A new prognostic scoring scale for patients with primary WHO grade III gliomas based on molecular predictors

DOI 10.1007/s11060-012-1026-x CLINICAL STUDY A new prognostic scoring scale for patients with primary WHO grade III gliomas based on molecular predictors Haihui Jiang Xiaohui Ren Wei Zhang Jun Ma Dali Sui Zhongli Jiang Xiangli Cui Song Lin Received: 10 August 2012 / Accepted: 7 December 2012 Ó Springer Science+Business Media New York 2012 Abstract This study was designed to select molecular markers associated with prognosis, and to propose a prognostic scoring scale for patients with primary WHO grade III gliomas based on these molecular predictors. A series of 83 grade III glioma patients surgically treated and pathologically confirmed in Beijing Tiantan Hospital between May 2009 and December 2010 were retrospectively reviewed in the study. Log-rank analysis was used to identify molecular markers associated with progression-free survival (PFS) and overall survival (OS), which were further assessed using Cox regression analysis. Based on the prognostic molecular markers, a scoring scale was proposed and Kaplan Meier plots were compared between different scoring levels by Log-rank method. Age \50, 1p/19q co-deletion, IDH1/2 mutation, negative MGMT and EGFR expression were correlated with longer PFS and OS. Cox regression confirmed age \50 and 1p/19q co-deletion as independent prognostic markers. This scoring scale mainly based on prognostic molecular markers stratified patients into four levels with different prognoses. Longer PFS and OS were correlated with higher scores (P \ 0.05). This scoring scale Haihui Jiang and Xiaohui Ren contributed equally to this study. Electronic supplementary material The online version of this article (doi:10.1007/s11060-012-1026-x) contains supplementary material, which is available to authorized users. H. Jiang X. Ren W. Zhang J. Ma D. Sui Z. Jiang S. Lin (&) Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing 100050, China e-mail: linsong2005@126.com X. Cui (&) Pharmacy, Beijing Tiantan Hospital, Capital Medical University, Beijing 100050, China based on prognostic molecular markers identified four levels with significantly different prognoses, and could be used to predict the prognosis of patients with primary WHO grade III gliomas. Keywords Scoring scale Glioma Grade III Molecular biomarker Prognosis Introduction Molecular pathological examination has been popularly used for glioma patients due to their merits in diagnosis and treatment. Many studies have been designed to investigate the correlation between molecular biomarkers and prognosis for glioma patients [1]. However, studies that focus on molecular biomarkers systematically and comprehensively in patients with grade III gliomas are few. Furthermore, the predictive value of some biomarkers on survival, such as O6- methylguanine-dna methyltransferase (MGMT), remains to be verified [2]. As a result, there has been an emphasis on the merits of these molecular biomarkers in survival prediction for patients with WHO grade III gliomas in this study. The single role of molecular biomarker has been widely reported for glioma patients; however, the combining role of multiple biomarkers is still unknown. Besides, the weight of each biomarker in prognostic prediction is unknown, either. In order to analyze the combining role of multiple molecular biomarkers systemically and comprehensively, molecular parameters routinely detected in our hospital were evaluated by univariate and multivariate methods in the study, including 1p/19q, IDH1/2, MGMT, EGFR, Gp170, PTEN, MMP-9, EGFR, P53, VEGF, TOPO-II, GST-p, GFAP, MDM-2, and Ki-67.

Although the prognoses of patients with WHO grade III gliomas are not so favorable, individual survival remains heterogeneous. This study was designed to evaluate the combining role of multiple molecular markers associated with prognoses and to propose a scoring scale based on these prognostic markers. And then the prognostic value of this scoring scale was assessed in patients with grade III gliomas. Materials and methods Patient selection A series of 83 patients (55 males and 28 females) underwent surgical resection for primary WHO grade III gliomas at Beijing Tiantan Hospital from May 2009 to December 2010. All specimens were independently re-evaluated by two experienced neuro-pathologists, who were blind to the clinical outcome of the patients, according to the 2007 WHO classification system of central nervous system tumors. In case of a discrepancy, the two observers simultaneously reviewed the slides to achieve a consensus. Patients were treated according to NCCN guideline for gliomas. Recorded variables The clinical, operative, and hospital course records of 83 patients who met the inclusion criteria were retrospectively reviewed. The following information was recorded including patient s age, gender, removal degree, adjuvant therapy, and molecular parameters. The molecular parameters in this study included 1p/19q, IDH1/2, MGMT, EGFR, P-gp, PTEN, MMP-9, EGFR, P53, VEGF, TOPO- II, GST-p, GFAP, MDM-2, and Ki-67. The removal degree was classified according to MR images obtained less than 72 h after tumor resection as GTR ([99 % resection), NTR (95 99 %), or STR (80 05 %) [3]. 