Intravitreal triamcinolone acetonide versus combined intravitreal bevacizumab and dexamethasone in diffuse diabetic macular oedema


Dr Saumil Sheth, 2/559, Crown Street, Surry Hills, Sydney CBD, NSW 2000, Australia. Email:


Background:  To compare the efficacy of a single injection of combined intravitreal dexamethasone and bevacizumab (Avastin) with that of intravitreal triamcinolone acetonide in eyes with diffuse cystoid diabetic macular oedema.

Design:  Prospective, non-randomized, masked, interventional case series.

Participants:  Twenty-four eyes of 24 subjects with centre-involved diabetic macular oedema extending over two disc-areas with predominant cystic changes on spectral domain optical coherence tomography were selected.

Methods:  Ten phakic and two pseudophakic, ocular hypertensive eyes received intravitreal dexamethasone and bevacizumab as against 12 pseudophakic, normotensive eyes that received intravitreal triamcinolone acetonide.

Main Outcome Measures:  Change in central macular volume on spectral domain optical coherence tomography and best-corrected visual acuity were measured at 6-week follow-up.

Results:  Baseline data were matched in both groups. Post-injection central macular volume (7.46 ± 0.73 mm3) was significantly lower (P < 0.001) in the intravitreal triamcinolone acetonide group when compared with its pre-injection central macular volume (9.11 ± 1.0 mm3) or when compared with the post-injection central macular volume (P = 0.02) of the intravitreal dexamethasone and bevacizumab group (8.42 ± 1.18 mm3). However, post-injection best-corrected visual acuity between the intravitreal triamcinolone acetonide (0.65 ± 0.15 logMAR) and the intravitreal dexamethasone and bevacizumab groups (0.685 ± 0.15 logMAR) was not significantly different (P = 0.06) at 6 weeks. No significant correlation was noted between change in central macular volume and change in best-corrected visual acuity (r = 0.35, P = 0.07) from the pooled data of both the groups. A fair correlation was noted between change in central macular volume and pre-injection central macular volume (r = 0.55, P = 0.005).

Conclusions:  Intravitreal triamcinolone acetonide may be more effective than intravitreal dexamethasone and bevacizumab in reducing macular volume in patients with diffuse cystoid diabetic macular oedema. A significant reduction in macular volume does not necessarily translate into a correspondingly significant improvement in best-corrected visual acuity.


Since the results of the early Treatment of Diabetic Retinopathy Study (ETDRS) on focal/grid laser (FGL) photocoagulation for clinically significant macular oedema (CSME) appeared, FGL has become the benchmark against which other treatments for diabetic macular oedema (DME) are compared.1 The Diabetic Retinopathy Clinical Research Network found that 1-mg and 4-mg doses of preservative-free intravitreal triamcinolone acetonide (IVTA) were only more effective than FGL in reducing centre-involved DME at 6-month follow-up, but failed to maintain this advantage at 2- and 3-years follow-up.2 This result has been attributed to the shorter duration of action and the need for multiple administrations of IVTA when compared with FGL.

With the advancement of optical coherence tomography (OCT), several patterns of DME have been classified.3 Investigators have reported that subtypes with cystoid DME extending over two disc-areas or DME with foveal detachment with or without vitreomacular traction (VMT) on OCT respond poorly to FGL.4 IVTA has been reported to be effective in these specific subtypes as well as in cases unresponsive to previous FGL.5,6

From above, one might conclude that initial treatment with IVTA could help reduce all subtypes of DME to an extent that is amenable to treatment and long-term maintenance with FGL. The side effect profile of IVTA, which includes increased intraocular pressure (IOP) and cataract progression, made us wonder if we could try alternative agents as initial therapy that would achieve efficacy similar to IVTA without incurring these potential side effects.7–9

Intravitreal bevacizumab (Avastin, IVA) has also been known to show promise in cases with centre-involved DME. The 1-year results of the Bevacizumab Or Laser Therapy study demonstrated a median of 8 ETDRS letters gained in the IVA group compared with a median of 0.5 ETDRS letters lost in the FGL group.10 A study directly comparing IVA with IVTA demonstrated equivocal results with respect to macular thickness reduction and best-corrected visual acuity (BCVA) improvement in cases of center-involved DME, but IVA had a better ocular side effect profile compared with IVTA.11 Nevertheless, contrary to the above studies where all subtypes of DME were included, in a trial comparing IVTA with IVA for cystoid DME refractory to FGL, IVTA demonstrated superior efficacy.12

Ultra-short acting intravitreal steroids like dexamethasone have less of an effect on IOP rise and cataract progression. However, they also tend to have less of an effect on the reduction of DME.13

In this study, we investigate whether or not a combination of intravitreal dexamethasone and bevacizumab (IVD + A) is synergistic enough to surpass the efficacy of IVTA in patients with cystoid DME extending over two disc-areas. In this way, we may also be able to avoid the ocular adverse effects of IVTA.


