The first three months after standard radiation therapy for a brain tumor must be hell for patients and their families. The established MacDonald criteria for assessing treatment force them to wait up to 10 weeks for follow-up CT or MR to determine if the size of the treated tumor has changed.
Using the Response Evaluation Criteria in Solid Tumors (RECIST), the physician will probably either stick with the treatment plan for patients with responding or stable disease or recommend surgery to remove tumors that do not respond. Between initial treatment and follow-up, patients can do little but imagine the worst and bide their time.
The days of this treatment paradigm may be numbered, however, if the work of department chair Dr. Brian Ross and his colleagues in the radiology department at the University of Michigan gains general support. They propose serial functional MR diffusion mapping performed a week before treatment and then one, three, and 10 weeks after the initiation of radiation therapy to give the oncologist an earlier and ongoing measure of treatment response.
Craig J. Galbán, Ph.D., a research investigator with the Michigan group, presented results at the ISMRM meeting based on the experiences of 60 brain tumor patients. They showed that the MR diffusion mapping technique gave them and their physicians earlier feedback about the effectiveness of therapy.
Patients underwent the diffusion-weighted MR brain scans on either 1.5T or 3T scanners. Diffusion-weighted images were acquired using a single short spin-echo diffusion-sensitized echo-planar sequence. DWI for the three orthogonal directions and B0 were used to calculate apparent diffusion coefficient maps.
All post-treatment MR images were coregistered to the pretreatment MR images. Radiologists defined regions of interest on the enhancing areas of the contrast-enhanced T1-weighted images. Software allowed a voxel-by-voxel comparison of ADC values within the tumor at one, three, and 10 weeks post-treatment with the values generated before treatment.
The resulting functional diffusion maps segmented the tumor into three categories: red voxels for tumor volumes where ADC increased significantly following treatment, blue voxels where ADC decreased significantly, and green voxels where ADC was unchanged.
DWI measures water mobility as inversely related to cellularity, Galbán said. Densely packed tumor cells hinder water movement to produce a low ADC, corresponding with blue voxels. Water moves more freely and ADC increases as the membranes of cells break down during a positive response to therapy. ADC increases even further during massive tumor cell necrosis, corresponding with red voxels.
"We are really interested in looking at the red voxels," Galbán said. "We are interested in the volume fraction of increasing ADC in the tumor."
The prospective trial involved 70 patients treated from 2000 to 2006. Of that total, 67 had had grade 3 or grade 4 gliomas that were treated with radiation therapy or both radiation and chemotherapy. Sixty were evaluated with functional brain mapping, and seven served as controls.
Receiver operator characteristic curve analysis found that functional brain mapping performed one week after the start of treatment was 40.7% sensitive and 89.5% specific for predicting one-year survival. Functional brain mapping repeated after three weeks produced 68.8% sensitivity and 85% specificity rates (p<0.0002). Mapping performed after 10 weeks produced 65.6% sensitivity and 70% specificity rates (p<0.04).
The ability of functional brain mapping to differentiate among responders, nonresponders, and stable disease after three weeks was similar to results produced with the conventional MacDonald criteria after 10 weeks, Galbán said.
Responders identified with the MacDonald criteria 10 weeks after the initiation of therapy survived threes times longer than patients with progressive disease. Using functional mapping data gathered three weeks after the initiation of therapy, patients who had a volume fraction of red that made up more than 5% of the tumor volume survived five times longer than patients who had a volume fraction of red that was less than 5% of tumor volume, he said.
Based on these results, a new paradigm for brain tumor treatment monitoring can be proposed, Galbán said. Instead of waiting 10 weeks to measure results, the oncologist can assess treatment after three weeks using functional diffusion mapping.
"Depending on the amount of red observed, therapy can be continued as planned, modified to target more treatment at nonresponding tumor, or changed for nonresponders," he said
The Michigan group is planning a multicenter clinical trial to validate the preliminary trial findings. It is also testing the value of functional diffusion mapping on other tumor types, Galbán said.