1p/19q was detected by fluorescence in situ hybridization (FISH) method, and IDH1/2 was sequenced as was described [4]. The other molecular parameters were detected by immunohistochemical method. Immunohistochemical analysis 5 lm sections were prepared. After deparaffinization and rehydration, the slides were boiled in 10 mm citrate buffer (ph 6.0) for 12 min or digested with pepsin at 37 C for 10 min to expose the antigen epitope according to the antigen retrieval technique in Supplementary Table S1. Endogenous peroxidase was blocked with 3 % aqueous hydrogen peroxide. The sections were incubated with primary antibody at 4 C overnight. Then the sections were washed with PBS for three times and incubated with the secondary antibody at 37 C for 20 min. The antibodies were then detected with diaminobenzidine as a chromogen. The slides were counterstained with hematoxylin. All the cases were assessed using a semi-quantitative scoring system. The expression levels were based on the percentage of immunopositive cells (low, -, \10 % of tumor cells; high,?, C10 % of tumor cells). Follow-up Recurrence of gliomas was defined according to the radiology after GTR, NTR or STR. Recurrence-free survival (PFS) was defined as the time between initial surgery and tumor recurrence on radiology. Patients who were recurrence-free at last follow-up were considered as a censored event in analysis. Overall survival (OS) was defined as the time between initial surgery and death. Patients who were still alive at last follow-up were considered as a censored event in analysis. Statistical analysis Survival as a function of time was plotted using the Kaplan Meier method. Log-rank analysis was used to compare Kaplan Meier plots and identify molecular markers associated with prognosis. The multivariate proportional hazards regression analysis was used to identify molecular markers independently associated with prognosis. Based on the molecular markers correlated with longer survival, a prognostic scoring scale was proposed. In detail, 2 points was assigned for each independently prognostic marker and 1 point was assigned for each of the other prognostic markers. Then patients were stratified into several levels according to the sum of prognostic scores. The PFS and OS between different levels were compared by Log-rank method to test the significance of this scoring scale. This was done after controlling for clinical, operative, and pathological factors that have been shown to be associated with prognosis including tumor location, removal degree, pathology, and adjuvant therapy [3]. Chi square (v 2 )or Fisher s exact test was used for the comparison of differences among four levels. SPSS 13.0 (SPSS for Windows, version 13.0 [SPSS Inc., Chicago, Illinois, USA]) was used for statistical analysis. Probability value was obtained from 2-sided tests, with a statistical significance of P \ 0.05. Results Basic characteristics Basic characteristics of 83 patients with WHO grade III glioma in this study were summarized in Table 1. Their ages

at onset ranged from 17 to 67 years; the mean age was 43.5 ± 12.4 years (all means are expressed ±SD). These patients included 55 male and 28 female, and the sex ratio was 2.0. This cohort included 17 anaplastic oligodendroglioma (AO), 35 anaplastic oligoastrocytoma (AOA), and 31 anaplastic astrocytoma (AA). GTR was achieved in 60 (72.3 %) patients, NTR in 19 (22.9 %) patients, and STR in 4 (4.8 %) patients. 