Prognostic value of metabolic response in breast cancer patients receiving neoadjuvant chemotherapy

Abstract

Background: Today's clinical diagnostic tools are insufficient for giving accurate prognosis to breast cancer patients. The aim of our study was to examine the tumor metabolic changes in patients with locally advanced breast cancer caused by neoadjuvant chemotherapy (NAC), relating these changes to clinical treatment response and long-term survival.

Methods: Patients (n = 89) participating in a randomized open-label multicenter study were allocated to receive either NAC as epirubicin or paclitaxel monotherapy. Biopsies were excised pre- and post-treatment, and analyzed by high resolution magic angle spinning magnetic resonance spectroscopy (HR MAS MRS). The metabolite profiles were examined by paired and unpaired multivariate methods and findings of important metabolites were confirmed by spectral integration of the metabolite peaks.

Results: All patients had a significant metabolic response to NAC, and pre- and post-treatment spectra could be discriminated with 87.9%/68.9% classification accuracy by paired/unpaired partial least squares discriminant analysis (PLS-DA) (p < 0.001). Similar metabolic responses were observed for the two chemotherapeutic agents. The metabolic responses were related to patient outcome. Non-survivors (< 5 years) had increased tumor levels of lactate (p = 0.004) after treatment, while survivors (≥ 5 years) experienced a decrease in the levels of glycine (p = 0.047) and choline-containing compounds (p ≤ 0.013) and an increase in glucose (p = 0.002) levels. The metabolic responses were not related to clinical treatment response.

Conclusions: The differences in tumor metabolic response to NAC were associated with breast cancer survival, but not to clinical response. Monitoring metabolic responses to NAC by HR MAS MRS may provide information about tumor biology related to individual prognosis.

Figure 1

Scores and loadings from multilevel PLS-DA. The score and loading plots show the two first latent variables separating treatment and control spectra based on (A) the whole data set, (B) 5-year survivors, and (C) non-survivors. The variables in the loadings are colored according to VIP scores, indicating the importance of each variable in the discrimination. The control spectra equal the difference between pre- and post-treatment spectra, while the treatment spectra equal the difference between post- and pre-treatment spectra. The most protruding differences in the loading plots of (B) survivors and (C) non-survivors show a decrease in GPC levels in survivors after treatment, while non-survivors have increased levels of lactate. PC appears to be present in high levels in some samples of both survivors and non-survivors. Lac, lactate; Gly, glycine; β-Glc, β-glucose.

Figure 2

PLS-DA of the MR spectra from biopsies excised post-treatment. (A) A score plot separating survivors and non-survivors, (B) Representative spectra showing the metabolic differences of the tumors of survivors and non-survivors. (C) The loadings of the PLS-DA model with variables colored according to the VIP scores. β-Glc, β-glucose; Lac, lactate, Gly, glycine; Cr, creatine; Ala, alanine.