Chemotherapy has an important role in the management of retinoblastoma. To improve treatment outcome, intensive research has focused on clinically relevant mechanisms of chemotherapeutic drug resistance in retinoblastoma.¹ Siegel et al² observed the presence of a small subpopulation of cancer stem cells (ABCG2 positive) and neural stem cells (MCM2 positive) in tumours from transgenic mice, human retinoblastoma cell lines and a small cohort of archival human retinoblastomas. They concluded that the presence of these cells in retinoblastomas might have significant impact on future treatment strategies.

ABCG2 is a half ATP binding cassette (ABC) transporter expressed on plasma membranes. Overexpression of ABCG2 in cell lines confers resistance on a wide variety of anticancer drugs including mitoxantrone, daunorubicin, doxorubicin, topotecan, and epirubicin.^3–5 MCM2 is one of six members of the family of minichromosome maintenance (MCM) proteins.⁶ MCM proteins are components of the prereplicative complex, which binds to replication origins in the G1 phase of the cell cycle and is essential for the initiation of DNA replication.⁷ MCM2 is a proved marker for detecting neural stem cells.⁸ Since primitive neuroectodermal cells are involved in retinoblastoma tumorigenesis,⁹ the presence of these neural stem cells could increase the aggressiveness of the original tumour.

There are no studies on ABCG2 and MCM2 proteins in a large cohort of archival retinoblastoma tumour samples. We investigated the expression of these two proteins and correlated their expression with clinicopathological parameters such as laterality, differentiation, and tumour invasiveness.

MATERIALS AND METHODS

Thirty nine tumours were available from 39 eyes for the study. Among them were tumours from 30 males and nine females. The age ranged from 4 months to 21 years (median 1 year). There were 31 unilateral retinoblastomas and eight bilateral retinoblastomas.

Tumour specimens

The study was reviewed and approved by the local ethics committee at Vision Research Foundation, Sankara Nethralaya, and the committee deemed that it conformed to the generally accepted principles of research, in accordance with the Helsinki Declaration. Tumours enucleated between 1997 and 2002 with a minimum follow up of at least 24 months were included in the study. Paraffin embedded blocks from 39 cases derived from enucleation of retinoblastomas were used for immunohistochemistry. For non-neoplastic retina donor eye balls, which were received in the pathology laboratory, were used.

The tumours were classified into three groups: group 1 (n = 18), tumour samples obtained from enucleated eyes as a part of the treatment and the patients were not subjected to either preoperative or postoperative chemotherapy (table 1); group 2 (n = 15), tumour samples obtained from enucleated eyes as a part of management and then the patients were subjected to postoperative chemotherapy because the histopathological study showed risk factors for metastasis such as invasion of choroids and optic nerve (table 2); and group 3 (n = 6), tumour samples obtained from enucleated eyes of patients who received preoperative chemotherapy (table 3)

Histopathological features

All tumour slides were reviewed and examined for invasion of choroid, optic nerve, and orbital invasion. Choroidal invasion was classified as either focal invasion or diffuse invasion of the choroid. For optic nerve invasion, post-laminar invasion and invasion of the surgical end of the optic nerve were considered.¹⁰

Invasion of tumours

There were 20 tumours with no invasion of the choroid or optic nerve and 19 tumours with invasion of the choroid, optic nerve, and orbit. Among the 19 tumours with invasion, eight had choroidal invasion (three diffuse and five focal choroidal invasion). There were four tumours with post-laminar optic nerve invasion. There were seven tumours with invasion of both choroid and optic nerve (five tumours with diffuse choroidal and post-laminar optic nerve invasion and two tumours with focal choroidal and post-laminar optic nerve invasion).

Differentiation of tumours

Retinoblastomas were graded microscopically into three groups according to the predominant pattern of differentiation.¹⁰ There were 11 well differentiated, five moderately differentiated, and 23 poorly differentiated tumours.

Treatment

Groups 1 and 2 tumours were not subjected to preoperative chemotherapy. Group 2 consisted of patients whose tumours were enucleated and then the patients were subjected to postoperative chemotherapy, because of risk factors for metastasis. Group 3 consisted of tumours, which were subjected to preoperative chemotherapy. In bilateral retinoblastomas, the eye where the tumour was small was treated with local therapy. The focal therapies employed were external beam radiation therapy (EBRT), diode laser photocoagulation (DLP), and transconjunctival cryopexy (TCC). The chemotherapy cycles ranged from two to eight cycles and drugs used included carboplatin, vincristine, and etoposide.