75 (90.4 %) patients underwent chemotherapy and 74 (89.2 %) underwent radiotherapy. Of 83 gliomas, 35 (42.2 %) had 1p/19q co-deletion, 32 (38.6 %) had IDH 1/2 mutation, 68 (81.9 %) had positive MGMT expression, and 60 (72.3 %) had positive EGFR expression. The median PFS for patients with AO, AOA, and AA was unavailable, unavailable, and 18.0 s, respectively. The Kaplan Meier PFS plots between them were significantly different (P = 0.029). The median OS for patients with AO, AOA, and AA was unavailable, 19.0 and 9 s, respectively. The Kaplan Meier OS plots between them were significantly different (P = 0.004). Molecular markers associated with prognosis All the molecular markers available in this study were evaluated including 1p/19q, IDH1/2, MGMT, EGFR, Gp170, PTEN, MMP-9, EGFR, P53, VEGF, TOPO-II, GST-p, GFAP, MDM-2, and Ki-67. Log-rank analysis was used to compare the Kaplan Meier plots and prognostic markers were identified, including 1p/19q, IDH1/2, MGMT and EGFR. Age was also identified as a prognostic marker. Data was shown in Table 2. Cox regression confirmed age \50 and 1p/19q co-deletion as independent markers associated with longer PFS and OS. Age \50 was associated with longer survival Age \50 was correlated with longer PFS and longer OS in univariate and multivariate analyses. Among 83 patients with grade III gliomas, 55 (66.3 %) patients were\50 years and 28 (33.7 %) C50. Of 28 patients with age C50, 20 (71.4 %) recurred with the median PFS of 8.0 s while 19 out of 55 (34.5 %) patients with age \50 recurred with unavailable median PFS. For patients with age C50, the 6.0-, 12.0-, 18.0-, and 24.0- PFS rates were 64.3, 42.9, 28.6, and 0.0 %, respectively, while the rates for patients with age \50 were 90.9, 76.4, 63.9, and 60.6 %, respectively, (P \ 0.001) (Table 2; Supplementary Fig. S1A). Of the 28 patients with age C50, 20 (71.4 %) died with the median OS of 18.0 s while 14 out of 55 (25.5 %) patients with age\50 died with unavailable median OS. The 6.0-, 12.0-, 18.0-, and 24.0- OS rates for patients with age C50 were 89.3, 64.1, 45.8, and 22.9 %, respectively, Table 1 Clinicopathological characteristics of 83 patients with primary WHO grade III gliomas Characteristic No. of patients (%) Age (years) Mean 43.5 ± 12.4 Range 17 67 Sex Male 55 (66.3) Female 28 (33.7) Tumor location Frontal 34 (41.0) Temporal 26 (31.3) Parietal 15 (18.1) Others 8 (9.6) Tumor resection GTR 60 (72.3) NTR 19 (22.9) STR 4 (4.8) Chemotherapy Yes 75 (90.4) No 8 (9.6) Radiotherapy Yes 74 (89.2) No 9 (10.8) Pathology AA 31 (37.3) AOA 35 (42.2) AO 17 (20.5) 1p/19q co-deletion Yes 35 (42.2) No 48 (57.8) IDH1/2 mutation Yes 32 (38.6) No 51 (61.4) MGMT Negative 15 (18.1) Positive 68 (81.9) EGFR Negative 23 (27.7) Positive 60 (72.3) while the rates for patients with age \50 were 100.0, 85.4, 74.1, and 71.0 %, respectively, (P \ 0.001) (Table 2; Supplementary Fig. S1B).

Table 2 Molecular markers correlated with PFS and OS in patients with primary grade III gliomas Characteristics Median PFS (INR) Progression-free survival rates P value Median OS (INR) 6-12- 18-24- Overall survival rates 6-12- 18-24- P value Age (years) \50 N/A 90.9 76.4 63.9 60.6 \0.001 N/A 100.0 85.4 74.1 71.0 \0.001 C50 8.0 (5.4 10.6) 64.3 42.9 28.6 0.0 18.0 (15.6 20.4) 89.3 64.1 45.8 22.9 1p/19q co-deletion Yes N/A 97.1 94.3 79.4 68.7 \0.001 N/A 100.0 94.3 91.0 86.7 \0.001 No 9.0 (6.0 12.0) 70.8 43.8 33.1 0.0 18.0 (16.8 19.2) 93.8 66.7 42.8 21.9 IDH1/2 mutation Yes N/A 96.9 84.4 75.4 63.7 0.002 N/A 100.0 93.4 81.2 71.0 0.004 No 14.0 (8.2 19.8) 72.5 52.9 38.6 34.3 20.0 (15.4 24.6) 94.1 68.6 52.9 40.4 MGMT expression Negative N/A 100.0 93.3 84.8 84.8 0.002 N/A 100.0 100.0 86.7 86.7 0.004 Positive 7.0 (12.1 21.9) 77.9 58.8 45.1 25.4 21.5(17.6 25.4) 95.6 73.3 58.6 42.7 EGFR Expression Negative N/A 87.0 82.6 76.7 65.7 0.01 N/A 100.0 87.0 81.2 74.9 0.022 Positive 16.5 (9.7 23.3) 80.0 58.3 42.2 35.2 21.5 (15.8 27.2) 100.0 74.8 56.7 43.4 1p/19q Co-deletion was associated with longer survival 1p/19q Co-deletion was correlated with longer PFS and OS in patients with grade III gliomas. Of 48 patients without 1p/19q co-deletion, 31 (64.6 %) recurred with the median PFS of 9.0 s while 8 out of 35 (22.9 %) patients with 1p/19q co-deletion recurred with unavailable median PFS. The 6.0-, 12.0-, 18.0-, and 24.0- PFS rates for patients without 1p/19q co-deletion were 70.8, 43.8, 33.1, and 0.0 %, respectively, while the rates for patients with 1p/19q co-deletion were 97.1, 94.3, 79.4, and 68.7 %, respectively, (P \ 0.001) (Table 2; Supplementary Fig. S2A). Of the 48 patients without 1p/19q co-deletion, 29 (60.4 %) died with median OS of 18.0 s while 5 out of 35 (14.3 %) with 1p/19q co-deletion died with unavailable OS. The 6.0-, 12.0-, 18.0-, and 24.0- OS rates for patients without 1p/19q co-deletion were 93.8, 66.7, 42.8, and 21.9 %, respectively, while the rates for patients with 1p/19q co-deletion were 100.0, 94.3, 91.0, and 86.7 %, respectively, (P \ 0.001) (Table 2; Supplementary Fig. S2B). IDH1/2 mutation is correlated with prolonged survival 30 out of 51 (58.8 %) patients with wild-type IDH1/2 recurred with the median PFS of 14.0 s, while 9/32 (28.1 %) with mutated IDH1/2 recurred with unavailable median PFS. The 6.0-, 12.0-, 18.0-, and 24.0- PFS rates for patients without IDH1/2 mutation were 72.5, 52.9, 38.6, and 34.3 %, respectively, while the rates for patients with IDH1/2 mutation were 96.9, 84.4, 75.4, and 63.7 %, respectively, (P = 0.002) (Table 2; Supplementary Fig. S3A). 