Major histocompatibility antigens and antigen-processing molecules in retinoblastoma

Authors


Abstract

BACKGROUND

Malignant transformation of cells is frequently associated with abnormalities in human leukocyte antigen (HLA) expression. These abnormalities may play a role in the clinical course of the disease, because HLAs mediate interactions of tumor cells with cytotoxic T lymphocytes (CTLs) and natural killer (NK) cells. Retinoblastoma is the most common intraocular malignant tumor in childhood and is characterized by direct spread to the optic nerve and orbit as well as hematogeneous and lymphatic spread. Little is known about the role of HLA expression in the progression of this malignant disease.

METHODS

HLA Class I antigen, β2-microglobulin (β2-m), HLA Class II antigens, and the antigen-processing molecules (APMs) of the HLA Class I pathway, including proteasomal subunits (low–molecular mass polypeptide 2 [LMP-2] and LMP-10), the transporter-associated protein (TAP-1) subunit, the binding protein tapasin, and the chaperone molecule calnexin, were studied in 30 archival retinoblastoma specimens by immunohistochemistry. Immunoanalysis was performed based on the International Histocompatibility Working Group Project Description.

RESULTS

HLA Class I antigen, β2-m, HLA Class II antigen, and APMs were positive in 12 tumors with no invasion and were decreased in 13 tumors with choroidal and optic nerve invasion. The difference in HLA and APM expression between the 2 groups was statistically significant (P < 0.001).

CONCLUSIONS

Decreased expression of HLA was observed in aggressive tumors and in poorly differentiated tumors. The current findings support a role for both CTLs and NK cell-mediated control of tumor growth in the clinical course of retinoblastoma. Cancer 2004;100:1059–69. © 2004 American Cancer Society.

Retinoblastoma is the most common primary intraocular tumor in children. In the United States, this disease presents most frequently as unilateral sporadic tumors and less frequently as bilateral hereditary tumors. When it is left untreated, retinoblastoma is almost always fatal. Prognosis is affected by many risk factors, the most important of which is the extent of invasion of the retinoblastoma into ocular coats and the optic nerve.1 Although many tumors are not detected until they are large enough to be visible to the parents, very small tumors sometimes may be treated effectively using laser therapy or cryotherapy. More commonly detected larger tumors often require removal by enucleation of the affected eye. When retinoblastoma is treated in the early stages by enucleation, the cure rates approach 95%.2–4 In addition to the loss of vision, this therapeutic approach may leave the child with a facial deformity that worsens throughout life.5 More advanced disease may require radiotherapy or chemotherapy in addition to enucleation. Both of these additional therapeutic regimes increase the probability that the surviving child will develop additional malignancies later in life. Currently, there is no successful therapy for the treatment of patients who develop metastatic disease. In recent years, however physicians have explored alternative therapies in an attempt to salvage the eyes and vision of the patient.

Recent advances in tumor immunology have provided the necessary information to develop immunotherapies to achieve this goal. There are several reasons why these therapies may be highly successful in treating patients with ocular tumors. Ocular tumors usually are detected early, when the tumor is relatively small. This allows initiation of treatment early in the development of the disease, which is important because it is believed that immunotherapies provide the greatest benefit when the tumor burden is low. Tumors also grow within a contained space, and surgical techniques are available for the delivery of intratumor injection of cytokine and/or lymphocytes.

Major histocompatibility complex (MHC) and antigen-processing molecules (APMs) have generated interest in immunotherapy in many tumors.6 MHC Class I antigens are responsible for presenting antigen to cytotoxic T lymphocytes (CTLs). In fact, reduced or eliminated Class I antigen expression on tumor cells is correlated inversely with their rejection. The pathway of antigen-processing machinery involved in Class I antigen recently was characterized. This pathway involves proteasomal complexes, consisting of low–molecular mass polypeptide 2 (LMP-2), LMP-7, and the recently identified LMP-10 or multicatalytic endopeptidase complex-like-1, which is necessary for LMP-2 expression; all generate antigenic peptide fragments. The transporter-associated proteins (TAP), TAP-1 and TAP-2, and the different endoplasmic reticulum (ER) resident chaperones, such as calnexin, calreticulin, the binding protein BiP, ER60, and tapasin, stabilize MHC Class I molecules during their folding and assembly in the ER or assist in their binding to peptides. Deficiencies in LMP, TAP, and the chaperone proteins reduce the supply and repertoire of peptides available for binding to MHC Class I molecules. Decreased human leukocyte antigen (HLA) Class I antigen levels resulting from a decrease in APMs have been observed in many tumors.7

