented,our study was underpowered to evaluate these characteristics adequately. However,because pupillary diameter and media quality were not specific exclusion criteriafor participation, our data suggest that most patients presenting for careat a tertiary retinal center have ocular media sufficient to perform bothcontact lens biomicroscopy and OCT.
Optical coherence tomography, performed by skilled technicians, wasable to produce interpretable thickness maps in 170 (99%) of 172 cases. Thisis consistent with findings of Polito et al,20 whorecently reported obtaining retinal thickness measurements in 55 of 55 eyeswith macular disease by means of OCT. The excellent technical performanceof OCT supports its potential utility in clinical practice.
Contact lens examination rated the presence of foveal edema as questionablein 14 (8%) of 172 cases. Equivocal examinations were strongly associated withthe presence of extrafoveal edema. This suggests that, for many cases withdefinite extrafoveal edema, it was not apparent whether edema was juxtafovealor actually extended into the macular center.
There are 2 potential systematic errors that could account for the pooragreement between contact lens examination and mild thickening on OCT in thisstudy. The first is the possibility that, for some cases without foveal edema,OCT demonstrated mild foveal thickening because of instrumentation errors.This is unlikely because Neubauer et al6 demonstratedthat OCT yields reproducible foveal thickness measurements with a coefficientof variation of only 10% within subjects. Even if we assume that all caseswith mild foveal thickening by OCT were 10% less thick than measured, agreementbetween contact lens examination and OCT would remain poor and the resultsof the study would be unaffected. The second possible systematic error isthat, for many cases without foveal edema, OCT could have measured parafovealrather than foveal thickness because of small amounts of scan decentration.This is unlikely for 3 reasons. First, the OCT technician took great careto ensure that that each scan was centered on the fovea both by asking thesubject to look at the fixation target and by monitoring fixation under directmacular visualization. Second, the spokelike scanning pattern used by OCTaffords a high density of thickness information at the center of the scan,making it unlikely that the foveal center would be omitted. Third, if thecases with mild thickening on OCT were not truly biologically edematous, contactlens examination would be expected to detect thickening in a proportion ofcases similar to cases without thickening on OCT. Instead, clinical thickeningwas detected in only 3% of cases with no thickening on OCT compared with 18%of cases with mild OCT thickening.
We propose the term subclinical foveal edema todesignate eyes with up to approximately a 50% increase in normal thickness(201-300 µm) by objective imaging methods, because we did not detectsuch thickening reliably by contact lens biomicroscopy. The short- and long-termclinical significance of subclinical edema is unknown. It does not seem reasonableat this time to extrapolate the results of the Early Treatment Diabetic RetinopathyStudy21 to cases of subclinical edema, becauseit is unlikely that such thickening was present in cases judged to have edemaand randomized to focal photocoagulation or deferral of treatment by the EarlyTreatment Diabetic Retinopathy Study investigators.
We recommend additional studies to determine whether subclinical fovealedema progresses to clinically apparent thickening more often than thickeningdevelops in cases without subclinical edema. If such an outcome were found,then the presence of subclinical foveal edema could serve as a marker foreyes at higher risk for conversion to clinically apparent foveal edema andthese patients could be monitored more frequently. To justify treatment ofsubclinical foveal edema, one would need to determine whether the treatmentof subclinical thickening reduces the risk of vision loss to a greater degreethan treatment at the time when the edema is apparent clinically.
In an era when ancillary testing is available to supplement clinicalexamination of diabetic patients, we recommend the following: (1) The presenceof macular edema can be detected with noncontact lens examinations. However,the absence of macular edema with noncontact lens examinations may not necessarilyindicate the absence with contact lens examinations. Therefore, if macularedema is not detected by a noncontact lens examination, the examiner may wantto consider ruling out the presence of edema with a contact lens examination.(2) If macular edema is not detected by contact lens examination, the examinermay want to consider obtaining an OCT examination to determine whether "subclinical"edema (as defined earlier) is present. Eyes with "subclinical" edema may bemore likely to progress to clinically evident edema for which focal laserphotocoagulation may be indicated to reduce the risk of additional visionloss.
Corresponding author and reprints: Neil M. Bressler, MD, RetinalVascular Center, Wilmer Eye Institute, The Johns Hopkins University Schoolof Medicine, Suite 115, 550 N Broadway, Baltimore, MD 21205-2002 (e-mail: firstname.lastname@example.org).
Submitted for publication June 5, 2003; final revision received September15, 2003; accepted October 1, 2003.
This research was supported in part by the Wilmer Eye Institute (JohnsHopkins) Macular Research Fund, Baltimore, Md.