On-treatment Biomarkers Can Improve Prediction of Response to Neoadjuvant Chemotherapy in Breast Cancer

Richard J. Bownes; Arran K. Turnbull; Carlos Martinez-Perez; David A. Cameron; Andrew H. Sims; Olga Oikonomidou


Breast Cancer Res. 2019;21(73) 

In This Article

Abstract and Introduction


Background: Neoadjuvant chemotherapy is increasingly given preoperatively to shrink breast tumours prior to surgery. This approach also provides the opportunity to study the molecular changes associated with treatment and evaluate whether on-treatment sequential samples can improve response and outcome predictions over diagnostic or excision samples alone.

Methods: This study included a total of 97 samples from a cohort of 50 women (aged 29–76, with 46% ER+ and 20% HER2+ tumours) with primary operable breast cancer who had been treated with neoadjuvant chemotherapy. Biopsies were taken at diagnosis, at 2 weeks on-treatment, mid-chemotherapy, and at resection. Fresh frozen samples were sequenced with Ion AmpliSeq Transcriptome yielding expression values for 12,635 genes. Differential expression analysis was performed across 16 patients with a complete pathological response (pCR) and 34 non-pCR patients, and over treatment time to identify significantly differentially expressed genes, pathways, and markers indicative of response status. Prediction accuracy was compared with estimations of established gene signatures, for this dataset and validated using data from the I-SPY 1 Trial.

Results: Although changes upon treatment are largely similar between the two cohorts, very few genes were found to be consistently different between responders and non-responders, making the prediction of response difficult. AAGAB was identified as a novel potential on-treatment biomarker for pathological complete response, with an accuracy of 100% in the NEO training dataset and 78% accuracy in the I-SPY 1 testing dataset. AAGAB levels on-treatment were also significantly predictive of outcome (p = 0.048, p = 0.0036) in both cohorts. This single gene on-treatment biomarker had greater predictive accuracy than established prognostic tests, Mammaprint and PAM50 risk of recurrence score, although interestingly, both of these latter tests performed better in the on-treatment rather than the accepted pre-treatment setting.

Conclusion: Changes in gene expression measured in sequential samples from breast cancer patients receiving neoadjuvant chemotherapy resulted in the identification of a potentially novel on-treatment biomarker and suggest that established prognostic tests may have greater prediction accuracy on than before treatment. These results support the potential use and further evaluation of on-treatment testing in breast cancer to improve the accuracy of tumour response prediction.


Chemotherapy is among the most common effective treatments for breast cancer, alongside radiotherapy, hormone therapy, and targeted treatments. Neoadjuvant chemotherapy is given prior to surgery with the aim to reduce the tumour burden and to provide early information on the response to treatment.[1] Studies have shown patients with tumours that have a pathological complete response (pCR) following neoadjuvant chemotherapy are much less likely to recur than those in women with residual disease.[2] Neoadjuvant chemotherapy is now considered as the standard of care in breast cancer and has seen a rise in recent years with data from powered studies suggesting that the pathological complete response achieved following neoadjuvant chemotherapy might be a surrogate of good prognosis.[3] A recent meta-analysis also showed significant tumour response and an increase in the rate of breast-conserving surgery following NACT with good rates of long-term local recurrence (5.5% vs. 15.9% adjuvant chemotherapy), however with an increase in the rate of short-term local relapses (1.35 RR 0–4 years, 1.53 RR 5–9 years).[4]

Neoadjuvant treatment provides a "window of opportunity" (Figure 1a), where sequential sampling of a tumour enables observation of the changes that occur in response to treatment to be measured and considered in the context of response and outcome.[5] Neoadjuvant therapy studies and pre-surgical treatments allow for a unique in vivo analysis of tumour treatment response,[6] as well as the possibility of predicting the response to treatment earlier in the treatment.[5] It has been suggested that on-treatment biomarkers may be superior to those measured before exposure to treatment.[3,7] On-treatment information has already been shown to be informative for the accurate prediction of response to endocrine therapy.[8] Here, it was found that patients with elevated Ki67 levels (higher than 10%) at 2 or 4 weeks exhibited resistance to endocrine therapy and were triaged to neoadjuvant chemotherapy.[8] We have also demonstrated the potential of on-treatment biomarkers by developing a four-gene signature which combined pre-treatment expression levels or two biomarkers (IL6ST and NGFRAP1) with patient-matched 2-week on-treatment expression levels of two proliferation markers (ASPM, MCM4) to accurately predict the response to endocrine therapy in a blinded independent validation set.[7]

Figure 1.

Unsupervised analysis cannot distinguish pre- and on-chemotherapy samples of breast tumours. a Schematic representation demonstrating sequential sampling of breast tumours during treatment. b PCA analysis of pre- and on-treatment samples from the Edinburgh NEO and I-SPY studies revealed no significant clustering of patients by time or response group. Red = non-responder, orange = partial responder, blue = complete responder

Gene expression-based studies of neoadjuvant chemotherapy treatment to date have largely been limited to studying the association of pre-treatment samples with pathological response.[9,10] Patient-matched sequential sampling gene expression studies have been previously attempted; however, they have not evaluated the predictive capacity or proposed new on-treatment predictive biomarkers.[11–13]

In this study, we present the largest sequentially sampled patient-matched analysis of neoadjuvant chemotherapy-treated breast cancer tumours to evaluate whether on-treatment biomarkers can improve the accuracy of predicting response before resection. Numbers of patients with sequential breast tumour samples are limited, but we compare and validate our results with the data from the I-SPY 1 Trial.