There have been several studies on HLAs in retinoblastoma. Previous studies focused on cell lines and used monomorphic antibodies W6/32 against HLA-A, HLA-B, and HLA-C heavy chains and BBM.1 against the associated β2-microglobulin (β2-m) light chain and on transgenic animal models of retinoblastoma. Studies showed that both Y-79 and WERI-Rb1 retinoblastoma cell lines exhibited significant levels of HLA-A, HLA-B, and HLA-C antigens; however, over time, subcultured Y-79 tumor cell lines lost Class I surface HLA expression, and there was a paucity of HLA expression in the Rb 355-7 cell line.8 In separate studies, it was found that MHC Class I and Class II antigen expression was negative in the Y-79 cell line but was induced by interferon (IFN) in Y-79 cells.9–11 Similarly, MHC Class I antigens were absent in RbSF81 cells (originally derived from Y-79 cells) but were induced later by retinoic acid.12 It is possible that the morphologic and genetic characteristics of cultures grown for extended periods may have contributed to the variation in the expression of HLAs in those cell lines. Studies in a transgenic mouse model of retinoblastoma showed that tumor cells expressed high levels of MHC Class I antigen and tumor-associated antigen (Tag), which induced a high T-cell response in which CD8 positive (CD8+) T cells recognized peptide fragments of Tag presented by the MHC Class I molecules. Further studies showed that CTLs failed to lyse transgenic retinoblastoma tumor target cells that had down-regulated HLA Class I molecules. However, it also was observed that retinoblastoma cells treated with IFN-γ up-regulated Class I expression and then were lysed by T cells from the transgenic mice.13, 14 Thus, these earlier studies showed that HLAs are expressed in retinoblastoma and that they are immunogenic tumors.

Nonetheless, to our knowledge, APMs and HLAs in retinoblastoma and their correlation with invasiveness have not been investigated. Therefore, in the current study, we investigated the immunoreactivity of HLA Class I antigen, β2-m, HLA Class II antigen, and APM (including proteasomal subunits LMP-2, LMP-10, the transporter protein TAP-1 subunit, and the chaperone molecules tapasin and calnexin) in archival specimens of retinoblastomas and correlated the results with regard to differentiation and invasion.

MATERIALS AND METHODS

Patients

Previously, we published a large series involving 232 Asian Indian children with retinoblastoma.15 In that study, a higher incidence of choroidal and optic nerve infiltration was noted among Asian Indian children than among children from the Western world. We concluded that this finding may have been due to delayed diagnosis or to a difference in the biologic behavior of tumors occurring in the Asian Indian population. Thirty retinoblastoma lesions were obtained from 12 male children and 18 female children with an age range of 1–8 years at the time of enucleation. All patients were evaluated in the Ocular Oncology Clinic at our hospital between 2000 and 2002. The tumors were divided into two groups: Group A tumors, with no invasion of the choroid or the optic nerve and no metastasis; and Group B tumors, with invasion of the choroid, optic nerve, and orbit and with metastasis.

Inclusion and Exclusion Criteria

The inclusion criterion was that all patients were treated by enucleation. Exclusion criteria included patients who had received preoperative adjunctive treatments, such as chemotherapy, which could influence the interpretation of immunohistochemistry.

Tumor Samples

Neoplastic tissues were obtained in enucleation material from the patients. Each sample was processed for conventional histopathologic diagnosis. Histologic sections were prepared from tissues fixed in 10% buffered neutral formalin for 48 hours and embedded in paraffin. Hematoxylin and eosin–stained, 6 μm sections were prepared from the central region of the tumor and were reviewed for differentiation and invasion. Blocks were obtained from consecutive patients.

