eAppendix. Diabetic Retinopathy Clinical Research Network
Diabetic Retinopathy Clinical Research Network. Macular edema after cataract surgery in eyes without preoperative central-involved diabetic macular edema. JAMA Ophthalmol. Published online April 18, 2013. doi:10.1001/jamaophthalmol.2013.2313.
Diabetic Retinopathy Clinical Research Network Authors/Writing Committee, Baker CW, Almukhtar T, Bressler NM, Glassman AR, Grover S, Kim SJ, Murtha TJ, Rauser ME, Stockdale C. Macular Edema After Cataract Surgery in Eyes Without Preoperative Central-Involved Diabetic Macular Edema. JAMA Ophthalmol. 2013;131(7):870-879. doi:10.1001/jamaophthalmol.2013.2313
The incidence of development or worsening of macular edema (ME) is variable in eyes without diabetic ME (DME) undergoing cataract surgery.
To estimate the incidence of central-involved ME 16 weeks following cataract surgery in eyes with diabetic retinopathy without definite central-involved DME preoperatively.
Design, Setting, and Participants
In a multicenter, prospective, observational study, 293 participants with diabetic retinopathy without definite central subfield thickening on optical coherence tomography (OCT) underwent cataract surgery.
Cataract extraction surgery performed within 28 days of enrollment of eyes without DME in individuals with diabetes mellitus.
Main Outcomes and Measures
Development of central-involved ME defined as the following: (1) OCT central subfield thickness of 250 μm or greater (time-domain OCT) or 310 μm or greater (spectral-domain OCT) with at least a 1-step increase in logOCT central subfield thickness preoperatively to the 16-week visit; (2) at least a 2-step increase in logOCT central subfield thickness preoperatively to the 16-week visit; or (3) nontopical treatment for ME received before the 16-week visit with either of the OCT criteria met at the time of treatment.
The median participant age was 65 years. The median visual acuity letter score was 69 letters (Snellen equivalent 20/40). Forty-four percent of eyes had a history of treatment for DME. Sixteen weeks postoperatively, central-involved ME was noted in 0% (95% CI, 0%-20%) of 17 eyes with no preoperative DME. Of eyes with non–central-involved DME, 10% (95% CI, 5%-18%) of 97 eyes without central-involved DME and 12% (95% CI, 7%-19%) of 147 eyes with possible central-involved DME at baseline progressed to central-involved ME. History of DME treatment was significantly associated with central-involved ME development (P < .001).
Conclusions and Relevance
In eyes with diabetic retinopathy without concurrent central-involved DME, presence of non–central-involved DME immediately prior to cataract surgery or history of DME treatment may increase the risk of developing central-involved ME 16 weeks after cataract extraction.
Diabetes mellitus increases the probability of developing cataract and may increase the risk of reduced visual outcomes after cataract surgery.1,2 As approximately 250 000 people with diabetes in the United States undergo cataract surgery each year, improvements in understanding the prognosis and management of these cases could confer a substantial benefit to many people.3
Past studies on patients with diabetes undergoing cataract removal using intracapsular and extracapsular cataract extraction techniques suggest that cataract surgery is a risk factor for incidence of macular edema (ME) or worsening of diabetic retinopathy.4- 6 Some reports suggest that ME after cataract surgery and in people with diabetes may occur predominantly in people with concurrent preexisting diabetic ME (DME) involving the center of the macula. However, other reports indicate that preexisting DME is not needed for ME to occur postoperatively.7 These studies, though, were completed prior to the availability of optical coherence tomography (OCT) technology. The use of OCT can provide both qualitative and quantitative data to explore the relationship of ME and cataract surgery in patients with diabetic retinopathy.
The Diabetic Retinopathy Clinical Research Network (DRCR.net) conducted an observational study to evaluate the incidence of central-involved ME, as defined by OCT, in eyes without definite central-involved DME immediately prior to cataract surgery. In addition, factors associated with development of central-involved ME are explored.
