Académique Documents
Professionnel Documents
Culture Documents
During the late 1970s to early 1990s, more rapid progres- ported to have less complications, particularly inflamma-
sion of diabetic retinopathy (DR) after cataract surgery was tion7,8 and postoperative astigmatism.
consistently documented when using the 2 most common Whether eyes operated using the phacoemulsification
surgical techniques at that time, intracapsular and extracap- technique still have more rapid progression of DR in pa-
sular cataract extraction.1– 4 The phacoemulsification surgi- tients with diabetes remains to be clarified,9,10 because few
cal technique was developed by Kelman in 1967, inspired small studies that assessed DR progression after phacoemul-
by dental ultrasonic techniques, and popularized in the early sification surgery included nonoperated eyes as con-
1980s. However, it was not widely accepted until 1996, trols.11–13 We aimed to assess this question in a cohort study
when phacoemulsification was used for 97% of cataract of cataract surgical patients with diabetes, including paired
procedures in the United States.5 This new technique rap- controls between operated and nonoperated eyes of the
same patients.
idly replaced the older methods and is currently the most
common surgical procedure for cataract surgery, in view of
the small incisions, reduced procedure time, and minimal
Materials and Methods
damage to ocular structures. The visual outcome after
phacoemulsification does not differ from that after extra- Patients aged 65 years or older undergoing phacoemulsification
capsular extraction.6 Phacoemulsification, however, was re- cataract surgery at Westmead Hospital, Sydney, Australia, from
1510 © 2009 by the American Academy of Ophthalmology ISSN 0161-6420/09/$–see front matter
Published by Elsevier Inc. doi:10.1016/j.ophtha.2009.03.003
Hong et al 䡠 Progression of Diabetic Retinopathy after Phacoemulsification
Table 1. Modified Early Treatment Diabetic Retinopathy Study betes. DR was graded by a single grader (TH) using the modified
Classification System ETDRS classification system, with DR levels ranging from levels
20 to 90 (Table 1). Questions about the lesions graded were
Diabetic Retinopathy Level adjudicated, and all cases exhibiting progression were confirmed
both by a senior grader who assessed images for the field study and
DR absent 10
Questionable 14 and 15
by a retinal specialist (PM).
Minimal NPDR 20 Both the preoperative and the 1-month postoperative visits
Mild NPDR 35 were considered baseline status, given the impaired quality of the
Moderate NPDR 43 preoperative photos in some patients, because of cataract. Side-
Moderately severe NPDR 47 by-side grading of the baseline or 1-month photographs with the
Severe NPDR 53 12-month postoperative visit photographs was performed to assess
Mild PDR 61 DR progression, defined as a worsening by 1 or more levels on the
Moderate PDR 65 ETDRS scale at the postoperative visits. This included progression
Severe PDR 71 from no DR to minimal DR (level 20). Incident DR was defined in
Advanced PDR 81 eyes without DR at baseline (level 10) in which DR developed
End-stage PDR 85 postoperatively and is a subgroup of the cases that progressed. We
Inactive PDR 90 defined patients who had panretinal laser treatment before baseline
Ungradable 99 as DR level 90 (inactive PDR) at baseline. These patients could not
progress any further and were therefore excluded from the analysis
DR ⫽ diabetic retinopathy; NPDR ⫽ nonproliferative diabetic retinopa- for progression of DR (n ⫽ 32).
thy; PDR ⫽ proliferative diabetic retinopathy. SAS (V9.13, SAS Institute, Cary, NC) was used for all analy-
ses. We assessed differences in DR prevalence between operated
and nonoperated eyes before the study baseline, and the incidence
2004 to 2006, were recruited into a cohort study, aiming to assess and progression of DR in operated compared with nonoperated
surgical outcomes, including the risk of age-related macular de- eyes 12 months after surgery. Generalized estimating equation
generation and DR after cataract surgery. The Study was approved models14 were used to estimate odds ratios (ORs) and 95% con-
by the University of Sydney and Sydney West Area Health Service fidence intervals (CIs) after adjusting for age, sex, baseline HbA1c
Human Research Ethics Committees and adhered to the tenets of level, and duration of diabetes.