27/51 (52.9 %) grade III gliomas with wild-type IDH1/2 died with the median OS of 20.0 s while 7 out of 32 (21.9 %) patients with mutated IDH1/2 died with unavailable median OS. The 6.0-, 12.0-, 18.0- and 24.0- OS rates of patients without IDH1/2 mutation were 94.1, 68.6, 52.9, and 40.4 %, respectively, while the rates for patients with IDH1/2 mutation were 100.0, 93.4, 81.2, and 71.0 %, respectively, (P = 0.004) (Table 2; Supplementary Fig. S3B). Negative MGMT expression was associated with longer survival Negative MGMT expression was correlated with longer PFS and OS. Of 68 patients with positive MGMT expression, 37 (54.4 %) recurred with the median PFS of 7.0 s while 3 out of 15 (20.0 %) with negative MGMT expression recurred with unavailable median PFS. The 6.0-, 12.0-, 18.0-, and 24.0- PFS rates for patients with positive expression were 77.9, 58.8, 45.1, and 25.4 %, respectively while the rates for patients with negative expression were 100.0, 93.3, 84.8, and 84.8 %, respectively, (P = 0.002) (Table 2; Supplementary Fig. S4A). 32/68 (47.1 %) with positive MGMT expression died with median OS of 21.5 s while 2/15 (13.3 %) patients with negative MGMT expression died with unavailable median OS. The 6.0-, 12.0-, 18.0-, and 24.0- OS rates for patients with positive MGMT expression were 95.6, 73.3, 58.6, and 42.7 %, respectively, while the rates for patients with negative MGMT expression were

100.0, 100.0, 86.7, and 86.7 %, respectively. The difference was significant (P = 0.004) (Table 2; Supplementary Fig. S4B). Negative EGFR expression was associated with longer survival Negative EGFR expression was correlated with longer PFS and OS. Of 60 patients with positive EGFR expression, 33 (55.0 %) recurred with the median PFS of 16.5 s while 6 out of 23 (26.1 %) with negative EGFR expression recurred with unavailable median PFS. The 6.0-, 12.0-, 18.0-, and 24.0- PFS rates for patients with positive expression were 80.0, 58.3, 42.2, and 35.2 %, respectively while the rates for patients with negative expression were 87.0, 82.6, 76.7, and 65.7 %, respectively, (P = 0.01) (Table 2; Supplementary Fig. S5A). 29/60 (48.3 %) with positive EGFR expression died with median OS of 21.5 s while 5/23 (21.7 %) patients with negative EGFR expression died with unavailable median OS. The 6.0-, 12.0-, 18.0-, and 24.0- OS rates for patients with positive EGFR expression were 100.0, 74.8, 56.7, and 43.4 %, respectively, while the rates for patients with negative EGFR expression were 100.0, 87.0, 81.2, and 74.9 %, respectively. The difference was significant (P = 0.022) (Table 2; Supplementary Fig. S5B). Independent factors predicting longer survival by Cox regression In Cox proportional hazards regression analysis for PFS, the factors that remained significantly correlated with longer PFS were age \50 (OR 0.443 [95 % CI 0.231 0.849)], P = 0.014 and 1p/19q co-deletion (OR 0.206 [95 % CI 0.087 0.486], P \ 0.001). In Cox proportional hazards regression analysis for OS, the factors that remained significantly correlated with longer OS were age \50 (OR 0.441 [95 % CI 0.216 0.900], P = 0.025) and 1p/19q co-deletion (0.179 [95 % CI 0.066 0.488], P = 0.001). A prognostic scoring scale for patients with WHO grade III gliomas Based on the above molecular markers associated with longer survival, a prognostic scoring scale was proposed (Table 3). In detail, 2 points was assigned for each independent prognostic marker (1p/19q co-deletion and age \50), while 1 point was assigned for each of the other prognostic markers (IDH1/2 mutation, negative MGMT and EGFR expression). Therefore, patients could have a prognostic score ranging from 0 to 7 points. Patients with a score of 0 1, 2 3, 4 5 and 6 7 were further stratified into level 1, 2, 3 and 4, respectively (Table 4). After controlling tumor location, extent of resection, histology subtype, and adjuvant therapy, Kaplan Meier plots were compared between four levels. The percentage of patients receiving chemotherapy was 85.7 % (12/14) in level 1, 88.2 % (15/17) in level 2, 90.6 % (29/32) in level 3 and 95.0 % (19/20) in level 4. There was no significant difference between four levels. Similarly, the percentage of patients receiving radiotherapy was 85.7 % (12/14) in level 1, 82.4 % (14/17) in level 2, 90.6 % (29/32) in level 3 and 95.0 % (19/20) in level 4. There was no significant difference between four levels. Gross total resection was achieved in 78.6 % (11/14) of the patients in