This prospective, noncomparative, multicenter, observational study was conducted by the DRCR.net at 45 clinical sites throughout the United States. The protocol and Health Insurance Portability and Accountability Act–compliant informed consent forms were approved by the institutional review board for each participating site. Each participant gave written informed consent to participate in the study. The study protocol (named “An Observational Study in Individuals With Diabetic Retinopathy Without Center-Involved DME Undergoing Cataract Surgery”) is available on the DRCR.net website (drcr.net) and summarized here.
Eligible study participants were at least 18 years of age with type 1 or type 2 diabetes and were receiving cataract surgery in an eye with diabetic retinopathy but without definite central-involved DME. Eyes were eligible provided the following criteria were met: (1) presence of cataract for which cataract surgery was scheduled within 28 days of study enrollment; (2) visual acuity (VA) of light perception or better; (3) OCT central subfield (CSF) thickness less than 250 µm on time-domain OCT (TD-OCT) (Stratus [Carl Zeiss Meditec]) or less than 310 µm on spectral-domain OCT (SD-OCT) (Cirrus [Carl Zeiss Meditec], Spectralis [Heidelberg], or Optovue RTVue [Optovue]); and (4) presence of microaneurysms or at least mild nonproliferative diabetic retinopathy (Early Treatment Diabetic Retinopathy Study [ETDRS] level 20 or higher8) on clinical examination. Eyes with major ocular surgery in the prior 4 months and eyes with an ocular condition (other than cataract or DME) that might affect VA during the course of the study were excluded. Only 1 eye from each study participant was enrolled.
Study participants were enrolled by investigators at participating DRCR.net clinical sites certified by the coordinating center in this protocol. Participants were either identified by a participating retina specialist and referred to the cataract surgeon or identified by a cataract surgeon and referred to the retina specialist for enrollment. As the actual cataract surgery was not part of the experimental design, the cataract surgeons were not considered investigators for this protocol. The cataract surgery, including preoperative and postoperative assessments and management, was conducted by and according to the cataract surgeon’s usual routine. With study participant permission, the DRCR.net clinical center obtained the surgical procedure report and postoperative records from the cataract surgeon for data collection.
Protocol study-related visits were specified as preoperative baseline visit (performed ≤28 days preceding surgery), 4-week post–cataract surgery visit, and 16-week post–cataract surgery visit. These visits were conducted at DRCR.net clinical sites by certified investigators and personnel. At all visits, procedures included protocol refraction, best-corrected electronic ETDRS VA test (E-ETDRS),9 fundus examination, and OCT of the study eye following a standardized protocol. When the metrics describing retinal thickness generated by OCT software were suspected to be erroneous by the clinical site or when the ratio of center-point thickness standard deviation to the mean center point thickness from TD-OCT Stratus scans was more than 10.5%, scans were forwarded to a central OCT reading center (Duke Reading Center, Duke University, Durham, North Carolina) for grading; otherwise, the automated OCT thickness measurements were used for analyses.
The DME status at baseline was categorized into 4 categories according to machine- and gender-specific retinal thickness values in the CSF and non-CSF from the OCT map (Table 1).
No DME was defined as thickness in each subfield less than the mean thickness from a diabetic cohort without DME.
Possible DME in the CSF was defined as thickness between the mean and the mean + 2 SDs from a diabetic cohort without DME.
Possible non–central-involved DME was defined as thickness of at least 1 non-CSF between the mean and the mean + 2 SDs from a diabetic cohort without DME.
Definite non–central-involved DME was defined as thickness of at least 1 non-CSF more than the mean + 2 SDs from a diabetic cohort without DME.
Any intervention performed on the study eye during the 16-week course of the study was at the discretion of the cataract surgeon or the DRCR.net investigator. When and if treatment for ME, other than topical medications, was given by the investigator for the first time postoperatively, best-corrected E-ETDRS VA, OCT, and fluorescein angiography were performed before administering the treatment. In addition, fluorescein angiography was required at the 4- and 16-week visits if OCT CSF thickness was at least 250 µm on TD-OCT (≥310 µm on SD-OCT) and sent to the Fundus Photograph Reading Center, University of Wisconsin, Madison, for assessment. Classification of ME patterns used fluorescein angiography and not OCT because, to our knowledge, classification of ME patterns as DME vs postsurgical (Irvine-Gass) cystoid ME (CME) on Stratus OCT is not well documented. Therefore, to have consistency across all eyes, fluorescein angiograms were used for ME classification. In addition, the central fundus photograph reading center has a validated method for assessing ME patterns on fluorescein angiography, but at the time the protocol was developed, it did not have such a validated mechanism for this assessment on OCT. Refraction, VA, OCT, and fluorescein angiography protocols can be found at drcr.net.