the Declaration of Helsinki. Written, informed consent was ob-
tained from all study participants. In this report, we included a
subset of this cataract surgical cohort of 1994 patients, 169 of 190 Results
patients with diabetes (type I or II) who were followed for at least
12 months after surgery. All 190 participants with diabetes underwent standard phacoemul-
Participants were examined and had retinal photography per- sification surgery at the time of recruitment or shortly after recruit-
formed at preoperative and 1-, 6-, 12-, and 24-month postoperative ment to the study, including 90 women and 100 men with a mean
visits. Before surgery, participants were interviewed using stan- age of 74.1 (⫾5.7) years. The mean duration since diagnosis of
dardized questionnaires to obtain information on medical and diabetes was 13.2 (⫾10.1) years, and the preoperative average
ocular conditions. Data collected included age, sex, race, and type, HbA1c level was 7.4% (⫾1.4%) at baseline (Table 2). Only 2
duration, and treatment for diabetes. Glycosylated hemoglobin patients were on insulin alone for 30 or more years. Of the 190
(HbA1c) values were collected preoperatively as a measure of patients, 3 had ungradable images at either the baseline or follow-
objective glycemic control. At each visit, a self-reported average up visits, 187 were followed at 6 months, but 18 were lost to
blood glucose level range was also collected. Snellen visual acuity follow-up after this time, leaving 169 followed for at least 12
and intraocular pressure were measured. Digital retinal photo- months postoperatively. Of these 169 participants, 278 eyes were
graphs of both eyes were taken after dilation, using a Canon pseudophakic and 60 eyes remained phakic up to the 12-month
CF-60DSi Digital Fundus Camera (Canon Inc, Tokyo, Japan). visit. Of these 60 phakic eyes, 45 were at risk of DR progression,
Photographs including Early Treatment Diabetic Retinopathy because 15 eyes had an ETDRS level of 90 (inactive PDR) at
Study (ETDRS) standard field 1 (optic disc), field 2 (macula), field baseline. There were a total of 32 eyes with ETDRS level 90
4 (superotemporal), field 5 (inferotemporal), and a nasal field, at baseline; 23 of these eyes were the study eyes of interest (to
including the optic disc, were taken on all participants with dia- undergo surgery at the study baseline). Apart from capsular tears,
1511
Ophthalmology Volume 116, Number 8, August 2009
1512
Hong et al 䡠 Progression of Diabetic Retinopathy after Phacoemulsification
whether subsequent surgery was performed on some so- active DR features, particularly signs of macular edema,
called nonoperated eyes) showed findings similar to those even when not yet clinically significant, and to reduce the
reported previously (32.0% in operated and 26.5% in non- risk of progression to proliferative DR or iris neovascular-
operated eyes; OR 1.34; 95% CI, 0.80 –2.25). ization in patients with significant DR8 undergoing cataract
The advantage of our study over these previous studies is surgery. The number of patients who reported previous laser
its relatively large sample size (n ⫽ 169 patients), providing treatment in our study sample was small (n ⫽ 32). Eyes that
information for comparison between any operated versus had previous panretinal laser treatment were no longer
nonoperated eyes. We also compared operated eyes with considered at risk of DR progression or development and
those that remained nonoperated for the entire 12-month were thus excluded from the analyses in this report.
follow-up period and conducted a paired-eye comparison in When comparing the reported DR progression rates after
45 patients who had unilateral surgery for at least 12 months phacoemulsification surgery with the rates after extracapsu-
prior; the only difference between the 2 eyes was that one lar cataract extraction surgery, the magnitude of the differ-
had undergone surgery and one had not. We therefore ences between operated and nonoperated eyes appears to
consider our findings for DR progression from the paired- have been reduced substantially. Jaffe et al2 reported 37%
eye comparison of the same subjects less likely to be biased and Pollack et al20 reported 38.2% progression rates 12 to
than findings from comparison of any operated versus non- 18 months after surgery, compared with progression rates of
operated eyes at baseline. The latter comparison may have approximately 30% or less after phacoemulsification sur-
incorporated differences between study subjects in the risk gery reported by recent studies,2,20 including ours.
estimates. In addition, using eyes that were nonoperated at
baseline but had had surgery during the follow-up period as
controls may have led to misclassification of operated and Study Limitations
nonoperated eyes, resulting in a bias toward the null.
