Pediatric central nervous system tumor with CIC::LEUTX fusion: a diagnostic challenge

Introduction

The fifth edition of the World Health Organization Classification of Tumors of the Central Nervous System (WHO CNS5) recognized CIC-rearranged sarcoma as a distinct mesenchymal, non-meningothelial tumor, designated as CNS WHO grade 4 [1, 2]. This entity is characterized by its CIC gene fusion with various partners, most notably NUMT1 or DUX4 [3, 4], serving as crucial molecular hallmarks and essential criterion for diagnosis [1, 4]. However, the genetic landscape, etiology and clinical implications of CIC-fused CNS tumors still remain elusive, posing significant challenges in routine diagnostic practices.

Recent studies have identified a set of pediatric high-grade neuroepithelial tumors with CIC fusions, which display a unique DNA methylation profile differing from CIC-rearranged sarcomas, suggesting an intermediate malignancy grade [5]. These findings elucidate that CNS tumors harboring CIC gene fusions may encompass various tumor types and divergent clinical outcomes.

DNA methylation profiling has emerged as a robust and dominant approach for the comprehensive classification of CNS tumors and the identification of new tumor entities [6,7,8,9,10]. WHO CNS5 also recommends this technology as a desirable tool for diagnosis, particularly in pediatric and embryonal tumors [1, 11,12,13]. Given the potential of CIC fusions to occur beyond CNS sarcomas, integrating methylation profiling is crucial for distinguishing other tumor entities that share CIC alterations.

In this multicenter study, we describe a cohort of tumors with CIC fusions, including 14 CNS cases (containing 6 CIC::LEUTX fusion tumors) and 5 peripheral sarcoma cases, with a particular focus on those with rare fusion partners and their diverse clinical outcomes. These tumors underwent comprehensive histological re-evaluation, RNA sequencing, DNA sequencing and genome-wide methylation profiling. Our aim is to deepen and broaden the clinical, histological, and molecular understanding of CIC-fused CNS tumors, as well as to assess whether CNS tumors with some specific CIC fusions should be considered as a distinct entity.

Materials and methods

Sample collection

Tumor specimens with CIC fusions were acquired from: the department of pathology of Sun Yat-sen University Cancer Center, Children’s hospital of Chongqing medical university, Guangzhou Women and Children Medical Center, Beijing Tiantan Hospital, Zhujiang Hospital of Southern Medical University, Nanjing Brain Hospital, Sun Yat-sen Memorial Hospital, Longgang District Central Hospital of Shenzhen, Meizhou People’s Hospital and the First Affiliated Hospital of Zhengzhou University. The whole series comprised 14 CNS tumors with CIC fusions including 6 cases each of CIC::LEUTX and CIC::NUTM1, one case of CIC::FOXO4, and one case of concurrent CIC::DUX4 & CIC::FRG2B, and 5 cases of peripheral CIC sarcomas, consisting of 3 cases of CIC::DUX4 (one of which also featured a CIC::DBET intergenic region fusion), one case of CIC::FBXO4, and one case of a CIC::LINC00854-LINC00910 intergenic region fusion. Reference group for methylation profiling was selected from Gene Expression Omnibus (GEO) database, specifically datasets GSE90496 and GSE223546. The collection of tumor samples and clinical data were processed in an accordance with standards approved by the ethical committees of department of pathology and center for molecular medicine testing, Chongqing medical university.

Histological analysis

Histological assessments were evaluated by three neuropathologists (Wanming Hu, Jing Zeng and Yanghao Hou) following the diagnostic guidelines of CNS WHO 5. Immunohistochemistry (IHC) staining was performed on 3 μm-thick formalin-fixed, paraffin-embedded (FFPE) sections using an automated BenchMark Ultra (Ventana Medical systems, Roche, SW). Antibodies were diluted against: GFAP (ZSGB-BIO, 1:200), Olig-2 (ZSGB-BIO, 1:200), synaptophysin (ZSGB-BIO, 1:200), WT-1 (ZSGB-BIO, 1:200), CD99 (ZSGB-BIO, 1:200), Ki67 (DAKO, RTU). Special reticulin staining was used VENTANA Silver ISH DNP Detection Kit.

