Tamoxifen response prediction: improving tailored endocrine treatment for estrogen receptor positive breast cancer

Whilst tamoxifen has been life saving for a extensive amount of patients, therapyresistance is still common with as much as 30% of patients not benefitting adjuvant therapy.

Expression of hormone receptors is essential, but dichotomous determination in breast cancer is not sufficient to accurately individually predict treatment response. In the quest for a reproducible, validated prediction tool a first step is to optimize results, already available. Furthermore, occurring therapy side effects might elucidate therapy efficacy information.

This thesis describes an alternative interpretation of already available techniques, with the intention to extract additional information by combining results.

Already in 2007 the St. Gallen consensus panel defined three endocrine response classes for decision-making on adjuvant treatment. Chapter 2 describes how we have investigated 6 different response class definitions to optimize the cut-off for PgR and ER. Additionally, we demonstrated a marked variability in endocrine therapy benefit between ER-I and ER-H patients. Patients with ER-H tumors have a larger benefit during tamoxifen and in the first years after completion of the adjuvant therapy, however ER-H patients suffer from late recurrences. So, endocrine response classes contain more predictive information than only the dichotomous classification of hormone receptor positive versus negative.

In the last decade several gene-signatures have been discovered. In chapter 3 we report the additional value of combining the 70-gene prognosis signature (MammaPrint ®) with St. Gallen response classes for prediction of outcome in tamoxifen treated women with ER+ breast cancer. In patients treated with adjuvant tamoxifen, both the 70-gene signature as well as the endocrine response classes were associated with breast-cancer-specific-survival (BCSS).

ThephysiologicalbehavioroftheERinvolvesbinding of the receptor to its natural ligand estradiol, after which the receptor associates to the chromatin. It is becoming apparent that mechanisms of resistance may result from intrinsic effects on ER/chromatin interactions. The refined technology, as described in chapter 4, to directly map the ER genome-wide chromatin binding landscape from only limited amounts of tumor tissue permits for a high-resolution transcription factor mapping even from clinical samples.

Frequently, side effects of therapy are suggested to be surrogate readouts for treatment success. In chapter 5 we investigate whether frequency or severity of hot flashes are correlated to serum concentrations of tamoxifen and its metabolites. No associations were observed between concentrations of active tamoxifen metabolites and hot flashes during treatment.

In chapter 6 we study the association between genetic variants of the gene coding for estrogen receptor alpha (ESR1) and hot flashes, since it was considered that a potential explanation for the inter-individual differences in occurrence of vasomotor symptoms, despite the same therapy, might be variability in receptor affinity by tamoxifen. Nevertheless we were unable to find evidence supporting the hypothesis that frequency of hot flashes is associated with rs9340799 and/ or rs2234693 polymorphisms in ESR1. Based on our results in chapter 5 and 6 occurrence of hot flashes should not be seen as a surrogate biomarker for therapeutic serum endoxifen levels nor for tamoxifen receptor affinity.