level 1, 70.6 % (12/17) in level 2, 71.9 % (23/32) in level 3 and 70.0 % (14/20) in level 4. There were no significant differences between four levels. There was no significant difference in tumor location or histology subtype, either (Supplementary Table S2). For patients in level 1, the median PFS and OS were 7.0, 18.0 s, respectively. For patients in level 2, the median PFS and OS were 14.0 and 20.0 s, respectively. For patients in level 3, the median PFS and OS were 24.0 s and unavailable, respectively. For patients in level 4, either median PFS or median OS was unavailable (Table 4). In log-rank analysis, the PFS plots for patients between different levels were significantly different (P \ 0.001, Fig. 1a). Patients in level 1 had a significantly longer PFS than patients in level 2 (P = 0.043), level 3 (P = 0.002), and level 4 (P \ 0.001). Similarly, patients in level 2 had a significantly longer PFS than patients in level 3 (P = 0.015) and level 4 (P \ 0.001). Finally, patients in level 3 had a significantly longer PFS than patients in level 4 (P = 0.034). Meanwhile, the OS plots for patients between different levels were significantly different (P \ 0.001, Fig. 1b). Patients in level 1 had a significantly longer OS than patients in level 2 (P = 0.095), level 3 (P \ 0.001), level 4 (P \ 0.001). Similarly, patients in level 2 had a significantly longer OS than patients in level 3 (P = 0.012) and level 4 (P = 0.001). Although the OS plots between patients in level 3 and level 4 was not statistically significant, it showed a trend towards being significantly different (P = 0.154). Table 3 A scoring scale for patients with primary WHO grade III gliomas based on molecular predictors Markers Yes No Age \50 2 0 1p/19q Co-deletion 2 0 IDH1/2 mutation 1 0 Negative MGMT expression 1 0 Negative EGFR expression 1 0

Table 4 Stratification of patients with WHO grade III gliomas based on the scoring scale Stratification Score Median PFS Progression-free survival rate* P value* Median 6-12- 18-24- OS Overall survival rate* 6-12- 18-24- P value* Level 1 (20 pts) Level 2 (32 pts) Level 3 (17 pts) Level 4 (14 pts) 0 1 7.0 55.0 30.0 20.0 0.0 \0.001 18.0 85.0 60.0 36.0 9.0 \0.001 2 3 14.0 81.3 56.3 41.3 0.0 20.0 100.0 68.0 54.6 48.5 4 5 24.0 100.0 94.1 79.0 45.1 N/A 100.0 100.0 92.9 82.5 6 7 N/A 100.0 100.0 91.7 91.7 N/A 100.0 100.0 100.0 100.0 * Log-rank survival analyses confirmed significant difference between each two levels Fig. 1 In the entire cohort of 83 patients, this scoring scale identified four levels with different PFS (A, P \ 0.001) and OS (B, P \ 0.001) Discussion This study was aimed to select molecular markers associated with prognosis and further to propose a scoring scale to predict prognosis for patients with primary WHO grade III gliomas. In this series of the 83 patients with primary WHO grade III gliomas, Log-rank analysis revealed that age\50, 1p/19q co-deletion, IDH1/2 mutation, negative MGMT and EGFR expression were prognostic markers. Furthermore, Cox regression confirmed age\50 and 1p/19q co-deletion as independent prognostic markers. Based on these prognostic molecular markers, a prognostic scoring scale was proposed. Prognostic scoring scale based on molecular markers This prognostic scoring scale is the first systematic prognostic scale based on molecular markers. Although some classification systems have been proposed for gliomas patients, all of them were based on clinical parameters. In 2004, Lamborn et al. [5] found that age\40, tumor in the frontal lobe, KPS [70, subtotal or total resection, and without chemotherapy were all independent factors associated with decreased survival. In 2008, Chang et al. [6] proposed a classification system for adult patients with hemispheric low-grade gliomas based on age [50 years, Karnofsky performance scale score B80, presumed eloquent location and tumor diameter [4 cm.in 2010, Chaichana et al. [7] used age[60 years, Karnofsky performance scale score B80, motor deficit, language deficit and periventricular tumor location to categorize glioblastoma patients preoperatively. In 2011, Chaichana et al. [3] found that KPS score\80, chronic obstructive pulmonary disease, motor deficit, language deficit, cognitive deficit, and tumor size [4 cm were all preoperative factors independently associated with decreased survival for older patients with GBM. In our present study, for the first time, we comprehensively and systematically analyzed the association between prognosis and molecular biomarkers and proposed a prognostic scale for patients with grade III gliomas based on molecular markers associated with prognosis. In this prognostic scale, independent molecular markers were assigned more impact weight in predicting prognosis. Previous studies utilized clinical factors to predict survival for patients with gliomas, and each predictor was assigned equal impact in survival prediction. In this study, more impact weight was assigned for independent factors. For example, 2 points was assigned for each factor (1p/19q co-deletion and

