Predicting O6-Methylguanine-DNA Methyltransferase Protein Expression in Primary Low- and High-Grade Gliomas Using Certain Qualitative Characteristics of Amide Proton Transfer-Weighted Magnetic Resonance Imaging.

OBJECTIVE:To demonstrate that certain qualitative amide proton transfer-weighted (APTw) characteristics can provide practical imaging clues for predicting O6-methylguanine-DNA methyltransferase (MGMT) protein expression in primary low- and high-grade gliomas, preoperatively and noninvasively. METHODS:Pathologically confirmed low- and high-grade gliomas with APT data and immunohistochemical (IHC) reports were recruited in this study. The MGMT protein expression status was classified by postsurgery specimen immunostaining. Subjects were divided into two groups, MGMT-positive and MGMT-negative group, according to the immunoreactivity of MGMT protein expression documented in IHC reports. APTw images scanned at 3T magnetic resonance preoperatively were retrospectively analyzed. Two neuroradiologists were trained to evaluate presence of certain APTw features. Kappa value was calculated to show the consistency between the 2 observers. The Mann-Whitney U test was used to evaluate relationships between the 2 groups on APTw features. Negative predictive value and positive predictive value was used to evaluate the ability of APTw characteristics in predicting MGMT protein expression. Receiver operating characteristic curve was used to evaluate the diagnostic performance of APTw characters. Two-tailed P < 0.05 was considered as statistically significant. RESULTS:Forty-two subjects were recruited in this study. Among them 38 specimens presented positive MGMT immunostaining (MGMT-positive group), 4 specimens were negative MGMT immunostaing (MGMT-negative group). There were, respectively, 37 and 5 APTw images appeared positive and negative APTw features. Differences between tumors of positive and negative MGMT expression on qualitative APTw features were significant (P = 0.020). The consistency coefficient of the 2 observers was 0.876 (kappa = 0.876). Three of five llgliomas with negative APTw features showed MGMT-negative immunostaining, leading to a negative predictive value of 60%, and 36 of 37 cases presenting positive APTw characteristics were tumors of MGMT-positive expression, generating a positive predictive value of 97.3%. The area under curve was 0.849. CONCLUSIONS:APTw characteristics could be promising imaging markers by which to predict IHC MGMT expression in primary low- and high-grade gliomas preoperatively and noninvasively.