Clinicopathological features and prognosis of malignant peripheral nerve sheath tumor: a retrospective study of 159 cases from 1999 to 2016

Objective To investigate the clinicopathological features and prognosis of malignant peripheral nerve sheath tumors (MPNST). Results A total of 159 patients with MPNST were enrolled in the study. The ratio of male to female was 1.04 to 1. The median age was 40 (range: 5–76) years at the time of diagnosis. The 3- and 5-year overall survival rates were 50.0% and 43.0%, respectively. The median follow-up period was 31.0 (range: 2.0–199.0) months. Multivariate analysis showed that AJCC stage and S-100 were independent factors affecting overall survival (p < 0.05 for both). 3- and 5-year tumor-free survival rates for 140 completely resected patients were 40.0% and 34.0%, respectively. Multivariate analysis showed that AJCC stage, S-100 and Ki67 staining were independent factors of tumor-free survival (p < 0.05 for all). Materials and Methods The clinical data of MPNST patients who were treated at Cancer Institute and Hospital, Chinese Academy of Medical Science from January 1999 to January 2016 was retrospectively reviewed. Conclusions MPSNT is a highly aggressive tumor with poor prognosis and this study may be useful for prognostic assessment and management decisions. This had been largest documented retrospective study of MPSNT among Chinese populations. Some characteristics were different from those of foreign populations which may suggest the specificity of Chinese patients.

Recently, research data have shown that NF1, surgical margin status, and tumor size are significant predictors of survival in patients with MPNST. Molecular predictors such as TP53 and S-100 are also suggested [4,[10][11][12]. However, large cohort studies on Chinese Research Paper www.impactjournals.com/oncotarget MPNST patients are totally absent [11,12]. Noteworthily, it had been the largest documented retrospective study of MPNST among Chinese populations up to now. Clinical and pathological prognostic predictors affecting local and/ or distant recurrence were analyzed.

RESULTS
A total of 159 patients with MPNST were enrolled into the study. Patient characteristics were summarized in Table 1. The median age was 40 (range: 5-76) years. The ratio of male to female patients was 1.04:1. In the series, 69 (43.4%) patients presented with primary tumors only, whereas 76 (47.8%) patients were accompanied with recurrent tumors and 14 (8.8%) with distant metastasis. The percentages of NF1 associated MPNST and sporadic MPNST patients were 44.0% and 47.8%, respectively. The remaining 8.2% of enrolled patients were RT-induced. Similar to the foreign current reports, most NF1 patients in the study showed typical clinical and pathological features, such as cutaneous neurofibromas, multiple café-au-lait spots ( Figure 1A), a large, lobulated soft tissue mass with heterogeneous signal intensity in axial T2-weighted magnetic resonance image ( Figure 1B) and spindle cell morphology with a fascicular pattern ( Figure 1C).
Among the 159 patients, 19 individuals received no surgery: 14 stage IV cases of the disease with distant metastasis that could not get any obvious benefit from surgery, and 5 patients received palliative treatment for unsuitable health conditions. 140 patients (88.1%) with localized tumors underwent a complete resection. The decision of receiving radiotherapy and/ or systemic chemotherapy was made by surgeons, radiologists, physicians, influenced by willingness of patients themselves. In the study, among the 140 patients, 60.7% underwent 15-76 Gy of radiotherapy to primary and/or recurrent lesions before, intra or after surgery. A total of 39 patients were administrated with regimens of doxorubicin and ifosfamide, with or without dacarbazine. 3 patients were stage I and their prognosis were quite well. 36 patients were stage II or stage III. So we conduct the role of chemotherapy among stage II and stage III cases.
The median follow-up time was 31.0 (range: 2.0-199.0) months. The 3-and 5-year overall survival rates of the whole group were 50.0% and 43.0%, respectively (Table 1). 2-and 3-year OS for patients with metastatic disease were 36.0% and 14.0%, which were significantly worse than the localized disease patients (p = 0.002, Figure 2A). NF1 associated MNPST did not show survival advantage compared with those without NF1 presentation (p = 0.671, Figure 2B and p = 0.995, Figure 3D). Additionally, the TFS and the OS of radiation-induced MPNST compared to the sporadic and NF1 -associated are similar (p > 0.05 for both, Table 2).

Prognostic factors for overall survival
Till the latest follow-up, 56.0% (89/159) of all patients died of MPNST. Death rate of patients with metastatic disease was much higher than that of patients with localized tumor (85.7% vs. 51.0%, p < 0.05). Since patients with metastases presented significantly worse survival outcomes, these 14 individuals were therefore excluded from further analysis. 5 patients who had inadequate organ functions received no surgery, and they were also excluded. The 5-year OS rate of 140 patients who underwent a complete resection was 45.0%.
To identify additional factors impacting the outcome of MPNST, further analysis was conducted among patients who received surgeries. Univariate analysis showed that the location and the depth of tumor, AJCC stage, margin status, S-100 and Ki67 ( Figure 3F) staining independently affected OS (p < 0.05, Tables 2 and 4). However, by multivariate analysis, only AJCC stage ( Figure 3A) and S-100 ( Figure 3B) were factors associated with prolonged OS (p < 0.05, Table 5).

