While levels of VEGF tended to be higher in those with proliferative diabetic retinopathy, some of their fluid had less VEGF than did the healthy participants.

But even the low-VEGF fluid from patients with proliferative diabetic retinopathy stimulated blood vessel growth in lab-grown cells.

"The results suggested to us that although VEFG clearly plays an important role in blood vessel growth, it's not the only factor, " Sodhi said in a statement.

A series of experiments in lab-grown human cells and mice revealed a second culprit, a protein called angiopoietin-like 4.

When the researchers blocked the action of both VEGF and angiopoietin-like 4 in fluid from the eyes of people with proliferative diabetic retinopathy, it markedly reduced blood vessel growth in lab-grown cells.

Sodhi suggested if a drug can be found that safely blocks the second protein's action in patients' eyes, it might be combined with the anti-VEGF drugs to prevent many cases of proliferative diabetic retinopathy.

The team is now investigating whether angiopoietin-like 4 might also play a role in other eye diseases, such as macular degeneration, which destroys the central portion of the retina.