Prognostic, predictive, and surrogate value of HER2 extracellular domain (ECD) for progression-free survival (PFS) in advanced breast cancer treated with lapatinib (lap): a meta-analysis

Background: We performed a meta-analysis to determine the clinical utility of HER2 ECD in advanced breast cancer patients undergoing lap therapy. Methods: Data from Phase 3 clinical trials (EGF30001 , EGF30008 , EGF100151) of patients randomized to receive lap-containing or control treatments were analyzed. Baseline HER2 ECD (bECD) and tissue HER2 status were examined for associations with PFS. We also performed a landmark analysis to quantify the association between ECD change, as measured at 10 weeks post-randomization, with PFS, and assessed the extent ECD change predicted lap benefit. Results: Levels of bECD were higher in HER2+ (n=378) than HER2- (n=1060) patients (61.59 vs 14.25ng/ml, p<0.0001). The effectiveness of lap was significantly associated with bECD (joint treatment-biomarker interactions P=0.001[bECD], P=0.02[HER2 tissue status]). At 10 weeks post-randomization, there was a mean 4.23ng/ml reduction in ECD with no significant difference between treatment arms (lap: 4.33ng/ml, control: 4.12ng/ml; P=0.92). In the control arms, change in ECD was prognostic for PFS (P<0.0001). The median PFS from landmark (mPFS) were 9.8 and 2.9 months for those with low ECD (≤15ng/ml) and high ECD (>15ng/ml), respectively. mPFS were 3.3 and 5.7 months for patients with ECD increase to >15ng/ml and ECD decrease to ≤15ng/ml, respectively. Similar finding was observed in the experimental arm (P<0.0001). In both treatment arms, there was 4% increase in risk of disease progression for every 10 unit ECD increment after adjustment of bECD and other factors (P<0.0001). Lap improved PFS over control (HR 0.83, P=0.002). Adjustment for ECD change yielded a similar result (HR 0.80, P<0.001), indicating that change in ECD did not explain the treatment effect of lap. Conclusions: Baseline HER2 ECD was predictive of lap benefit. ECD increase during lap and non-lap therapies was associated with a shorter PFS. Change in ECD had additional prognostic information in addition to baseline ECD, but was not a valid surrogate marker for lapatinib benefit as the treatment effect of lap was largely independent of ECD change.