Introduction. Prediction of response to preoperative breast cancer chemotherapy may offer a substantial optimization of medical management of this disease. The most efficient prediction would be done a priori, before the start of chemotherapy and based on the biological features of patient and tumor. Numerous markers have been proposed but none of them has been applied as a routine. The role of MKI67 and HSP90 expression has been recently suggested to predict treatment sensitivity in HER2-positive breast cancer. The aim of this study was to validate the utility of proliferation based markers (MKI67 and CDK1) and heat shock proteins (namely HSP90) to predict response to chemotherapy in cohort of breast cancer patients treated preoperatively.

Material and methods. Ninety-three patients with breast cancer, all females, mean age 42.2 years, among them 32% T1-T2 patients, 49% T3 patients and 13% with T4 tumor stage, 27% N0, 42% N1, 16% N2, 15% N3 were subjected to initial chemotherapy. The majority of patients (86%) received anthracycline and taxane chemotherapy. Among the patients there were 9 individuals with metastatic disease (M1) at initial presentation, and 11 patients were not treated surgically after initial chemotherapy (no sufficient disease response). From 82 patients operated on, 20 patients (24%) showed pathological complete response (pCR), while in 62 patients there was no pCR. 42% of patients were hormone-sensitive HER2-negative, 20% hormone-sensitive HER2-positive, 9% only HER-positive and 29% with triple negative breast cancer. Four gene transcripts (MKI67, cyclin-dependent kinase 1 [CDK1], heat shock proteins HSP90AA1 and HSP- 90AB1) were analyzed in total RNA isolated from single core obtained during preoperative core needle biopsy by quantitative real-time PCR with fluorescent probes (Universal Probe Library, Roche). Results were normalized to the panel of reference genes.

Results. There were no statistically significant differences in MKI67 and CDK1 expression between pCR and no pCR groups (p = 0.099 and 0.35, respectively), although the median expression of both genes was slightly higher in pCR group. In contrast, both HSP90AA1 and HSP90AB1 transcripts showed decreased expression in pCR group (medians 0.77 and 0.55) when compared to no p CR group (median 0.86 and 0.73), statistically significant for HSP90AA1 (p = 0.031) and of borderline significance for HSP90AB1 (p = 0.054). The most significant predictor of pCR was the ratio of CDK1 transcript to HSP90AA transcript. This ratio was significantly higher in CR group (median 0.99) than in no CR group (median 0.68, p = 0.0023), and showed a potential diagnostic utility (area under receiver operating characteristic [ROC] curve 0.72).

Conclusions. HSP90AA1 and AB1 genes exhibit low expression in breast cancers highly sensitive to chemotherapy and may indicate the patients with higher probability of pathological complete response. The ratio of HSP90AA1 to proliferation-related markers (CDK1 or MKI67) may be even better predictor of pCR chance, with higher expression of proliferation genes and lower stress response in patients sensitive to chemotherapy.