P-glycoprotein expression in retinoblastoma

We have read with interest the paper by Krishnakumar et al.¹ We have studied 18 children with retinoblastoma using immunohistochemical detection of P-glycoprotein by the mouse monoclonal antibody HYB-241 in frozen section tumour samples. Like Krishnakumar et al, we found limited expression of P-glycoprotein in retinoblastoma cells since only four of the 18 samples were positive. However, six of 18 samples had P-glycoprotein positivity in tumour associated endothelial cells. Contrary to Krishnakumar et al, we had the opportunity of evaluating children who received previous therapy with drugs that can be affected by P-glycoprotein and no obvious difference in its expression was evident. Only one of the three tumour samples coming from patients who had been exposed to P-glycoprotein mediated chemotherapeutic agents (vincristine and/or etoposide) were positive in retinoblastoma cells. In addition, of the six cases in which P-glycoprotein was detected in tumour associated endothelial cells, only one of them came from a patient who had been exposed to P-glycoprotein mediated chemotherapeutic agents (etoposide). Also, one patient whose tumour and tumour associated endothelium were negative had endothelial cells in the optic nerve that were positive. The patient had not been exposed to P-glycoprotein mediated chemotherapy.

There is little information regarding its expression in normal eye blood vessels, but P-glycoprotein expression has been detected in the normal human retinal pigment epithelium and post-laminar optic nerve,² and it is absent in the permeable microvessels of the choroid and the ciliary process.³ P-glycoprotein expression in endothelial cells of newly formed capillaries induced by tumours has also been described.⁴ Its prognostic relevance in retinoblastoma is unknown, and it cannot be determined from our data, since all of these eyes were already enucleated and no patient had an extraocular relapse.

The potential role of P-glycoprotein in tumour associated endothelium creating a blood-tumour barrier is uncertain. Our results suggest that if cyclosporine is indeed improving the efficacy of chemotherapy, it may be acting by improving its delivery through P-glycoprotein expressing tumour associated endothelial cells or through altering pharmacokinetics of chemotherapeutic drugs, rather than exclusively acting to reverse drug resistance at the level of the tumour cell.

Our data do not support the hypothesis that cyclosporine will be a useful adjunct to chemotherapy for the treatment of retinoblastoma via competitive inhibition of P-glycoprotein in tumour cells.