The primary outcome, incidence of central-involved ME, was predefined as any of the following events: (1) OCT CSF thickness of at least 250 μm (TD-OCT) or at least 310 μm (SD-OCT) with at least 1 unit increase in logOCT from baseline to the 16-week visit; (2) OCT CSF thickness increase at least 2 logOCT units from baseline to the 16-week visit irrespective of absolute thickness at the 16-week visit; or (3) any nontopical treatment for ME performed at any time after cataract surgery during the study, provided that either of the previous 2 OCT criteria were met prior to starting treatment. An increase of 1 logOCT unit is approximately equivalent to a 26% increase in thickness, and an increase of 2 logOCT units is approximately equivalent to a 36% increase in thickness.10 It was shown that a 1-step log scale change exceeds the measurement error for all degrees of retinal thickness in current instruments; subsequently, a 1-step logOCT change likely represents a real change beyond variability limits. The development (or progression) of non–central-involved ME was defined based on changes in either the inner subfields or the outer subfields on the OCT 6-mm-diameter map (Table 1).
Primary analysis was conducted within predefined DME subgroups based on preoperative OCT retinal thickness as defined in Table 1 and based on history of DME treatment.
Point estimates for binary outcomes and corresponding exact 95% CIs were computed. Association of baseline factors with the primary outcome was evaluated using Fisher exact test; continuous variables were categorized if necessary. All reported P values were 2-sided. For all analyses, SAS version 9.2 statistical software (SAS Institute, Inc) was used.
Between October 2, 2009, and December 31, 2010, 45 clinical sites enrolled 329 study participants scheduled for prompt cataract surgery. Twelve study participants had their surgery canceled and subsequently were not considered part of the enrolled cohort. In addition, 24 eyes were considered ineligible owing to nongradable baseline OCT scans (n = 7), baseline OCT CSF thickness greater than 250 µm on TD-OCT machines or greater than 310 µm on SD-OCT machines (n = 10) after manual adjustment by the Duke Reading Center, or surgery conducted more than 28 days from baseline (n = 7). The mean (SD) age of the remaining 293 participants was 65 (9) years. Women composed 58% of the cohort, and the majority of participants (65%) identified themselves as white. One hundred twenty-eight study eyes (44%) had prior DME treatment, among which only 29 (23%) had received intravitreal injection of steroid within 4 months or anti–vascular endothelial growth factor within 2 months of the cataract surgery. Of 293 eligible eyes at baseline, 18 (6%) did not have complete OCT data (nongradable or missing) for non-CSF thickness. Subsequently, 275 eyes (94%) were classified into 4 DME categories based on status of central and noncentral OCT thickness: (1) no central-involved DME and no non–central-involved DME (ie, no DME) (17 eyes [6%]); (2) no central-involved DME and definite or uncertain non–central-involved DME (100 eyes [36%]); (3) possible central-involved DME and no non–central-involved DME (5 eyes [2%]); and (4) possible central-involved DME and definite or possible non–central-involved DME (153 eyes [56%]). Owing to small group size, 5 eyes originally in the category of possible central-involved DME and no non–central-involved DME were reassigned to other categories based on morphologic features after the protocol chair individually examined their baseline OCT scans; 1, 1, and 3 eyes were reassigned to no DME, no central-involved DME and definite or possible non–central-involved DME, and possible central-involved DME and definite or possible non–central-involved DME groups, respectively. Twenty-four percent of study eyes had a history of panretinal photocoagulation, of which 80% had the panretinal photocoagulation at least 1 year before enrollment. The mean (SD) CSF thicknesses from TD-OCT (n = 281) and SD-OCT (n = 12) machines were 201 (26) µm and 244 (32) µm, respectively. Additional baseline study participant and ocular characteristics are reported in Table 2. Of the 293 participants eligible for analyses, 14 study participants (5%) did not complete the 16-week visit, including 1 who died prior to the 16-week visit, 7 who withdrew from the study, and 6 who were lost to follow-up.
