The crossover of the bevacizumab and panitumumab survival curves in the PARADIGM study: A clue to time-related effects of bevacizumab on the risk of tumor progression.

Abstract

159 Background: The PARADIGM study in CRC patients demonstrated a cross over of the bevacizumab survival curve below the panitumumab survival curve at 28 months, followed by a steady decrease of the bevacizumab survival curve below the panitumumab survival curve up to 11% by 5 years. Similar differences were seen in the FIRE3, CALGB-80405 and PEAK trials starting at 20 months, suggesting an unsuspected time-related effect of bevacizumab on patient’s survival starting at about 2 years after its 1st administration. Methods: Review of time related effects of bevacizumab in 1st and 2nd line trials conducted in CRC, breast, ovary and renal cancer showed a significant PFS but no OS survival benefit, with a detrimental effect on survival in the adjuvant setting (AVANT, NSABP-08 studies). Preclinical data on tumor biology, intratumor-heterogeneity, clonal evolution and the cross talk between bevacizumab and the diverse tumor microenvironments, was correlated to the clinical data, in an attempt to explain the time-related effects of the biological. Results: In the NSABP-08 trial, bevacizumab reduced relapse for 6 months, effect diminished and lost at 12 months. At 24 months after bevacizumab's 1st administration, the relapse of the bevacizumab treated population increased steadily over the control. Identical kinetics were shown in AVANT study and metastatic ovarian patients suggesting that these time-related effects of bevacizumab are universal. The biologic explanation of the clinical data is that short-term treatment with bevacizumab (plus chemotherapy) induces initial tumor stasis and shrinkage, followed by selection and emergence of more aggressive and resistant tumor clones, characterized by an accelerated growth, invasiveness and metastasis shifting the overall tumor growth kinetics to a steeper angle. Bevacizumab appears to be effective for the time of its administration, followed by a slow rebound after its discontinuation. The initial survival benefit will be lost at about 20-24 months after its first administration, as clearly demonstrated in the PARADIGM, FIRE-3, CALGB 80405 and the PEAK studies. Conclusions: As bevacizumab benefit persist for about 2 years after its 1 st administration, it explains the "survival benefit" obtained in studies with with median survival of 20 months but not in studies with survivals >30 mo. (PARADIGM, FIRE-3, CALGB 80405 and the PEAK). This phenomenon also explains the crossover of the survival curves in PARADIGM study and the benefits of bevacizumab obtained preferentially in patients with short survival (rt. sided tumors), high tumor burden, and 2 nd and 3 rd line CRC.