Volumetric 3D CT analysis - an early predictor of response to therapy

Abstract

4576 Background: Evidence suggests that changes in unidimensional measurements (using RECIST criteria) may not accurately reflect actual changes in tumor size, and do not correlate well with other biomarkers of response, tumor progression, or patient outcome. We evaluated changes in tumor volume of target lesions with volumetric CT, and compared them with response assessments derived from RECIST criteria. Methods: We evaluated target lesions, including lymph node, liver, peritoneal and lung metastases in 25 patients with metastatic gastric cancer or gastroespohageal junction adenocarcinoma treated on a phase II clinical trial with irinotecan, cisplatin and bevacizumab. All patients underwent thin-section multidetector CT at baseline and 6-week follow-up for 10 cycles. Target lesions were measured unidimensionally and volumetrically, using validated automated and semiautomated segmentation algorithms. For initial analysis, correlation was made with the RECIST response using a cut-off value of 65% volume change. A ratio of change in RECIST measure to volume measure was calculated at time of maximal clinical response for each patient, as well as for responders versus non-responders. Results: 18 of 25 (72%) patients showed a clinical response. Of these 18 responders, 5, 6, 4, and 3 were identified as responders using RECIST criteria after cycles 1, 2, 3, and 4, respectively. Using a cut-off of 65% volume change, 14 of these responders were indentified after cycle 1, and 4 were identified after cycle 2. Using a cut-off value of 65%, volume measurements identified responders a mean of 50.3 days earlier than did RECIST criteria. There was a statistically significant (p<0.01) change in ratio of volume measurement change to RECIST measurement change for responding versus stable patients. Conclusions: Volumetric change in tumor size appears to predict clinical response earlier than do RECIST criteria in the majority of cases. Changes in volume are more sensitive because they exhibit a wider dynamic range. Ratios of size changes volumetrically appears to better discriminate RECIST responders from those with stable disease. Further work is needed to define the biologically relevant cut-off value for volume change, and to correlate volume change with other biomarkers. No significant financial relationships to disclose.