In 98% of study eyes, cataract extraction was performed using phacoemulsification, and 99% had posterior chamber intraocular lens implantation. One eye (<1%) developed endophthalmitis. At the time of cataract surgery, 2 eyes underwent anterior vitrectomy, 1 eye developed mild to moderate positive vitreous pressure, 1 eye had a small corneal abrasion, and 1 eye developed floppy iris syndrome, possibly due to concomitant use of tamsulosin hydrochloride (Flomax). There were no cases of choroidal hemorrhage, ruptured capsule, dropped nucleus, or wound leak. During follow-up, anterior and posterior segment inflammation was noted in 13 eyes (5%) and 3 eyes (1%), respectively. Increased intraocular pressure was reported in 4 eyes (2%) and posterior capsular opacification was reported in 3 eyes (1%). Other reported complications included phacoanaphylactic glaucoma (n = 1), iris prolapse (n = 1), keratitis (n = 1), and wrinkling of the posterior capsule (n = 1). No study eyes were excluded from the analysis owing to any of the complications reported here.
Of 293 eligible eyes, the primary outcome was calculated for 261 eyes (89%) that had complete OCT data at both baseline and the 16-week visit. No eyes without DME at baseline (n = 17) developed central-involved ME (95% CI, 0%-20%). Of the remaining eyes in which ME in the inner and outer subfields could not be ruled out (ie, any subfield more than the mean normal thickness for that subfield), 10 of 97 eyes (10%; 95% CI, 5%-18%) without central involvement (ie, thickness less than the mean normal thickness) progressed to central-involved ME and 18 of 147 eyes (12%; 95% CI, 7%-19%) in which central involvement was possible (ie, between the mean normal thickness and 2 SDs above the mean normal thickness) progressed to central-involved ME. The rate of development of central-involved ME 16 weeks following cataract surgery differed by history of DME treatment (P < .001). Eyes with DME at baseline and no history of DME treatment (n = 140) had a 4% incidence (95% CI, 2%-9%) of central-involved ME and eyes with DME at baseline with a history of DME treatment (n = 104) had a 21% incidence (95% CI, 14%-30%) at 16 weeks following cataract surgery (Table 3).
Of 279 study eyes that completed the 16-week visit, 23 eyes (8%) received nontopical postoperative treatment for ME prior to or at the 16-week visit. The median number of days from surgery to the first postoperative DME treatment was 50 (range, 15-113). Central-involved ME rates at 4 weeks and 4 or 16 weeks were largely comparable to rates at 16 weeks (Table 4). Table 4 also shows rates of development or progression of non–central-involved ME.
The median change in OCT CSF thickness and the median relative change in retinal total volume from baseline to the 16-week visit (or at the first ME treatment visit) for eyes with no DME, eyes with DME not involving the center, and eyes with DME possibly involving the center at baseline were +11 μm (95% CI, +4 to +21 μm) and 5% (95% CI, 2% to 7%), +16 μm (95% CI, +12 to +20 μm) and 5% (95% CI, 4% to 7%), and +8 μm (95% CI, +6 to +11 μm) and 3% (95% CI, 2% to 4%), respectively.
Table 5 provides data on the association of baseline factors with the development of central-involved ME and the development or progression of non–central-involved ME at 16 weeks. History of DME treatment was significantly associated with development of central-involved ME and progression or development of non–central-involved ME (P < .001 for both). Eyes with better VA and those with less severe diabetic retinopathy at baseline tended to have a lower incidence of development of central-involved ME (P = .06 for both).
At the 16-week visit, the median VA score was 81 letters (Snellen equivalent approximately 20/25) (Table 6) with a median change from baseline of +10 letters. Two-hundred forty-one eyes (86%) had a VA of 20/40 or better (approximate Snellen equivalent) at the 16-week visit. A VA loss of at least 10 letters occurred in 4%. Compared with eyes that did not meet criteria of incident central-involved ME, eyes that developed central-involved ME had, on average, less improvement in VA, with a median change of +11 and +5 letters, respectively. In eyes that did not develop central-involved ME and in those that developed central-involved ME, VA at the 16-week visit was 20/40 or better (approximate Snellen equivalent) in 89% and 67%, respectively; however, the former group had better VA at baseline (mean E-ETDRS letter score, 67) and a higher proportion of DME treatment history (80%) than the latter group (mean E-ETDRS letter score, 60; history of DME treatment, 40%).