Prognostic factors for tumor-free survival
The 3-and 5-year TFS rates for the 140 patients were 40.0% and 34.0%, respectively. Most patients experienced local recurrence or distant metastasis during their followup. 42.9% experienced local relapse and the median time to relapse was 6.0 months. The 3-year distant relapse rate was 49.3%, and 47.9% developed distant metastases at a median time of 8.0 months. The most common site of distant metastasis was the lungs, which was documented in 35 patients (25.0%). Other common metastatic sites were the bone, brain, and liver, which occurred in 12.1%, 7.1%, and 5.0% of all patients.
Several possible prognostic factors predicting disease progression by analyses of univariable Cox proportions were found. The extremities and the superficial location of MPNST, their early AJCC stage, margin negative, administration of radiotherapy, S-100 positive and Ki67 < 20% were associated with lower recurrence/metastatic rates and prolonged TFS ( Table 2). On multivariate analyses, patients with advanced AJCC stage ( Figure 3C) and S-100 positive ( Figure 3E) had a higher tendency of developing local recurrence and metastases (Table 5).

Biomarker expression in patients with MPNST
It is important to identify MPNST molecular markers for diagnostic value. To achieve this goal, firstly, histologic specimens of 50 NF1 patients diagnosed in the hospital in the same period were collected. Univariate analysis was performed to evaluate possible correlations between biomarker expression and histologic diagnosis (NF1 vs. MPNST). The percentage of NF1 and MPNST www.impactjournals.com/oncotarget   (Table 3). In summary, the expression of S-100 and Ki67 staining differentiated MPNST from NF1 at a significance of p < 0.05. Though Vimentin, NF and GFAP were slightly higher in positivity in NF1 specimens, they did not reach statistical significance (Table 4).

