MR spectroscopy results can help determine if brain tumor patients with nonspecific lesions may bypass invasive biopsy for conservative management. A study by a multidisciplinary group in Ann Arbor, Mich., found that elevated levels of brain metabolites correlated with evidence of tumor recurrence. The researchers then developed a tumor recurrence model based on their data.
“Often the site of the primary tumor has been subjected to radiation, chemotherapy, and...surgical resection, causing post-treatment imaging features that are often nonspecific and difficult to interpret,” the authors wrote (Am J Roentgenol 192: W45-W52, 2009).
For this study, they enrolled 33 patients who underwent treatment for primary intracranial neoplasm, including conventional fractionated radiation therapy (54 to 70 Gy) and concomitant chemotherapy. Twenty-eight of the 33 had gliomas (WHO grades II-IV).
Routine follow-up MRI on a 1.5T scanner showed contrast-enhancing lesions (see Tables 1 and 2). The decision to perform MRS (2D chemical shift; point-resolved spectroscopy sequence; 16 x 16 phase-encoding matrix) was made in consensus by a multidisciplinary team including a radiation oncologist and a medical oncologist. The spectroscopic data were
retrospectively reviewed by a neuroradiologist (see Table 3). The mean interval between the completion of radiotherapy and development of new, contrast-enhancing lesions on follow-up MRI was 24.6 months, the authors reported. The patients with tumor recurrence (20 of 33) had higher mean values of Cho/CR and Cho/NAA than those with radiation change (13 of 33), according to the results. These patients also had lower mean values of NAA/Cr.
Based on the data, Dr. Smith’s group created a model for predicting tumor recurrence. “The final logistic regression model using Cho/NAA for predicting tumor recurrence had an area under the ROC curve of 0.92,” they said. The model achieved a sensitivity of 85%, a specificity of 69.2%, a positive predictive value of 91%, and a negative predictive value of 75%.
Finally, they devised a hypothetical scenario demonstrating the potential clinical application of this model to stratify patients into different management strategies. If the acceptable limit for recurrence is greater than or equal to 80%, then patients with a lower Cho/NAA ratio (≤1.1) could be assigned for imaging follow-up while those with a higher Cho/NAA ratio of (≥2.3) would undergo immediate treatment. Patients with a Cho/NAA ratio between those values would undergo biopsy.
“In the future, prediction models combining multiple metabolic ratios with or without clinical data may prove to be even more effective decision-making tools and (reduce) the number of patients subjected to unnecessary invasive procedures or treatment,” the group concluded.
Dr. Smith told Oncology News International that his group is “actively collecting data from two additional sites in a multi-center observational study to validate the prediction model we developed in different populations, which would add strength to the model. Our goal is to test the prediction rule in a prospective trial.”
He also pointed out that while the use of the prediction model is fairly new; the use of MRS as a problem-solving tool has been accepted by the multidisciplinary team at his institution. “MRS is frequently employed in clinical decision-making, particularly in the differentiation of recurrent tumor from radiation injury, the clinical population in which we developed our prediction model. MRS have shown to be helpful in many cases both as a diagnostic tool and as a guide for biospy,” Dr. Smith explained. “We routinely participate in multidisciplinary conferences with the radiation oncologists, medical oncologists, and neurologists where patients discussed are referred for MRS, particularly in difficult cases.
However, MRS is not reimbursed across the board, which hampers its use. “Our work is meant to provide evidence-based support for the use and reimbursement of MRS,” he said.
