Methods

This retrospective study was conducted under the auspices of the Massachusetts Eye & Ear Infirmary’s Institutional Review Board (IRB protocol number 602340-2, reference number 14-064H), in coordination with those of the Wills Eye Hospital, Duke University, and the University of Heidelberg, in compliance with the US rules and regulations of the Health Insurance Portability and Accountability Act and all applicable US federal and state laws, and in adherence to the tenets of the Declaration of Helsinki.

After a search for the diagnosis of intraocular medulloepithelioma in the regular diagnostic and consultation files in the ophthalmic pathology laboratories of the Massachusetts Eye and Ear Infirmary and the Wills Eye Hospital and in the Duke University Pathology Department from January 1996 to April 2014, 10 cases were found. An additional 10 cases were contributed by the Clinical Cooperation Units of the German Cancer Research Center (DKFZ) and the Department of Neuropathology, Heidelberg University Hospital, Heidelberg, Germany, bringing the total up to 20 cases. Criteria necessary for inclusion in this study were based on the availability of glass slides used in the original diagnosis and of archived blocks for the preparation of additional unstained paraffin sections for immunohistochemical staining or FISH testing.

The histopathologic features displayed in formalin-fixed, paraffin-embedded, and hematoxylin-eosin-stained sections on glass slides were reexamined in all cases (by F.A.J. and A.K.), with complete concordance. The paraffin blocks were used to obtain additional immunohistochemical staining and to perform DNA analysis. Five cases have been separately reported ; the use of a different immunohistochemical stain in the present study and the introduction of DNA analysis with FISH testing and a new investigative focus were the bases for the inclusion of these cases in the present study. Patients’ medical records were evaluated for demographic information. Follow-up information was available for 19 of the 20 cases.

Immunohistochemical staining was performed on all cases employing a polyclonal antibody (A117, #3978, Cell Signaling Inc, Boston, Massachusetts, USA) for the detection of LIN28A. Immunostaining was done using an automated stainer (Benchmark; Ventana XT, Tucson, AZ, USA) with appropriate controls to validate antibody quality. Endogenous peroxidase activity was blocked by H ₂ O ₂ before antibody incubation. The chromogen diaminobenzidine was used and the tissues were counterstained with hematoxylin. Positive staining of LIN28A was cytoplasmic and not nuclear. The staining was graded as weak, moderate, or strong, and as unifocal, multifocal, or diffuse. Such staining was evaluated on the basis of a single slide. It is possible that multiple-step sections throughout a given lesion might have disclosed slightly different results, although most of the tumors were small and therefore the slides were likely to be representative. In parallel, FISH testing was performed on all samples studied for an abnormality in the 19q13.42/ C19MC locus, according to the methodology for FISH previously described. Fluorescein isothiocyanate–labeled locus probe 634C1 (19q13.42/ C19MC) and digoxigenin-labeled reference probe 2658N19 (19p13) were applied. Signals were scored in 200 nonoverlapping intact nuclei. Specimens were considered amplified for C19MC locus when more than 10% of tumor cells showed either more than 8 signals of this locus probe or innumerable tight clusters of signals. The forgoing tests were conducted in the Department of Neuropathology, University Hospital, Heidelberg, Germany.

Results

Patient demographics, salient histopathologic features, and immunohistochemical and genetic findings are summarized in the Table . Of the 20 patients included in the study, 11 were female and 9 were male, ranging in age from 3 months to 53 years (4 were older than 20 years). The median age was 4.5 and the average age was 11.8 years. The right and left eyes were equally affected. All but 1 tumor arose in the ciliary body region; the exception filled the posterior half of the vitreous cavity and was associated with a postequatorial retinochoroidal coloboma. Follow-up information was available on 19 of the 20 cases. Detailed ocular history was also available on approximately 60% of cases. One patient with a prolonged delay in diagnosis experienced spontaneous extraocular extension and eventually died with distant metastases ( Table , Case 19). Eighteen other patients are alive, and 1 patient ( Table , Case 8) was lost to follow-up. Another patient had a lensectomy performed at the time of a subtotal intraocular biopsy of a tumor, just prior to enucleation ( Table , Case 11). None of the other patients were known to have had previous ocular surgery (cataract or glaucoma) prior to enucleation. Of the 10 cases from Europe, however, 1 ( Table , Case 5) had undergone histopathologically detectable evidence of glaucoma drainage device implantation with extraocular extension around the plate.