age\50) independently associated with longer survival, while 1 point for each of other factors (IDH1/2 mutation, negative MGMT and EGFR expression). Regardless of the status of IDH1/2 mutation and MGMT/EGFR expression, patients with 1p/19q co-deletion or age \50 years had significantly longer survival than those without such markers. Therefore, 1p/19q and age were assigned more impact weight on survival than IDH1/2, MGMT, and EGFR. The prognostic scoring scale based on these molecular markers could be used to predict prognosis in patients with primary WHO grade III gliomas. According to the scale, patiens were stratified into four levels and patients of different levels had different PFS and OS (P \ 0.05). For patients in level 1, the median PFS and OS were 7.0 s, 18.0 s, respectively. For patients in level 2, the median PFS and OS were 14.0 s and 20.0 s, respectively. For patients in level 3, the median PFS and OS were 24.0 s and unavailable, respectively. For patients in level 4, either median PFS or median OS was unavailable (Table 4). This is the first prognostic scoring scale for patients with primary WHO grade III gliomas based on molecular markers. And it could be used in prognostic stratification and therapeutic decision. Molecular biomarkers correlated with prognoses Younger age was an independent prognostic factor for longer survival in patients with primary WHO grade III gliomas. This was in accordance with many prior studies [5, 7, 8]. Older age may result in a decreased ability for patients to withstand neurological insults caused by the tumor, surgery, and/or adjuvant therapies [5, 8, 9]. Additionally, older patients were believed to be with more mutations, causing these tumors to behave more aggressively and infiltrate more widely [10 12]. Older patients also tended to have more medical co-morbidities, making it more difficult for them to survive long surgical time and prolonged hospital course [10 12]. So these may be the reasons for younger age serving as an important predictive factor that was associated with longer survival in Chinese patients with WHO grade III gliomas. 1p/19q co-deletion was an independent prognostic factor for longer survival in patients with primary WHO grade III gliomas. Aldape et al. [13] found that 1p/19q co-deletion was associated with improved prognosis and could serve as a clinically useful marker for glioma patients. Co-deletion of chromosome arms 1p and 19q have been associated with favorable response to chemotherapy and good prognosis, but the molecular mechanisms responsible for this correlation are as yet unknown [14]. Due to the association of 1p/19q co-deletion with favorable chemo-sensitivity and prognosis, and the histological diagnostic facilities with such a marker, such genetic testing should be performed routinely for these tumors [15]. Our results demonstrated that patients with IDH1/2 mutations had longer survival than those without such mutation in patients with primary WHO grade III gliomas. Yan et al. [16] found that glioblastoma patients with IDH1/2 mutations had a better outcome than those with wild-type IDH1/2 genes. Moreover, grade II and III gliomas with IDH1 mutation also have a better survival compared with those without such mutations [17 20]. Biochemical studies have provided evidence that mutant IDH1/2 may act in a dominant-negative fashion to inhibit catalytic activity of such enzymes. The IDH1/2 enzymes catalyze the oxidative decarboxylation of isocitrate to a-ketoglutarate generating NADPH from NADP?. Biochemical studies of recombinant IDH1 proteins revealed that mutant IDH1 had a dramatically reduced affinity for isocitrate [21]. Furthermore, by assaying the reduction of NADP? to NADPH, all IDH1/2 mutations identified in patients were found to abrogate their enzymatic activity [21, 22]. Therefore, patients with IDH1/2 mutations have a longer survival; IDH1/2 mutations can also serve as a favorable prognostic marker in grade III gliomas. Negative MGMT expression was a predictive factor for longer survival in patients with primary WHO grade III gliomas. Negative expression of MGMT was associated with an improved response to alkylating agents, and glioma patients with this trait had better prognosis [23]. Positive MGMT expression may cause resistance to