P2-09-02: Predicting Response to Bevacizumab in Primary Breast Cancer Using 18F-Fluorothymidina (FLT) and 18F-Misonidazole (MISO) Positron Emission/Computed Tomography (PET/CT) as Imaging Biomarkers

Background: To investigate the hypothesis that early changes in tumor proliferation and hypoxic status induced by bevacizumab and assessed by imaging biomarkers might predict response to bevacizumab therapy.

Methods: 73 chemotherapy naïve, stage II-III breast cancer (BC) patients (pts) were enrolled in the training set of this phase II, single-arm, multicenter and prospective clinical trial from October 2009 until November 2010. Pts received single infusion of bevacizumab (15 mg/kg) (C1) 3 weeks prior to the beginning of neoadjuvant chemotherapy (NAC) consisting in 4 cycles of docetaxel (60 mg/mq), doxorubicin (50 mg/mq) and bevacizumab (15 mg/kg) every 21 days (C2-C5) following by surgery. Tumor proliferation and hypoxic status were evaluated using FLT and MISO PET/CT at baseline and 14–21 days after bevacizumab (C1). Standardized uptake values (SUV) for FLT and MISO and ratios to reference tissues, mediastinum (T/Me) or muscle (T/Mu), for MISO were calculated. Pathological response on surgical specimens was assessed according to Miller/Payne grading system. Pts with reduction in tumor cells >30% (G3-G4-G5) or <30% (G1-G2) were respectively considered as responders and no-responders. Association between pathological response, baseline and changes induced by bevacizumab (C1) in imaging biomarkers was analyzed using Mann-Whitney test. Receiver operating characteristic (ROC) curve was performed to test sensitivity and specificity of the biomarker found associated to response. Its value as independent predictor was tested in multivariate analysis using logistic regression. Results: Median baseline MISO and FLT SUV values in tumors were 1.2 (range 0.69−2.39) and 2.89 (range 0.97−7.18). Significant change after C1 was observed in FLT (2.7 vs 1.8, p<0.001) but no in MISO uptake. Fifty-two (74%) pts achieved response (G3-G4-G5) whether 18 (24%) were considered as no responder (G1-G2); for 3 (4%) patients Miller/Payne tumor evaluation was not available. Response showed a trend toward an association with negative estrogen receptors (ER) expression (p=0.08) and triple negative tumors (11/73) (p=0.05). FLT SUV baseline and changes after C1 in MISO SUV, T/Mu and T/Me were all significantly associated with pathological response (p=0.057, 0.03, 0.016, 0.010). ER expression and T/Mu change remained significantly associated with response in multivariate analysis (OR=24.8, IC95% 1.8-334, p=0.01 and OR=0.95, IC 95% 0.92−0.99, p value=0.02). Decrease in MISO T/Mu uptake >20% yielded a ROC curve area of 0.7 (95% CI: 0.56 - 0.85) with 94% sensitivity and 87% specificity. Conclusion: Bevacizumab determined a marked decrease in tumor proliferation. Interestingly, a decrease greater than 20% in tumor hypoxic status after C1 and assessed by MISO was found significantly associated with pathological response suggesting a potential value of early decrease in hypoxic tumor status as predictive biomarker of response. Bevacizumab, causing normalization of the tumor microvasculature, seems to potentiate the effect of cytotoxic agents on primary BC. A validation set is warranted to confirm these findings.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-09-02.

P2-08-05: Use of Dynamic Contrast-Enhanced MR Imaging To Predict Pathological Response in Primary Breast Cancer

Purpose: To investigate the effect of bevacizumab infusion on vascular parameters assessed by dynamic contrast-enhanced magnetic resonance (DCE-MR) imaging and to test their association with pathological response in primary breast cancer.

Materials and Methods: 73 patients (median age, 47 ys; age range, 29–70 ys) with biopsy-proven, previously untreated, primary breast cancer were recruited from October 2009 to November 2010 in this phase II, multicenter and non-randomized clinical trial. Patients (pts) received single infusion of bevacizumab (15 mg/kg) (C1) 3 weeks prior to the beginning of neoadjuvant chemotherapy consisting in 4 cycles of docetaxel (60 mg/mq), doxorubicin (50 mg/mq) and bevacizumab (15 mg/kg) every 21 days (C2-C5) following by surgery. All pts underwent DCE-MR imaging before and 14–21 days after C1. Quantitative and semiquantitative kinetic parameters were calculated at baseline and after C1, including the volume transfer constant (K(trans)), which primarily reflects the wash-in of the contrast agent, the backflow rate contrant (K(ep)), extracellular volume fraction (V(e)) and the initial area under the gadolinium concentration-time curve over 60 seconds (IAUGC(60)). Changes in the DCE-MRI kinetic parameters K(trans), K(ep), V(e) and IAUGC(60) were calculated and Wilcoxon test was used to assess significant effects induced by bevacizumab on kinetic parameters. Pathological response on surgical specimens after C5 was assessed according to Miller and Payne classification. Pts with tumor reduction >30% were considered as responders (G3-G4-G5) whether tumor reduction <30% were considered as no responders (G1-G2). DCE-MR imaging parameters and clinical-pathological characteristics were correlated with pathological response using Mann-Whitney test in univariate and logistic regression in multivariate analyses. Receiver operating curves (ROC) was used to define the best cut-off of the parameter found associated with pathological response.

Results: DCE-MRI was performed before (n=72) and after (n=71) C1. K(trans), K(ep), V(e) and IAUCG(60) values were significantly different at the baseline and after C1 (p<0.01). Median changes were, respectively, −51, −101, −52.5 and −4.8. Fifty-two (74%) pts achieved response (G3-G4-G5) after C5 whether 18 (24%) were considered as no responder (G1-G2); for 3 (4%) patients Miller/Payne tumor evaluation was not available. At univariate analysis, negative estrogen receptor (ER) status and higher post-C1 K(ep) (p=.057) showed a trend toward an association with response. At multivariate analysis, only ER status remains a significant predictor of response (p= .04). Area under ROC curve for K(ep) was 0.65 (IC95% 0.05−0.8, p=.057)

Conclusion: Bevacizumab affects tumor vasculature, perfusion and permeability as showed by the significantly reduction in all kinetic vasculature parameters obtained in DCE-MR imaging after C1. However, in our population these changes were not associated with pathological response. On the other hand, backflow rate constant, K(ep), a perfusion-related parameter derived from DCE-MRI yielded after C1 major than 80, may be associated with higher pathological response with a specificity of 88% and sensitivity of 90%. Future studies are warranted to confirm these findings.

Citation Information: Cancer Res 2011;71(24 Suppl):Abstract nr P2-08-05.