Enrolment of subjects, inclusion and exclusion criteria

Before screening, a detailed written and informed consent of every subject was taken as per the tenets of the declaration of Helsinki. The enrolment period extended over 9 months from a single centre (AJEH Pvt. Ltd.).

Eligible study subjects in this prospective, non-randomized, masked, comparative, interventional case series were divided into two groups. The first group was chosen to receive IVTA whereas the other group was chosen to receive IVD + A. The selection of subjects was deliberately made non-randomized in order to minimize the known ocular risks associated with IVTA. Each subject chosen to be in the IVTA group was pseudophakic with a normal appearing optic nerve on fundus examination and an IOP measurement less than 22 mmHg. The inclusion/exclusion criteria were as follows:

Inclusion criteria

  • 1A diagnosis of diabetes mellitus, any subtype.
  • 2The presence of CSME (ETDRS definition) in one or both eyes, either primary (previously untreated) or refractory (persistent CSME> 3 months following one or more focal/grid photocoagulation sessions). When both eyes had CSME, the worse of the two eyes was chosen if both eyes met the criteria below:
    • a >50% of area with cystoid changes on 5 × 5 grid raster (E-MM5) protocol (see below)
    • b Central foveal thickness > 300 microns on cross-hair protocol (see below)
    • c BCVA worse than 0.3 but better than 1.0 on the logMAR chart
    • d Diffuse macular thickening of greater than two disc-areas on slit lamp biomicroscopy
  • 3Serum lipids, renal function and blood pressure controlled with or without medication as per investigations done less than 1 month prior to study enrolment.
  • 4Haemoglobin A1c < 8% as per investigation done less than 3 months prior to study enrolment.
  • 5Haemoglobin > 10 gm/dl as per investigation done less than 1 month prior to study enrolment.
  • 6Patient amenable to at least 6 weeks of follow-up care.

Exclusion criteria

  • 1Prior history of intraocular/periocular anti-vascular endothelial growth factor or steroid injection.
  • 2Evidence of proliferative diabetic retinopathy on fundus fluorescein angiography (FFA) or clinical examination.
  • 3Area of foveal avascular zone in arteriovenous phase of FFA > 1000 µ in diameter.
  • 4Presence of a taut posterior hyaloid, VMT or epiretinal membrane on OCT.
  • 5Presence of subfoveal hard exudates.
  • 6Any other ocular cause that in the opinion of the investigator would interfere with BCVA such as significant cataract, glaucoma, age-related macular degeneration, branch retinal vein occlusion, etc.
  • 7Patient undergoing systemic steroid therapy.
  • 8Personal history of previous thromboembolic events.


Baseline evaluation

On initial examination, a detailed clinical history was taken and the most recent investigations pertaining to lipid profile, renal profile, blood pressure, haemoglobin and haemoglobin A1c were noted from each subject. If reports of these investigations were not available as per the inclusion criteria, appropriate blood and urine tests were performed.

Trained optometrists masked to the treatment group measured the BCVA using a logMAR chart. IOP was recorded using Goldmann aplanation tonometry. A single vitreoretinal fellow performed the slit lamp, fundus and IOP evaluations. The phakic lens status was graded as per Lens Opacity Classification System III specifications.14

Baseline FFA examinations were obtained. Each FFA was used to rule out the presence of retinal neovascularization and to measure the diameter of the foveal avascular zone in the arteriovenous phase.