At the 16-week visit, diabetic retinopathy severity as assessed on clinical examination by the investigator did not change from baseline in 229 eyes (82%) on a scale that included the following: no diabetic retinopathy, microaneurysms only, mild to moderate nonproliferative diabetic retinopathy, severe nonproliferative diabetic retinopathy, and proliferative diabetic retinopathy. At least 1-step and at least 2-step worsening on this clinical scale were reported in 9% and 1%, respectively. Similarly, at least 1-step and at least 2-step improvements on this clinical scale were reported in 9% and 1%, respectively (Table 6). Of the 198 eyes without proliferative diabetic retinopathy at baseline, 4 (2%) progressed to proliferative diabetic retinopathy or had panretinal photocoagulation applied at follow-up.
Of the 36 fluorescein angiographic images graded by the Wisconsin reading center from eyes with a CSF thickness of 250 μm or more, 16 eyes (44%), 15 eyes (42%), and 5 eyes (14%) were classified as DME only, combined CME and DME, and CME only, respectively.
This prospective, multicenter, observational study demonstrates that in eyes without definite central-involved DME within 1 month prior to undergoing cataract extraction, the chance of manifesting central-involved ME and/or development or progression of ME in the non-CSFs 16 weeks following cataract surgery may be influenced by the presence of preexisting DME and history of DME treatment. In this study, eyes with a history of DME treatment had a higher rate of central-involved ME (20%) than eyes with no history of DME treatment (4%) (P < .001) 16 weeks following cataract surgery. The analysis of baseline OCT groups showed that no eyes without DME at baseline developed central-involved ME (95% CI, 0%-20%). The size of this subgroup was limited (n = 17 at 16 weeks) and might not reflect the true rate of ME development in this subgroup. On the other hand, the incidence rate in the subgroups with DME not involving the CSF and the subgroup with DME with possible center involvement were comparable at about 10%.
In this study, similar to more recent studies, OCT was used to evaluate progression of ME following cataract surgery. Earlier studies, however, reported rates of ME progression from other methods; for example, the proportion of eyes manifesting angiographic CME was 9% after cataract surgery using fluorescein angiography in people without diabetes.11 Romero-Aroca et al12 reported that 6.06% of 132 eyes of diabetic patients developed DME on evaluation by fluorescein angiography and OCT following uneventful phacoemulsification. In a single-center study of 50 eyes, Kim et al7 reported an incidence rate of 22% (95% CI, 13%-35%) for DME exacerbation (defined as ≥30% increase in OCT center-point thickness compared with presurgical OCT) 1 month after cataract surgery. All progressing eyes showed cystoid abnormalities on OCT. The variability in reported progression rates of ME following cataract surgery in diabetic persons may be explained by the lack of a unified definition of clinically important progression of ME and/or different population settings or pathologic parameters between studies. In our study, we defined progression of central-involved DME based on the log change in OCT CSF thickness from baseline, which takes into account baseline thickness and requires OCT change beyond variability of the OCT machine itself.10
One of the difficulties of assessing and managing ME after cataract surgery arises from the fact that 2 clinical forms of ME can be present, either alone or in combination, in this setting: DME and postsurgical (Irvine-Gass) CME. Cystoid abnormalities in the macula in the absence of microaneurysms and lipid on clinical examination and petaloid accumulation of fluorescein around the fovea with staining of the optic disc during fluorescein angiography characterize postsurgical CME. Retinal capillary hyperpermeability from intraocular inflammation is thought to be a major pathway in the development of this ME. In the present study, eyes with CSF thickness greater than 250 μm at any study visit were graded by a central reading center in which ME was classified 44% of the time as DME only, compared with approximately 14% CME and approximately 42% a combined mechanism. Depending on the type of ME present, the course of treatment may be different. For example, CME, which shares clinical findings with ME associated with uveitis, may be treated with anti-inflammatory treatments (both corticosteroid and nonsteroidal). On the other hand, anti–vascular endothelial growth factor and steroid products have increased the treatment options for DME. None of these treatments, however, have been proven definitively to have a role in the management of postsurgical ME.
