In this study, we assessed for the presence of mutations within key BC genes utilising the UltraSEEK® technology in a paired cohort of BCs and BMs. A mutation in at least one of the five genes (TP53, PIK3CA, ERBB2, ESR1, AKT1) was identified in most cases, with 90.6% of primary BC and 87.5% of the paired BMs having at least one mutation. The observed similarities regarding the frequency of the mutated genes reflect the nature of the UltraSEEK® BC panel that was designed to target actionable genes present in all the key BC subtypes. The actionability of these mutated genes in primary BC and BM is presented in Table 2 and in a recent systematic review . The detection of ESR1 mutations in the primary tumours also highlights the sensitivity of the assay . Although ESR1 mutations were more commonly identified in metastatic BCs [22, 23], the use of the sensitive droplet-digital PCR reported higher frequencies in primary tumours [24, 25]. The identification of therapeutic strategies in breast cancers harbouring ESR1 mutants is an area of active interest. Fulvestrant has demonstrated poor clinical activity in ESR1-mutated BC [26, 27], whereas bazedoxifene and lasofoxifene have demonstrated activity in pre-clinical models of ESR-mutated BC [28, 29]. The efficacy of lasofoxifene is currently being explored in the ELAINE trial (NCT03781063) in patients with ESR1-mutated BCs (Supplementary table 4). Interestingly, in our cohort, ESR1 mutations were seen in 28.1% of BMs despite the loss of ER expression by IHC, indicating that the ESR1-mutant clones are likely dominant clones, resistant to therapy [22–25, 27]. These results are in accordance with others who detected a high ESR1 mutation frequency (34.3–44.9%) in BMs .
While TP53 is mutated in all BC subtypes, it is most common in TNs and HER2 [3, 30–32]. TP53 mutations have been associated with worse clinical outcomes and poor response to hormonal therapy, chemotherapy and/or radiotherapy [30–33] and our samples have been acquired from patients who had progressed and developed brain metastasis despite prior treatment. TP53 can be re-activated by targeting molecules that modulate its posttranslational modifications, localization and degradation and several ongoing clinical trials are using TP53-reactivating compounds in combination with chemotherapeutic drugs (Supplementary table 4) [34, 35].
ERBB2/HER2 mutations were identified in only 2 samples, a primary ER-positive BC and a HER2-positive BM. These ERBB2/HER2 mutations are associated with resistance to lapatinib but are sensitive to neratinib, highlighting the importance of treating HER-mutated cancers with the appropriate HER-targeted drugs (Table 2, Supplementary table 4) [36, 37]. The identification of mutations in PIK3CA and AKT1 are also of clinical significance since both genes are druggable (Table 2, Supplementary table 4). The PI3Kα-specific inhibitor alpelisib, has shown activity in PIK3CA-mutant breast cancers (NCT02437318) and recently granted FDA and European commission approval while its potential in the regression and stabilisation of progressive BCBM has been highlighted [38, 39]. The brain-penetrant inhibitor paxalisib (GDC-0084) has demonstrated activity in pre-clinical models of BCBM . Several inhibitors targeting the AKT1 pE17K mutation, an oncogenic driver in BC, have shown efficacy as monotherapy or in combination with other drugs (Table 2, Supplementary table 4) [41–43]. We recently presented a summary of current clinical trials on mutated BCBMs .
Several studies have identified large similarities in the mutational profiles of primary tumours and their metastases including BMs [5–10], whereas others showed clear differences between primary and metastasis in the numbers and types of mutations [11–14]. It was recently suggested that the systemic metastatic seeding can begin early during primary tumour growth and that the clonal architecture is remodelled by treatment that may select for disseminated cells harbouring resistant mutations . Treatment was also associated with high gene heterogeneity and monoclonal metastases . Similarly, within our cohort of paired BC and BM cases, the median BMFS time (irrespectively of mutation status) was 26 months and certain gene mutations were detected with similar confidence levels in 56.3% of paired cases (monoclonal metastases). We also identified cases where the mutation was absent or present with low confidence in the primary and with high confidence in the BM and vice versa. The fact that only 37.5% of mutations in the 5 chosen genes are conserved between the primary and just one major metastatic site, the brain, suggests that, in general, compounds targeting these mutations identified solely in the primary breast tumour are unlikely to be successful in the majority of advanced/metastatic patients. Therefore, it seems to be necessary to sample metastases as well as the primary tumour to identify somatic mutations common to tumours at both sites, to predict more accurately whether a patient will respond to such chemical interventions. The differences in the mutational landscape could be attributed to the clonal evolution process, the selective pressure of different therapeutic regimes and the receptor switching between primary BC and metastasis. Evidence of clonal remodelling between primary tumours and metastases associated with the clinical subtype conversion was recently presented, but the most frequently mutated genes in primary tumours were also identified in metastases independent of the tumour subtype . We did not observe significant differences in the frequently mutated genes in relation to the subtypes in BC and BM and there was no significant association between receptor switching and number of mutations. This could also be attributed to the small number of genes present in the targeted BC mutation panel and the small number of mutations identified in our cohort. Nevertheless, within our cohort, patients carrying ≤ 3 mutations had a significantly better BMFS (p = 0.0001), than patients with > 3 mutations indicating that the higher number of mutations correlates to worst prognosis (metastasis-free survival).
A limitation of the MassARRAY UltraSEEK technology in comparison to next generation sequencing on the identification of mutations is the use of predefined mutations across different oncogenes. Although, it cannot detect unknown mutations and copy number alterations, it is more cost-effective and could be easier applied in a clinical setting [20, 45, 46]. Moreover, the challenges to obtain BM samples for sequencing including the inherent risks of neurosurgery, as samples can only be taken when surgical resection is clinically indicated, make longitudinal studies of the changes in the BM genome unethical. The alternative use of circulating cell-free DNA (cfDNA) in the CSF and plasma is under investigation and in the future patients will be able to have tailored treatments based on the results of sequencing cfDNA from CSF rather than relying on the genomics of the primary lesion [19, 45, 46].
In summary, our data highlights the presence of clinically important and actionable mutations in AKT1, ESR1, PIK3CA, TP53 and ERBB2 genes in BC and in BCBM as identified by the UltraSEEK BC panel that provides a powerful tool to investigate low abundance mutations and could be potentially useful in a clinical environment [24, 25]. These mutations could be used to identify patients resistant to certain therapeutic regimes and enable the development of more tailored clinical studies utilising targeted agents or combinations of them in the brain metastatic setting.