Sir,

In the May 2006 issue of Eye, Lee et al1 present their findings from a small study which adds to our current understanding of pathological neovascularization in diabetic retinopathy. However, certain issues related to their study merit further discussion.

Their investigation of systemic (rather than local) angiogenic factors in diabetic retinopathy is not novel, as claimed by the authors. In 2000 and 2004, our research group reported plasma levels of vascular endothelial growth factor (VEGF), soluble VEGF receptor (Flt-1), angiopoietin-2 and its receptor tie-2 in patients with and without retinopathy.2, 3 Our studies confirmed the presence of abnormally high levels of circulating plasma markers of angiogenesis/vasculogenesis in these patients, suggesting an association between plasma indices of angiogenesis and the presence of diabetic retinopathy.

Of note, Lee et al1 found no correlation between the severity of diabetic retinopathy and serum VEGF levels, and cite the EUCLID substudy as corroborative evidence.4 In that particular study, the researchers measured plasma as apposed to serum VEGF levels,4 making a comparison difficult. Indeed, the use of serum to measure systemic VEGF levels has to be interpreted with caution, because activated platelets (during blood clotting) release VEGF into serum, and thus, results based on serum samples may be inaccurate in view of the possible artifact relating to the source of VEGF levels.5 Also, the EUCLID study did report an upward trend in plasma VEGF levels, with increasing retinopathy (no retinopathy vs nonproliferative vs proliferative) (mean levels pg/ml: 11.5 vs 12.9 vs 16.1, respectively), an association that approached statistical significance (P=0.06) but was severely confounded by the disproportionate numbers in the groups (eg for nonproliferative n=167, proliferative n=8). In our own study, plasma levels of VEGF were elevated in diabetic patients, with the highest levels seen in more severe retinopathy (grade 2 and 3) (P<0.05).3

Furthermore, the title of their article1 states circulating ‘endothelial progenitor cells’ but in fact, the study concentrates on stem cells (c-Kit+) and haematopoietic progenitor (CD34+) mononuclear cells. Increasingly, evidence is emerging that endothelial precursor cells are a heterogenous population with potentially different parent cell lines, but with some overriding phenotypic similarities in markers such as CD34, CD133, and VEGF receptor-2 (KDR). Furthermore, to prove endothelial lineage of these cells, uptake of acetylated low-density lipoprotein and Ulex Europaeus lectin should be demonstrated.6 Thus, Lee et al1 do not positively confirm that the increase in c-Kit+ and CD34+ cells in their study is due to a concurrent increase in the endothelial progenitor cell fraction, and we suggest that they should be more circumspect in their conclusions.

Lastly, the study by Lee et al1 suffers from the obvious limitations of a cross-sectional design, and the many associated co-morbidities frequently seen with diabetic patients (hypertension, obesity, cardiovascular disease, etc) and drug treatments have to be fully taken into consideration, especially in a relatively small study. Many comorbidities in diabetics (as well as drugs such as angiotensin converting enzyme inhibitors and statins) could have a large effect on the various indices measured, making their results difficult to interpret.