Patients and tissue samples
The clinical records of MPNST patients diagnosed between January 6th, 1991 and December 1st, 2011 were evaluated to include only patients with sporadic MPNST and to exclude patients with NF1-associated or radiotherapy-induced MPNST. Clinicopathologic information was collected, including sex, age, tumor site, tumor size, American Joint Committee on Cancer (AJCC) stage, radiotherapy, chemotherapy, surgery type, disease recurrence, metastasis, and patient survival.
Tissue samples were obtained from the Department of Pathology, Tianjin Medical University Cancer Institute and Hospital. All tissues were collected before patients received radiotherapy and/or chemotherapy and were formalin-fixed and then embedded in paraffin. Two independent pathologists confirmed the diagnosis and ensured that at least 90% of cells in each specimen were tumor cells.
Tissue microarrays and immunohistochemistry
Tissue microarrays (TMAs) were constructed with the most representative areas of the tumor region as previously reported [21, 22]. Immunohistochemical (IHC) staining of TMAs was performed using the streptavidin–peroxidase method  with the following biotinylated primary antibodies: anti-CXCR4 antibody (1:100, ab124824, Abcam, Cambridge, UK), anti-CXCL12 antibody (1:100, sc-28876, Santa Cruz Biotechnology, Santa Cruz, CA, USA), and anti-Cyclin D1 antibody (1:50, Beijing Zhong Shan Golden Bridge Biotechnology Co. Ltd, Beijing, China).
The stained specimens were evaluated independently by two pathologists who were blinded to the clinicopathologic information and patient prognoses. Slides were examined by light microscopy, and high-power fields of five random areas per tumor were evaluated. Staining intensity was scored as 0 for no staining, 1 for light yellow, 2 for brownish yellow, and 3 for brownish black staining. The percentage of positive cells was scored as 0 for ≤ 10%, 1 for 11%–25%, 2 for 26%–50%, 3 for 51%–75%, and 4 for > 75%. The summation of the two scores was taken as the final IHC score: 0–1 (negative), 2–3 (weak positive), 4–5 (moderate positive), and 6–7 (strong positive). The expression level was further categorized into low (negative and weak positive) and high (moderate and strong positive) expression based on the final IHC score.
Associations between CXCR4, CXCL12, and Cyclin D1 expression levels and clinicopathologic variables were evaluated using the χ2 or Fisher’s exact tests. Correlations among the three protein levels were determined using Pearson’s correlation coefficient. The follow-up was stopped at April 14th, 2015. OS was defined as the duration from the date of diagnosis to the date of death or the last follow-up. Disease-free survival (DFS) was defined as the duration from the date of diagnosis to the confirmed disease progression or the last follow-up. Relationships between CXCR4, CXCL12, or Cyclin D1 expression levels and survival were evaluated using the Kaplan–Meier method and a log-rank test. The data of patients who were lost to follow-up or dead because of non-tumor-related diseases were censored. The Cox regression model was used for univariate and multivariate analyses. Analyses were performed using SPSS 19.0 software for Windows (SPSS Inc., Chicago, IL, USA). A P value < 0.05 was considered statistically significant.