Differentiation

Retinoblastomas were graded microscopically into three groups according to the predominant pattern of differentiation: 1) poorly differentiated cells, with high nuclear-to-cytoplasmic ratios and high mitotic indices (pseudorosettes and Homer–Wright rosettes were found in such areas); 2) moderately differentiated cells, with moderate nuclear-to-cytoplasmic ratios, moderate mitotic indices, and possible pseudorosettes and Flexner–Winter Steiner rosettes; and 3) well differentiated cells, with low nuclear-to-cytoplasmic ratios, low mitotic indices, and the presence of Flexner–Winter Steiner rosettes and florets. There were 6 well differentiated tumors, 6 moderately differentiated tumors, and 18 poorly differentiated tumors.

Invasion

All tumor slides were reviewed, and choroidal invasion was staged according to the new grading system devised by Schilling et al.16

Staging of Choroidal Invasion

Choroidal invasion was staged as follows: Stage 0, no changes in retinal pigment epithelium and choroids; Stage 1, changes in retinal pigment epithelium without involvement of Bruch membrane; Stage 1a, simple defects of the retinal pigment epithelium cell layer; Stage 1b, tumor growth under the elevated retinal pigment epithelium cell layer with or without reactive cell proliferation; Stage 2, complete defect of the retinal pigment epithelium, including the basement membrane, without tumor invasion into the choriocapillaries; Stage 3, proof of tumor cells in the choriocapillaries and superficial infiltration around the large choroidal vessels; and Stage 4, complete infiltration of the choroid with lateral extension.

Optic Nerve Invasion

For optic nerve invasion analyses, prelaminar invasion, postlaminar invasion, and invasion of the surgical end of the optic nerve were considered. There were 12 tumors with no invasion and 18 tumors with invasion. Three tumors had invasion of the choroid only, six tumors had invasion of the choroid and the optic nerve, and nine tumors had invasion of the optic nerve. There was one tumor that had metastasized to the submandibular lymph node. The various stages of choroidal invasion and optic nerve invasion are shown in Table 1, and the immunochemistry results are shown in Table 2.

Table 1. General Characteristics of the Cohort with Retinoblastoma Exhibiting Invasion
Patient no.Age (yrs)GenderClinicopathologic parametersChoroid invasionOptic nerve invasion
  1. Mod: moderately; diff: differentiated; OS: left eye; OD: right eye; OU: both eyes; OD: right eye.

17.0FemaleOS, mod diffSurgical end
21.0MaleOD, poorly diffPrelaminar
32.0FemaleOD, poorly diffStage 4Surgical end
44.0MaleOU, OS, poorly diffStage 4Surgical end
54.0FemaleOD, mod diffLamina cribrosa
62.0MaleOD, poorly diffLamina cribrosa
71.1 mosFemaleOD, poorly diffStage 4Postlaminar
83.0MaleOD, poorly diffPostlaminar
94.0FemaleOD, mod diffStage 3
102.5FemaleOD, poorly diffStage 4
112.0FemaleOD, poorly diffPostlaminar
123.0MaleOD, poorly diffPostlaminar
135.0MaleOD, poorly diffStage 4Surgical end
145.0FemaleOU, (OD) well diffStage 3Prelaminar
152.0FemaleOU, (OS) poorly diffStage 4
162.0MaleOD, poorly diffPostlaminar
171.0FemaleOU, (OD) well diffLamina cribrosa
182.0FemalePoorly diff metastasis to submandibular lymph nodeStage 4Postlaminar
Table 2. Immunohistochemistry Results from Group B: Retinoblastoma with Invasion
Patient no.% stained cells (staining intensity)
HLAProteasomeTransporter proteinChaperone protein
HLA Class Iβ2-mHLA Class IILMP-2LMP-10TAP-1TapasinCainexin
  1. HLA: human leukocyte antigen; LM: low–molecular mass polypeptide; β2-m: beta 2 microglobulin; TAP: transporter-associated protein; (−): absent staining intensity; (+/−): dull staining intensity.

1< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 ()
240 (+/−)40 (+/−)30 (+/−)40 (+/−)30 (+/−)40 (+/−)50 (+/−)40 (+/−)
30 (−)0 (−)0 (−)0 (−)0 (−)< 25 (+/−)< 25 (+/−)40 (+/−)
4< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)30 (+/−)30 (+/−)
550 (+/−)50 (+/−)40 (+/−)50 (+/−)30 (+/−)50 (+/−)50 (+/−)50 (+/−)
640 (+/−)40 (+/−)30 (+/−)40 (+/−)60 (+/−)30 (+/−)60 (+/−)40 (+/−)
7< 25 (+/−)< 25 (+/−)< 25 (+/−)0 (−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)
8< 25 (+/−)< 25 (+/−)0 (−)< 25 (+/−)0 (−)< 25 (+/−)< 25 (+/−)< 25 (+/−)
9< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)
10< 25 (+/−)< 25 (+/−)< 25 (+/−)0 (−)< 25 (+/−)0 (−)< 25 (+/−)< 25 (+/−)
11< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)0 (−)< 25 (+/−)
1250 (+/−)50 (+/−)40 (+/−)50 (+/−)30 (+/−)40 (+/−)50 (+/−)50 (+/−)
130 (−)0 (−)10 (+/−)0 (−)0 (+/−)0 (−)< 25 (+/−)< 25 (+/−)
14< 25 (+/−)< 25 (+/−)< 25 (+/−)< 25 (+/−)0 (−)< 25 (+/−)< 25 (+/−)< 25 (+/−)
1550 (+/−)50 (+/−)60 (+/−)30 (+/−)50 (+/−)40 (+/−)50 (+/−)50 (+/−)
1650 (+/−)50 (+/−)50 (+/−)50 (+/−)40 (+/−)30 (+/−)50 (+/−)50 (+/−)
1740 (+/−)40 (+/−)40 (+/−)30 (+/−)40 (+/−)30 (+/−)40 (+/−)40 (+/−)
180 (−)0 (−)0 (−)0 (−)0 (−)0 (−)0 (−)0 (−)

Monoclonal Antibodies

The affinity purified mouse antihuman, locus-specific monoclonal antibody (mAb) HC-10, which recognizes a determinant expressed on HLA-A10, HLA-A28, HLA-A29, HLA-A30, HLA-A31, HLA-A32, and HLA-A33 heavy chains and on virtually all HLA-B heavy chains; the anti-β2-m mAb L368; the anti-HLA Class II mAb LGII-612.1417–19; and the APMs in the HLA Class I pathway, including the anti-LMP-2 mAb SY-1, the anti-LMP-10 mAb TO-6, the anti-TAP-1 mAb TO-1, the anti-tapasin mAb TO-4, and anticalnexin mAb TO-5, which are target-specific and exhibit no cross reactivity, were used.20 The antibodies were gifts from Dr. Soldano Ferrone (Department of Immunology, Roswell Park Cancer Institute, Buffalo, NY). A labeled streptavidin kit was purchased from DAKO Laboratories (Copenhagen, Denmark).

Immunohistochemistry

Immunostaining of tissue sections was performed using labeled streptavidin with an indirect immunoperoxidase technique. Briefly, 4-μm, formalin fixed paraffin sections were used for the study. Tissue sections were then deparaffinized and rehydrated. Endogenous peroxidase was blocked with hydrogen peroxide for 10 minutes at room temperature. Antigen retrieval was performed before antibody incubation using a microwave oven. Tissue sections were then rinsed in Tris-buffered saline (TBS), pH 7.6, and incubated with respective primary antibodies over night at 4 °C. This was followed by a sequential, 40-minute incubation with biotinylated secondary antibody and streptavidin-labeled horseradish peroxidase (DAKO Laboratories). Sections were washed with TBS between incubations. The peroxidase reaction was developed for 5 minutes using commercially available 3,3′-diaminobenzidine and was counterstained with Harris hematoxylin.

Assessment of Immunohistochemical Results

Tissue sections were read independently by two ocular pathologists (S.K. and J.B.), each without knowledge of the results obtained by the other investigator. Furthermore, each investigator read all slides twice without the knowledge of the results obtained in the previous reading. The staining intensity was scored as absent (−), dull (+/−), or bright (+). The tumors were graded as follows: positive (> 75% of cells stained with + intensity), heterogeneous (25–75% of cells stained, mainly at +/− intensity; percentages of cells are expressed to the nearest 10%), and negative (− staining or < 25% of cells stained with +/− intensity).21

Variations in the percentage of stained cells enumerated by the 2 investigators were smaller than 10%. The staining intensity of adjacent normal structures (i.e., lymphoid and endothelial cells) was used as an internal control to evaluate the staining intensity of malignant cells. For negative controls, the primary antibody was omitted, and nonimmune serum was used in the immunostaining. The study was reviewed and approved by the local ethics committee of our institute, and the committee deemed that it conformed to the generally accepted principles of research, in accordance with the Helsinki Declaration

Statistical Analysis

For statistical analyses, negative, heterogeneous, and positive expression of HLAs and APM expression were compared between tumors with no invasion and tumors with invasion using the Mann–Whitney U test.