DISCUSSION
MPNSTs are rare soft tissue sarcoma. Knowledge of their clinical outcome is limited, which impedes the ability to construct tumor-specific sensitive prognostic paradigms. This is one of the largest studies published within the last 30 years evaluating MPNST populations consisting of at least 100 patients [4,6,12,13]. Data showed the unfavorable outcome of MPNST, as well as the diagnostic value of S-100 and Ki67 in MPNST. Most importantly, it had been the largest retrospective study of Chinese populations to identify clinical and molecular predictors for MPNST to date.
In the study, the 3-and 5-year OS rates were 50.0% and 43.0%, respectively for the whole group, and 54.0% and 45.0%, respectively for patients who received tumor resection. The survival of Chinese patients with MPNST after multidisciplinary treatments was unsatisfactory, which was no difference from the published data with the 5-year survival rates ranging from 35 to 52% [4][5][6]12].
The percentages of NF1 associated MPNST and RT-induced were 47.8% and 8.2%, respectively. 44% of patients in the series had NF1. Other studies in the    [4][5][6]13]. It is no doubt that patients with NF1 are at increased risk of developing MPNST. There are conflicting reports as to whether malignant MPNST in NF1-patients with have worse prognoses than MPNST in non-NF1 patients. Porter and colleagues found that NF1 (p = 0.007) remained independent predictors of poor outcome and they recommended that NF1 be taken into account during MPNST staging [14]. Kolberg conducted a survival meta-analysis for > 1800 MPNST patients with and without NF1 [10]. The compiled literature from 1963 to 2012 indicated a significantly worse outcome of MPNST in patients with NF1 syndrome compared with that in non-NF1 patients. However, survival for the NF1 patients has improved in the last decade, and the survival difference is diminishing. In the study, there was no significant difference between the survival of patients with and without NF1 (p > 0.05). Additionally, in our study, it has no significance of the TFS and the OS of radiationinduced MPNST compared to the sporadic and NF1 -associated MPNST. However, LaFemina and colleagues reported on 105 patients with MPNST and found that NF1-associated and sporadic MPNSTs may be associated with improved disease specific survival compared to RTinduced tumors [15].
Noteworthily, the study found several prognostic factors for survival of MPNST patients. Data showed that the location, the depth, the size, the AJCC stage of the tumor, and S-100 were associated with the tumor-free survival. Whereas, late AJCC stage and S-100 negative were independent unfavorable factors affecting OS. Local recurrence or distant metastasis were much common in patients whose tumor was deeply located, especially in the head and neck region, tumor size > 5 cm, advanced AJCC stage, and/or Ki67 ≥ 20%. These results from Chinese patients in the report were in accordance with the reported data worldwide.  Some molecules including S-100, Ki67 may be useful in assessing the prognosis of MPNST. S-100 was considered a marker of neural crest differentiation and widely used to identify nerve sheath tumors from other soft-tissue neoplasms [16]. However, S-100 was negative in some MPNST patients as reported previously [12], which was considered the differentiation of Schwann cells [17]. In the series, 15.2% of the tumors were identified as S-100 negative and 84.8% positive. Further, it was found that S-100 negative was an independent prognosis factor, with a 3.24-fold increased risk of recurrence or metastasis, and a 5.62-fold increased risk of mortality. Ki67 was reported as a marker of cell proliferation, and had been used for predicting the prognosis of some tumors, such as breast cancers and lymphomas. Previous studies also confirmed upregulation of Ki67 in MPNST when compared with benign schwannoma [5]. In the report, 58.0% of MPNST tumors were identified as Ki67 ≥ 10%, compared with 3.3% of NF1. In prognostic analysis, Ki67 ≥ 20% was proven to be an independent prognostic factor, with a 2.81-fold Oncotarget 104792 www.impactjournals.com/oncotarget increased risk of mortality, compared with Ki67 < 20% MPNST (p = 0.001). The above conclusion was quite similar to the study from Kolarov [18]. It was an agreement that Vimentin was a maker of epithelial-mesenchymal transition. Therefore, it is higher in many malignant tumors, such as prostate cancer, breast cancer and colorectal cancer [19]. In the study, Vimentin, NF, and GFAP did not express significant differences in content between MPNST and NF1. Also, it was confirmed that Vimentin, NF, and GFAP were not prognostic factors for postoperative TFS and OS.
The mainstay of therapy for MPNST is surgical resection with the goal of achieving complete removal with negative margins. Many authors confirmed that total resection with a clear margin could reduce the recurrence rate and improve the prognosis of MPNST [4,[20][21][22]. In the series, negative margins could improve both the OS (p = 0.002) and the TFS (p = 0.003) on analyses of univariable Cox proportions (Table 4). But it was not an independent prognostic factor via multivariate analysis (p > 0.05, Table 5). Radiation therapy is an important adjunct to surgery in improving local control, and may be administered in the neoadjuvant or adjuvant setting, and as intraoperative therapy in centers with the available resources [23]. Although improvement in rates of local control have been seen with adjuvant radiation therapy, only one study out of Milan, Italy has been able to demonstrate that lack of radiation therapy predicts decreased disease specific survival [4]. In the data, adjuvant radiation therapy (HR: 0.688) could improve the OS rate, but it did not reach statistical significance (p = 0.078). The administration of adjuvant radiation therapy was associated with an improved TFS on univariate analyses (p = 0.002), though it had no impact on OS on multivariate analyses. The role of chemotherapy remains controversial. Some studies supported the effect of chemotherapy in MPNSTs [12,24], while others considered it as useless [4,6]. Generally, firstline chemotherapy regimen was doxorubicin-based [25].
Kroep conducted a research analyzing the response and survival of different chemotherapy regimens in patients with advanced MPNST [26]. They came to the conclusion that the doxorubicin-ifosfamide combination had the best response rate (HR: 6.283, 95% CI: 2.342-16.852). In that study, the administration of chemotherapy in MPNST patients had no impact on the survival. That patient sample group was quite small, though ( Table 2). In summary, despite combined multimodality therapy, MPNST behaves as an aggressive sarcoma with a propensity to recur locally or to metastasize to distant sites. It is believed that better survival outcomes were directly related to the high rate of negative margins combined with some certain clinical characteristics. From data of the hospital, tumor complete resection with MPNST patients is encouraged. The role of multidisciplinary approaches including adjuvant chemotherapy and/or radiotherapy need more randomized clinical studies to confirm.

MATERIALS AND METHODS
The clinical data of MPNST patients who were pathologically diagnosed and treated at Cancer Institute & Hospital, Chinese Academy of Medical Science from January 1999 to January 2016 was retrospectively reviewed. The study has been approved by the Ethics Committee in our hospital. For a tumor to be considered a NF1-associated MPNST. Firstly, its presence was determined on the basis of established NIH criteria [4], ie, patients with NF1 had at least 2 of the following: ≥ six cafeau-lait macules (> 5 mm before puberty, > 15 mm after puberty), skin-fold freckles (groin, axilla, base of neck), ≥ 2 neurofibromas (1 plexiform), skeletal dysplasia (orbital or tibial), Lisch nodules (iris hamartomas), optic gliomas, and family history. For a tumor to be considered a sporadic MPNST, it must have met at least one of the following criteria: [5] (1.) originated in a peripheral nerve; (2.) originated in a pre-existing nerve sheath tumor (rare, other than neurofibroma in NF1); (3.) exhibited ultrastructural features of Schwannian differentiation. Multiple systems