Background
Malignant peripheral nerve sheath tumor (MPNST) is characterized by differentiation toward peripheral nerve sheath tissue. Approximately 50% of MPNST cases are associated with neurofibromatosis 1 (NF1), which is the most important clinical parameter for a definitive diagnosis of MPNST. Diagnosis of MPNST without NF1 is sometimes challenging because there are currently no established markers for MPNST on immunohistochemistry (IHC). Classically, MPNSTs express S-100 protein sparsely, but clearly, on IHC. However, the intensity of S-100 protein expression is sometimes markedly diminished and non-specific S-100 protein expression is often observed in other sarcomas. Recent studies have revealed that histone 3 trimethylated on lysine 27 (H3K27me3) is a useful diagnostic marker for MPNST [
1,
2]. MPNST often shows complete or so-called “mosaic” loss of H3K27me3 expression on IHC, and loss of H3K27me3 is the basis for diagnosing MPNST. However, approximately one-third of cases retain H3K27me3 expression, and the evaluation of H3K27me3, especially the detailed ratio of mosaic loss, is empirically difficult and seems to be somewhat non-objective. In the clinical setting, we make a final diagnosis of MPNST by the combination of H3K27me3 expression and classic diagnostic hallmarks including S-100 protein expression and association with NF1.
It has been reported that
NF1 deletion is characteristic of MPNST.
NF1 deletion can be detected by fluorescence
in situ hybridization (FISH) and is useful in pathological diagnosis using formalin-fixed and paraffin-embedded specimens. Perry et al. found that
NF1 was characteristically deleted in MPNST cases in a study of benign peripheral nerve sheath tumors and MPNSTs [
3]. In addition, Suzuki et al. showed that
NF1 deletion FISH was useful for diagnosis in diagnostically challenging cases of intraosseous MPNST without NF1 [
4].
Perry et al. first revealed the diagnostic utility of a FISH assay for
NF1 and
p16 in MPNST [
5]. They performed a dual-color FISH assay with 22 MPNSTs and benign and malignant non-peripheral nerve sheath tumors (non-MPNSTs) including 13 plexiform schwannomas, 5 cellular schwannomas, 8 synovial sarcomas, 6 fibrosarcomas, and 13 hemangiopericytomas. They demonstrated the specific deletion of
NF1 and
p16 in MPNST and indicated that homozygous and heterozygous deletion of
p16 might distinguish MPNST from benign nerve sheath tumor and other histologically mimicking sarcomas. From the genetic aspect,
p16/CDKN2A inactivation is considered a key event in the occurrence and progression of MPNST. Loss of the
p16/CDKN2A locus at 9p21 is one of the earliest events in the malignant transformation of neurofibromas. The expression of p16 on IHC is sometimes decreased in MPNST [
6]. Atypical and conventional neurofibromas also exhibit the loss of nuclear p16 expression. Therefore,
p16/CDKN2A inactivation may be an early change in malignant progression [
7].
In this study, aiming for more accurate differential diagnosis, we examined the diagnostic utility of FISH for NF1 and p16 deletions in the differential diagnosis of MPNST and its mimicking spindle cell and pleomorphic sarcomas, especially in mosaic loss cases in which the interpretation of H3K27me3 can be difficult.