DNA and RNA extraction

Areas rich in tumor cells (> 70% tumor cell content) were identified on hematoxylin and eosin (H&E) stained slides. Necrotic and lymphocyte-rich areas was avoided to ensure the quality of methylation array and next generation sequencing. DNA and RNA were obtained separately from 10 individual 10 μm-thick FFPE sections, with areas precisely matched the selected regions identified by H&E staining to ensure optimal tumor cell content and quality for molecular analysis. Extractions were performed using QIAamp DNA FFPE Tissue kit (QIAGEN, Germany) and RNeasy FFPE kits (QIAGEN, Germany), according to the manufacturer’s protocols.

DNA methylation and copy number profiling

The raw DNA methylation data (.idat) were obtained from the Infinium MethylationEPIC (850 K) or Infinium MethylationEPIC2.0 (935 K) BeadChip array (Illumina, San Diego, USA), DNA extracted from FFPE tissue were all repaired by Infinium FFPE QC and DNA Restoration Kits (WG-321-1002, Illumina, San Diego, USA) following the manufacturer’s instructions as previously described [14]. BeadChip were scanned by the iScan (Illumina, San Diego, USA). Raw idat files were proceeded in R by package “minfi” and “sesame”, “limma” package was used to remove batch effect. The same probe that preserved in both 450k (reference set), 850k (reference set and local sample set) and 935k (local sample set) were selected for beta-value calculation. In total, five samples failed to perform methylation array due to insufficient DNA quantity. Copy number profiling were derived from the methylation raw data in R version 4.3 (https://www.R-project.org) by using the package “conumee” (http://bioconductor.org/packages/conumee/).

Next generation sequencing

A panel-based NGS assay was used to detect gene alterations in formalin-fixed, paraffin-embedded (FFPE) tumor samples. First, DNA was extracted from undyed FFPE sections with the proportion of tumor cells more than 20% and whole-blood samples and then purification and library preparation were performed. Second, a probe with 2.29 Mbp in size was used for hybrid capture and enrichment in gene-specific regions where various aberrations of 360 cancer-related genes including single nucleotide variants, copy number variations, small insertions, deletions and gene arrangements were covered. R package “maftools” was used for generating tumor mutation plots.

RNA sequencing and fusion calling

The quality of each RNA sample was tested using Qubit 4.0 and Agilent 4200 TapeStation system prior to library preparation and sequencing. cDNA synthesis, and library preparation were performed using the KAPA RNA HyperPrep Kit (Kapa Biosystems, KK8540). The total volume of the final library was at least 40 ng. No obvious joint contamination was detected in the final library using the Agilent 4200 TapeStation system, and the main peak was between 300 and 500 bp. After quantification, NGS was performed on an Illumina Novaseq 6000 instrument (Illumina).

The software fastp (v.2.20.0) was used for adapter trimming. The software STAR (v2.7.6a) was used to align reads to the reference genome (UCSC’s hg19 GRCh37). Fusion expression was calculated based on fusion fragment per million (FFPM) using the raw data from the RNA fusion panel (166 genes). We use STAR-Fusion software (v1.9.1) to perform fusion detection. The coverage of gene exons for probe was calculated based on the fragments per kilobase of exon model per million mapped fragments (FPKM).

Statistical analysis

For T-Distributed Stochastic Neighbor Embedding (t-SNE) and unsupervised hierarchical clustering analysis, 15,000 most variable CpG sites with the highest median absolute deviation were been selected. 2500 iterations and a perplexity value of 5 was configured for t-SNE plotting. Graphic visualizations were conducted by R packages “ggplot2”. Displaying reads of fusion genes using the IGV (The Integrative Genomics Viewer, https://igv.org/doc/desktop/) tool. Kaplan–Meier survival curves was performed using R packages “survival” and “surminer”.

