Molecular features of untreated breast cancer and initial metastatic event inform clinical decision-making and predict outcome: long-term results of ESOPE, a single-arm prospective multicenter study

Abstract

Background: Prognosis evaluation of advanced breast cancer and therapeutic strategy are mostly based on clinical features of advanced disease and molecular profiling of the primary tumor. Very few studies have evaluated the impact of metastatic subtyping during the initial metastatic event in a prospective study. The genomic landscape of metastatic breast cancer has mostly been described in very advanced, pretreated disease, limiting the findings transferability to clinical use.

Methods: We developed a multicenter, single-arm, prospective clinical trial in order to address these issues. Between November 2010 and September 2013, 123 eligible patients were included. Patients at the first, untreated metastatic event were eligible. All matched primary tumors and metastatic samples were centrally reviewed for pathological typing. Targeted and whole-exome sequencing was applied to matched pairs of frozen tissue. A multivariate overall survival analysis was performed (median follow-up 64 months).

Results: Per central review in 84 patients (out of 130), we show that luminal A breast tumors are more prone to subtype switching. By combining targeted sequencing of a 91 gene panel (n = 67) and whole-exome sequencing (n = 30), a slight excess of mutations is observed in the metastases. Luminal A breast cancer has the most heterogeneous mutational profile and the highest number of mutational signatures, when comparing primary tumor and the matched metastatic tissue. Tumors with a subtype change have more mutations that are private. The metastasis-specific mutation load is significantly higher in late than in de novo metastases. The most frequently mutated genes were TP53 and PIK3CA. The most frequent metastasis-specific druggable genes were PIK3CA, PTEN, KDR, ALK, CDKN2A, NOTCH4, POLE, SETD2, SF3B1, and TSC2. Long-term outcome is driven by a combination of tumor load and metastasis biology.

Conclusions: Profiling of the first, untreated, metastatic event of breast cancer reveals a profound heterogeneity mostly in luminal A tumors and in late metastases. Based on this profiling, we can derive information relevant to prognosis and therapeutic intervention, which support current guidelines recommending a biopsy at the first metastatic relapse.

Trial registration: The trial was registered at ClinicalTrials.gov ( NCT01956552 ).

Keywords: Breast cancer; Metastasis; Next generation sequencing; Prognosis; Targetable genes; de novo metastases.

Conflict of interest statement

The authors declare that they have no competing interests.

Figures

Fig. 1

Mutational load in breast cancer primary tumors and their metastatic counterparts by whole-exome sequencing. a Average mutation rate in primary tumors (PT) and their matched metastatic tissue (M). b Average mutation rate in primary tumors (PT) and their matched metastatic tissue (M), according to the subtype of the primary tumor. The mutational burden in all paired samples (a) and by molecular subtypes (b) is represented as the number of mutations per megabase of sequence. The violin plots represent the distribution of the data. Asterisks indicate a statistically significant difference among TNBC and luminal B patients (Wilcoxon rank-sum test false discovery rate)

Fig. 2

Examples of phylogenetic trees of different tumor subtypes, based on mutational signatures. The mutational signatures were generated separately for variants shared between PT and M or private to each sample. The relative contribution of the signature were then annotated to the phylogenetic tree of 12 breast cancers cases of all 4 subtypes (luminal A, luminal B, Her2, and TNBC). HR, homologous recombination; MMR, mismatch-repair deficiency

Fig. 3

Mutational landscape and mutational load in matched primary tumors and metastases. The respective mutational landscape of primary tumors and matched metastases is shown as captured by targeted sequencing, according to primary tumor subtype and metastatic site. Gray cells indicate a sequencing depth 

Fig. 4

Number and distribution of mutations in driver genes as captured by whole-exome sequencing. a Number of shared (light gray) and private (dark gray) mutations in driver genes. *p < 0.01. b Number of private mutations in driver genes, in the primary tumor (light gray), or in metastases (dark gray). *p < 0.01. c Fraction of overall or driver mutations detected in both primary tumors and metastases. *p<0.01 Lines correspond to the mean

Fig. 5

Overall survival. Numbers of patients at risk are indicated beneath the curves. a Overall survival according to primary tumor IHC subtype. b Overall survival according to metastasis subtype. c Overall survival according to LAMA2 mutational status

Fig. 6

Mutational profile of de novo versus late metastases. a Landscape of private versus shared mutations in primary tumors and their matched metastatic tissue, according to metastasis onset pattern. b Pie plot showing the proportion of private versus shared mutations in primary tumors and their matched metastatic tissue. c Box-plot showing the mutation number in late versus de novo metastases. White box: de novo metastases; black box: late metastases. *p < 0.01. d Box-plot showing the median of shared and private driver mutations identified in late and de novo breast cancers. Dark boxes: shared mutations; gray box: private mutations; ***p < 0.0001