Diabetic Macular Edema at the time of Cataract Surgery trial: a prospective, randomized clinical trial of intravitreous bevacizumab versus triamcinolone in patients with diabetic macular oedema at the time of cataract surgery – preliminary 6 month results

Introduction

Diabetic retinopathy (DR) is the leading cause of legal blindness in the working age population in developed countries such as Australia.1 As the prevalence of diabetes increases, the number of people with sight-threatening DR is estimated to triple in the United States by 2050, from 1.2 million in 2005 to 3.4 million, with similar predictions for Australia.2 Diabetic macular oedema (DME) is the most common cause of visual loss in DR.3, 4

Cataract also occurs more frequently in people with diabetes, and at an earlier age.5, 6 Although the removal of the cataract is usually a quick and effective means of restoring vision, the outcomes of cataract surgery in people with DR are worse compared to people without diabetes due to the associated development or acceleration of DME post-operatively.7-11 Indeed, patients with co-existent DME at the time of cataract surgery are most at risk of a poor outcome, with studies showing that only 53% of these patients achieve a post-operative visual acuity of ≥6/12, in comparison to >95% in those without DR.9, 12, 13

Recent large randomized controlled trials have shown that anti-vascular endothelial growth factor (anti-VEGF) agents such as ranibizumab,14, 15 aflibercept16, 17 or bevacizumab (BVB)18 given as intravitreous injections are an effective treatment for DME with better visual results than the previous standard treatment of laser photocoagulation10, 11, 17. However, recent studies have indicated that these agents may not be as effective for DME aggravated or induced by cataract surgery.19-21

One of the key mechanisms in the development of DME is the breakdown of the blood–retinal barrier, resulting in the egress of fluid into the retina (macula).22 Cataract surgery is an inflammatory insult to the eye,23 resulting in the release of soluble inflammatory mediators that lead to further breakdown of this already compromised blood retinal barrier, and could be the key factor in the causation, or worsening, of DME postoperatively.

The use of intravitreous depot-steroid injections such as triamcinolone acetonide (TA), at the time of surgery, may therefore be a better option for treatment in this group, given that TA has both anti-VEGF and anti-inflammatory effects.24

Despite several studies comparing the use of intravitreous injections of anti-VEGF agents and TA in the treatment of laser-resistant DME,25-28 none have directly compared these two agents in the setting of cataract surgery in an eye with pre-existing DME.

This paper reports the study methodology and 6-month preliminary follow-up data of the Diabetic Macular Edema at the time of Cataract Surgery (DiMECat) study, a prospective, single-masked, randomized clinical trial comparing the outcomes of cataract surgery in patients with current or past DME who receive either intravitreous bevacizumab (BVB, Avastin, Genentech Inc., San Francisco, CA, USA; 1.25 mg) or triamcinolone (TA, Triesence; Alcon Pharmaceuticals, Ft. Worth, TX; 4 mg) at the time of surgery.

Methods

Statistical methods

Data from all participants who completed at least 6 months of follow-up has been included in this intention-to-treat analysis. Two subjects were lost-to-follow-up (one from each treatment group) but are included in these analyses as described below. All analyses were undertaken using Stata IC 12.1 for Windows (StataCorp LP, College Station, TX).

Continuous baseline characteristic variables were summarized as means with 95% confidence intervals (95%CI), or as medians (interquartile range, IQR) if they did not appear to be normally distributed. The variables were then compared across treatment groups using two-sample t-test (with or without unequal variance as needed), and the two-sample Wilcoxon rank-sum (Mann–Whitney) test, respectively. Categorical variables were summarized as percentages and compared across treatment groups using Fisher's exact test.

Both treatments were expected to have an initial rapid effect that then levels off, so a per-eye summary mean of all post-baseline measurements was generated.30 The per-person summary mean was calculated as the sum of all post-baseline measurements divided by the number of post-baseline measurements, thus:

As such, subjects with missing data (either intermittent or sustained through loss-to-follow-up) were included in the calculations of post-baseline means and post-baseline change variables by providing the average of all available data divided by the number of available data points.

A linear regression model was then constructed with this summary mean and the baseline measurement and treatment group. The Stata margins command was utilized to test for a significant difference between treatment groups and to calculate the adjusted mean of summary means (adjusted for baseline measurement). The interaction term between treatment and baseline measurement was included in the model if it was found to be significant. The interaction term was significant for CMT (P = 0.007) but not for VA (P > 0.05).

The change from baseline was also calculated as measurement at time less baseline measurement. These change variables were then summarized into per-person means as above. The regression models for mean change included only the mean change and treatment group.