Baseline OCT examinations were obtained from a single spectral domain OCT (SD-OCT) machine (RTVue, Optovue Inc, Fremont, CA, USA). The OCT data collected included high-definition cross-hair and 5 × 5 mm grid raster (E-MM5) protocols. The central foveal thickness as well as the presence of an epiretinal membrane, VMT, taut posterior hyaloid or foveal detachment were noted on the high-definition cross-hair protocol. The area of cystoid changes and volumetric measurements were noted on the E-MM5 protocol. The E-MM5 protocol divides 5 × 5 mm of the macula (centred at the fovea) into 16 × 16 small squares by 17 × 17 perpendicularly intersecting lines. For measuring area of cystoid changes, the central 8 × 8 small squares (Area = [8 × 5/16 mm] × [8 × 5/16 mm] = 6.25 mm2) corresponding to two disc-areas (2 × π × r2 = 2 × 3.14 × [≈1000 µ]2 = 6.28 mm2) must have at least ≥50% of the intersecting points with cystoid changes (Fig. 1). Another single masked vitreoretinal fellow performed each OCT evaluation.

Figure 1.

E-MM5 grid raster protocol on RTVue spectral domain optical coherence tomography showing the central two disc-areas that include 8 × 8 squares or 2.5 × 2.5 (6.25) mm2 or 9 × 9 intersecting lines. At each intersecting point, the presence or absence of cystoid changes is noted.


Within 1 week of the baseline examination, each subject was given either IVTA or IVD + A by a single surgeon in the operating room under sterile conditions. Preservative-free IVTA (Inj. Trilon 40 mg, Ajanta Pharma, Mumbai, India) was injected in a dose of 4 mg/0.1 mL via the pars plana inferotemporally with a 27-gauge needle. IVD + A was injected in doses of 0.4 mg/0.1 mL (dexamethasone) and 1.25 mg/0.05 mL (bevacizumab) via the pars plana superotemporally through separate sites using 30-gauge needles. An aqueous paracentesis was performed if central retinal artery pulsations were obliterated. Postoperatively, antibiotic eyedrops were prescribed (Ciprofloxacin 0.3%) four times daily for 7 days in both groups.

Follow-up evaluation

Each subject underwent BCVA, IOP and SD-OCT measurements with slit lamp and fundus evaluations on the first postoperative day and thereafter at the end of first, second, third, fourth and sixth week.

All subjects, irrespective of the intravitreal agent used, showing a clinically significant reduction in macular volume (by at least 1 mm3) underwent maintenance FGL at 6 weeks. All subjects not showing a clinically significant reduction in macular volume were allowed to choose between observation and IVTA with maintenance FGL at a later stage. Beyond 6 weeks, IOP was monitored in all subjects until the IOP stabilized <22 mmHg (two occasions 1 week apart) without medication.

Outcome measures

The primary outcome measure was total macular volume change using the SD-OCT E-MM5 protocol. The secondary outcome measure was ETDRS BCVA expressed in logMAR.

Statistical analysis

All tests were two-tailed and set at P-value of significance of <0.05. Categorical data were analysed using the Fisher's exact test. Continuous data were analysed using the paired and the unpaired ‘t’-test for paired and unpaired data, respectively. Correlation analysis was done using Pearson's product moment correlation coefficient.


Twelve eyes of 12 subjects were treated with IVD + A, and 12 eyes of another 12 subjects were treated with IVTA. In the IVD + A group, 10 subjects were phakic. The two remaining subjects were pseudophakic with a baseline IOP > 21 mmHg. In the IVTA group, all 12 subjects were pseudophakic with a baseline IOP < 22 mmHg. None of the study subjects was lost to follow-up for the 6-week duration of the study.

The mean age was 64.5 ± 11.3 (mean ± SD) years. The baseline characteristics between the two groups were found to be similar as shown in Table 1, except for the proportion of foveal detachments, which was significantly higher in the IVD + A group.

Table 1.  Baseline characteristics, IVTA group versus IVD + A group
Baseline characteristicsIVTA groupIVD + A group P-value
  1. Independent samples t-test. Fisher's exact test. DM, diabetes mellitus; FAZ, foveal avascular zone; IVD + A, intravitreal dexamethasone and bevacizumab; IVTA, intravitreal triamcinolone acetonide; NPDR, non-proliferative diabetic retinopathy.

