158P: PET/CT based imaging biomarkers for response prediction of stage IV NSCLC treated with paclitaxel–carboplatin–bevacizumab with or without nitroglycerin

Background:

To predict early response to treatment, various imaging biomarkers defined on repeated [18F] FDG-PET/CT imaging were investigated in a randomized phase II study in patients with stage IV non-small cell lung cancer (NSCLC) receiving paclitaxel-carboplatin-bevacizumab with or without nitroglycerin patches (NCT01171170). The hypothesis was that changes within the tumor, due to response to treatment, are visible earlier on PET than on CT. PET-based imaging biomarkers could therefore enable identification of responders at an earlier time-point than CT-based parameters.

Methods:

For 60 patients in the trial, two PET/CT scans were available, one at baseline and one at week 3. For these patients, the primary tumor volume was delineated and used as the volume of interest for calculating the imaging biomarkers. Response of the primary tumor was assessed at week 3 using a 15% and 30% decrease in CT diameter, CT volume, SUVmax, SUVmean, SUVpeak and Total Lesion Glycolysis (SUVmean x CT volume). The change in imaging biomarkers at week 3 was compared to true response, defined by RECIST response at week 6, by calculating the sensitivity and specificity as well as the area-under-curve (AUC).

Results:

There were more PET-based responders in week 3 than CT-based responders. Furthermore, PET showed a high sensitivity whereas CT showed a high specificity (Table 1). PET parameters (max and peak SUV) reached a significant AUC to predict RECIST analysis.

Conclusions:

The higher number of PET responders combined with the higher AUC values compared to CT indicates that PET is able to predict treatment response earlier than CT, though typically at a lower specificity. The next step will be to define the optimal thresholds for better response prediction including the combination of PET-based and CT-based criteria together with clinical variables.