Results

Methylation signature of CIC-fused tumors

We performed t-SNE analysis by using our cohort (n = 14) together with 1889 published reference cases [7], encompassing nearly all CNS tumor types and subtypes, including the recently published methylation class (MC): high-grade neuroepithelial tumor (HGNET) CIC fusion-positive [5]. We further refined t-SNE analysis for better visualization (Fig. 1) by narrowing representative MC, the clustering results of each case were consistent with the broader t-SNE analysis.

Fig. 1

Methylation based t-distributed stochastic neighbor embedding (t-SNE) analysis of 12 CIC-fused CNS tumours (colored in gray) and 2 peripheral tumors harboring CIC::DUX4 fusion (colored in red). All cases were clustered to conventional CIC-altered sarcoma except for case 1, case 2 and case 3 with CIC::LEUTX fusion. AB_MN1 MC Astroblastoma MN1-altered, BCOR_ITD MC CNS tumour with BCOR internal tandem duplication, CIC_Sarcoma MC CIC-altered sarcoma, CNSNB_FOXR2 MC CNS neuroblastoma FOXR2-activated, CIC_HGNET MC high-grade neuroepithelial tumor CIC fusion-positive, EWS MC Ewing sarcoma, EPN_SPINE MC spinal ependymoma, EPN_ZFTA MC ependymoma ZFTA fusion-positive, ETMR MC Embryonal tumour with multilayered rosettes, GG MC Ganglioglioma, GBM_RTKII MC glioblastoma RTKII subtype, LGG_MYB MC diffuse astrocytoma MYB altered, PXA MC pleomorphic xanthoastrocytoma, SFT MC solitary fibrous tumor

Of the 12 CNS tumors with various CIC fusion partners, nine cases clustered with the reference MC: CIC-rearranged sarcoma, alongside 2 peripheral sarcomas with CIC::DUX4 gene fusion, serving as internal controls. Notably, three of the six CNS tumors with CIC::LEUTX fusions clustered elsewhere: Case 1 clustered to the reference set “HGNET CIC fusion-positive”. Case 2 clustered closely to the MC ganglioglioma, and Case 3 exhibited an independent methylation signature near the reference methylation group for low-grade glioma, MYB-altered.

Invariable LEUTX locus aberrations on chromosome 19

In all six primary CNS tumors featuring CIC::LEUTX gene fusions, methylation array-based copy number profiling consistently revealed LEUTX locus aberrations at chromosome 19q13.2, demonstrated by visual inspection of LEUTX locus loss or gain (6/6, 100%) in CNV results (Fig. 2a, b). Among the tumors with CIC::NUTM1 fusion, alterations were observed at the NUTM1 locus (15q14) in two cases and at the CIC locus (19q13.2) in one case (Fig. 2c). Case 12 (with CIC::FOXO4 fusion) and Case 18 (CIC intergenic rearrangement) exhibited aberrations at the CIC locus (Fig. 2d). Furthermore, gains on chromosome 8 (4/14, 28%) were frequently observed. Detailed descriptions of these chromosome changes are provided in Table 1.

Fig. 2

Representative copy-number profiles derived from methylation data. Loss (a) and gain (b) were observable at the LEUTX locus (19q13.2) on chromosome 19q. Loss of NUTM1 (15q14) in case 10 with NUTM1::CIC fusion (c). Gain of CIC locus (19q13.2) in Case 12 harboring CIC::FOXO4 fusion (d)

Table 1 Molecular characteristics of 19 CIC-fused tumors

Clinical findings

The clinical data for the cohort are detailed in Table 2. Of the 14 patients with CIC-fused CNS tumors, 8 were male and 6 were female, with a median age at diagnosis of 5 years (range:1–17). The tumors manifested across various brain regions and parts of the spinal cord, most commonly in cerebral hemispheres.

Table 2 Clinical findings of 19 tumors harboring CIC fusions

Histological and immunohistochemical characteristics of CIC and CIC::LEUTX fused tumors

Histological examination of the CIC-fused CNS tumors revealed a diverse morphological spectrum (Fig. 3). Typical features included highly undifferentiated, small to medium-sized blue round cells with brisk mitotic activity, microvascular proliferation, and necrosis. A range of cellular morphologies, including gemistocytic, epithelial, giant, triangular and spindle cells, was noted across different cases, with perivascular pseudorosette occasionally observed.

