Clinical and imaging features
Eight neuroepithelial tumors with PLAG-family genetic alterations were encountered at St. Jude (EWSR1-PLAGL1 fusion n = 6; PLAGL1 amplification n = 1; PLAGL2 amplification n = 1), with the clinical and imaging features summarized in Table 1 and Supplementary Table 1. Supratentorial PLAGL1-fused cases (4F:2 M) were detected by RNA sequencing with a corresponding methylation profile of neuroepithelial tumor PLAGL1 fused (DKFZ 12.5, Table 2), ranging in age from 9 months to 14 years at time of initial diagnosis. Amplified cases included a cerebellar mass with PLAGL2 amplification in a 2-year-old female, and a left temporal mass with PLAGL1 amplification in a 4-year-old male.
All of the PLAG-family tumors in this series were well-defined with involvement of the adjacent dura/meninges, occurring as large (> 5 cm) supratentorial tumors, except for the one tumor with PLAGL2 amplification located in the cerebellum. The shared imaging findings included heterogeneous enhancement and diffusion restriction of the solid component, except for patient #5 whose tumor was multicystic [Fig. 1]. Hemorrhage was present in 6/8 cases, peritumoral edema was present in 3/8 cases, and a cystic component was present in 3/8 cases. Four of the 8 tumors had associated bone remodeling (either thinning or osteopenia).
Representative pre-operative brain MRI images of neuroepithelial tumors with PLAG-family genetic alterations are shown. Cases #1–6 have EWSR1-PLAGL1 rearrangement, while Case #7 and #8 have PLAGL2 amplification or PLAGL1 amplification respectively. The top row demonstrates post-contrast T1-weighted sequences, the middle row demonstrates T2-weighted sequences, and the bottom row shows apparent diffusion coefficient maps. Heterogeneous enhancement and diffusion restriction of the solid component are the most common findings, except for tumor #5, which was multicystic and showed no enhancement or diffusion restriction. Tumor cysts were present in case #5, 6, and 8. Additional imaging details are available in supplementary Table 1
Histologic and immunohistochemical features
The EWSR1-PLAGL1 cases appeared predominantly as solid glial neoplasms, with areas of infiltration. In a few of the cases clear demarcation from adjacent brain parenchyma could be seen in selected areas. The architecture was either ependymal-like with perivascular anucleate zones (perivascular pseudorosettes), subtly ependymal with less pronounced perivascular zones, or in one case a combination of ependymal-like areas and areas with nuclear clusters reminiscent of an ependymoma/subependymoma.
While there was not an appreciable ganglion cell component on initial resection for case #1, which was cellular with perivascular anucleate zones [Fig. 2L], occasional dysmorphic ganglion cells were seen within the tumor on the recurrences. The degree of ganglion cell involvement on initial resection varied [Fig. 2, Table 2] from prominent clusters of ganglion cells within ependymoma/subependymoma-like areas [Case #4, Fig. 2A–E], ganglion cells in in areas of lower-cellularity with adjacent areas of increased cellularity and smaller cells with ependymal features [Case #3, Fig. 2H–I], to interspersed ganglion cells of low to moderate frequency among otherwise subtle ependymal-like histology [Case #5, Fig. 2F–G], and focal areas containing ganglion cells within an ependymal-like neoplasm [Case #6, Fig. 2J–K].