The need for re-treatment outcome was assessed up to but not including the 6-month visit to account for the fact that subjects in the pilot phase of this trial were exited at 6 months and, thus, could not have been retreated even if clinically indicated. A sensitivity analysis was undertaken whereby subjects with missing data for retreatment up to 6 months were:

1. Excluded from analysis.
2. Included and all missing values were assumed to be ‘retreated’ to a maximum of two retreatments for TA group and to a maximum of five for BVB group (the maximum that each drug's protocol would allow).
3. Included and all missing values were assumed to be ‘no treatment’.

The per-subject number retreatments were summarized as median (IQR) or percentages and compared between treatment groups using to a two-sample Wilcoxon rank-sum (Mann–Whitney) test and a Fisher's exact test, respectively.

Adverse events were defined as any IOP measurement greater than 21, 25 and 30 mmHg at any visit over the 6 months. Subjects with missing values for IOP were included only for those visits where there IOP data was recorded; in effect, a last-observation-carried-forward approach. The number of subjects who had one or more instances of raised IOP were compared between treatment groups using the Fisher's exact test.

Results

Seventy patients were enrolled between June 2012 and December 2014. Eighteen patients are yet to reach 6 months follow-up, and 11 patients were excluded due to concurrent medical illnesses or intraoperative complications that necessitated withdrawal from the trial (Fig. 1). Results presented here are for 41 patients enrolled prior to April 2014 who were randomized, treated and eligible to have reached 6 month follow-up. There were no statistically significant differences between the treatment groups for any of the demographic or baseline characteristics (Table 2). The mean age of all patients was 66.4 (95% CI 63.7 – 69.1), 73.2% were male and the majority of patients had Type II diabetes requiring insulin (65.0%). Most participants had mild to moderate DR (53.7%), with 17.1% having severe DR and 29.3% treated PDR. Most participants (82.9%) had macular oedema present at baseline. The average BCVA was 51.4 (95% CI 46.5 – 56.2) letters, a Snellen equivalent of approximately 6/30. Twenty-two eyes (53.7%; 10 in BVB group, 12 in TA group) had received at least one macular laser treatment prior to study enrolment,12 eyes had received prior PRP (29.3%; 6 in BVB group, 6 in TA group) and five (12.2%) having prior injection treatment (range one to five). Two participants had both eyes enrolled in the study.

Following the initial intravitreous injection given at the time of cataract surgery, 70.6% of participants in the BVB group required further injections, compared to 16.7% in the TA group (P < 0.0001; refer Fig. 2). The means for number of retreatment injections were 1.71 and 0.17, respectively. No participants met the treatment failure guidelines or required macular laser during follow-up.

Over time, vision improved significantly from baseline in both treatment groups (mean +21.4 (95% CI [14.9, 27.9]) letters in TA group P < 0.0001, +12.5 (95%CI[4.8, 20.2]) letters in BVB, P = 0.002), with the overall mean Snellen equivalent VA at six months improving to between 6/12 and 6/15. However, the mean of all post-baseline letter score measurements did not differ significantly between the groups (P = 0.085; refer Fig. 3).

At baseline, one subject in each treatment group had vision in their study eye which would allow them to drive (≥69 letters, 6/12); they both retained this VA at 6 months. At 6 months, 60.9% of eyes receiving TA achieved driving vision, compared to 73.3% in the BVB group (P = 0.501).

The proportion of patients who achieved a visual acuity improvement of 15 letters or more at six months was 69.6% in the TA group and 60.0% in the BVB group (P = 0.728). The proportions for a 10-letter improvement were 82.6% and 73.3%, respectively (P = 0.687).

There was a significant difference in mean post-baseline change in CMT between the two groups over six months. The BVB group experienced a mean post-baseline increase of +37.3 microns (95% CI [−6.6, +81.3]), while the TA group experienced a mean post-baseline decrease of −43.8 microns (95% CI [−6.8, −80.8]), P = 0.006 (refer Fig. 4).

Discussion

With the projected increasing prevalence of diabetes, the co-existence of cataract and DR is set to increase over the next few decades. Our preliminary results have demonstrated that either intravitreous TA or BVB administered at the time of cataract surgery resulted in improved visual acuity at six months, with a trend towards greater improvement in the TA group. The TA group also experienced a significantly better mean CMT than the BVB group with a reduction in CMT (−43.8 µm) while the BVB group experienced slight thickening from baseline (+37.3 µm).