Age (years ± SD)65.8 ± 11.663.5 ± 11.10.63
Men (%)58.3366.660.24
Duration of DM (years ± SD)10.2 ± 2.010.6 ± 2.20.65
Insulin takers (%)58.3366.660.24
Severe NPDR (%)5041.660.32
FAZ abn (%)33250.27
Foveal detachment (%)2558.330.02
Prior laser (%)25251
Pre OCT volume (mm3 ± SD)9.10 ± 1.019.02 ± 0.960.84
Pre BCVA (logMAR ± SD)0.76 ± 0.130.77 ± 0.140.91

In the IVTA group, the post-injection macular volume showed a statistically significant improvement from a baseline of 9.11 ± 1.0 mm3 to 7.62 ± 0.7 mm3 (reduction of 1.49 mm3, P < 0.001) at 1 week and to 7.46 ± 0.73 mm3 (reduction of 1.65 mm3, P < 0.001) at 6 weeks. However, the post-injection BCVA showed only a trend towards improvement from a baseline of 0.76 ± 0.13 to 0.66 ± 0.14 (gain of 0.1 logMAR units, P = 0.065) at 1 week to 0.65 ± 0.15 (gain of 0.11 logMAR units, P = 0.06) at 6 weeks. In the IVD + A group, the post-injection macular volume did not show a statistically significant improvement from a baseline of 9.03 ± 0.96 mm3 to 8.39 ± 1.1 mm3 (reduction of 0.64 mm3, P = 0.1) at 1 week and to 8.42 ± 1.18 mm3 (reduction of 0.61 mm3, P = 0.21) at 6 weeks. The post-injection BCVA also did not show any significant improvement from baseline of 0.77 ± 0.14 to 0.71 ± 0.14 (gain of 0.06 logMAR units, P = 0.2) at 1 week and to 0.69 ± 0.15 (gain of 0.08 logMAR units, P = 0.17) at 6 weeks (Figs 2,3).

Figure 2.

Intravitreal triamcinolone acetonide (IVTA) group, optical coherence tomography (OCT) macular volume pre- and post-injection, reduction = 1.65 mm3, P < 0.001, lines joining paired cross markers (n = 12) showing a parallel orientation with down sloping. Intravitreal dexamethasone and bevacizumab (IVD + A) group, OCT macular volume pre- and post-injection, improvement = 0.61 mm3, P = 0.21, lines joining paired cross markers (n = 12) showing a criss-cross orientation.

Figure 3.

Intravitreal triamcinolone acetonide (IVTA) group, best-corrected visual acuity (BCVA) pre- and post-injection, improvement = 0.11 logMAR units, P = 0.06, lines joining paired triangular markers (n = 12) showing a criss-cross orientation. Intravitreal dexamethasone and bevacizumab (IVD + A) group, BCVA pre- and post-injection, improvement = 0.08 logMAR units, P = 0.17, lines joining paired triangular markers (n = 12) showing a criss-cross orientation.

The difference in post-injection macular volumes between the two groups at 6 weeks was statistically significant (0.96 mm3, P = 0.02). The difference in post-injection BCVA between the two groups at 6 weeks was not statistically significant (0.04 logMAR units, P = 0.06) (Fig. 4).

Figure 4.

Intravitreal triamcinolone acetonide (IVTA) versus intravitreal dexamethasone and bevacizumab (IVD + A) group, optical coherence tomography (OCT) macular volume post-injection, reduction = 0.96 mm3, P = 0.02; best-corrected visual acuity (BCVA) post-injection, improvement = 0.04 logMAR units, P = 0.06.

For correlation analysis, we clubbed the data from the two groups. The change in BCVA showed poor correlation (r = 0.35, P = 0.07) with the change in macular volume. The pre-injection macular volume showed a fair, significant and positive correlation with change in macular volume (r = 0.55, P > 0.05).

No serious ocular adverse events such as vitreous haemorrhage, retinal detachment or endophthalmitis were seen in either group. Six out of 12 patients (50%) in the IVTA group had a post-injection rise of IOP > 21 mmHg. None of the subjects in the IVTA group had a rise above 30 mmHg on any occasion and were managed with a single anti-glaucoma medication (Timolol maleate 0.5% eyedrops twice daily). The IOP measurements of all subjects in the IVTA group returned to <22 mmHg after a median duration of 7 weeks post injection. In the IVD + A group, none of the 10 phakic patients had cataract progression as per Lens Opacity Classification System III or a rise of IOP > 21 mmHg on any occasion. No systemic adverse events were noted in either group.