Histology of neuroepithelial tumors with EWSR1-PLAGL1 rearrangement. Concurrent ependymal and ganglionic differentiation was observed on initial resection in 4/6 cases with EWSR1-PLAGL1 rearrangement in this series. The most prominent example occurred in a tumor with combined ependymoma/subependymoma-like histology (Case #4) including areas with intermixed clusters of ganglion cells (A–E). Paranuclear dot-like staining for EMA was only convincingly present in Case #4 (B). Case #5 also showed lower-grade histologic features and had interspersed dysmorphic ganglion cells of low to moderate frequency among otherwise subtly ependymal-like histology; the ganglion cells were small in size (F–G). Case #3 contained ganglion cells in regions of lower-cellularity with occasional microcalcifications, the adjacent areas with greater ependymal quality showed smaller sized cells and variable cellularity (H–I). Case #6 was predominantly ependymal-like with focal areas containing ganglionic cells, there was regionally increased cellularity (J–K). Case #1 lacked a ganglion cell component on initial resection, and showed higher-grade histologic features with increased cellularity (L), elevated Ki-67 labeling and mitotic index, as well as areas of necrosis
Case #2 [Fig. 3A–E] was also a solid cellular predominantly glial appearing neoplasm with ependymal-like architecture [Fig. 3A], but was distinctive in that there was subclonal INI1 loss by immunostaining [Fig. 3C]. The area of INI1 loss involved a more cellular area in which there were small clusters of embryonal/rhabdoid cells [Fig. 3B]. This patient developed multifocal intracranial disease with leptomeningeal involvement. Subsequent resection showed a prominent ganglion cell component at the cortical surface [Fig. 3D], and intermixed smaller cells that were similar in appearance to those of the prior ependymal-like areas but with slightly greater pleomorphism. The vast majority of tumor cells at recurrence, including the ganglion cells and the small cell component, showed INI1 loss [Fig. 3E]. The development of ganglion cells on recurrence in this case may have in-part been treatment related. The region of INI1 inactivation could indicate either ATRT transformation within an ependymal-like neoplasm, or an acquired accompanying mutation associated with tumor progression in an EWSR1-PLAGL1 neuroepithelial tumor. Methylation profiling separately performed in the region of INI1 loss demonstrated a low subthreshold calibrated score of 0.39 for the methylation class atypical teratoid rhabdoid tumor SHH activated, further supporting a transitional area of the neoplasm with a greater degree of epigenetic homology for AT/RT. The methylation class neuroepithelial tumor PLAGL1 fused was not represented among the lower calibrated score results in this area of the tumor, though an extremely low score for CNS embryonal tumor with PLAG family amplification was found.
Neuroepithelial tumor with EWSR1-PLAGL1 fusion and subclonal INI1 loss. On initial resection of case #2 a majority of the neoplasm demonstrated ependymal-like architecture (A). However, a selected region of the tumor with increased cellularity contained small clusters of embryonal or rhabdoid cells (B) and showed loss of INI1 by immunostaining (C). The patient received chemotherapy per ACNS0334 and developed multifocal intracranial disease, with additional resection at 7 months showing dense leptomeningeal involvement including a prominent ganglion cell component (D). There was loss of INI1 staining in most neoplastic cells of the second resection, including the small cell component and ganglionic cells (E). The patient died of disease progression 11 months after initial resection
In 2/6 EWSR1-PLAGL1 fusion cases (#4, #5) mitotic activity was low with infrequent or inconspicuous mitotic figures, which correlated with the absence of necrosis, and lower Ki-67 labeling index (ranging from 5 to 10%). In contrast, 3/6 cases (#1, #2, #6) had an elevated mitotic index with 12, 16, or 19 mitotic figures per 10 high power fields, areas of non-palisading necrosis, and a higher Ki-67 labeling index at or above 30%. While case #3 had variable areas of mitotic activity and Ki-67 on the initial resection and was histologically considered a high-grade neuroepithelial tumor or potentially a type of anaplastic ganglioglioma, the two subsequent resections had lower-grade histologic features. Microvascular proliferation, though present in three cases at least focally, wasn’t a prominent finding in any of the cases. The overall histologic impression when considering cellularity, Ki-67 labeling, mitotic index, and areas of necrosis supports a lower-grade neuroepithelial tumor in 2/6 cases (#4, #5), a higher-grade neuroepithelial tumor in 3/6 cases (#1, #2, #6), and in one case intermixed low and higher-grade areas (#3). The mitotic activity did not specifically correlate with age at diagnosis, as elevated mitotic activity was observed both in patients less than 1 year of age and at 11 years of age.
In EWSR1-PLAGL1 fusion cases the absence of OLIG2 and SOX10 staining with relatively solid growth pattern are ependymal-like features, though ependymoma type EMA staining was only convincingly present in one case [Fig. 2B] with focal/rare paranuclear dot-like staining in two cases. Neurofilament staining supported a predominantly solid growth pattern in 3/5 cases. There was a mixed solid and infiltrative pattern initially in case #3, with more solid growth in subsequent resections. In case #5 neurofilament and synaptophysin staining suggested both areas of infiltration and a component of neural antigen expression within the tumor. L1CAM did not demonstrate significant strong or diffuse staining within the four cases tested. Desmin staining in two of the higher-grade appearing PLAGL1 rearranged cases showed positivity in a smaller to intermediate percentage of neoplastic cells and cellular processes.
