Effect of calcium dobesilate on progression of early diabetic retinopathy: a randomised double-blind study
- First Online:
- Cite this article as:
- Ribeiro, M.L., Seres, A.I., Carneiro, A.M. et al. Graefe's Arch Clin Exp Ophthalmol (2006) 244: 1591. doi:10.1007/s00417-006-0318-2
- 254 Views
The study was carried out to confirm the effect of calcium dobesilate (CaD) compared to placebo (PLA) on the blood-retinal barrier (BRB) permeability in early diabetic retinopathy (DR).
Adults with type II diabetes and early diabetic retinopathy (below level 47 of ETDRS grading and PVPR between 20 and 50×10−6/min, plasma-free fluorescein) were included in this double-blind placebo-controlled study. Treatment was 2 g daily for 24 months. The primary parameter, posterior vitreous penetration ratio (PVPR), was measured every 6 months by fluorophotometry. Secondary parameters were fundus photography, fluorescein angiography and safety assessments. Metabolic control was performed every 3 months.
A total of 194 patients started the treatment (98 CaD, 96 PLA) and 137 completed the 24-month study (69 CaD, 68 PLA). Both treatment groups were comparable at baseline, with ETDRS level 10 in about 59% of patients. Mean PVPR change from baseline after 24 months was significantly (P=0.002) lower in the CaD group [−3.87 (SD 12.03)] than in the PLA group [+2.03 (SD 12.86)], corresponding to a 13.2% decrease in the CaD group and a 7.3% increase in the PLA group. PVPR evolution was also analysed by HbA1c classes (<7%, between 7 and 9%, ≥9%) and results confirmed the superiority of CaD independently of the diabetes control level. A highly significant difference [CaD: −3.38 (SD 13.44) versus PLA: +3.50 (SD 13.70)] was also obtained in a subgroup of patients without anti-hypertensive and/or lipid-lowering agents (P=0.002 at 24 months). A further analysis of the secondary parameters showed significant changes in favour of CaD in the evolution from baseline to the last visit of haemorrhages (P=0.029), DR level (P=0.0006) and microaneurysms (P=0.013). Regarding safety, only 2.5% (n=5 patients/events) of all adverse events reported were assessed as possibly or probably related to the test drug, while all serious adverse events were reported as unlikely. There was no statistical difference between groups.
Calcium dobesilate 2 g daily for 2 years shows a significantly better activity than placebo on prevention of BRB disruption, independently of diabetes control. Tolerance was very good.
KeywordsDiabetic retinopathyBlood-retinal barrierPosterior vitreous penetration ratioCalcium dobesilate
Type II diabetes mellitus is one of the most prevalent and costly chronic diseases in Western society, with a rising prevalence [34, 47, 49]. One complication is diabetic retinopathy (DR), in itself a leading cause of visual impairment and blindness in working age adults . Progression of DR can be delayed by tight glycemic control , but this can be difficult in practice .
Retinal laser photocoagulation can prevent and sometimes improve visual loss , but the quality of life of the patient may be altered, e.g. some patients may be unable to drive after surgery due to loss of night and peripheral vision [36, 41]. Moreover, patients undergoing laser photocoagulation may experience immediate worsening after the intervention .
Clearly, pharmacological treatment in the early stages of DR to slow or stop its progression would be desirable [14, 24]. Some treatments have been suggested, including aspirin [2, 15], ACE inhibitors [9, 38], angiotensin I and II receptor antagonists [10, 23, 29, 45] and inhibition of the beta isoform of protein kinase C . However, the therapeutic benefit of such agents has yet to be demonstrated in large controlled clinical trials.
Several double-blind placebo-controlled clinical studies performed more than 20 years ago have demonstrated the effectiveness of CaD at slowing progression of diabetic retinopathy [4, 20, 42], with efficacy endpoints such as leakage, fluorescein angiography and fundus photography parameters in relatively small cohorts. Except for one study with a treatment duration of 2 years , all lasted less than 1 year. Fluorescein leakage, also known as posterior vitreous penetration ratio (PVPR), was the primary endpoint of a pilot study at the high dose of 2 g/day for 12 months and in which the permeability of the blood-retinal barrier (BRB) was stabilised as reflected by PVPR .
The objective of the present study was to confirm the efficacy of calcium dobesilate to restore the BRB in a larger patient population at an oral dose of 2 g/day over 2 years.