Fig. 3

Variable morphological features of CIC::LEUTX fused CNS tumors. Case 1 (a, b) showed well-defined boundaries from the surrounding brain tissue, and a high-grade neuroepithelial tumor appearance with vascular endothelial hyperplasia; Case 2 (c, d) was initially diagnosed as an embryonal tumor due to small round blue cell embryonal tumor morphology; Case 3 (e, f) exhibit low-grade morphological features with calcifications and pleomorphic GFAP & Olig-2 positive tumor cells; Case 4 (g, h) exhibited spindled to gemistocytic/epithelioid cytology. Case 5 (i, j) is characterized by epithelioid tumor cells with prominent nucleolus, mimicking metastatic carcinoma; Case 6 (k, l) shows a nodular and diffuse sheeting growth pattern, clear-cell cytology is focally presented

In tumors harboring the CIC::LEUTX fusion, a glial fibrillary matrix could be found, with robust expression of glial markers (GFAP, Olig-2) and neuronal markers (synaptophysin). Notably, two distinct immunophenotypic patterns emerged. One was a neuroepithelial tumor pattern (Case 1–3) characterized by a lack of WT-1 expression (Fig. 4a) and absence of reticular fibers, except in perivascular spaces (Fig. 4d). The other pattern, resembling sarcomas (Case 4–6), showed positive WT-1 staining (Fig. 4b) and reticular fibers encircling individual cells (Fig. 4e), similar to peripheral CIC sarcomas (Fig. 4c, f). Detailed IHC results are provided in Table 3.

Fig. 4

Representative immunohistochemistry and reticulin staining. Total negativity of WT-1 staining in Case 1 (a) in comparison to focal positivity in Case 4 (b) and diffuse WT-1 expression in non-LEUTX CIC-rearranged tumors (c, Case 8); Absence of reticular fibers in Case 1 (d) in comparison to rich reticulin staining in Case 4–19 (e, Case 4; f, Case 8); GFAP, Olig-2 and Synaptophysin expression of Case 1 (g, h, i) and Case 5 (j, k, l)

Table 3 Immunohistochemistry results

The detailed description of 3 CIC::LEUTX fused CNS tumors (not aligned to MC CIC-rearranged sarcoma in methylation profiling) were listed below.

Case 1 (Fig. 3a, b)

Histologic examination revealed a high-grade neuroepithelial tumor characterized by densely packed, poorly differentiated cells displaying nuclear atypia and marked pleomorphism. The mitotic activity is brisk, accompanied by prominent microvascular proliferation and necrosis. IHC showed focal GFAP expression (Fig. 4g) and strong positivity for Olig-2 (Fig. 4h) and synaptophysin (Fig. 4i). WT-1 was negative (Fig. 4a). Additionally, reticulin staining revealed a pattern typical for neuroepithelial tumors, with reticular fibers confined to vascular areas and absent within the tumor parenchyma (Fig. 4d). These morphological features were consistent with a malignant glioma of WHO grade 4.

Case 2 (Fig. 3c, d)

The tumor also presented as a high-grade neuroepithelial tumor. It consisted of poorly differentiated or undifferentiated cells interspersed with neuropil-like structures and abundant vascular proliferation. No typical true rosettes were found. Some tumor cells were small, round, and poorly differentiated, whereas others displayed atypia and rough, dark chromatin. Mitotic figures and apoptotic bodies were readily apparent under HPF. IHC showed negative staining for both GFAP and Olig-2. WT-1 was also negative. Reticular staining did not reveal significant desmoplasia but maintained a pattern consistent with neuroepithelial tumors. Remarkably, the tumor cells exhibited strong positivity for synaptophysin. The preliminary diagnosis was CNS embryonal tumor, NOS, supported by morphology and its strong synaptophysin immunoreactivity. This case was previously described and published as a case study [15] but without methylation analysis.