DNA-alkylating drugs commonly used in glioma chemotherapy [14]. The transcriptional activity of the MGMT gene may be downregulated by hypermethylation of a CpG island located in the 5 0 -region of the gene [24, 25]. Although the methylation of MGMT promoter was not detected in our study, it is proved that immunohistochemical staining for MGMT protein was a prognostic marker for longer survival. Besides, MGMT promoter hypermethylation has been closely associated with negative MGMT protein expression in various types of human tumors [24, 26]. In this study, negative EGFR expression is correlated with longer survival in patients with primary WHO grade III gliomas. Barker et al. [27] undertook a research to investigate the correlation between EGFR expression and radiation response in glioblastoma multiforme, and found that 33 % of tumors with negative EGFR immunoreactivity had better responses to radiation compared with 18 % of tumors with intermediate staining and 9 % of tumors with strong staining. Barker et al. [27] also found that positive EGFR expression correlated with increased tumor radioresistance (P = 0.046). Study limitation Some limitations existed in this study. This series of patients was not so large. In the future we would expand our sample for further research. Besides, methylation of MGMT promoter

was not included in this study due to its high expense, which was proved to be a prognostic marker in some researches. In the following research, the role of MGMT promoter methylation will be detected and assessed in Chinese patients with primary WHO grade III gliomas. In conclusion, the molecular markers associated with longer PFS and OS are age\50, 1p/19q co-deletion, IDH1/ 2 mutation, negative MGMT and EGFR expression in patients with primary WHO grade III gliomas. Patients with an increasing number of these factors would have a longer survival. A prognostic scoring scale based on these molecular markers was proposed and patients were stratified into four distinct levels with different prognoses. This scoring scale may provide patients and doctors with useful prognostic information, and help choose individualized therapeutic strategy. Statement of translational relevance This study was designed to select molecular markers associated with prognosis, and propose a prognostic scoring scale for patients with primary WHO grade III gliomas based on these molecular predictors. Age \50, 1p/19q codeletion, IDH1/2 mutation, negative MGMT and EGFR expression were correlated with longer PFS and OS. Cox regression confirmed age \50 and 1p/19q co-deletion as independent prognostic markers. This scoring scale based on these prognostic markers stratified patients into four levels with different prognoses. Longer PFS and OS were correlated with higher scores. So, this scoring scale based on prognostic molecular markers identified four levels with significantly different prognoses, and could be used to predict prognosis and choose individualized therapeutic strategy for patients with primary WHO grade III gliomas. Acknowledgments This work was supported by Beijing Natural Science Foundation (7122061). Conflicts of interest No potential conflicts of interest were disclosed. References 1. Masui K, Cloughesy TF, Mischel PS (2012) Review: molecular pathology in adult high-grade gliomas: from molecular diagnostics to target therapies. Neuropathol Appl Neurobiol 38:271 291 2. Bourne TD, Schiff D (2010) Update on molecular findings, management and outcome in low-grade gliomas. Nat Rev Neurol 6:695 701 3. Chaichana KL, Chaichana KK, Olivi A, Weingart JD, Bennett R, Brem H, Quinones-Hinojosa A (2011) Surgical outcomes for older patients with glioblastoma multiforme: preoperative factors associated with decreased survival. Clinical article. J Neurosurg 114:587 594 4. Ren X, Cui X, Lin S, Wang J, Jiang Z, Sui D, Li J, Wang Z (2012) Co-deletion of chromosome 1p/19q and IDH1/2 mutation in glioma subsets of brain tumors in Chinese patients. PLoS ONE 7:e32764 5. Lamborn KR, Chang SM, Prados MD (2004) Prognostic factors for survival of patients with glioblastoma: recursive partitioning analysis. Neuro Oncol 6:227 235 6. Chang EF, Smith JS, Chang SM, Lamborn KR, Prados MD, Butowski N, Barbaro NM, Parsa AT, Berger MS, McDermott MM (2008) Preoperative prognostic classification system for hemispheric low-grade gliomas in adults. J Neurosurg 109:817 824 7. Chaichana K, Parker S, Olivi A, Quinones-Hinojosa A (2010) A proposed classification system that projects outcomes based on preoperative variables for adult patients with glioblastoma multiforme. J Neurosurg 112:997 1004 8. Laws ER, Parney IF, Huang W, Anderson F, Morris AM, Asher