Response monitoring with DCE-MRI, H₂¹⁵O-PET and ¹⁸FDG-PET in patients with advanced NSCLC treated with bevacizumab and erlotinib

Abstract

3638

Background:
Efficacy of targeted drugs by RECIST criteria may underestimate their clinical value. The aim of the current study was to evaluate whether early response to treatment with bevacizumab and erlotinib in patients with advanced NSCLC can be assessed with ¹⁸FDG-PET, H₂¹⁵O-PET and DCE-MRI.
Methods:
In a multicenter phase II study, chemonaive patients with advanced non-squamous NSCLC were treated with bevacizumab 15 mg/kg q 3 week and erlotinib 150 mg daily until progression. Primary endpoint was non-progression rate (NPR) at 6 weeks posttreatment.
Patients underwent dynamic ¹⁸FDG-PET, H₂¹⁵O-PET and CE-MRI scanning as well as CT at baseline and after 3 and 6 weeks of treatment.
RECIST measurements at CT, standard uptake value (SUV) at ¹⁸FDG-PET, tumor perfusion (F) at H₂¹⁵O-PET and the endothelial transfer constant (K^ps) at DCE-MRI were defined at each time point.
Results:
A total number of 44 patients were included. NPR after 6 weeks was 75%.
At baseline and after 3 weeks of treatment complete data sets were available for 15 patients. H₂¹⁵O-PET, ¹⁸FDG-PET and DCE-MRI data at both baseline and after 3 weeks were available for 19, 32 and 29 patients, respectively.
Results for all patients: M/F 22/22; median age 59 (range 34-80); stage IIIB/IV 9/35; PS 0/1/2: 21/18/5. At the time of analysis 32/44 patients have died and 40/44 progressed. Median time to follow up was 462 days, median time to progression 92 days and median survival 209 days.
Radiological response by RECIST after 3 weeks of treatment (stable disease/partial response vs. progressive disease) did not significantly correlate with survival (p=0.1) and showed borderline significance after 6 weeks of treatment (p=0.048).
K^psshowed a median decrease of 32% (IQR -60% to +26%) after 3 weeks (p=0.09), but did not correlate with TTP and overall survival (p=0.5, p=0.9).
Residual SUV and F after 3 weeks of treatment predicted TTP (p=0.001, p=0.029) and survival (p=0.02, p=0.039).
At baseline and after treatment, no significant correlations were found between MRI derived parameter K^ps and PET derived parameters F and SUV, as well as between SUV and F (pearson correlation coefficient, p>0.1).
Conclusion:
¹⁸FDG-PET and H₂¹⁵O-PET seem valuable biomarkers in monitoring early response to antiangiogenic treatment in patients with NSCLC.
DCE-MRI derived K^ps reflected treatment effect, but was not predictive of patient outcome.
The lack of correlation between glucose metabolism and perfusion, as measured with PET, indicates that both have independent prognostic value. Multivariate analysis can only be done in a larger sample. Therefore, these results should be confirmed in a prospective study.