Materials and methods
Inclusion criteria were subjects of both sexes, aged 40–75 years, with stable adult onset type 2 diabetes, minimal evidence of retinopathy (score less than 47 on the Wisconsin grading scale ) and evidence of mild fluorescein leakage in vitreous fluorometric examination (PVPR between 20 and 50×10−6/min, plasma-free fluorescein, normal value=15.9±4.7×10−6/min ).
Exclusion criteria were signs of other eye disorders such as macular oedema, retinal vascular diseases or vitreous syneresis and previous laser therapy.
This randomised double-blind placebo-controlled study was conducted in eight European centres. The trial was approved by the competent ethics committees and national authorities, and all patients provided written informed consent prior to enrollment.
After final selection, the patient was randomly assigned to active treatment (two capsules of calcium dobesilate 500 mg, Doxium 500, OM PHARMA, Meyrin/Geneva, Switzerland) or a matching placebo (PLA), administered once in the morning and once in the evening after meals, for the duration of the study. Treatment compliance was assessed at each monthly visit by counting the capsules in returned medication bottles. The study lasted 24 months and was monitored regularly to ensure compliance with Good Clinical Practice.
The primary efficacy endpoint was the change in PVPR (measurement of fluorescein leakage) over the entire study, at 24 months and at the last available visit. Leakage was measured every 6 months with a Fluorotron Master (OcuMetrics, Mountain View, Calif., USA). Scans were taken before intravenous administration of fluorescein (14 mg/kg) and 60 min after injection. Blood samples were collected after 10, 15 and 50 min to measure plasma fluorescein concentration. The details of the procedure have been described elsewhere [11, 30].
Retinographic assessments by fundus photography according to Wisconsin Grading [16, 28] and angiographic assessments according to Early Treatment Diabetic Retinopathy Study (ETDRS) Grading , determined every 6 months, were used as secondary efficacy endpoints. Visual acuity and intra-ocular pressure were also monitored.
Adverse events were assessed and laboratory analyses were performed every 3 months to evaluate the safety of the drug. These laboratory assessments included measurement of fasting blood glucose and glycosylated haemoglobin (HbA1c) to assess the degree of metabolic control.
All patients who received at least one dose of study medication were included in the safety analysis. Efficacy was analysed for the intent-to-treat (ITT) population (all patients entering the treatment phase with at least one efficacy measure after baseline). The worst eye of each patient, as assessed by PVPR and fundus photography, was used for the primary efficacy analysis.
A further analysis of the secondary parameters was conducted to investigate if the choice of the worst eye based on PVPR levels introduced a bias which denied the unveiling of a difference between CaD and PLA in the results of the stereoscopic retinography. This analysis was based on the “worst” fundus eye instead of the “worst” PVPR eye of the previous analysis. The “worst” fundus eye for each patient was defined as the one with the higher value of DR level at baseline. The following fundus parameters were investigated: haemorrhages, microaneurysms and DR level. Exudates were not analysed since too few patients showed these signs during the trial.
Demographic and baseline characteristics of ITT population
Calcium dobesilate (n=87)
Onset of diabetes (years)
Duration of diabetes treatment (years)
Pulse rate (beat/min)
BP systolic (mmHg)
BP diastolic (mmHg)
Dietary restrictions (Y/N)
Conc. disease (Y/N)
Conc. medication (Y/N)
Eye considered (RE/LE)
DR level (10/15/20/35)
For the efficacy analysis, ANCOVA by treatment group was performed with all the slopes of the individual PVPR values by time and with the intercept as covariate. The slopes were weighted according to the duration of treatment. The 24-month point and the last time point available for change from baseline of PVPR were also compared between treatment groups by unpaired Student t-test or Mann-Whitney test and within groups by paired Student t-test or Wilcoxon signed rank test, according to the normality of the distributions.
A subgroup analysis was included in the statistical plan while the data were still blinded depending on the average levels of HbA1c: HbA1c<7.0%, 7.0%≤HbA1c<9.0%, HbA1c≥9.0%.
Secondary efficacy parameters expressed as raw values and changes from baseline were compared for the two groups with the Mann-Whitney test.
Robust multiple regression analyses were also performed for each study group to investigate the influence of other factors on the permeability of the BRB.
The safety variables (adverse events and laboratory values) were displayed descriptively, and the two treatment groups compared using the Fisher exact test or the Mann-Whitney test.
