Presently, both triamcinolone acetonide and dexamethasone are widely used in the treatment of diabetic macular edema refractory to anti-VEGF treatment with or without focal or grid macular photocoagulation. Triamcinolone is administered as an injection approximately every three months, while dexamethasone is an implant advertised to last six months. The zero-order kinetics of dexamethasone implant theoretically improves its IOP profile in contrast to large-bolus triamcinolone injections. Both treatments have been assessed separately to measure efficacy with regard to diabetic macular edema [2, 3]. They have also been directly, prospectively compared in a comparative effectiveness trial for the treatment of uveitic macular edema and retrospectively for retinal vein occlusions [4, 5]. Data comparing the two for diabetic macular edema are surprisingly very limited. This retrospective study sought to compare the efficacy and adverse effects of triamcinolone with the dexamethasone implant for treating DME.
The main outcome measure for this study was change in central macular thickness at one month. While a large amount of additional and still informative data was collected, this outcome was chosen because it represents an important clinical outcome that was free of any confounding by combination treatments. The results clearly demonstrate a superior effect of IVTA over dexamethasone in treating DME at month one, with a change in CMT of -143 µm versus − 39.3 µm (p = 2.5 x10− 4). This difference is not only highly statistically significant, but is clinically significant as well. CMT changes beyond month one cannot be reliably attributed to the effects of these two corticosteroids because combination treatment was allowed at the discretion of the treating physician. Thus, overall CMT trends across the six months are difficult to interpret. However, one important observation in the dexamethasone group is that there is an inflection point of mean CMT at month three. It falls for the first three months and then begins to rise thereafter. Again, one cannot definitively conclude that the dexamethasone implant’s peak effect is three months because some of these patients received anti-VEGF from month one onwards, but it suggests that the agent’s efficacy declines after month three. These results are consistent with the anatomical data in the Ozurdex PLACID Study comparing the dexamethasone implant plus macular laser photocoagulation to sham treatment plus laser [6]. Peak effect was seen at approximately month one, and there was a steady increase in CMT thereafter. To the authors’ knowledge, anatomical data between injections during the three-year Ozurdex MEAD Study were never published [3, 7]. Interestingly, the POINT Trial comparing the dexamethasone implant to IVTA or periocular triamcinolone for uveitic macular edema, a different indication for corticosteroids but similarly an exudative maculopathy, did not find greater size or duration of effect for the dexamethasone implant over triamcinolone [4].
Visual acuity showed no significant improvement within either group. Both groups contained a mix of phakic and pseudophakic patients. This result is not particularly surprising given the well documented experience with each of these agents in mixed phakic and pseudophakic populations. The Diabetic Retinopathy Clinical Research Network’s Protocol I study, comparing ranibizumab with prompt or deferred laser, triamcinolone with laser, or sham plus laser found no visual acuity benefit at two years with IVTA [8]. In a later DRCR trial, adding dexamethasone to regular anti-VEGF therapy similarly did not improve visual acuity over ranibizumab alone [9].
To assess the potential impact of lens status on visual acuity, visual acuity data was compared for phakic and pseudophakic patients, both within each treatment group and overall. From month zero to month one, there were statistically significant differences favoring improved visual acuity in both pseudophakic triamcinolone patients (p = 0.01). From month zero to month three, dexamethasone patients demonstrated the same trend (p = 0.044). Overall, pseudophakes had better visual outcomes. Although we did not specifically collect data on lens grading in the phakic group, a notoriously subjective assessment, we strongly suspect this was due to corticosteroid-induced cataract formation in these patients, effectively nullifying the expected visual gains from improved DME.
Because all efficacy data after month one are of course confounded by the use of combined anti-VEGF therapy, we sought alternate methods to assess the effect of these two corticosteroids on treatment burden for DME after month one in this real-world setting. In other words, to what extent does use of either of these drugs impact the need for future injections? To answer this question, we created a survival curve of time to next treatment of any kind (anti-VEGF or corticosteroid) and compared the two curves in general by a log-rank (Cox-Mantel) test and in particular for median time to next treatment. The mean time to next treatment was 19 weeks for IVTA and 20.3 weeks for dex. Similarly, the log-rank test showed no difference between the two curves. When only corticosteroid treatments were taken into account, there were on average 1.45 corticosteroid treatments administered within six months when patients were treated with triamcinolone, and 1.38 corticosteroid treatments administered within six months when patients were treated with dexamethasone.
With regard to corticosteroid response, both groups experienced modest increases in IOP, but with a trend towards more significant elevation in the triamcinolone group. Of course, post-injection IOPs were influenced by concurrent ocular hypotensive treatment. 16.7% of IVTA group patients were on IOP medications prior to treatment, which increased to 38.9% after one triamcinolone injection. In comparison, there were no patients in the dexamethasone treatment group receiving IOP-lowering medications prior to treatment, and only 6.6% were prescribed IOP-lowering medications after one implant. Considering the statistically and clinically significantly greater fraction of patients requiring ocular hypotensive treatments post-injection, it is possible that adverse IOP effects induced by triamcinolone may be greater than those of the slow-releasing dexamethasone implant.
This study was limited by several factors. First, it suffers from the usual weaknesses of retrospective studies. The groups were not randomized, and thus there may be imbalances between them. Indeed, the IVTA group had slightly greater CMT at baseline, which increases this group’s ability to improve CMT, the so-called ceiling or floor effect. Similarly, they differed in number of prior anti-VEGF treatments, 3.9 in the IVTA group and 6.53 in the dexamethasone group. However, the implications of this difference are unclear. On one hand, it may suggest DME that is more resistant to treatment. On the other hand, a suboptimal response to eight anti-VEGF injections should not necessarily preclude response to intravitreal corticosteroids because of their different mechanism of action, which is precisely why they remain a second-line treatment option. Moreover, physicians treated their patients per clinical judgment, and thus times to next treatment reflect the treating physicians’ degree of aggressiveness and not simply the efficacy of these different agents. Last, as already mentioned, efficacy data after month one were confounded by combination treatment with anti-VEGF in most cases, and IOP data by the use of ocular hypotensive treatments. However, analyzing several metrics like use of ocular hypotensive agents in each group and time to next treatment “survival” curves allow us to make meaningful if limited real-world conclusions about comparative IOP profiles and treatment burdens with these agents.
Last, we concede on two points. We realize that the retrospective design of our study allowed for there to be less chronic cases in one group over the other. This is something that we tried to control for by following records back to the first recorded injection at our facility, assumedly documenting all anti-VEGFs used as well as the first corticosteroid administered. Many of these were 2012 or prior, so paper records were used at that time. It is possible that we did not catch the first injection for everyone, but we strived to ensure that we did.
Secondly, 29 patients (35 eyes) meeting inclusion and exclusion criteria do not constitute a large group. One example of this is that there was only one treatment-naïve patient per group, which prevented comparison of treatment-naïve to non-naïve. While certain limited conclusions can be reached, other questions can only be answered with larger numbers and in a prospective fashion. For instance, determining the comparative duration of effect of the two agents can only be determined in a randomized, head-to-head trial without the use of combination therapy.
In summary, intravitreal triamcinolone and the intravitreal dexamethasone implant both show efficacy at improving DME and reduce treatment burden similarly, but both had little effect on visual acuity. Triamcinolone had superior anatomical efficacy at one month, but possibly at the expense of a worse IOP profile. While this study could not determine which drug has the greater duration of effect, the dexamethasone implant showed waning efficacy after three months. Additionally, pseudophakic patients receiving either treatment demonstrate better outcomes with regard to visual acuity measures.