RESULTS

The clinical and immunohistochemical data on patients who had retinoblastoma with invasion are shown in Tables 1 and 2, and the data for patients who had retinoblastoma without invasion are shown in Tables 3 and 4.

Table 3. General Characteristics of the Cohort with Retinoblastoma Exhibiting No Invasion
Patient no.Age (yrs)GenderClinicopathologic parameters
  1. OU: both eyes; OD: right eye; OS: left eye.

12.0MaleOU, (OD) well differentiated
25.0MaleOS, poorly differentiated
31.0FemaleOD, well differentiated
42.0MaleOU, (OD) well differentiated
58.0MaleOS, poorly differentiated
64.0FemaleOS, poorly differentiated
71.0FemaleOS, poorly differentiated
85.0FemaleOD, poorly differentiated
91.0FemaleOU, (OD) moderately differentiated
102.0FemaleOU, OD, moderately differentiated
112.0FemaleOU, OS, moderately differentiated
121.5MaleOD, well differentiated
Table 4. Immunohistochemistry Results for Group A: Retinoblastoma with No Invasion
Patient no.HLA antigensProteasomeTransporter proteinChaperone protein
HLA Class Iβ2-mHLA Class IILMP-2LMP-10TAP-1TapasinCainexin
  1. HLA: human leukocyte antigen; LM: low–molecular mass polypeptide; β2-m: beta 2 microglobulin; TAP: transporter-associated protein; Pos: positive.

1PosPosPosPosPosPosPosPos
2PosPosPosPosPosPosPosPos
3PosPosPosPosPosPosPosPos
4PosPosPosPosPosPosPosPos
5PosPosPosPosPosPosPosPos
6PosPosPosPosPosPosPosPos
7PosPosPosPosPosPosPosPos
8PosPosPosPosPosPosPosPos
9PosPosPosPosPosPosPosPos
10PosPosPosPosPosPosPosPos
11PosPosPosPosPosPosPosPos
12PosPosPosPosPosPosPosPos

Immunoreactivity of HLA Class I Antigens and β2-m in Retinoblastoma

Immunoreactivity for HLA Class I antigens and β2-m was concurrent. HLA Class I antigens and β2-m stained positively (> 75% of cells stained with + intensity) in the 12 tumors in Group A with no invasion. In Group B (n = 18), heterogeneous expression was observed in 8 tumors (30–50% of cells stained with +/− intensity) and was not observed in 10 tumors (7 tumors with < 25% of cells stained with +/− intensity and 3 tumors with absent staining intensity).

Immunoreactivity of HLA Class II Antigens in Retinoblastoma

HLA Class II antigen staining was positive (> 75% of cells stained with + intensity) in the 12 tumors in Group A with no invasion. In Group B (n = 18), heterogeneous expression was observed in 7 tumors (30–50% of cells stained with +/− intensity) and was negative in 11 tumors (8 tumors with < 25% of cells stained with +/− intensity and 3 tumors with absent staining intensity).

Immunoreactivity of Proteasomal Subunits in Retinoblastoma

Expression levels of LMP-2 and LMP-10 were concordant. LMP-2 and LMP-10 expression levels were positive in 12 tumors in Group A (> 75% of cells stained with + intensity). In Group B (n = 18), heterogeneous expression was observed in 7 tumors (30–50% of cells stained with +/− intensity). LMP-2 was negative in 11 tumors (6 tumors with < 25% of cells stained with +/− intensity and 5 tumors with absent staining intensity). LMP-10 was negative in 11 tumors (5 tumors with < 25% of cells stained with +/− intensity and 6 tumors with absent staining intensity).

