ESR1 and endocrine therapy resistance: more than just mutations

Estrogen receptor (ER)-positive breast cancer accounts for 70%–80% of all diagnosed breast cancers [1]. The adoption of endocrine therapies, including ER modulators/degraders (SERMs/SERDs), which antagonize ER, and aromatase inhibitors (AIs), which suppress estrogen synthesis, as the mainstay of treatment of ER-positive breast cancer patients has resulted in substantial survival benefit for patients with early stage disease [2]. Treating ER-positive metastatic breast cancer (MBC), however, remains a significant clinical challenge, due to the development of secondary resistance to all modalities of endocrine therapy [3]. Recently, studies have identified recurrent somatic mutations within the ligand-binding domain (LBD) of ESR1 (encoding ER) in >30% of ER-positive MBC [4–8]. These mutations alter the conformation of ER and produce a constitutively active form of the protein. Mutations at residues 536–538, in particular, promote ER activity in the absence of ligand, resulting in resistance to AIs and reduced sensitivity to SERMs/SERDs [4, 5]. ESR1 fusion genes have also been reported in ER-positive MBCs; however, a detailed description of their manifestations and clinical prevalence is lacking [9]. In this issue of Annals of Oncology, Hartmaier et al. reported the identification of recurrent hyperactive ESR1 fusion genes in breast cancers resistant to endocrine therapy [10], adding to the diversity of reported ESR1 alterations.

Hartmaier et al. carried out a retrospective study to discover genomic rearrangements involved in the acquired resistance to ER-targeted therapies [10]. Using mate-pair DNA sequencing and/or RNA sequencing of matched primary-metastasis-normal samples from 6 patients, the authors identified an ESR1-DAB2 in-frame fusion transcript that fused exons 1–6 of ESR1 to exons 3–15 of DAB2. This fusion was found only in the lymph node metastasis but not in the primary. RNA sequencing analysis of an additional 51 breast cancer metastases revealed an ESR1-GYG1 fusion gene in a bone metastasis, comprising the ESR1 exons 1–6, the same involved in the ESR1-DAB2 fusion gene, and the 3′ end of GYG1. Importantly, both fusions were also detectable at the protein level [10].

Prompted by the discovery of recurrent ESR1 fusion breakpoints, the authors analyzed 9542 breast cancers (including 5216 from metastases) and 254 circulating tumor DNA (ctDNA) samples from advanced breast cancer patients, and identified 7 additional ESR1 fusion genes. Including the initial cohorts subjected to mate-pair and/or RNA sequencing, 5 fusions were identified in metastatic disease (5/5, 272, 0.09%), 1 in local recurrence after endocrine therapy (1/4, 329, 0.02% of primary tumors) and 3 in ctDNA (3/254, 1.2%). For the 4/9 patients with available clinical histories, all had been treated extensively with AIs. Of note, the ESR1 breakpoints were all in or between exons 6 and 7, disrupting the LBD of ESR1. In vitro analysis of 3 of the fusions identified (ESR1-DAB2, ESR1-GYG1, and ESR1-SOX9) demonstrated that all had ligand-independent activity and two were hyperactive [10]. These observations suggest a potential role for the distinct 3′ gene partners in determining resultant ER activity.

With genomic breakpoints frequently located in intronic regions, capture-based targeted sequencing of exons does not always detect fusion genes. Structural rearrangements, however, are frequently associated with copy number alterations. Based on this notion, the authors devised a novel algorithm copyshift to detect intra-genic fusion junctions associated with copy number changes in targeted sequencing. When tested in a cohort of lung cancer with known ALK rearrangements, the authors showed that copyshift was specific (>89% positive predictive value), albeit with limited sensitivity (∼85% false negative rate). Applying copyshift to the cohort of 9542 breast cancers to interrogate the recurrent ESR1 breakpoint region, the authors found 83 copyshift-positive tumors. These tumors were enriched for ER-positive and metastatic disease, and were associated with the presence of activating ESR1 mutations. Taken together, the authors estimated that at least 1% of MBCs harbored ESR1 fusion genes. With the 85% false negative rate of copyshift, the prevalence of 1% may represent an underestimate, since copyshift prioritizes specificity over sensitivity.

The discovery of recurrent ESR1 fusion genes reinforces the concept that resistance to targeted therapies often represents a convergent phenotype (i.e. that resistance, the phenotype observed, may be caused by distinct genetic alterations) [11]. In the case of resistance to endocrine therapy, the identification of recurrent rearrangements adds to the previously described activating mutations to expand the repertoire of genetic alterations affecting ESR1 (Figure 1). A similar phenomenon has been described for resistance to PARP inhibitors/platinum salts in BRCA1/2 germline mutation carriers, with distinct BRCA1/2 intragenic deletions or reversion mutations [12–14] restoring the reading frame and thus resulting in resistance. In fact, multiple activating ESR1 mutations have been detected in the ctDNA samples of patients harboring activating ESR1 fusion gene [10], suggesting the existence of polyclonal resistance mechanism, akin to the recent reports of polyclonal BRCA1/2 reversion mutations in therapy-resistant patients [15, 16].

Figure 1.

Development of resistance to estrogen suppression via distinct somatic genetic alterations in the ESR1 gene. Structural/functional representation of the somatic genetic alterations in ESR1 leading to resistance to estrogen suppression in breast cancer. The structural domains of the ESR1 gene are shown in different colors, including the transcription activation function 1 (AF-1) domain, the DNA-binding zinc finger, C4 type domain (zf), ligand-binding domain (LBD), and the nuclear receptor C-terminal domain (NR). The positions of the coding exons 3–10 are illustrated beneath the protein domains. Whilst wild-type ESR1 consists of intact coding exons 3–10, ESR1 fusion genes and/or hotspot activating mutations disrupt the ligand-binding domain.

The results reported by Hartmaier et al. [10] have important and immediate clinical implications. The emergence of ESR1 mutations as a resistance mechanism to AIs has led to the development of novel molecules and combination regimens that are more effective in suppressing ESR1-mutant tumor growth [17–19]. Even these new agents and regimens, however, rely on targeting the LBD. The loss of the LBD by genomic rearrangements may represent a resistance mechanism not readily detectable by common approaches such as ddPCR/BEAMing and targeted sequencing of exonic regions, arguing for the use of methods such as copyshift or RNA sequencing based methods to identify such alterations. It should be noted, however, that the biological and clinical significance of the findings by Hartmaier et al. [10] remains to be fully elucidated, given that limited clinical history was available. Furthermore, the authors demonstrated the detection of the chimeric proteins in only a handful of patients and it is possible that some of the ESR1 fusion genes are not transcribed and/or translated, or may have limited impact on the resistance to endocrine therapies. It would be important to investigate the mutual exclusivity or co-occurrence between ESR1 mutations and/or fusions and other mechanisms of resistance [20–23] and the existence of multiple subclonal resistance mechanisms in individual patients, and whether the mechanisms resulting in the acquisition of distinct modalities of resistance to endocrine therapy would differ.

Funding

Swiss National Science Foundation (Ambizione grant number PZ00P3_168165) to SP. SC and JSR-F are funded in part by the Breast Cancer Research Foundation. BW, SC and JSR-F are supported in part by the MSKCC Cancer Center Support grant of the National Institutes of Health/National Cancer Institute (CCSG, P30 CA08748). The content is solely the responsibility of the authors and the funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Disclosure

The authors have declared no conflicts of interest.