Case 3 (Fig. 3e, f)

The tumor specimens documented spindle cells and large pleomorphic cells with lipidized cytoplasm, alongside notable calcification, suggesting a low-grade glioma. The tumor cells were positive for both GFAP and olig-2. Staining for Synaptophysin, BRAF, and H3K27M were negative. ATRX and INI1 were retained. The Ki67 proliferation index was only 1%, and WT-1 was also negative. In summary, it displayed a low-grade tumor morphological appearance which could potentially mimic pleomorphic xanthoastrocytoma (PXA) or polymorphous low-grade neuroepithelial tumour of the young (PLNTY), However, unlike PXA or PLNTY, reticular fibers were only observed surrounding blood vessels, as well as the CD34 expression. It also did not exhibit BRAF V600E mutations or CDKN2A/B deletions. The primary diagnosis was PXA-like LGG, NEC.

Notably, different from peripheral CIC sarcomas, the nucleoli of the tumor cells in the above three cases were not obvious.

Case 4 (Fig. 3g, h), 5 (Fig. 3i, j) and 6 (Fig. 3k, l)

These three cases were CIC::LEUTX fused CNS tumors which clustered to MC CIC-rearranged sarcoma in methylation profiling, Their morphology was also similar to that of peripheral CIC sarcoma. The tumor displayed variable cell morphology ranging from spindled to gemistocytic/epithelioid, including sheets of monotonous cells with high nuclear: cytoplasmic ratios and distinct nucleoli. Some areas showed high cellularity with uniform, round to oval nuclei, while large nucleoli were prominent in these tumor cells. Areas resembling fibro/sarcomatoid spindle cells may also be seen, which looks more like a sarcoma. Case 5 even looked like angiosarcoma, for there was a significant presence of small blood vessel proliferation, hemorrhage, and red blood cell exudation. The tumor cells were negative for GFAP (Fig. 4j), Olig-2 (Fig. 4k), and only individual tumor cells expressed synaptophysin (Fig. 4l). WT-1 showed focal positive expression (Fig. 4b), but not diffusely strong as other non-LEUTX::CIC-rearranged tumors (Fig. 4c). Reticular fibers were abundant (Fig. 4e) and distinct from case 1–3 (Fig. 4d), with more fibers encircling individual cells, similar to the pattern observed in peripheral CIC sarcoma (Fig. 4f).

Discussion

In our study of 19 tumors featuring CIC fusions, spanning both CNS and peripheral tissues, we observed that only CNS tumors harboring the CIC::LEUTX fusion exhibited neuroepithelial differentiation with better outcomes compared to those of CIC-rearranged sarcomas. The distinct diagnostic features for these tumors included positive GFAP and Olig-2 expression, negativity of WT-1 and reticulin, and a methylation profile incompatible with conventional MC CIC-rearranged sarcomas.

The LEUTX gene, also known as Leucine Twenty Homeobox, is integral to embryonic development and early cellular differentiation [16]. Recent research has linked gene fusions involving LEUTX to oncogenesis, particularly in primary CNS sarcomas and high-grade neuroepithelial tumors that harbor CIC::LEUTX fusions [5, 17], as well as in embryonal tumors with BRD::LEUTX fusions [18]. These rare entities, especially affecting young children, typically exhibit poor clinical outcomes. Notably, within our cohort, we discovered two cases of neuroepithelial tumors with CIC::LEUTX fusion that demonstrated prolonged progression-free survival (PFS) and presented with unique, unassigned methylation signatures. This includes one case with low-grade morphological features which have not been previously reported in CIC-fused CNS tumors.

Diagnostic implications for tumors harboring CIC fusion

The variable morphological features of pediatric CNS tumors with CIC fusions frequently pose diagnostic challenges. Nonetheless, molecular evidence of a CIC fusion (the first essential diagnostic criteria of CIC-rearranged sarcoma listed by WHO CNS5) or a methylation profile matched to MC CIC-sarcoma generally confirms the diagnosis [1].