For the sample size calculation, the difference in penetration ratio between calcium dobesilate and placebo was assumed to be 3.4×10−6/min (adjusted from 0.6 reported in the pilot study based on total plasma fluorescein) . Sixty-four patients were required in each treatment group to obtain an alpha of 0.05 and a beta of 0.8, but because of the expected high rate of drop-outs in a 24-month study with out-patients, 240 patients were planned to provide approximately 130 analysable cases.
For all comparisons, P-values of 0.05 or less were considered statistically significant. Statistical calculations were conducted using NCSS version 2004 (NCSS, Kaysville, USA), Systat v.9.0 (SPSS Inc, Chicago, Ill., USA) and Testimate v.5.02 (IDV, Gauting, Germany).
The demographic and baseline characteristics of the two treatment groups were similar except for sex, which showed a trend to dissimilarity with a higher proportion of men in the active treatment group. DR level at baseline was similar in both groups, with level 10 in 59.3% of CaD group and 58.5% of PLA group (Table 1). Mean treatment duration was 676.5 (SD 136.5) days for the CaD group and 679.4 (SD 131.8) days for the PLA group. Compliance, measured as a percentage of theoretical consumption of 4 capsules a day was 95% for both groups.
Factors affecting posterior vitreous penetration ratio (PVPR) progression: a robust multiple regression analysis
PVPR at 24 months (vs baseline)
Factor full model R2=0.2951
Diastolic blood pressure (mmHg)
Systolic blood pressure (mmHg)
Previous duration of diabetes (years)
Glycosylated haemoglobin (%)
Body mass index (kg/m2)
Intra-ocular pressure (mmHg)
Other concomitant medications (Y=1/N=0)
PVPR at baseline
There were no consistent or significant differences between groups in the evolution of best corrected visual acuity, intra-ocular pressure, glycaemia or HbA1c.
The fundus or angiography parameters were also similar for both groups at baseline and at study end, with the worst eye based on PVPR. The further analysis of the fundus parameters, with worst eye based on DR level revealed, however, the following results for haemorrhages, microaneurysms and DR level.
The evolution of microaneurysms count from baseline until the last visit showed a trend between groups in favour of CaD (P=0.066). Moreover, the number of microaneurysms was improved in significantly more patients with the active treatment compared to placebo at the last visit (P=0.038). All these results are confirmed by the highly significant worsening for PLA (P<0.001) from baseline to the last visit, compared to no significant variations for CaD. By ranking microaneurysms by classes , the difference of progression between groups appears even more evident (Fig. 6) and also gives a significant result in favour of CaD (P=0.013).
Finally, DR level evolution from baseline until the last visit showed a highly significant difference between groups in favour of CaD (P<0.001) as shown by Fig. 6. Again, DR level improved in significantly more patients in the active treatment group than in the placebo group (P<0.001).
This analysis of fundus parameters, considering the worst eye based on DR level, confirmed the significant difference in favour of CaD for haemorrhages, microaneurysms and DR level.
Adverse events with probable or possible relationship with the study drug
Vascular leakage is a very early process in the pathophysiology of diabetic retinopathy [31, 32], and can be a prelude to a cascade of events that may lead to eventual blindness. Fluorescein leakage into the vitreous is a surrogate endpoint that may potentially predict outcome of clinically relevant endpoints such as visual loss from macular oedema or the need for photocoagulation. Indeed, previous studies have found high rates of leakage to be associated with an increased risk of macular oedema and need for photocoagulation [8, 13, 50].
Calcium dobesilate was recently found to act on basic biochemical mechanisms involved in diabetic retinopathy. In vitro and in vivo results suggest that CaD has antioxidant properties via a direct scavenging effect of free radicals [5, 6, 50]. Moreover, CaD strongly reduces retinal levels of AGEs and VEGF and reversed retinal hyper-permeability in diabetic rats with pre-proliferative retinopathy . The most plausible explanation of these results is that calcium dobesilate first reduced carboxymethyl-lysine-advanced glycation end product (CML-AGE) formation via its antioxidant properties and that this in turn decreased retinal VEGF over-expression and secondary albumin leakage. The above pharmacological evidence suggests that CaD can stabilise the blood-retinal-barrier in patients with diabetic retinopathy by an antioxidant mechanism .