In contrast, prior studies have shown that visual outcomes after cataract surgery in patients with co-existing DME are poor, with one recent large study finding that only 53% of this group achieved a post-operative BCVA of 6/12 or better, with more than half either having no meaningful improvement in vision, or worse, loss of vision.9 This is in stark contrast to the visual results seen in individuals undergoing cataract surgery who do not have diabetes, where greater than 95% achieve a BCVA of 6/12 or better.12, 13

The development, or worsening, of DME after cataract surgery is the major cause for the observed poorer visual outcomes post cataract surgery.10, 31-33 Despite recent advances in the treatment of DME with anti-VEGF agents,17, 26, 34 the efficacy of these agents with treating or preventing post-operative DME are more mixed. Although some studies in patients with DR but no DME at the time of surgery have shown a short term benefit of a reduced CMT and BCVA at one to three months post-surgery in comparison to controls with no adjunctive treatment, this benefit was then largely lost at six months post operatively.20, 21, 35

In those with DME at the time of surgery, the results are even worse. In a prospective, non-comparative study of standard of care of individuals with DME undergoing cataract surgery, Bressler and colleagues found only 32% of patients had a significant improvement in vision (at least four lines), with 10% losing at least two lines of vision.9 This study also highlighted the lack of a consensus of how best to manage this group of patients, as only 7% of study eyes received intraoperative treatment, while 35% of all eyes did not receive any treatment for their DME throughout the entire study period. This variability in treatment may also account for the markedly different results achieved in comparison with our current study.

Outcomes with the use of anti-VEGF agents are also variable. Although one prospective study found a benefit with BVB treatment with reduced CMT and improved BCVA in comparison to controls at up to three months post operatively,36 two further prospective studies found that concurrent treatment with BVB or ranibizumab only prevented the worsening of DME post operatively in comparison to controls, with the actual CMT largely remaining unchanged from baseline, indicating that the anti-VEGF treatment failed to reduce the severity of the pre-existing DME post operatively.19, 37 Our pilot data supports these findings, as the CMT in our BVB group tended to increase slightly over the 6 month post-operative period, rather than improve, despite multiple re-treatments.

As increased inflammation induced by cataract surgery has been postulated to be the cause of DR and DME progression, the use of intravitreous TA (IVTA) at the time of cataract surgery has been advocated as potentially the best treatment for those with high risk DR or pre-existing DME undergoing surgery.38

The mechanism of action of IVTA in DR is thought to be due to its ability to both inhibit the actions of VEGF and to its anti-inflammatory effects.24 It has been found to reduce the breakdown of the blood-retinal barrier,39 inhibit the production of pro-inflammatory prostaglandins and also down regulate the production of VEGF.39-42 This theory is further supported by preliminary studies that have shown a beneficial effect on DME with a single ocular steroid injection administered in patients with diabetes undergoing cataract surgery.40, 43, 44

Our preliminary 6-month results support this hypothesis, as only the TA group maintained a sustained reduction in CMT in the 6-month post-operative period. Although not statistically significant, we also demonstrated a trend towards a greater improvement in BCVA post-surgery in the TA group in comparison to the BVB treated group.

Various adverse events have been reported with IVTA injections, with the most significant being endophthalmitis and raised intraocular pressure (IOP).45 Although complications such as endophthalmitis may occur with any intraocular procedure, raised IOP, which has been reported to occur in up to 23.5%,38, 40 is unique to TA injections and is not seen with BVB. Thus far, only 12.5% of our TA group have experienced an increased IOP rise after treatment – however time will tell as to whether this will increase with repeated treatments over an extended duration of follow up.

It is clear that the coexistence of visually significant cataract and DME will become increasingly common. Given our trial design, and the high proportion of patients who had DME at the time of surgery (89%) our results relate best to patients with concurrent DME undergoing cataract surgery. A significant proportion of our population were treatment naïve, and therefore these results may not be generalizable to patients who undergo intravitreous treatment in the lead up to surgery. Although such a treatment approach may also result in better outcomes, this approach has not yet been formally tested in a trial and may be logistically more difficult in a large, public hospital setting. A potential future study could incorporate this pre-operative treatment approach compared with, or in combination with, intra-operative treatment.

In conclusion, our preliminary findings have shown that treatment with either BVB or TA will result in a significant improvement in BCVA post cataract surgery. However, only the TA group maintained a sustained reduction in CMT through to 6 months, whereas the CMT tended to increase in the BVB group despite repeated postoperative injections. Despite these differences in CMT, no significant difference in BCVA or number of letters gained was found; however, longer term outcomes in a larger number of patients, as we intend to do over the next 6 months, will determine whether the differences in CMT will translate into better BCVA gains in the TA group.