Figure 1.
Different size-based criteria and Choi criteria with relative geometrically derived-cut-offs for partial response (PR) and progressive disease (PD). Corresponding geometrical shapes are shown at the left side of the figure. Stable disease (SD) corresponds to intermediate changes between PR and PD. Abbreviations: CT, computed tomography; D, tumor density; HU, Hounsfield unit; r, ri (i = 1,2,3), radius; RECIST, Response Evaluation Criteria in Solid Tumors; S, size. Symbols: *appearance of a new lesion, **new intra-tumoral nodules or increase in the size of the existing intra-tumoral nodules, *** the sum of the longest diameters of target lesions as defined in RECIST [30].
Table 1.
Main clinical characteristics of patients in the training and validation cohorts.
Figure 2.
Example of manual and semi-automated images evaluation.
This figure shows a case where 3D measurements could detect a size reduction >20%, and 1D measurements did not, because the shrinkage was mostly in the shortest diameters of the target lesion. To note, the shortest diameter (not taken into account by RECIST) decreased of 20% and 22% at 6 and 12 months respectively. Images are shown in the axial plane and in the coronal plan, and 3D reconstruction with OsiriX® Imaging Software is shown in the upper, middle, and lower part of the figure, respectively. The number of CT slices where the lesion could be identified at different time-points is indicated. Interestingly, this lesion was classified as SD according to RECIST and 3D-sphere criteria, and as PR according to 3D-ellipsoid and Choi criteria at both time-points.
Table 2.
Sensitivity in discrimination of size change # (%) (unadjusted/unscaled method).
Table 3.
Response assessment by different criteria # (%) (adjusted/scaled method).
Figure 3.
Overall survival by different criteria in the training cohort (2 curves).
Overall survival (OS) shown up to 72 months (mo) of all assessable patients by response criteria, divided in progressed versus not progressed (SD+PR) at 3, 6 and 12 months of imatinib therapy. To note, survival curves related to Choi criteria include only patients assessable with both manual and semi-automated method, thus they should be evaluated separately and not directly compared with the other criteria (Panels D,H,L). Panels A–C: When the tumor response is evaluated on the basis of RECIST and 3D criteria, a statistically significant difference is observed in the long-term prognosis between patients with and without PD at 3 months. Panels E–G: Significant difference is observed in the long-term prognosis between patients with and without PD at 6 months only by 3D-ellipsoids. Panels I–K: A difference is observed in the long-term prognosis between patients with and without PD at 12 months only by 3D-spheres and 3D-ellipsoids, but no significant difference is observed by RECIST. Panels D,H,L: Significant difference is observed in the long-term prognosis between patients with and without PD by Choi criteria only at 6 months.
Figure 4.
Overall survival by different criteria in the training cohort (3 curves).
Overall survival (OS) shown up to 72 months (mo) of all assessable patients by response criteria, divided in PD versus SD versus PR, at 3, 6 and 12 months of imatinib therapy. To note, survival curves related to Choi criteria include only patients assessable with both manual and semi-automated method, thus they should be evaluated separately and not directly compared with the other criteria (Panels D,H,L).