Utility of standardized T2-weighted MRI intensity ratio for prediction of pituitary macroadenoma consistency

OBJECTIVE: The physical consistency of pituitary adenomas (PAs) is highly variable, ranging from soft/cystic to firm/calcified. The association between PA consistency and surgical/endocrinological outcomes has been well established, with firm tumors demonstrating poorer outcomes and higher complication rates. However, to date, no reliable means to determine PA consistency preoperatively exist, although T2- and diffusion-weighted imaging show early promise. As such, this study aimed to quantitatively analyze the value of normalized T2-weighted imaging (T2WI) and diffusion-weighted imaging ratios in the preoperative prediction of PA consistency. METHODS: The authors reviewed a prospectively maintained database of all patients undergoing PA resection at a single institution between 2011 and 2024. Inclusion criteria included the following: 1) a PA with a minimum diameter ≥ 20 mm in at least one dimension; 2) a consistency grade assigned at the time of surgery using a previously validated grading scale; and 3) preoperative MRI performed prior to surgery. Normalized tumor to cerebellar T2-weighted imaging intensity (TCTI) ratios were calculated using 10-mm2 regions of interest. Data were analyzed using Kruskal-Wallis tests, multivariable ordinal logistic regression models, and receiver operating characteristic (ROC) curve analyses with 4-fold cross-validation. RESULTS: A total of 189 patients (mean age 55.4 years, 49.2% female) were included, of whom 77 (40.7%) had PAs with a consistency grade of 1+2, 76 with a consistency grade of 3 (40.2%), and 36 (19%) with a consistency grade of 4+5. When using mean T2WI intensity measurements, TCTI ratios were lower for grade 4+5 tumors (1.48 ± 0.24, p < 0.001) than for grade 3 tumors (1.77 ± 0.44) and grade 1+2 tumors (2.16 ± 1.10, p = 0.001). When using maximum T2WI measurements, TCTI ratios were also lower for grade 4+5 tumors (1.50 ± 0.24, p < 0.001) compared with grade 3 tumors (1.79 ± 0.45), and grade 1+2 tumors (2.15 ± 0.81, p < 0.001). No association between apparent diffusion coefficient values and tumor consistency was observed. In the ROC analysis, comparing soft (grade 1+2) with firm (grade 4+5) tumors, an area under the curve of 0.877 was observed when using maximum signal intensity measurements within the ROI. A TCTI ratio cutoff of 1.682 was associated with a sensitivity of 80.6% and specificity of 85.7% in predicting firm tumors (grade 4+5) versus all other grades in the test dataset. CONCLUSIONS: The T2WI TCTI ratio is predictive of PA consistency where higher ratios are associated with softer tumors. Preoperative prediction of PA consistency using the TCTI ratio might improve patient selection and outcome predication, and guide the excision technique.