Principal factor analysis of predictive markers of response and resistance to primary chemo-endocrine treatment in elderly breast cancer (BC) patients.
Generali DG., Fox S., Buffa FM., Berruti A., Brizzi M., Allevi G., Bonardi S., Dogliotti L., Bottini A., Harris A.
618 Background: This study was undertaken to identify protein expression patterns able to predict response and resistance to chemo-endocrine therapy in elderly BC patients enrolled in a randomised phase II study. METHODS: 114 women with T2-4 N0-1, estrogen receptor positive BC were randomly assigned to 6 months of primary Letrozole (L) (2,5 mg/daily) or L plus oral "metronomic" cyclophosphamide (50 mg/daily) (LC). Expression of 24 markers was assessed before treatment on TMAs. Markers were involved in signalling and angiogenic/hypoxia pathways as follow; erbB2, T regulatory cells, caspase3, BNIP3, phosphorylated ERaplha, CCDN1, mTor, Hif-1α, phd1, phd2, phd3, CA9, COX2, p38, p44, EGFr, PI3k, pAkt, CD31, VEGF, Ki67, p53, bcl2, herb2. Principal factor (PF) analysis attempts to identify underlying factors that explain the correlation patterns within a set of observed variables; here, PF analysis was used to reduce the marker expression data by identifying a small number of PFs that explained most of the variance observed. These PFs were introduced in a multivariate logistic regression (MLR) to study basal protein expression profiles with response to chemo-endocrine treatment. Clinical variables such as treatment, age, tumor size, nodal status, grading and histotype were also included in MLR. RESULTS: 91 out of 113 evaluable patients (80.5%) attained a disease response, 48 patients a complete response (CR) (42.5%) whereas 22 did not respond (19.5%). The first 12 extracted PFs from PF analysis explained 80% of the expression data variance; these were considered in a MLR together with clinical variables. The 4(th) PF, mainly representing Hif-1α and p44 expression, and at lower degree EGFr expression, was the only independent predictor of disease response (OD=0.22, p=0.003). The 6(th) PF, mainly representing phosphorylated ERalpha expression, was the only independent factor associated with CR (OD=2.036, p=0.023). There was no interaction between these PFs and treatment randomisation. CONCLUSION: The activated form of ERalpha is with complete response, whereas Hif-1α and p44 were able to discriminate patients resistant to chemo-endocrine treatment. In this latter cohort, specific target therapies can be recommended. No significant financial relationships to disclose.