Methods

A retrospective cohort study was performed. A chart review was performed at an urban teaching hospital of all patients from January 2008 through May 2013 with active wet AMD who had undergone cataract surgery. The selected years represent the timeframe in which an electronic medical record was available at the institution. Active wet AMD was defined as eyes that received at least 1 anti–vascular endothelial growth factor (anti-VEGF) injection in the 1-year study window. The eyes that met the criteria formed the surgical cohort. A nonsurgical control group was formed by reviewing the charts of patients in the same health system who were receiving anti-VEGF injections for active wet AMD in the year 2012. Eyes were included if they were phakic and had not undergone cataract surgery in that year. The year 2012 was selected as it represented a time period where the majority of OCT images in the study setting were obtained using a Stratus OCT. Non-Stratus OCTs were not included in quantitative analyses. The “midpoint” of the study was defined as the date of cataract extraction in the surgical group and July 1, 2012 in the nonsurgical group.

The cases were identified by reviewing the medical records of the practices of 5 retina specialists. The patients were predominately receiving anti-VEGF injections using a treat-and-extend protocol, which has been described in detail elsewhere. In brief, eyes were typically treated with an induction phase of 3 consecutive monthly injections and extended based on OCT findings. The study was prospectively approved by the Henry Ford Hospital Institutional Review Board and is in accordance with the principles outlined in the Declaration of Helsinki. Handling of patient data was in compliance with Health Insurance Portability and Accountability Act (HIPAA) guidelines.

Patients with both occult and classic wet AMD were included. Patients were excluded if they had concomitant diagnoses that could contribute to neovascularization, such as proliferative diabetic retinopathy or retinal vascular occlusion. Eyes with significant ocular pathology such as advanced glaucoma or retinal detachment that may limit visual acuity were also excluded. Eyes with complicated cataract extractions were noted and separate analyses were performed both including and excluding these eyes. Data collected from medical records included demographic characteristics and ocular history. The number of anti-VEGF injections, type of anti-VEGF injection administered, and changes in anti-VEGF medication administered were recorded in the 6 months before and after the midpoint. Best-corrected visual acuity (BCVA) was recorded 3 months before and after the midpoint. Central OCT thickness was recorded only in Stratus OCTs in the 3 months before and after the midpoint. Presence of cysts and subretinal fluid on OCT was recorded in the 3 months before the midpoint. Only worsening or new cysts and subretinal fluid were recorded in the 3 months after the midpoint. Qualitative OCT findings were recorded in all OCT types, including Stratus, Cirrus, and Spectralis. The presence of macular hemorrhage on examination or fundus photography in the 3 months before and after the midpoint was noted. The presence of subfoveal atrophy was noted on OCT, examination, or fundus photography 3 months before, 3 months after, and 6 months after the midpoint.

The data were collected and entered into a computerized database (Excel 2013; Microsoft Corporation, Redmond, Washington, USA). All eligible patients in the surgical group were included in the database. Eyes eligible for the control group were far more prevalent, so a representative sample was taken. The list of eligible control eyes was randomized using the “randomize list” function in Excel. Eyes were then included consecutively from the list until a group of comparable size was formed.

For statistical analysis, SAS software version 9.2 (SAS Institute, Cary, North Carolina, USA) was used. Snellen BCVA was converted to equivalent logMAR values. The group comparisons of the numeric data were performed using the Wilcoxon rank sum test if the distributional normality assumption was violated, otherwise using 2-sample t tests. The group comparisons of the categorical data were performed using the Fisher exact test in the presence of sparse data, otherwise using the χ ² test. A P value less than .05 was considered statistically significant.

Results

Forty of 40 eligible eyes were included to form the surgical group (39 patients), and 42 of 179 eligible eyes were included by computer-generated randomization to form the nonsurgical group (38 patients). The mean age in the surgical and nonsurgical groups was 80.8 ± 6.5 years and 79.2 ± 9.1 years, respectively. The majority of patients were white, representing 34 patients (85%) in the surgical group and 28 patients (67%) in the nonsurgical group.