In the last 20 years, several double-blind placebo-controlled clinical studies demonstrated the efficacy of calcium dobesilate at slowing DR progression. Most of them were performed with small cohorts, low dosage and short study duration and their evaluation parameters were only based on fluorescein angiography or fundus photography, but none of them indeed evaluated PVPR.
In the 1990s, a controlled study against no therapy , with a dosage of 1.5 g CaD /day in the treated group and a follow-up of 2 years, did not show any reduction in DR progression based on the same parameters. According to the authors, the intake of study drug was poorly controlled, which could explain the lack of efficacy.
On the other hand, a pilot double-blind placebo-controlled study with 47 patients  reported that PVPR remained stable in patients with early diabetic retinopathy on calcium dobesilate 2 g/day for 12 months. As in the pilot study, we studied the effect of calcium dobesilate on PVPR in patients with diabetic retinopathy in its early phases, but over 2 years and in a much larger patient collective.
In the present study, the demographics and baseline characteristics of the two treatment groups were comparable, except that there were more men than women in the CaD group, which did not influence the outcome of the analyses.
This study showed a progressive significant improvement in PVPR with respect to baseline in the group treated with calcium dobesilate during the 24 months of the study (P<0.0001). The placebo group, on the other hand, showed no significant change with respect to baseline. Direct comparison of calcium dobesilate treatment with placebo showed a significant effect of active treatment (P=0.006 for change from baseline to last available value, P=0.002 for change from baseline to 24-month value). These results were further confirmed by ANCOVA comparison of the corrected slopes for the individual regression lines.
As expected in a population with minimal retinopathy, treatment with calcium dobesilate did not influence visual acuity or intra-ocular pressure.
Fundus parameters, however, showed significant differences in favour of CaD in the evolution of haemorrhages, microaneurysms and DR level, when considering the worst eye based on DR level at entry.
Higher levels of glycosylated haemoglobin correlate with a faster rate of progression of diabetic retinopathy [33, 43]. Given the importance of levels of HbA1c on progression of diabetic retinopathy, we analysed PVPR evolution by degree of metabolic control. In this analysis, control of BRB leakage was significantly better in the group treated with calcium dobesilate compared to placebo, regardless of degree of metabolic control. The fundus parameters evolution by subgroup of HbA1c also gave significant differences in favour of CaD at the last visit, mainly for patients with HbA1c>9%.
Hypertension correlates with poor prognosis of diabetic retinopathy . According to one study , high systemic blood pressure contributes to abnormal blood-retinal barrier permeability but anti-hypertensive therapy reverses this abnormal permeability. Hypercholesterolemia has been described as an additional independent risk factor in diabetic retinopathy, thus lipid-lowering agents may improve prognosis of diabetic retinopathy. The possible influence of anti-hypertensive and/or lipid-lowering agents on the main outcome of the current study was therefore investigated. Almost 40% of our ITT population were taking at least one of these agents. Use of such agents appeared to reduce the PVPR ratio in both the calcium dobesilate group and the placebo group. Such treatment had a favourable influence on PVPR only in placebo-treated patients, without affecting the efficacy of CaD on this parameter. However, a significant reduction with respect to baseline was seen only in the active treatment group.
We conclude that calcium dobesilate significantly lowered the permeability of the BRB as measured by the PVPR. Its effect was manifest regardless of the degree of metabolic control and the use of anti-hypertensive and/or lipid-lowering agents. Moreover, CaD had a significant beneficial effect in controlling the haemorrhages and the global evolution of DR. Our findings suggest that calcium dobesilate can slow progression of leakage in diabetic retinopathy.
The experience gained in this study, together with previous ones, could be used to design further studies on the efficacy of CaD on visual acuity outcome and the need for photocoagulation in macular oedema.
The authors would like to thank the statistician, Giancesare Gamba, PhD (Biometrix SA, CH-1196 Gland) and the other members of the DX-Retinopathy Study group: Austria – N. Maar, M. Tittl (Vienna); France – G. Coscas, G. Soubrane, L. Perlemuter, Z. Kahal, H. Oubraham (Créteil); Germany – C. Ohrloff (Frankfurt); Hungary – I. Süveges, A. Borbandy (Budapest); Portugal – E. Leite, C. Lobo (Coimbra) and JF. Castro-Correia, F. Falcão-dos-Reis (Porto); Switzerland – P. Leuenberger, A. Dosso, A. Golay, F. Ustun, L. Sekkat, M. Bagnoud (Geneva) and E. Messmer (Zürich).