Methods
Sample selection
This study was performed with the approval of the Institutional Review Board (IRB) of Sapporo Medical University Hospital (No. 272-107). The IRB approved an opt-out informed consent approach for a retrospective, non-interventional study. We selected 55 MPNST cases from our pathological archives. Twenty cases were NF1-associated MPNSTs (NF1Ms) and 35 cases were non-NF1-associated MPNSTs (NNF1Ms). We determined histological grade using the parameters of tumor differentiation, mitotic figures, and tumor necrosis according to the Fédération Nationale des Centres de Lutte Contre le Cancer grading system. We performed hematoxylin and eosin staining using 3-μm-thick sections. We reviewed all hematoxylin and eosin-stained slides and previously stained IHC slides in individual cases. Next, we selected 35 non-MPNST sarcomas that needed to be distinguished from MPNST, comprising 9 synovial sarcomas (SSs), 8 leiomyosarcomas (LMSs), 10 myxofibrosarcomas (MFSs), and 8 undifferentiated pleomorphic sarcomas (UPSs). The histopathological criteria for non-MPNSTs were as follows. SS showed fascicular proliferation of uniform spindle cells. The tumor cells were positive for cytokeratin and/or epithelial membrane antigen to varying degrees on IHC. All SS cases had been shown to have SS18 rearrangement by FISH. LMS exhibited interlacing fascicules of spindle cells with cigar-shaped nuclei and eosinophilic cytoplasm. The tumor cells expressed at least two of three myogenic markers (α-smooth muscle actin, desmin, and muscle-specific actin) on IHC. MFS was characterized by multinodular growth of hypocellular proliferating atypical spindle cells with prominent elongated, curvilinear, thin-walled blood vessels in a myxoid background. UPS exhibited a pattern-less growth of severely atypical spindle or pleomorphic cells without any differentiation toward specific tissue on IHC.
We checked the immunoreactivity of previously performed IHC for S-100 protein in all 55 MPNST cases. As a result, 50 cases (approximately 90%) expressed S-100 protein on IHC. Twenty-eight cases demonstrated sparse or focal expression of S-100 protein. On the other hand, 22 cases were diffusely positive for S-100 protein.
H3K27me3 IHC
We performed IHC for H3K27me3 using representative sections from formalin-fixed paraffin-embedded tissues from MPNST and non-MPNST cases. These tissues were sliced into 3-μm-thick sections and examined with an automated IHC system at Sapporo Medical University Hospital. All slides were loaded into a PT Link Module (Agilent Technologies, Santa Clara, CA) and subjected to a heat-induced antigen-retrieval protocol with EnVision FLEX Target Retrieval Solution (Agilent Technologies) before being transferred to an Autostainer Link 48 (Agilent Technologies). We used antibodies against H3K27me3 (C36B11, 1:200 dilution; Cell Signaling Technology, Danvers, MA). We determined H3K27me3 loss when we recognized the loss of H3K27me3 nuclear staining in the tumor cells. We evaluated the percentage of cells with H3K27me3 loss using measured values at 10% intervals and categorized it as complete loss (100% of tumor cells lost staining), partial loss (10% to 90% of tumor cells lost staining), and intact (no tumor cells lost staining).
FISH
We performed FISH using the commercially available probes MD NF1 (17q11)/MPO (17q22) (Leica, Wetzlar, Germany) for
NF1 deletion and Vysis LSI CDKN2A SpectrumOrange/CEP9 SpectrumGreen Probes (Abbott, Abbott Park, IL) for
p16 (CDKN2A) deletion. Each of the probes was labeled with red or green dye for the target or control locus, respectively. FISH was performed as described previously [
8]. In brief, the specimens were tumor tissues in 4-μm-thick slices on glass slides. We first selected an area showing representative histology and marked a 5-mm circle with a marker pen on the glass slide. We used a PathVysion HER-2 DNA Probe Kit (Abbott) and followed the manufacturer’s procedure, with modifications: baking (60°C for 1 h), deparaffinization, target gene activation (20 min with 0.2 M HCl followed by 80°C for 30 min with pretreatment solution), enzyme treatment (37°C for 60 min with protease solution), re-fixation (10 min with 10% formalin neutral buffer solution), denaturation (72°C for 5 min with denaturation solution), washing and dehydration (1 min each in 70%, 85%, and 100% ethanol), hybridization with 10 mL DNA probe solution (90°C for 5 min, followed by 37°C for 48 h), and washing with post-hybridization wash buffer (72°C for 2 min). For counterstaining, 10 μL of 4,6-diamidino-2-phenylindole was added. The slides were cover-slipped for viewing under a fluorescence microscope. We counted 50 tumor nuclei and calculated the percentage of cells with deletion signals. Deletion signals were categorized as homozygous deletion, heterozygous deletion, and monosomy. Homozygous deletion was defined as complete deletion of both alleles of the target locus and cells showed only control green signals. Heterozygous deletion was defined as alternate deletion of one of the alleles of the target locus, and so the number of red signals for the target locus was less than the number of green signals for the control locus. Tumor cells with monosomy had one allele, which showed a pair of red and green signals. Monosomy was considered a variation of heterozygous deletion. We defined “deletion” if the percentage of either the
NF1 or
p16 deletion signals was greater than 50% of tumor cells.