Our study reveals that pediatric CNS tumors harboring the CIC::LEUTX fusion represent a heterogeneous set of tumors. These occur across a spectrum including conventional CIC-altered sarcoma, high-grade neuroepithelial tumor, and even rare lower-grade glial tumors, indicating that not all the intracranial CIC-fused tumors are CIC-altered sarcomas, especially those with CIC::LEUTX genetic fusion. In contrast to peripheral CIC-rearranged sarcomas, which typically express WT-1, lack GFAP and Olig-2, and have abundant reticular fibers [19], intracranial CIC::LEUTX fused tumors may be considered as neuroepithelial tumors if they exhibit following features, especially the last two items: (1) GFAP & Olig-2 expression (at least focally positive); (2) Synaptophysin expression; (3) Absence of WT-1 expression; (4) Lack of reticular fibers. Given our limited cohort, these observations should be seen as preliminary and indicative rather than definitive criteria.

For unresolved cases, DNA methylation profiling has proven to be the most effective molecular approach for the precise classification of CIC-fused CNS tumors [20]. The diagnosis can be assigned if the methylation profile aligns to its representative MC: CIC-rearranged sarcoma [7] or the novel entity “HGNET CIC fusion positive” [5]. If the methylation result is paradoxical to their morphological or molecular features, as seen in case 2 and case 3 in our cohort, it remains debatable to classify these tumors as ganglioglioma and LGG MYB. However, these cases have demonstrated PFS of 56 and 34 months, respectively, despite their CIC alterations and mild to aggressive morphological appearances.

Additionally, methylation derived copy-number profiling can provide crucial evidence of CIC-related fusions. In our series, 10 out of 12 CNS tumors showed gene locus aberrations related to their fusion partners, particularly those with LEUTX, where all six cases demonstrated LEUTX locus aberrations.

Clinical values of subclassification CIC-fused tumors

The distinction between CIC-rearranged sarcomas and HGNET fusion-positive is crucial, given that the former typically presents a worse prognosis [19]. This has been highlighted in the study by Philipp Sievers, which suggested the latter as intermediated malignancy [5]. Consequently, the accurate classification of sarcoma or neuroepithelial types of tumors harboring CIC fusions should be encouraged in routine diagnostics. Noteworthy, in our cohort, Case 2 and 3 demonstrated methylation profiles that were inconsistent with either CIC-rearranged sarcoma or HGNET CIC fusion-positive, but were closer to the MC of lower-grade entities. The outcome data appear to support their methylation signature: Case 2, despite displaying high-grade neuroepithelial tumor appearance, achieved a PFS of 56 months; Case 3 showed a lower-grade neuroepithelial tumor morphology with low Ki67 index. Without chemotherapy or radiotherapy, the patient has reached a PFS of 34 months post-GTR. The overall survival rate (Sup Fig. 1a) and PFS (Sup Fig. 1b) of CNS tumors with CIC::LEUTX fusions show better clinical outcomes compared to those with non-LEUTX fusions. However, given the limited cases numbers, more outcome data is needed to draw concrete conclusions.

These cases further indicate that the confirmation of CIC-rearranged sarcoma requires differential diagnosis of neuroepithelial tumors or rare lower-grade entities. Such differentiation should ideally integrate both morphological evidence and DNA methylation profiling rather than relying solely on the presence of CIC fusion alone, especially for those tumors harboring CIC::LEUTX fusion.

Conclusion

In summary, our study expands the knowledge of CIC-rearranged pediatric CNS tumors, specifically those tumors harboring CIC::LEUTX fusions, which may be a heterogeneous group of tumors consisting of CIC-rearranged sarcomas, HGNET CIC fusion-positive, and rare lower-grade neuroepithelial tumors with undefined methylation signatures. The combination of GFAP, Olig-2, synaptophysin, WT-1 and reticulin staining can help differentiate sarcoma and neuroepithelial tumors. For unresolved cases, DNA methylation profiling serves as an ideal approach for precise and efficient classification. Studies on larger cohorts are still required for a better understanding these tumors.