Background: We have previously demonstrated how transcriptional pathway activity and the molecular subtypes of breast cancer metastases significantly influence patient post-relapse survival. Here we extend our analysis to determine whether the prognostic information provided by gene expression signatures in primary breast tumours is also relevant in the metastatic setting. Specifically, we test the research versions of the Genomic Grade Index (GGI), Mammaprint, Recurrence score (RS) and PAM50 gene signatures along with our own cell-cycle based classifier (CCS).

Methods: 287 patients with morphologically confirmed loco-regional or distant breast cancer relapse were enrolled in the Swedish multicenter TEX trial from December 2002 until June 2007. Of these, sufficient tumour RNA for gene expression profiling was obtained from metastatic tissue by fine needle aspiration from 111 patients (totalling 120 relapse biopsies). Gene signatures were applied as described in the original research articles and their relationship to short (1.5 year) and long-term (5 year) post-relapse survival was assessed using likelihood ratio, Kaplan-Meier and Cox regression analysis.

Results: As anticipated from an aggressive metastatic cohort, the majority of samples (> 70%) were classified into intermediate or high risk groups by all signatures. In both short and long-term survival analysis only PAM50 provided statistically significant prognostic information (short: LRχ2 = 14.7, p = 0.005 and long: LRχ2 = 13.2, p = 0.010), with the cell cycle score signature displaying a prognostic trend in long-term survival only (LRχ2 = 5.2, p = 0.074). Kaplan-Meier curves and Cox regression analysis suggest that the strength of both signatures resides in their ability to select a group of low-risk patients with better long-term survival.

Conclusions: Our findings demonstrate the prognostic utility of the multi-level PAM50 and to a lesser extent, cell cycle score signatures in predicting survival of patients with metastatic breast cancer. Simpler binary gene expression signatures (GGI and Mammaprint) do not appear to capture the same prognostic information and as such may have limited utility in a metastatic setting.

Short and long term survival Likehood Ratios for five gene expression signatures in metastatic breast cancer

  Short term survival (1.5 year) Long term survival (5 year) 
Gene Signature LRχ2 P-value LRχ2 P-value 
GGI 1.3 0.251 0.5 0.500 
Mammaprint 1.7 0.190 0.6 0.427 
RS 3.9 0.143 4.4 0.110 
CCS 2.8 0.242 5.2 0.074 
PAM50 14.7 0.005 13.2 0.010 
  Short term survival (1.5 year) Long term survival (5 year) 
Gene Signature LRχ2 P-value LRχ2 P-value 
GGI 1.3 0.251 0.5 0.500 
Mammaprint 1.7 0.190 0.6 0.427 
RS 3.9 0.143 4.4 0.110 
CCS 2.8 0.242 5.2 0.074 
PAM50 14.7 0.005 13.2 0.010 

GGI: Genomic grade index; RS: Recurrence score; CCS: Cell cycle score

Citation Format: Tobin NP, Lundberg A, Lindström LS, Harrell JC, Egyhazi Brage S, Frostvik Stolt M, Einbeigi Z, Loman N, Malmberg M, Perou CM, Bergh J, Hatschek T. Multi-level gene expression signatures provide significant prognostic information in metastatic breast cancer patients [abstract]. In: Proceedings of the 2016 San Antonio Breast Cancer Symposium; 2016 Dec 6-10; San Antonio, TX. Philadelphia (PA): AACR; Cancer Res 2017;77(4 Suppl):Abstract nr P1-07-16.