Our study sample has characteristics similar to those in Limitations of our study should be noted. First, the assess-
other reported study samples. The mean ages at cataract ment of DR in nonoperated eyes may have been hampered
surgery were similar across all the studies.11–13 Mean by lens opacity, so that we may have missed recognizing
HbA1c levels at preoperative visits were also similar across mild DR lesions in nonoperated eyes. However, eyes that
these studies, including ours. Some previous studies found remained unoperated for 12 months would have been
that poor diabetic control contributed to DR progression unlikely to have advanced cataract. For eyes with mod-
after surgery.17,18 Paired-eye comparisons of the same pa- erate cataract, although a few microaneurysms may have
tients who had unilateral surgery eliminates the confound- been undetected, this would not be as likely for retinal
ing effect from poor diabetic control and is therefore the hemorrhages.
most efficient way to investigate the effect of surgery on DR Second, the fellow eyes may not be entirely comparable.
incidence and progression. Although the decision for surgery was not made randomly,
In keeping with our study findings, other studies have factors including the duration of diabetes and glycemic,
reported an increased risk of DR progression after phaco- blood pressure, and blood lipid control would be likely to
emulsification surgery in patients with more advanced pre- affect both eyes similarly. So in this regard, we have made
operative stages of DR.9,10,19 The blood–aqueous barrier the best and most feasible comparison. It must be acknowl-
has been found to be impaired more severely in patients edged, however, that (1) local factors can result in asym-
with proliferative DR than in patients without diabetes or in metry of DR between the 2 eyes; and (2) decisions about
those with nonproliferative DR, as indicated by substan- which eye is operated may reflect differences in DR
tially higher levels of aqueous flare after surgery in eyes levels between the 2 eyes, of which we have no specific
with proliferative DR.10 We had few patients with active knowledge.
proliferative DR in our sample and so were not able to Third, eyes that underwent cataract surgery may already
assess whether the detrimental effect of surgery was greater be at a greater risk of DR progression than nonsurgical
in patients with proliferative DR. Given the previously fellow eyes, regardless of their surgical status. The devel-
documented rapid DR progression after cataract surgery opment of cataract in people with diabetes is strongly re-
using older surgery techniques, there is coherence in the lated to poor glycemic control, which can also accelerate
evidence relating to DR progression after cataract surgery. DR. Although a randomized controlled trial is an ideal study
One possibility that this study cannot easily address is design for this study question, it would not be feasible or
the concern that both DR and cataract are complications of ethical. A longitudinal observational study is a feasible
diabetes, and therefore the presence of cataract may be a design with evidence quality close to that from random-
marker for greater severity or increased likelihood of pro- ized controlled trials, and paired-eye comparison of the
gression of DR. Our study findings might, in part, reflect same patients is the most comparable analysis that we can
this. A second concern is that there might be an “early conduct.
worsening” effect during the first 12 months, which may We did not collect renal function data in our study
potentially reflect efforts to tighten blood glucose control sample or complete surgical data on the types of intraoc-
before or after the surgery. Our paired-eye comparison of 45 ular lens, width of the incision, and skill level of the
subjects does account, at least partly, for both possibilities. surgeons. Although these factors are unlikely to have had
It is now well-accepted clinical practice that preoperative a major effect on DR progression, they may have had
laser treatment should be considered to adequately control minor influences.
1513
Ophthalmology Volume 116, Number 8, August 2009
1514