| PURPOSE: The objective of part I of the study was to accurately predict survival in patients with metastatic melanoma treated with anti-angiogenic therapy using a combination of a serum biomarker (lactate dehydrogenase) and initial post-therapy CT findings of tumor devascularization. The objective of part II of the study was to use computed tomography (CT) texture analysis (CTTA) to predict overall survival (OS) in patients with metastatic melanoma and Response Evaluation Criteria in Solid Tumors RECIST Stable Disease (SD) on initial post-therapy CT images. MATERIALS AND.;METHODS: In part I of the study, consent was waived for this institutional review board (IRB)-approved retrospective secondary analysis of 44 patients with metastatic melanoma who received bevacizumab therapy in the context of a randomized prospective phase II trial. Target lesions on the initial post-therapy CT were evaluated by Response Evaluation In Solid Tumors (RECIST) 1.1, Choi Criteria, and Morphology, Attenuation, Size, and Structure (MASS) Criteria. Cox proportional hazards models were used to assess the association of baseline clinical variables, including serum lactate dehydrogenase (LDH), and imaging findings with progression free survival (PFS) and overall survival (OS). The receiver operating characteristics (ROC) curve with area under the curve (AUC) was used to evaluate predictive accuracy. The part II analysis included 42 patients with metastatic melanoma who received bevacizumab therapy in the context of a randomized prospective phase II trial. Target lesions on the baseline and initial post-therapy contrast-enhanced CT were evaluated by CTTA using TexRAD software in patients with RECIST SD on initial post-therapy images. Cox proportional hazards models were used to assess the association of CTTA measurements, initial tumor size changes, and baseline serum lactate dehydrogenase (LDH.) level with OS. The area under the receiver operating characteristics curve (AUC) was used to evaluate predictive accuracy.;RESULTS: In part I of the study, a multivariate analysis revealed high baseline serum LDH was associated with decreased PFS (HR=1.29 for each increase of 100 1U/L, p=0.002) and OS (HR=1.44 for each increase of 100, p=0.001). MASS Criteria response on the first post-therapy CT strongly predicted PFS (p<0.001) and OS (p<0.001). Baseline serum LDH has a moderate degree of accuracy for predicting PFS at 9 months (AUC=0.793) and OS at 18 months (AUC=0.689). A combination of baseline serum LDH levels and MASS Criteria response on the first post-therapy CT has a high degree of accuracy in predicting PFS at 9 months (AUC=0.969) and OS at 18 months (AUC=0.813), significantly higher than the AUC value for baseline serum LDH alone predicting PFS (p=0.020),In part II of the study, a multivariate analysis indicated change in mean positive pixels (MPP) at spatial scale filter (SSF) of 4mm (available on 23 patients with RECIST SD and median OS 1.51 years), change in tumor size, and baseline LDH level were predictors of OS (HR=10.4 for ≥ 10% decrease in MPP vs. <10% decrease or increase in MPP, p=0.004; HR=3.28 for every 5% increase in size, p=0.058; HR=1.45 for every 100 IU/L increase in baseline LDH, p=0.009). A prognostic index containing these three factors was highly accurate for predicting OS at 18 months (AUC = 0.977).;CONCLUSION: In part I of the study, a combination of a serum biomarker (baseline serum LDH) and MASS Criteria imaging response on the first post-therapy CT had the highest accuracy for predicting survival in patients with metastatic melanoma treated with bevacizumab therapy. In part II of the study, patients with metastatic melanoma and RECIST SD on initial post-therapy CT images, a model incorporating baseline LDH levels and initial changes in tumor size and CTTA of target lesions was highly accurate in predicting OS. |
