Drs Dan and Mihai Călugăru make several comments regarding our study, to which we provide the following responses.

First, they state that there were no data regarding anatomic types of diabetic macular edema (DME) or presence of vitreomacular traction. All subjects had intraretinal fluid, most with large cystoid spaces, and 3 had small collections of subretinal fluid as well as cystoid spaces. None of the patients had vitreomacular traction, but 4 had incidental epiretinal membranes that did not contribute to retinal thickening.

Second, they state that there were no comparative data referring to the demographics and baseline ocular characteristics of patients treated with dexamethasone implant and injections of a vascular endothelial growth factor (VEGF)-neutralizing protein. Table 1 of the subject article shows the demographics and baseline characteristics for all subjects; and since all received both treatments, there is no comparison of baseline characteristics.

Third, they question if the 2 groups were comparable, but the comparison is between 2 treatments given to the same patient population, so comparability among groups is not an issue.

Fourth, they question the use of a crossover design. A crossover design was used because it allowed correlation of changes in vasoactive proteins with changes in edema after a dexamethasone implant or a VEGF-neutralizing protein in eyes from the same subjects. We suspect, from the types of questions posed, that the Călugărus may be thinking that the study was comparing efficacy of dexamethasone implant and anti-VEGF agents, which is not the case. The purpose of the study was to correlate changes in intraocular vasoactive proteins with changes in edema after a dexamethasone implant. We also obtained similar correlative data after injection of a VEGF-neutralizing protein as a type of control, because that type of injection can only specifically reduce 1 protein, VEGF. This was informative because 2 proteins, MMP-9 and IGFBP-3, which were reduced after dexamethasone implant, were also reduced after injection of a VEGF antagonist; thus, their reduction cannot be attributed to the direct pharmacologic activity of dexamethasone. Instead, MMP-9 and IGFBP-3 may be produced by edematous retina and reduction in edema by any means may indirectly reduce them. This allowed us to eliminate these 2 proteins from consideration as contributors to DME.

Fifth, they speculate that because many of our subjects had chronic/recurrent DME, they had permanent damage to retinal capillaries with incurable pathology. Some subjects had complete elimination of intraretinal fluid and some had marked improvement, indicating that many of the subjects did not have irreversible edema. Some subjects had little reduction in edema after dexamethasone implant or injection of a VEGF-neutralizing protein and it is possible that these subjects have permanent damage to retinal vessels that prevents a response to any therapeutic, but it is also possible that the duration of treatment was insufficient to see an optimal response or the eyes of these patients may contain pro-permeability factors that are not reduced by a dexamethasone implant.

Sixth, they speculate that pro-permeability factors may differ between patients with chronic/recurrent DME and those with recent-onset DME. It is reasonable to hypothesize that chronic/recurrent edema may cause increased production and/or release of pro-permeability factors (and our findings with regard to MMP-9 and IGFBP-3 support that), but our study did not compare vasoactive proteins in recent-onset vs chronic/recurrent DME and there are no published data that provide insight regarding this issue.

Finally, they suggest that efficacy of any treatment depends primarily on the duration of DME at the time treatment is started. It is clear that delay in treatment onset can have a negative impact on visual outcomes, but studies examining the effect of baseline characteristics on visual outcomes indicate that it is not the only factor influencing prognosis. There are some patients who have a suboptimal response despite timely treatment onset. From a practical standpoint, there are many patients with chronic/recurrent DME who are seen in clinical practice and it is important to identify the vasoactive proteins that contribute and hopefully this will help identify new treatments to improve outcomes in these patients.