Statistical analysis
To examine whether there was a difference in the IHC and FISH results between low-grade and high-grade MPNST, we performed statistical analysis by Fisher's exact test using IBM SPSS Statistics 25 software (IBM SPSS Statistics, Chicago, IL). For all analysis, differences at P < 0.05 were considered statistically significant.
Discussion
In general, a definitive diagnosis of MPNST is often difficult or challenging because the tumor fundamentally shows spindle cell morphology and does not possess any IHC markers that are completely specific for MPNST. The differential diagnosis of MPNST ranges from benign to malignant tumors [
7], and so we investigated SS, LMS, MFS, and UPS. SS shows fascicular proliferation of spindle cells, especially in monophasic fibrous SS, and expresses epithelial markers (cytokeratin and epithelial membrane antigen) on IHC. Finally, SS has a specific chimeric
SS18-SSX fusion gene. LMS also shows fascicular proliferation of spindle cells with cigar-like nuclei and eosinophilic cytoplasm. The tumor cells often express myogenic markers (α-smooth muscle actin, desmin, and muscle-specific actin) on IHC. MFS exhibits loose fascicular proliferation of atypical spindle cells with abundant myxoid matrix. UPS consists of markedly atypical spindle and pleomorphic cells. Neither MFS nor UPS have characteristic markers on IHC. Therefore, MPNST should always be differentiated from these tumors.
On IHC, MPNST often shows complete or mosaic loss of H3K27me3 expression. However, the specificity of the loss of H3K27me3 expression in the differential diagnosis of MPNST is not complete and one study reported that H3K27me3 cannot be used safely to differentiate between MPNST and malignant melanoma [
9]. In addition, the loss of H3K27me3 expression has been confirmed in other histologic mesenchymal tumors including some extraskeletal osteosarcomas and dedifferentiated chondrosarcomas [
10,
11].
A recent study revealed the diagnostic utility of histone H3K27 dimethylation (H3K27me2) loss in the differential diagnosis of MPNST [
12]. Marchione et al. investigated H3K27me2 IHC in MPNST, K27M-mutant glioma, ependymoma, and Merkel cell carcinoma, which were characterized by the loss of H3K27me3. They demonstrated that while H3K27me3 loss is common across these tumor types, H3K27me2 loss is limited to MPNST. Moreover, they also investigated H3K27me2 IHC in cases of malignant melanoma and SS, which were histologic mimics of MPNST and showed various degrees of H3K27me3 loss. As a result, while global H3K27me3 loss was not observed in these tumors, weak and limited H3K27me3 staining was common. On the other hand, H3K27me2 staining was more clearly retained in all cases. Thus, H3K27me2 may be a useful marker for the differential diagnosis of MPNST instead of H3K27me3, the partial loss of which is particularly difficult to assess using immunoreactivity.
Indeed, the present study revealed that some non-MPNSTs showed H3K27me3 partial loss ranging from 10% to 90% of tumor cells. Among them, MPNSTs tended to exhibit NF1 and/or p16 deletion more frequently than non-MPNSTs. In particular, the combination of H3K27me3 IHC and NF1/p16 deletion FISH was able to reach an accurate diagnosis in approximately 90% of MPNST cases.