The BCVA at 3 months prior to the midpoint was 0.65 ± 0.64 logMAR (20/89 Snellen acuity) in the surgical group and 0.65 ± 0.90 logMAR (20/89 Snellen acuity) in the nonsurgical group ( P = .097). The mean change in BCVA was statistically significant between the 2 groups 3 months after the midpoint, with surgical eyes improving 0.23 ± 0.65 logMAR (with improvement to 20/53 Snellen acuity) and nonsurgical eyes improving 0.11 ± 0.59 logMAR (with improvement to 20/69 Snellen acuity, P = .049). Table 1 shows the distribution of eyes based on visual acuity 3 months before and after cataract surgery. Table 2 shows the distribution of eyes based on visual acuity 3 months before and after the midpoint in the control group.

There was no significant difference in the number of injections pre- and post-surgery in the surgical group, with eyes receiving 2.31 ± 1.40 injections before and 2.30 ± 1.45 injections after the midpoint ( P = .921). Nonsurgical eyes received 3.00 ± 1.45 injections before and 2.57 ± 1.45 injections after the midpoint, and this decrease approached statistical significance ( P = .057). There was no statistically significant difference in the change in the number of injections received by the surgical (−0.03 ± 1.60) and nonsurgical (−0.43 ± 1.42) groups before and after the midpoint ( P = .233).

In the preoperative period of the cataract extraction group, 28 eyes (70%) were receiving predominantly bevacizumab and 12 eyes (30%) ranibizumab. Before the midpoint of the nonsurgical group, 33 eyes (79%) were receiving predominantly bevacizumab and 9 eyes (21%) ranibizumab. There was no significant difference in the proportion of eyes that switched the type of anti-VEGF medication after the midpoint of the study (7 eyes [16.7%] in the surgical group vs 7 eyes [17.5%] in the nonsurgical group, P > .999).

Analysis of OCT imaging showed no significant difference in starting central thicknesses between the surgical and control groups (228.8 ± 60.9 μm vs 223.1 ± 60.6 μm respectively, P = .649). A significant difference was found in the final central thickness, with surgical eyes measuring 265.4 ± 98.4 μm and nonsurgical eyes measuring 216.4 ± 58.3 μm ( P = .011). Surgical eyes had significantly more new or worse cysts after the midpoint, with worsening noted in 13 surgical eyes (54.2%) vs 9 nonsurgical eyes (28.1%) ( P = .048). There was no significant difference in subretinal fluid between the 2 groups, with worsening subretinal fluid in 7 surgical eyes (30.4%) and 4 nonsurgical eyes (12.5%) ( P = .171). There was also no significant difference in the development of subfoveal atrophy at 3 and 6 months ( P = .187 and P = .422, respectively) or incidence of new macular hemorrhage between the 2 groups ( P = .432).

A subanalysis was performed comparing surgical eyes to control eyes while excluding surgical cases that had complicated cataract extraction. Reasons for exclusion included anterior vitrectomy, posterior capsular tear, retained lens fragments, or intraoperative floppy iris requiring the use of a Malyugin ring (MicroSurgical Technology, Redmond, Washington, USA). In total, 6 eyes were excluded from the subanalysis. The results again showed no significant difference in the change in the number of injections before and after the midpoint (−0.43 ± 1.42 for the nonsurgical group and −0.03 ± 1.63 for the surgical group, P = .262). A significant difference was again found in mean OCT thickness in the second half of the study (216.4 ± 58.3 μm in nonsurgical eyes and 263.0 ± 101.0 μm in surgical eyes, P = .019) and the presence of worsening cysts (9 nonsurgical eyes [28.1%] vs 12 surgical eyes [60.0%], P = .023). There was no significant difference in the development of subfoveal atrophy, worsening subretinal fluid, worse or new hemorrhage, or change in the administered anti-VEGF medication between the surgical and nonsurgical eyes.

An analysis of statistical power was performed using a sample size of 82 (40 surgical eyes, 42 control eyes) along with a 2-sided alpha level of 0.05. A 2-sample t test had a power of 0.80 to detect an effect size of about 0.63. This corresponded to the ability to detect an underlying mean group difference of at least 0.39 logMAR for the pre- to postsurgery change in BCVA and at least 0.95 for the pre- to postsurgery change in number of injections. Furthermore, using the sample size along with a 2-sided alpha level of 0.05, the χ ² test had a power of 0.80 to detect a mean rate difference between the 2 groups of about 8% vs 32% for a characteristic that was present in 20% of the total study sample.