Immunoreactivity of the TAP-1 Subunit in Retinoblastoma

TAP-1 expression was positive in 12 tumors in Group A (> 75% cells stained with + intensity). In Group B (n = 18), heterogeneous expression was observed in 7 tumors (30–50% of cells stained with +/− intensity) and was negative in 11 tumors (8 tumors with < 25% of cells stained with +/− intensity and 3 tumors with absent staining intensity).

Immunoreactivity of Chaperone Molecules in Retinoblastoma

Tapasin and calnexin expression levels were positive in 12 tumors in Group A (> 75% of cells stained with + intensity). In Group B (n = 18), tapasin exhibited heterogeneous expression in 8 tumors (30–50% of cells stained with +/− intensity) and no expression in 10 tumors (8 tumors with < 25% of cells stained with +/− intensity and 3 tumors with absent staining). Calnexin exhibited heterogeneous expression in 9 tumors (30–50% of cells stained with +/− intensity) and no expression in 9 tumors (8 tumors with < 25% of cells stained with +/− intensity and 1 tumor with absent staining intensity).

Correlation of Immunoreactivity of HLA Class I Antigens and LMP-2, LMP-10, TAP-1, Tapasin, and Calnexin in Retinoblastomas

The expression levels of HLA Class I antigens and of LMP-2, LMP-10, TAP-1, tapasin, and calnexin were concordant in most of tumors from Group A and Group B. Staining for HLA Class I antigens and LMP-2, LMP-10, TAP-1, tapasin, and calnexin was positive in tumors without invasion and was decreased in tumors with invasion of the choroid and optic nerve. The difference in HLA and APM expression between the 2 groups was statistically significant (P < 0.001). Figure 1A–D shows positive immunoreactivity findings for HLA Class I, HLA Class II, LMP-2, and TAP-1; Figure 1E–H shows heterogeneous immunoreactivity findings for HLA Class I, LMP-2, and TAP-1; and Figure 1I,J shows negative immunoreactivity findings for HLA Class I antigens.

Figure 1.

Immunohistochemistry of retinoblastoma. (A) Microphotograph showing the positive immunoreactivity of human leukocyte antigen (HLA) Class I antigens, with bright intensity in tumor cells with no invasion (3,3′-diaminobenzidine [DAB] chromogen with hematoxylin counterstain). (B) Microphotograph showing the positive immunoreactivity of HLA Class II antigens, with bright intensity in tumor cells with no invasion (DAB chromogen with hematoxylin counterstain). (C) Microphotograph showing the positive immunoreactivity of low–molecular mass polypeptide 2 (LMP-2) in tumor cells arranged around the blood vessels (DAB chromogen with hematoxylin counterstain). (D) Microphotograph showing positive immunoreactivity of the transporter-associated protein 1 (TAP-1) subunit in tumor cells of well differentiated retinoblastoma with no invasion (DAB chromogen with hematoxylin counterstain). (E) Microphotograph showing the heterogeneous immunoreactivity (50% of cells stained with dull intensity) of HLA Class I antigens in tumor cells arranged around the blood vessels in retinoblastoma with choroidal invasion (DAB chromogen with hematoxylin counterstain). (F) Microphotograph showing the heterogeneous immunoreactivity (40% of cells stained with dull intensity) of LMP-2 in retinoblastoma with choroidal invasion (DAB chromogen with hematoxylin counterstain). (G) Microphotograph showing the heterogeneous immunoreactivity (50% of cells stained with dull intensity) of the TAP-1 subunit in retinoblastoma with choroidal invasion (DAB chromogen with hematoxylin counterstain). (H) Microphotograph showing the heterogeneous immunoreactivity (40% of cells stained with dull intensity) of the TAP-1 subunit in tumor cells invading the optic nerve (DAB chromogen with hematoxylin counterstain). (I) Microphotograph showing the negative immunoreactivity of HLA Class I antigens in tumor cells invading the postlaminar portion of the optic nerve (DAB chromogen with hematoxylin counterstain). (J) Microphotograph showing the negative immunoreactivity of HLA Class I antigens in tumor cells at the metastatic site (submandibular region; DAB chromogen with hematoxylin counterstain). Original magnification ×20 (C); ×40 (D,F,G); ×100 (A,B,H–J); ×200 (E).