Nevertheless, there were some non-MPNST cases that were difficult to differentiate from MPNST according to the status of H3K27me3 loss and
NF1 and/or
p16 deletion, but these cases can be diagnosed by combining other IHC and FISH findings. For example, SS-3 and SS-9 cases showed
NF1 deletion, and SS-9 cases also exhibited complete loss of H3K27me3. SS is especially important for differential diagnosis from MPNST because of the monomorphic appearance of spindle cell proliferation. However, we were able to reliably distinguish these two cases from MPNST by cytokeratin expression and identifying the
SS18-SSX fusion gene. In another example, some LMSs (LMS-1, 3, 5) also showed
NF1 and/or
p16 deletion, although positivity for myogenic markers on IHC helped us to reach a correct diagnosis. We often find that the differential diagnosis of MFS and UPS is the most challenging because these tumors have no specific markers on IHC. A recent study revealed that somatic tumor genomic profiles by FoundationOne® CDx showed a high prevalence of
p16/CDKN2A alterations in MPNST, MFS, and UPS [
13]. These sarcomas may have some overlap of genetic features. From the clinical point of view, MFSs are relatively distinguishable according to their superficial localization and characteristic radiographic findings showing extension along the fascia. Ultimately, it seems that there are certain UPS cases that are tremendously difficult to distinguish from MPNST. We re-evaluated the MFS and UPS cases showing co-deletion of
NF1 and
p16 (MFS-5, 10, UPS-2, 5, 6) and confirmed that they were morphologically typical MFS and UPS and different from MPNST when judged comprehensively. In such a situation, we need a multidisciplinary approach that combines not only pathological findings but also epidemiological, clinical, and radiographic findings for an accurate diagnosis.
It has been reported that loss of p16 expression on IHC is associated with poor prognosis in patients with high-grade sarcoma including MPNST [
6,
14]. In addition,
p16 homozygous deletion is a marker of poor prognosis in patients with Ewing sarcoma [
15]. Moreover,
p16 deletion has been revealed to be a biomarker for poor prognosis in patients with soft tissue sarcoma [
13]. In this way,
p16 is an important marker of the clinical behavior of patients with soft tissue sarcoma. In this study, we revealed that the presence of
NF1 and/or
p16 deletion was associated with the histological grade of MPNST. Although we did not investigate the association between
p16 deletion and prognosis directly,
p16 deletion might be an important biological factor associated with the histological grade of MPNST that can be used to predict prognosis.
Acknowledgements
The authors thank the following pathologists and clinicians for kindly contributing case materials and clinical follow-up information: Akihisa Saito, Institute for Clinical Research, National Hospital Organization Kure Medical Center and Chugoku Cancer Center, Hiroshima, Japan; Arafumi Maeshima, Department of Pathology, National Hospital Organization Tokyo Medical Center, Tokyo, Japan; Daisuke Noda, Department of Otolaryngology Head and Neck Surgery and Takashi Tsuchiya and Masato Sugawara, Department of Orthopedic Surgery, Yamagata University, Faculty of Medicine, Yamagata, Japan; Hajime Umezu, Division of Pathology, Niigata University Medical and Dental Hospital, Niigata, Japan; Hiroyuki Iwaki, Department of Pathology and Yasuharu Kunishima, Department of Urology, Sunagawa City Medical Center, Hokkaido, Japan; Hiroyuki Kawashima, Department of Orthopedic Surgery, Graduate School of Medical and Dental Sciences, Niigata University, Niigata, Japan; Johji Imura, Department of Diagnostic Pathology, Graduate School of Medicine and Pharmaceutical Sciences, University of Toyama, Toyama, Japan; Katsushige Yamashiro, Division of Pathology, National Hospital Organization, Hokkaido Cancer Center, Hokkaido, Japan; Michiko Yamashita and Yoshiyuki Fujii, Division of Pathology, Tokushima Red Cross Hospital, Tokushima, Japan; Satoko Morohashi, Department of Pathology and Bioscience, Hirosaki University, Graduate School of Medicine, Aomori, Japan; Shigeri Ooba, Ooba Clinic, Hokkaido, Japan; Shingo Sakashita, Department of Pathology, Tsuchiura Kyodo General Hospital, Ibaraki, Japan; Takahiro Sugiyama, Division of Surgical Pathology, Chiba Cancer Center, Chiba, Japan; Takashi Kawasaki, Department of Pathology, Niigata Cancer Center Hospital, Niigata, Japan; and Yuji Harada, Department of Surgical Pathology, Shimane University Faculty of Medicine, Shimane, Japan.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.