Fig 1.
Lesion heterogeneity in CT image for Case A.
(A): Illustrative image data of a progressive lesion for Case A. The lesion was target lesion 1 in lung and bronchus from three time points and demonstrated eventual progression of disease by RECIST criteria. The tumor growth pattern was also displayed in Fig 2A in black color dashed line. (B): Illustrative image data of a regressive lesion for Case A. The lesion was target lesion 2 in lymph node at three timepoints and showed continued decrease in size of lesion consistent with response to treatment. The tumor growth pattern was also displayed in Fig 2B in red color dashed line.
Fig 2.
(A): Case A had three lesions with two in lymph node and one in lung site. In the lymph node site, one lesion had at least 30% reduction (ratio<0.7), but the other lesion progressed with more than 60% increase (ratio>1.6). The lesion in the lung site also increased the tumor size about 30%. Results showed completely different treatment reaction in each lesion with one regressive lesion and one progressive lesion in the lymph node and one progressive lesion in the lung site, indicating significant variation of lesion growth pattern within and between organ sites. However, the RECIST sum averaged these three heterogeneous lesions and classified as SD, suggesting all lesions were under control with no change of tumor growth. (B): In contrast, case B had also three lesions with two in lymph node and one in lung site. All lesion sizes remained relatively stable over time with minor changes from the baseline. The RECIST sum took an average of these three homogeneous lesions and also classified as SD. In comparison of both cases, while they reached the same SD classification, lesion variation to treatment reaction was quite different. Lesions in case B were comparable and did not progress or regress. On the other hand, case A had diverse tumor growth patterns composed of progressive and regressive lesions. While the RECIST sum might well characterize the stable condition in case B, classification of SD in case A seriously distorted lesion variation and could misguide clinical decision.
Fig 3.
LeHeC algorithm.
Table 1.
Demographic characteristics of 36 patients with stage 4 NSCLC.
Fig 4.
Comparison of MSD among lesion heterogeneous subgroups.
PD&PR: mix of progression and regression; PD&SD: mix of regression and stability; PR&SD: mix of regression and stability; Non-Heter: non-heterogeneity.
Table 2.
Frequency of the heterogeneity associated subgroups in each RECIST response category.
Fig 5.
Mixed response.
Fig 6.
Mix of pseudo-progressive and regressive lesions.
(A): Case ID 3 had 4 chest lesions with two chest lesions (lesion 3-4) experienced early progression and then developed response with a ratio below 0.7 after day 100. The other two chest lesions had a ratio below 0.7 after the 1st CT scan. (B): Case ID 4 had a similar pattern where an initial ratio >1.2 and then disappeared after 100 days in two upper lobe nodule lesions (lesion 2 and 4). The two lower lobe nodule lesions had tumor reduction at least 30% with one after day 100 and the other one after day 300.
Fig 7.
*: median OS; m: month.