Histopathologically, all tumors showed a varying prominence of multilaminar rosettes or neurotubules, cords or festoons of cells, and small cellular units that were either solid or displayed different-sized lumens and were distributed in an Alcian blue-positive (hyaluronic acid) matrix. Owing to the vagaries of the selection process for the inclusion of tumors in this study, only 1 lesion in this series ( Table , Case 13) was judged to be teratoid based on the presence of myogenin-positive spindle cells. In consideration of the number of mitotic figures per 5 high-power fields (generally >10 for malignancy), the degree of cytologic atypia, focal necrosis, and evidence of invasion or destruction of intraocular structures ( Figure 1 , Upper left), 16 lesions were diagnosed as malignant according to standard criteria in routine hematoxylin-eosin sections before LIN28A testing was performed. Noncytologic features included the following: iris neovascularization (6 cases); spread onto and into the retina (6 cases), 1 extending backward to the optic nerve head ( Figure 1 , Upper right); ciliary body invasion or destruction (6 cases); tumor extrusion or extension through a corneoscleral rupture ( Figure 1 , Middle left) (4 cases); extensive and subtotal tumor necrosis (4 cases); premedullary cuboidal strands of epithelium, usually located behind or on the iris (4 cases); partial pigmentation (3 cases); iris destruction (3 cases); choroidal invasion (2 cases); retrolenticular neoplastic cyclitic membrane (1 case); pseudoadenomatous hyperplasia of the nonpigmented ciliary epithelium contiguous to the tumor (1 case); and neuropil formation (1 case). Regarding the identification of a neuropil, care had to be taken not to confuse the background of invaded retina or reactive retinal gliosis adjacent to the tumor for this feature (observed in 3 cases).

Histopathologic and LIN28A immunohistochemical findings of intraocular medulloepithelioma. (Upper left) Neurotubular formations of medulloepithelioma display aggressive behavior by invading inner corneoscleral tissue (CS). (Upper right) An epipapillary tumor (EPT) nodule located at the edge of the optic nerve head. The arachnoid space (AR) behind the sclera (S) is widened and the edge of the optic nerve with its dura (arrow) is also identifiable; the meninges are uninvolved by tumor. (Middle left) An intraocular medulloepithelioma (IM) has erupted (arrows) though a rupture in the cornea (C) to create an epibulbar tumor (EBT). The inset discloses that the epibulbar mass is composed of small nests of undifferentiated round cells in a fibrous stroma. (Middle right) LIN28A strongly and uniformly stains the cytoplasm of the tumor cells in this neoplasm. (Bottom left) Top panel: Moderate staining by LIN28A takes the form of some intense foci within a background of weaker positivity. Bottom panel: In these moderately staining round clusters some cells are more positive than others. (Bottom right) Weak LIN28A immunostaining. (Upper left, Upper right, Middle left, hematoxylin-eosin; ×50, ×20, ×12.5; Middle right, Bottom left, Bottom right, immunoperoxidase reaction, diaminobenzidine chromogen, hematoxylin counterstain; ×50, ×100 [top], ×200 [bottom], ×200.)