Clinical outcome

In group 1, where the tumours did not have any invasion nor had only focal choroidal invasion, there was no recurrence. In group 2, all the 15 tumours were subjected to postoperative chemotherapy because of the risk factors for metastasis. There was orbital invasion in four tumours (cases 4, 6, 8, 10), bone marrow metastasis in one tumour (case 5), and submandibular node metastasis in one tumour (case 12). In group 3, the tumours were subjected to preoperative chemotherapy and then enucleated.

Antibodies and chemicals

Mouse monoclonal anti-ABCG2 (5D3); sc-18841 (200 μg/ml) antibody for the detection of ABCG2 protein was obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and also mouse monoclonal antibody MAAB4146 Ms X BRCP (0.25 mg/ml) was obtained from Chemicon International. Primary mouse monoclonal human MCM protein MCM2 (150 μg/ml); clone: CRCT2.1) antibody was obtained from Novocastra, Newcastle, UK. labelled streptavidin biotin (LSAB) kit and developing solution (DAB) were obtained from Dakocytomation Corp, Glostrup, Denmark.

Immunohistochemistry

Paraffin sections of 39 tumour samples of retinoblastoma were washed in phosphate buffered saline (PBS) and incubated for 1 hour with the ABCG2/MCM2 antibody or with 1% BSA-PBS as a negative control. After washing, the slides were incubated with biotinylated anti-mouse immunoglobulin (LSAB) for 30 minutes, washed again, and incubated with horseradish peroxidase conjugated streptavidin for 30 minutes. The reaction was revealed by 3, 3′-diaminobenzidine and counterstained with haematoxylin.

Evaluation of slides

Antigen expression was defined as the presence of membrane staining for ABCG2 and nuclear staining on the tumour cells for MCM2. All stained cells were considered positive, irrespective of staining intensity. Because ABCG2 and MCM2 were expressed heterogeneously, 20 vital tumour fields were evaluated (under 20× magnifications) and a final mean score for each tumour was obtained. The staining was scored as the percentage of positively stained cells. Both ABCG2 and MCM2 were scored on a three tiered scale: – 0 (when ABCG2 or MCM2 negative); 1%–⩽50% cells positive and >50% cells positive for ABCG2 or MCM2. ABCG2 and MCM2 immunoreactivity was correlated with the invasiveness and differentiation of the tumours and in the three groups of tumours.

Statistical analysis

Statistical analyses were performed using the non-parametric Mann-Whitney U test. p Values <0.05 were considered statistically significant. Well differentiated and moderately differentiated tumours were grouped and compared against poorly differentiated tumours.

RESULTS

Tables 1–3 show the clinicopathological information and immunoexpression of ABCG2 and MCM2 in the three groups of tumours. The immunoreactivity data of ABCG2 from two antibodies was concurrent. Figures 1A–C show ABCG2 positivity in the non-neoplastic retina obtained from donor cadaveric eyes and figure 1D–G show ABCG2 immunoreactivity in the retinoblastoma tumour samples. Figures 2A–C show MCM2 positivity in the non-neoplastic retina obtained from donor cadaveric eyes and figure 2D–G show MCM2 immunoreactivity in the retinoblastoma tumour samples.

ABCG2 and MCM2 in non-neoplastic retina

Both ABCG2 and MCM2 were sporadically positive in the ganglion cells, inner nuclear layers, and in the photoreceptor layers in the retina obtained from donor eyes and in the non-neoplastic portion of the retina with retinoblastoma.

ABCG2 expression in retinoblastoma

Among the 20 non-invasive retinoblastomas, ABCG2 was positive in 12 tumours (two tumours with >50% cells positivity and 10 tumours with 1%–⩽50% cells positivity) and negative in eight tumours. Among 19 invasive retinoblastomas, ABCG2 was positive in 15 tumours (10 tumours with >50% cells positivity and five tumours with 1 to ⩽50% cells positivity) and negative in four tumours. Among the six tumours which had either orbital recurrence or metastasis, ABCG2 was positive in six tumours (all >50% cells positivity). The percentage of ABCG2 positive cells was more in the invasive tumours (Mann-Whitney U test; p<0.01).

MCM2 expression in retinoblastoma

Among 20 non-invasive retinoblastomas, MCM2 was positive in 14 tumours (six tumours with >50% cells positivity and eight tumours with 1%–⩽50% cells positive) and negative in six tumours Among the 19 invasive retinoblastomas, MCM2 was positive in 16 tumours (13 tumours with >50% cells positivity, three tumours with 1%–⩽50% cells positivity) and negative in three tumours. Among the six tumours, which had, either orbital recurrence or metastasis MCM2 was positive in five tumours (all >50% cells positivity). The percentage of MCM2 positive cells was more in the invasive tumours (Mann-Whitney U test; p<0.01).