This study uniquely evaluates macular volume on SD-OCT as the primary outcome measure in DME rather than central point thickness or central subfield macular thickness (CSMT) on time domain OCT.15,16 The rationale behind this selection is that a change in macular volume over a 5 × 5 mm area should be more reflective of the overall treatment effect when measured against a change confined to either a single central point or a small central area, especially in cases of diffuse (over two disc-areas) DME. The volumetric analysis on 5 × 5 grid raster scan on SD-OCT may also be more reproducible in cases of cystoid DME where localization of the central fovea is difficult. In any case, it has been shown that global retinal volume on fast macular scans of time domain OCT showed good correlation (r = 0.75–0.77) with CSMT.

Most studies in the literature have been unable to demonstrate a good correlation between changes in BCVA to changes in CSMT.17,18 Likewise, our study was unable to show any significant correlation between change in macular volume and change in BCVA. However, we found that a higher macular volume on OCT pre-injection was related to greater change in macular volume on OCT post-injection. Therefore, it may be concluded that a more oedematous macula in a patient with DME responds somewhat better to either IVTA or IVD + A.

This study is the first to compare directly the use of IVD + A with IVTA in subjects with DME. A single injection of IVTA has been shown to significantly reduce DME and improve BCVA across various studies, albeit over a short-term period.19,20 Most reports in literature included all subtypes of DME, except for Paccola et al.,12 which had select cases of DME refractory to laser and diffuse in nature with mean central macular thickness greater than 250 microns. Our study specifically targeted similar cases of DME of the predominantly cystoid variety and found IVTA to reduce macular volume significantly.

Haritoglou et al. reported a significant reduction in macular thickness at 2 (15%), 6 (17%) and 12 weeks (25%) following monthly IVA in patients with DME.21 Soheilian et al. found that combination therapy with IVTA and IVA demonstrated no additional benefit when compared with IVA alone.22 These studies, however, included all subtypes of DME. No significant macular volume change or improvement in BCVA was observed in our study in the IVA group even though supplemented with intravitreal dexamethasone. The cystoid DME treated by this study best explains the lack of response to IVD + A.

There was a possibility that IVD + A could have shown a very immediate significant reduction in macular volume (1- or 2-week follow-up), because of the ultra-short acting nature of dexamethasone. It is worthwhile pointing out that OCT measurements were taken at weekly intervals for the first 4 weeks and none showed a significant reduction in macular volume. We presented our final results at 6-week follow-up assuming that IVTA and IVD + A would have ceased to exert their effects thereafter.

The absence of a longer follow-up remains a limitation of our study. Moreover, some may argue in favour of administering multiple injections of IVD + A to achieve a progressive stepwise reduction in macular thickening and attain an OCT appearance more amenable to treatment with FGL. Keeping in mind the off-label use and the short duration of action of IVTA and IVA as well as the ocular adverse effects of IVTA, we did not consider using multiple injections of any agent. The primary purpose of using these intravitreal agents was to achieve a temporary reduction in macular thickening that could then be treated with the more permanent standard-of-care FGL. This was confirmed by the fact that all cases registering a clinically significant reduction in macular volume received maintenance FGL at 6 weeks. Any further conclusion about the long-term efficacy of injections would have been confounded by this major intervention.

Finally, our study has a non-randomized allocation of its subjects to either group. This was perhaps more permissible on ethical grounds, as we tried to avoid the ocular adverse effects of IVTA in phakic and ocular hypertensive patients. An attempt was made to match every confounder known to interfere with either the OCT or BCVA outcomes post enrolment, and we felt that lens status should not skew our results. It can be argued that the significantly greater number of subjects with subfoveolar fluid in the IVD + A group could have accounted for its poor response. Nevertheless, the pre-injection macular volumes that take the volume of subfoveolar fluid into account matched well in the two groups.

In conclusion, a single injection of IVTA demonstrated better efficacy compared with IVD + A in terms of OCT macular volume reduction in select cases of primary or refractory, diffuse, cystoid DME after 6-week follow-up. However, neither treatment modality significantly improved BCVA. Until the efficacy of IVTA compared with focal/grid photocoagulation over long-term is established for this select group of refractory, diffuse, cystoid DME by a head-to-head trial, the latter still remains the mainstay of treatment.