Nineteen of 20 tumors were immunohistochemically LIN28A positive. The immunostaining was localized to the cytoplasm and was strongly positive in 9 cases ( Figure 1 , Middle right), moderately positive in 6 cases ( Figure 1 , Bottom left), and weakly but unequivocally positive in 4 cases ( Figure 1 , Bottom right). Moderately staining tumors manifested either intermediate staining density or heterogeneous staining with some cells being denser than others ( Figure 1 , Bottom left). The staining was diffuse in 8 cases ( Figure 1 , Middle right) or multifocal ( Figure 2 , Upper left, top panel) or unifocal ( Figure 2 , Upper left, bottom panel) in 11 cases. Staining was totally negative in all components of the 1 case with a permeative neuropil and scattered multilaminar rosettes. In correlating these results with local ocular behavior, as revealed in hematoxylin-eosin-stained sections, large size, retinal extension, invasion of the choroid or ciliary body, iris destruction, and extraocular extension ( Figure 2 , Middle left) were encountered only in those tumors exhibiting diffuse cytoplasmic staining that was either moderately or strongly intense. For example, 3 of 4 tumors with extraocular extension stained strongly positive and 3 diffusely positive. The 3 partially pigmented areas of 3 tumors were LIN28A positive, as determined by a lighter brown staining by the chromogen in comparison with the black melanin granules ( Figure 2 , Middle right, left and right panels). The adjacent pseudoadenomatous hyperplasia of the nonpigmented ciliary epithelium discovered next to a medulloepithelioma did not stain for LIN28A.

Additional LIN28A immunohistochemical findings and fluorescence in situ hybridization (FISH) studies of intraocular medulloepithelioma. (Upper left) Top panel: Multifocal LIN28A positivity. Bottom panel: Unifocal LIN28A positivity on the left in a tumor that is otherwise negative, shown on the right. I, iris. (Upper right) Strong and diffuse LIN28A immunopositivity of the epipapillary nodule illustrated in Figure 1 , Upper right. (Middle left) Strong and diffuse LIN28A immunopositivity of an intraocular tumor (IOT) that has eroded ( arrows ) through a break in the corneal stroma (C) to produce an epibulbar tumor (EBT). This is the same lesion illustrated in Figure 1 , Middle left. (Middle right) Left panel: The tumor cells that are less heavily pigmented display a light brown LIN28A immunopositivity that is not obscured by the dense black melanin collections that many other clusters show. Right panel: A looser collection of cells reveals more clearly the positive light brown immmunoreactivity of LIN28A. This finding establishes that the pigmented cells are truly neoplastic rather than fellow-traveling or reactive nonneoplastic cells. (Bottom left) Absence of C19MC amplification in ocular medulloepithelioma detected by FISH: balanced profile for 19q/19p loci with a similar number of target/reference probes signals. The target C19MC locus on 19q13.42 was labeled in green, and the reference probe located on 19p13 was labeled in red. (Bottom right) An example of C19MC amplification within an intracranial embryonal tumor with multilayered rosettes detected with FISH analysis. The locus of common focal amplification (19q13.42) was labeled in green and disclosed innumerable signals and clouds in comparison to the reference probe (19p13) labeled in red. (Upper left, Upper right, Middle left, Middle right, immunoperoxidase reaction, diaminobenzidine chromogen, hematoxylin counterstain; ×50 [top], ×20 [bottom], ×20, ×20, ×200 [left], ×200 [right].)

By means of FISH testing all tumors were found to be negative for 19q13.42 ( C19MC ) amplification ( Figure 2 , Bottom left). Balanced profile for 19q/19p was detected in all ocular medulloepitheliomas, with a similar number of locus/reference probes signals (from 2 to 4 per nuclei) and with reference/control ratio in a range from 0.8 to 1.3 per tumor. For comparison, an example of a positive FISH test of an embryonal tumor with multilayered rosettes is also illustrated ( Figure 2 , Bottom right).

Ten retinoblastomas, 5 almost completely undifferentiated and 5 with both prominent rosettes and undifferentiated round cells, were evaluated for LIN28A and failed to disclose any immunoreactivity in both components in the cells’ nuclei or cytoplasm. DNA rearrangement at 19q13.42 leading to amplification of C19MC was also not discovered in any case.