ABCG2 and MCM2 expression in tumours not subjected to preoperative chemotherapy

Among 33 tumours (groups 1 and 2) that had no preoperative chemotherapy, ABCG2 was positive in 22 tumours (11 tumours with >50% cells positivity and 11 tumours with 1%–50% cells positivity) and MCM2 was positive in 28 tumours (18 tumours with >50% cells positivity and 10 with 1%–50% cells positivity).

ABCG2 and MCM2 expression in tumours subjected to preoperative chemotherapy

Among the six tumours, ABCG2 was positive in five tumours (four tumours with 1%–0% cells positivity and one tumour with >50% cells positivity) and negative in one tumour. MCM2 was positive in one tumour (>50% cells positivity) and negative in five tumours.

ABCG2 and MCM2 expression and correlation with differentiation and laterality of tumours

There was no correlation observed with MCM2 expression against differentiation and laterality of tumours.

DISCUSSION

The current study compares the immunoexpression of ABCG2 and MCM2 in human archival retinoblastoma tumours with an earlier study by Siegel et al² where they observed the immunoexpression of ABCG2 and MCM2 in retinoblastoma tumours from mice transgenic for the SV40 T antigen (driven by the β luteinising hormone promoter), cell pellets of human Y79 and WERI-Rb27 retinoblastoma cell lines, and three archival human retinoblastoma pathological specimens. In their study, small numbers of retinoblastoma cells exhibited immunoreactivity to stem cell markers ABCG2 and MCM2. MCM2 positivity was observed in tumour tissue surrounding blood vessels, adjacent to the optic nerve, and in the inner nuclear layer and photoreceptor layers of the retina in eyes with retinoblastoma.

Our results are concurrent with their study in that the human archival retinoblastoma tumour samples showed the immunoexpression of both ABCG2 and MCM2. Both ABCG2 and MCM2 immunoreactivity was observed in both well and poorly differentiated tumours and in tumour cells arranged around blood vessels and also in the invading portion of the tumour cells in the optic nerve and orbit. We also had the opportunity of correlating the expression of these two stem markers with invasiveness and with chemotherapy in our cohort. The percentage of ABCG2 and MCM2 positive cells was more in the invasive tumours (Mann-Whitney U test; p<0.01 for both). There was no correlation with respect to laterality and differentiation of the tumours. These data support the earlier observation that stem cells expressed in retinoblastoma could have a role in tumorigenesis.²

Among the 33 tumours not subjected to preoperative chemotherapy, ABCG2 was positive in 21 tumours and MCM2 was positive in 29 tumours. This observation suggests that retinoblastoma expresses ABCG2 and MCM2 proteins even previous chemotherapy. This could have implications with regard to chemotherapy response. Among the six tumours subjected to preoperative chemotherapy (table 3) ABCG2 was positive in five tumours and MCM2 was positive in one tumour. The increased percentile of ABCG2 positive cells in tumours, which were enucleated after preoperative chemotherapy, suggests that ABCG2 positive cells could escape the chemotherapy drugs. Regarding the correlation of ABCG2 and MCM2 protein with response to chemotherapy, in the 15 tumours in group 2 (table 2) where the patients were subjected to postoperative chemotherapy, ABCG2 was positive in 12 tumours. In the six patients who had developed orbital recurrence or developed metastasis, ABCG2 was positive in all six tumours and MCM2 was positive in two tumours.

Overexpression of ABCG2 has been observed in several human cancer cell lines selected for drug resistance,^3–5,11 as well as in tumour samples of cancer patients.^12,13 Tumorigenic cells with characteristics similar to neural stem cells with MCM2 positivity have been isolated from paediatric brain tumours.⁷ Presence of neural stem cells could increase the aggressiveness of the original tumour. Thus the expression of MCM2 in retinoblastoma is similar to other tumours where MCM2 is seen more in aggressive tumours.^14–18

We also observed the immunoreactivity of ABCG2 and MCM2 in ganglion cells, inner nuclear layer, and the photoreceptor cells of non-neoplastic retinas obtained from donor eyes. ABCG2 and MCM2 positivity was also observed in the retinal tissue adjacent to the tumour. In normal tissue, high expression of the ABCG2 is found in stem cells, epithelial cells of small and large intestines, ducts and lobules of the breast, endothelial cells of veins and capillaries, and synchitiotrophoblastic cells of the placenta.¹⁹ The localisation of ABCG2 suggests that it could have a potential role in protection against toxins.

In summary, we confirm and expand upon the findings that cancer stem cell and drug efflux protein ABCG2 and neural stem cell and proliferative marker MCM2 are expressed in archival retinoblastoma samples. These could add to the existing drug resistance proteins²⁰ in retinoblastoma. However, further studies are needed to understand the role of these stem cells in retinoblastoma and their contribution to drug resistance and tumour progression.