DISCUSSION

The current study is the first to examine HLAs and APMs on archival paraffin sections in retinoblastoma, with results analyzed for correlations in terms of invasion of the choroid and the optic nerve. Our results are in accordance with the results from previous studies on HLAs in retinoblastoma.8–14 The current study demonstrates that HLA Class I, β2-m, and HLA Class II antigens are expressed in retinoblastomas with no invasion and exhibit decreased expression when invasion of the choroid and the optic nerve are present. The difference in immunoreactivity was significant (P < 0.001). There was no correlation with tumor differentiation. The components of the antigen-processing pathway exhibited concurrent expression with HLA Class I antigen. Similar decreased expression of HLA Class I antigen with decreased expression of the components of the APMs has been reported in numerous other tumors7 and also in uveal melanoma, which is another type of intraocular tumor found in adults.22

Thus, APMs are necessary for efficient peptide delivery for MHC Class I antigen expression. These linked patterns of expression have led to the use of the term HLA Class I coordinome to indicate a set of collaborating gene products. The coordinome is onionlike and is organized hierarchically in multiple layers, suggesting the description of global changes in the antigen-processing machinery of neoplastic cells.23

The majority of invading tumors in our cohort expressed HLA Class I antigen and members of the antigen-processing machinery at low levels. This gives the tumor an advantage, because retinoblastoma has multiple modes of invasion and metastasis, unlike uveal melanoma, which metastasizes hematogenously to the liver.24 Human retinoblastomas exhibit four patterns of invasion and metastasis: 1) direct, invasive spread along the optic nerve to the brain, which also can seed the orbital tissue and adjacent bone, the nasopharynx via the sinuses, or the cranium via the foramina; 2) tumor cells that have invaded the optic nerve and leptomeninges and then disperse into the circulating subarachnoid fluid are characteristic of the second pattern of metastasis; 3) hematogenous dissemination that results in widespread metastasis to the lungs, bones, brain, and other viscera (metastasis after orbital invasion and, to a lesser degree, choroidal invasion often is through this route); and 4) lymphatic spread, which occurs when the tumor is located anteriorly or when massive extraocular invasion has occurred. Only tumors with these characteristics can spread through the lymphatic system, because there are no lymphatic vessels in the eye or orbit. When disease reaches the regional lymph nodes, hematogenous spread also can occur.

Thus, both CTLs and natural killer (NK) cells must play a role in the spread of retinoblastoma. MHC Class I antigen down-regulation in tumor cells is an important factor in the ability of these cells to escape recognition and destruction by HLA Class I antigen-restricted, tumor-associated, antigen-specific CTLs.25 In contrast, MHC Class I antigen loss may result in increased sensitivity of malignant cells to NK cells.26 However, the majority of invading tumor cells had some amount of HLA expression in the current study. A global loss of MHC Class I alleles, which occurs infrequently in tumors, may be due to the expression of killer-inhibitory receptors on NK cells.27 Killer-inhibitory receptor molecules recognize MHC Class I, and when they engage their ligand molecules, NK cell–mediated lysis is inhibited.28, 29 Therefore, a tumor cell that is completely devoid of MHC Class I expression may be able to evade specific CTL recognition but would remain a good NK cell target. Some tumors seem to lose expression of single HLA-A or B alleles while retaining the expression of other HLA alleles.30–32 Such a partial HLA loss may be advantageous to the tumor, because it potentially allows for the escape of tumor cells from CTLs directed against immunodominant epitopes, presented by the lost HLA allele, while maintaining resistance to NK cell lysis.

Thus, tumors only gain time for the accumulation of critical mutations and/or derangements in the expression of malignancy-causing genes, which eventually will provide these tumors with uncontrollable invasive potential. Early-stage tumors manage to survive the immune attack by maneuvering HLA expression levels within a rather narrow optimal window and by modifying the expression of the individual HLA alleles, LMP, TAP, and tapasin without deviating excessively from the balance inherited from normal cells.