While areas of the PLAGL1 amplified tumor showed strong positivity for GFAP or synaptophysin, sometimes with overlapping areas of staining, other areas of the tumor were largely negative for both lineage markers with small clusters or occasional cells showing positivity. Despite this variation for GFAP and synaptophysin staining, desmin was positive in a significant percentage of neoplastic cells throughout the neoplasm [Fig. 4E], with rare cells showing densely eosinophilic cytoplasm or elongation [Fig. 4F]. The histology varied from crowded primitive appearing cells with hyperchromatic nuclei [Fig. 4A–B], to glial areas with a greater degree of cytoplasm [Fig. 4C], and less prevalent spindled areas [Fig. 4D]. This histology and staining pattern is similar to that described by Keck et al. [8], which reported desmin staining in 9/12 tumors ranging from rare positive cells to diffuse strong positivity. The PLAGL2 amplified solid cellular tumor similarly showed desmin staining in a larger subset of neoplastic cells [Fig. 4I], with a greater degree of desmin positivity than for either GFAP or synaptophysin. The PLAGL2 histology showed a vaguely perivascular arrangement of smaller embryonal appearing cells with hyperchromatic nuclei [Fig. 4G–H]. A similar loose perivascular/pseudopapillary architecture of primitive cells was also seen within areas of the PLAGL1 amplified case [Fig. 4A]. Mitotic activity was elevated within primitive areas in both PLAG amplified cases, with 16 or 19 mitoses per 10 high power fields respectively.
Histology of high-grade neuroepithelial/embryonal tumors with PLAGL1 or PLAGL2 amplification. Case #8 (PLAGL1 amplification) demonstrated divergent differentiation with a prominent embryonal component, glial elements, and myogenic differentiation (A–F). A loose perivascular/pseudopapillary architecture of primitive cells was seen in selected areas (A), with higher magnification showing small hyperchromatic crowded nuclei with nuclear molding (B). The histology varied with other areas having a greater degree of eosinophilic cytoplasm and glial fibrillarity (C) or spindled cells in fascicular arrangement (D). Desmin immunostaining was positive in a significant percentage of neoplastic cells (E), with myogenic qualities in rare cells with densely eosinophilic cytoplasm or elongation (F). Case #7 (PLAGL2 amplification) showed primitive cells with a vaguely perivascular arrangement and occasional foci of myxoid material (G–H); desmin positivity also involved a significant percentage of neoplastic cells (I)
Molecular features
For cases with EWSR1-PLAGL1 fusion, in 5/6 cases EWSR1 (NM_05243.4) exon 8 was fused to PLAGL1 (NM_001080951.2) exon 8, with one case having fusion of EWSR1 (NM_05243.4) exon 7 to PLAGL1 (NM_001080951.2) exon 8. Copy number changes showed focal alterations on chromosome 6q and 22q, near the locations of the PLAGL1 (6q24.2) and EWSR1 (22q12.2) genes. One EWSR1-PLAGL1 case was found to have a SMARCB1 truncating pathogenic mutation (p.R53* at 34% allelic frequency), which corresponded to the subclonal loss of INI1 observed by immunohistochemistry. Despite the presence of p53 nuclear positivity in the vast majority of neoplastic cells in case #1, a corresponding TP53 mutation was not found. In cases with PLAG-family amplification, RNA sequencing demonstrated increased RNA expression of PLAGL1 and PLAGL2 respectively. The molecular findings for all cases, including methylation profiling calibrated score results, are summarized in Table 2.
Clinical outcomes
Clinical follow-up was available for 5/6 patients with PLAGL1 fusion (range: 11 months to 10 years) with various treatment modalities and extent of resection [Table 1]. Three patients developed recurrence, one at 7 months following treatment per ACNS0334 with intracranial multifocal disease and death at 11 months, one patient with two recurrences at 1.6 years and 9 years after chemotherapy, and one patient with residual/recurrent disease at 6 months and local rapid recurrence at 1.2 years. Outcome data for the PLAGL2 amplified case showed no evidence of disease at 4.8 years, and available outcome data for the PLAGL1 amplified case showed no definitive residual disease at 7 months.
Add Comment