Not surprisingly, after cataract surgery the majority of eyes gained VA. Overall, 86% of eyes had VA of 20/40 or better. A VA of this level is generally attained in a larger proportion of eyes undergoing cataract surgery among persons without diabetes. Similar to our study, Somaiya et al13 reported that 82% of 106 eyes of diabetic persons achieved VA of 20/40 after small-incision phacoemulsification in a retrospective review. On the other hand, VA of 20/40 was achieved in 95% of 55 eyes of nondiabetic persons in the same study. In our study, VA of 20/40 or better occurred less often in eyes that developed central-involved ME (67%). However, these eyes had a lower mean baseline VA (60 letters [approximate Snellen equivalent 20/63]) than eyes that did not develop central-involved ME (67 letters [approximate Snellen equivalent 20/50]). Other studies have also reported less improvement of VA in eyes of diabetic persons in comparison with eyes of nondiabetic counterparts.14
The large sample size and the multicenter nature of this study provide progression rates with more precision and broader representation of an overall US diabetic population when compared with previous studies on progression of ME after cataract surgery. A limitation of this study is that the treatment and prophylaxis for DME and CME prior to, during, and after surgery were not standardized. In addition, because measurements of blood glucose, glycated hemoglobin, and blood pressure were not obtained during follow-up in this observational study, we cannot determine from this study whether sudden worsening of these features accounted for the results. The use of topical nonsteroidal anti-inflammatory drugs or topical steroids was also not standardized; rather, these treatments were left to the discretion of the DRCR.net investigator and cataract surgeon. The use of topical nonsteroidal anti-inflammatory drugs in this study (31%) was lower than previously reported.15,16 This rate may have been underreported as the data were extracted from surgical records that were often separate.
This study represents a collaborative effort between anterior segment and posterior segment ophthalmologists in both academic and community-based institutions. The participants recruited for this study were known to be at risk for ME development or progression. Preoperative, perioperative, and postoperative clinical management was allowed in a manner that reflects common approaches for ME prophylaxis found in a mix of community-based and academic institutions. This study demonstrated that collaboration among DRCR.net retina specialists and cataract surgeons is possible—and might be realistically incorporated into the future design and conduct of experimental studies.
This study shows that eyes with a history of DME treatment and/or DME immediately prior to cataract surgery are at higher risk for developing central-involved ME 16 weeks after the cataract extraction than those with no history of DME treatment or DME. The effect of preoperative DME treatment on postoperative ME could not be evaluated in this study. Clinicians should continue to maintain vigilance in diabetic patients after cataract extraction even when central ME is not present immediately prior to cataract surgery, especially in eyes with prior DME treatment or non–central-involved DME that may be at particularly high risk for development of central-involved ME after cataract surgery. These data may help researchers plan future clinical trials that attempt to minimize this clinically important event. It may also help physicians to explain the risk of DME progression after cataract surgery to their patients, tailored according to the level of DME and history of DME treatment in a particular eye.
Authors/Members Writing Committee and Investigators: A complete list of the Diabetic Retinopathy Clinical Research Network appears in the Supplement.
Submitted for Publication: July 19, 2012; final revision received November 7, 2012; accepted November 27, 2012.
Corresponding Author: Talat Almukhtar, MBChB, Jaeb Center for Health Research, 15310 Amberly Dr, Ste 350, Tampa, FL 33647 (email@example.com).
Published Online: April 18, 2013. doi:10.1001/jamaophthalmol.2013.2313.
Conflict of Interest Disclosures: A complete list of all DRCR.net investigator financial disclosures is available at drcr.net.
Funding/Support: This work was supported through cooperative agreements EY14231 and EY018817 from the National Eye Institute and the National Institute of Diabetes and Digestive and Kidney Diseases, National Institutes of Health, US Department of Health and Human Services.
Role of the Sponsor: The National Institutes of Health participated in oversight of the conduct of the study and review of the manuscript but not directly in the design or conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation or approval of the manuscript.