In the current study, HLA Class II antigens were detected in retinoblastoma localized to the eye with no invasion. HLA Class II molecules play an important role in tumor progression, due to their ability to present antigenic tumor peptides to T lymphocytes. In this regard, experimental findings have shown that selected Tags are presented effectively by HLA Class II antigens to CD4+ T cells. Furthermore, it has been demonstrated that solid tumor cells themselves can act as nonprofessional antigen-presenting cells (APCs) and are able to stimulate directly the proliferation of autologous CD4+ T cells. However, it also has been suggested that tumor cells drive T cells in an anergy state that results in immunosuppression of tumor-reactive T-cell clones.33

HLA Class II antigens have been found in a number of malignancies, albeit with variable frequency. Solid malignancies of different histology can express HLA Class II antigens. However, limited data are available on their functional role. HLA Class II antigen expression is associated with poor prognosis and disease progression in patients with cutaneous melanoma34 and osteogenic sarcoma.35 However, an association with improved prognosis has been described in patients with squamous cell carcinoma, breast carcinoma, colorectal carcinoma, cervical carcinoma, laryngeal carcinoma, and conjunctival carcinomas.36–42

In the current study, we observed low HLA Class II expression levels in all invasive tumors. What are the possible implications of these low expression levels in retinoblastoma cells? HLA Class I expression on tumor cells is essential for killing by CD8+ cytotoxic T cells. Nonetheless, recent studies also suggest an important role for CD4+ helper-T cells and, consequently, for HLA Class II expression on tumor cells.43–45 In addition, cytotoxic CD8+ cells are effective only after activation by CD4+ helper-T cells (i.e., when both CD8+ and CD4+ cells recognize tumor-specific antigenic determinants on the same APCs).46 Thus, loss of both HLA Class I and Class II expression may result in impaired activation of cytotoxic CD8+ T cells and may facilitate the growth of tumor cells. This finding suggests that not only host factors but also tumor cell characteristics, such as loss of HLA Class I and II expression, contribute to the phenomenon of immune privilege.

To date, the molecular basis of low HLA Class I antigen expression in retinoblastoma has not been elucidated. In this regard, the concordant expression of HLA Class I antigen and APMs in our study suggests defects in the regulatory mechanisms that control their expression. However locus-specific down-regulations may be mediated transcriptionally or may be caused by genetic defects. Immunohistochemical defects cannot easily discriminate between these two possibilities. Analysis using cell lines have shown that MHC Class I and Class II antigen expression was up-regulated after retinoic acid and IFN, suggesting that transcriptional mechanisms are implicated in the modulation of HLA expression in retinoblastoma.

Although our understanding of the genetic events associated with retinoblastoma metastasis is beginning to emerge, information on the mechanisms of tumor invasion and metastasis, which may allow the development of new diagnostic and therapeutic strategies to counteract the development of metastatic disease, remains limited. Several factors, such as early genetic events (including increased copy numbers of chromosomes 6p and 1q) and late events (such as high levels of telomerase activity, loss of chromosome 1p, and p53 inactivation), contribute to the aggressiveness of retinoblastoma. In addition, increased angiogenesis, deregulation of cell-to-cell adhesion molecules, and changes in integrins also contribute to tumor aggressiveness.47 In our earlier work on retinoblastoma, we observed that levels of Fas (APO-1 or CD95, a type I membrane protein of 43 kD and a member of the tumor necrosis factor receptor family),48 nm23 protein (a metastasis suppressor protein),49 and tetraspanin protein KAI1/CD82 were decreased in retinoblastoma exhibiting invasion of the choroid and the optic nerve.50

Because the eye is an immune privileged site, tumors must use multiple mechanisms to escape from immune-mediated rejection. These include ignorance, impaired antigen presentation, expression of immunosuppressive factors and molecules, tolerance, apoptosis resistance, and counterattack. Our preliminary data suggest that by expressing decreased levels of HLAs on their surfaces, aggressive tumor cells escape recognition and destruction by HLA Class I antigen-restricted, tumor-associated, antigen-specific CTLs, and they also escape NK cells in traveling through the blood stream.

Clinically, HLAs are of limited interest compared with known histologic factors; such as invasion of the choroid, optic nerve, and orbit; which currently are used to generate prognoses for patients with retinoblastoma. Nonetheless, biologically, our findings suggest a potential role for HLAs in tumor progression. Thus, further studies are required to elucidate the mechanism of expression of HLAs in retinoblastoma. Resolution of this issue is of great importance to the development of potential therapeutic regimens.

Ancillary