A, Lillehei KO, Bernstein M, Brem H, Sloan A, Berger MS, Chang S (2003) Survival following surgery and prognostic factors for recently diagnosed malignant glioma: data from the glioma outcomes project. J Neurosurg 99:467 473 9. Krex D, Klink B, Hartmann C, von Deimling A, Pietsch T, Simon M, Sabel M, Steinbach JP, Heese O, Reifenberger G, Weller M, Schackert G (2007) Long-term survival with glioblastoma multiforme. Brain 130:2596 2606 10. Alonso ME, Bello MJ, Gonzalez-Gomez P, Arjona D, Lomas J, de Campos JM, Isla A, Sarasa JL, Rey JA (2003) Aberrant promoter methylation of multiple genes in oligodendrogliomas and ependymomas. Cancer Genet Cytogenet 144:134 142 11. Rickert CH, Strater R, Kaatsch P, Wassmann H, Jurgens H, Dockhorn-Dworniczak B, Paulus W (2001) Pediatric high-grade astrocytomas show chromosomal imbalances distinct from adult cases. Am J Pathol 158:1525 1532 12. Sung T, Miller DC, Hayes RL, Alonso M, Yee H, Newcomb EW (2000) Preferential inactivation of the p53 tumor suppressor pathway and lack of EGFR amplification distinguish de novo high grade pediatric astrocytomas from de novo adult astrocytomas. Brain Pathol 10:249 259 13. Aldape K, Burger PC, Perry A (2007) Clinicopathologic aspects of 1p/19q loss and the diagnosis of oligodendroglioma. Arch Pathol Lab Med 131:242 251 14. Mollemann M, Wolter M, Felsberg J, Collins VP, Reifenberger G (2005) Frequent promoter hypermethylation and low expression of the MGMT gene in oligodendroglial tumors. Int J Cancer 113:379 385 15. Idbaih A, Omuro A, Ducray F, Hoang-Xuan K (2007) Molecular genetic markers as predictors of response to chemotherapy in gliomas. Curr Opin Oncol 19:606 611 16. Yan H, Bigner DD, Velculescu V, Parsons DW (2009) Mutant metabolic enzymes are at the origin of gliomas. Cancer Res 69:9157 9159 17. Dubbink HJ, Taal W, van Marion R, Kros JM, van Heuvel I, Bromberg JE, Zonnenberg BA, Zonnenberg CB, Postma TJ, Gijtenbeek JM, Boogerd W, Groenendijk FH, Smitt PA, Dinjens WN, van den Bent MJ (2009) IDH1 mutations in low-grade astrocytomas predict survival but not response to temozolomide. Neurology 73:1792 1795 18. Sanson M, Marie Y, Paris S, Idbaih A, Laffaire J, Ducray F, El HS, Boisselier B, Mokhtari K, Hoang-Xuan K, Delattre JY (2009) Isocitrate dehydrogenase 1 codon 132 mutation is an important prognostic biomarker in gliomas. J Clin Oncol 27:4150 4154 19. van den Bent MJ, Dubbink HJ, Marie Y, Brandes AA, Taphoorn MJ, Wesseling P, Frenay M, Tijssen CC, Lacombe D, Idbaih A, van Marion R, Kros JM, Dinjens WN, Gorlia T, Sanson M (2010) IDH1 and IDH2 mutations are prognostic but not predictive for

outcome in anaplastic oligodendroglial tumors: a report of the European Organization for Research and Treatment of Cancer Brain Tumor Group. Clin Cancer Res 16:1597 1604 20. Weller M, Felsberg J, Hartmann C, Berger H, Steinbach JP, Schramm J, Westphal M, Schackert G, Simon M, Tonn JC, Heese O, Krex D, Nikkhah G, Pietsch T, Wiestler O, Reifenberger G, von Deimling A, Loeffler M (2009) Molecular predictors of progression-free and overall survival in patients with newly diagnosed glioblastoma: a prospective translational study of the German glioma network. J Clin Oncol 27:5743 5750 21. Zhao S, Lin Y, Xu W, Jiang W, Zha Z, Wang P, Yu W, Li Z, Gong L, Peng Y, Ding J, Lei Q, Guan KL, Xiong Y (2009) Glioma-derived mutations in IDH1 dominantly inhibit IDH1 catalytic activity and induce HIF-1alpha. Science 324:261 265 22. De Carli E, Wang X, Puget S (2009) IDH1 and IDH2 mutations in gliomas. N Engl J Med 360(2248):2249 23. Sasai K, Nodagashira M, Nishihara H, Aoyanagi E, Wang L, Katoh M, Murata J, Ozaki Y, Ito T, Fujimoto S, Kaneko S, Nagashima K, Tanaka S (2008) Careful exclusion of non-neoplastic brain components is required for an appropriate evaluation of O6- methylguanine-dna methyltransferase status in glioma: relationship between immunohistochemistry and methylation analysis. Am J Surg Pathol 32:1220 1227 24. Esteller M, Herman JG (2004) Generating mutations but providing chemosensitivity: the role of O6-methylguanine DNA methyltransferase in human cancer. Oncogene 23:1 8 25. Gerson SL (2004) MGMT: its role in cancer aetiology and cancer therapeutics. Nat Rev Cancer 4:296 307 26. Esteller M, Hamilton SR, Burger PC, Baylin SB, Herman JG (1999) Inactivation of the DNA repair gene O6-methylguanine- DNA methyltransferase by promoter hypermethylation is a common event in primary human neoplasia. Cancer Res 59: 793 797 27. Barker FN, Simmons ML, Chang SM, Prados MD, Larson DA, Sneed PK, Wara WM, Berger MS, Chen P, Israel MA, Aldape KD (2001) EGFR overexpression and radiation response in glioblastoma multiforme. Int J Radiat Oncol Biol Phys 51: 410 418