Recurrence is relatively common in meningioma patients post-treatment, occurring in up to 20% of cases, even in histologically benign (WHO grade 1) cases . In order to minimize the rate of recurrence as well as the side effects from excessive treatment to normal brain tissue, an ideal imaging modality should be specific enough to define the correct tumor volume from tumor-free tissue, thereby aiding in surgical and radiation planning to achieve the maximal safe target volume. In the post-treatment setting, it should also be sensitive to detect any residual or recurrent tumor from post-treatment scarring and inflammation. The most recently updated European Association of Neuro-Oncology guideline for the management of meningioma continues to highlight contrast-enhanced MRI as the gold standard for diagnosing and follow up imaging in meningioma patients . However, MRI appearance is often limited in accuracy particularly in the presence of post-treatment scarring and inflammation. Rachinger and colleagues demonstrated that standard MRI has sensitivity of 79% and specificity of 65% in distinguishing meningioma from tumor-free tissues . Another group showed that MRI alone can achieve sensitivity of 95% and specificity 88% in the diagnosis of meningioma in a cohort of 57 patients, but noted that diagnostic accuracy diminishes in cases of small lesions (< 0.5 cm3) and certain difficult locations such as skull base . For transosseous growing meningiomas, MRI’s sensitivity was even lower at 54% .
The recently updated European Association of Neuro-Oncology guidelines also highlight the role of [68Ga]-DOTATATE PET in distinguishing meningioma from healthy tissue and post-surgical changes . Histology-controlled studies showed that the extent of meningiomas is better delineated with [68Ga]-DOTATATE PET compared to contrast-enhanced MRI alone [5, 17]. During radiation planning, [68Ga]-DOTATATE and DOTATOC PET alter target volume delineation for stereotactic fractionated radiation therapy, often resulting in a reduction of the gross tumor volume compared with results from MRI or CT [11, 24, 25]. The utility of adjuvant RT, compared to active surveillance, in resected meningiomas is currently being evaluated in the NRG0539 trial, in which our preliminary analysis suggests [68Ga]-DOTATATE’s efficacy for RT response assessment with a marked reduction in [68Ga]-DOTATATE SUV in meningioma post-RT. Additionally, a case series of 20 patients demonstrated the clinical utility of [68Ga]-DOTATATE PET/MR in identifying additional meningiomas not previously identified on contrast-enhanced MRI and in differentiating disease from reactive enhancement, thus facilitating treatment planning in such cases . Further reinforcing its clinical utility, [68Ga]-DOTATATE PET has shown efficacy in predicting progression in non-benign meningioma as well as predicting clinical outcome for SSTR targeted radionuclide therapy such as Lu-DOTATATE [18, 19]. Within our cohort, we confirmed that all 3 approaches to [68Ga]-DOTATATE PET quantification (absolute SUV, SUVRsss and SUVRpit) aid in the differentiation of meningioma from post-treatment changes, confirming the reliability of the SSTR2 targeted imaging in meningioma patients. SUV, SUVRsss, and SUVRpit did not correlate with WHO grade, consistent with prior histopathological studies, and suggesting that SSTR2 expression is independent of the differentiation status of meningioma tumor cells [17, 26]. There was a significant relationship observed between tumor size and SUV, SUVRsss, and SUVRpit.
In order to effectively utilize [68Ga]-DOTATATE PET in the clinical context of meningioma, an SUV threshold that constitutes as the diagnostic threshold must be established. Prior studies often employed reference tissue SUVR approaches, such as contralateral brain parenchyma, contralateral subarachnoid space, liver, gluteal muscle, and superior sagittal sinus [4, 5, 13-15]. One prospective study of 21 patients reported absolute SUV threshold of 2.3 with 90% sensitivity and 73% specificity for the purpose of diagnosing meningioma from tumor free tissue, which have been utilized in several other studies [5, 17, 18]. One other well-known method is the Krenning score system, derived from [111In]Octreotide scintigraphy of gastrointestinal neuroendocrine tumors, which uses the liver and spleen as the reference regions and has been validated in neuroendocrine tumors that are SSTR2 positive for the purpose of assessing candidacy for PRRT . However, the Krenning score has not been utilized in meningioma and requires a body PET. Therefore, we evaluated PET/MRI-based approaches to quantitative [68Ga]-DOTATATE PET analysis that do not require whole-body imaging, and instead rely on dynamic brain PET imaging. Amongst the anatomic regions that can be obtained from brain PET, we chose the SSS or cranial blood pool as a background reference region, as previously published, and the pituitary gland, a notably SSTR2 positive intracranial organ .
In our cohort of 62 patients, at the pre-determined threshold, the SUV method with the threshold of 2.3 demonstrated the highest sensitivity (98.2%) but much lower specificity than the prior study of 21 patients (56.1% vs. 73.5%) . The optimal threshold for SUV in our cohort was 4.65, much greater than the threshold of 2.3 in the prior study . This discrepancy may be explained by the different sample sizes, varying acquisition time and technicality of imaging, and random variation in SUV across patients. The highest specificity was achieved with SUVRpit (87.8%) with the threshold of 0.260 but it was not significantly different from the specificity of SUVRsss at the pre-determined threshold of 3, which had a greater sensitivity than SUVRpit. We then recalculated sensitivity and specificity of each classification method at the optimal threshold as determined by Youden’s J statistics. Notably, the optimal threshold for SUVRsss was 3.23, similar to 3 as set by the prior study and as tested in our analysis of the pre-determined thresholds . Interestingly, at the optimal thresholds, the methods with the highest sensitivities and specificities were flipped; SUV had the highest specificity (97.6%) and SUVRsss had the highest sensitivity (86.1%). At the optimal thresholds, the SUVRpit method was not statistically different from the SUV method, however it had a significantly lower sensitivity than the SUVRsss method with no difference in specificities, suggesting that SUVRsss is a superior method overall compared to SUVRpit. AUCs of the ROC for each method were comparable. Based on this result, we conclude that SUVRsss with the threshold of 3.23 may be used in clinical settings where greater sensitivity is desired such as in the post treatment setting in the evaluation of recurrence or progression. The SUV threshold of 4.65 may be more appropriate for instances where high specificity is desired such as surgical and radiation planning. However, it is important to note that SUVR that is normalized to a region of interest in the same patient may be more reproducible and thus a more reliable metric of SSTR expression, allowing more robust comparison across time points, scanners, and patients. Furthermore, given greater variability of SUV of the pituitary gland compared to SUV of the SSS, SUVRsss may be the more robust of the two SUVR methods and thus the preferred method in the clinical setting.
Our study has several limitations. It is important to note that a significant number of individual lesions included in the study (77% of meningiomas) were confirmed based on clinical follow ups rather than biopsy. For the purpose of correlating WHO grades and SUV in meningioma, multiple meningiomas in a given patient were assigned the same WHO grade as the WHO grade of the pathology proven meningioma in the given patient, as applied in previously published studies [19-21]. Additionally, the number of the pathology proven post-treatment change is relatively low given that they are most often confirmed with longitudinal follow ups with MRI as the gold standard. Notably, 17% of the post-treatment change lesions included in the study were confirmed with the MR appearance of the [68Ga]-DOTATATE PET/MR in the absence of both biopsy and clinical follow ups, which may affect the accuracy of the classification. Additionally, while we excluded meningiomas that were previously irradiated from our analysis, recurrence in prior RT fields might affect SUV of the lesions when compared with de novo or untreated meningiomas, an important issue which we plan to study in future work. Finally, a SSTR2 negative meningioma, although exceedingly rare, may contribute to potential misclassification of the lesions .