Concordant with previous published data, we confirm that TP53 and PIK3CA is the most common mutations in BCBM, as well as, the matched PTs, in to our very best knowledge so far one of the largest cohorts of genetically characterized BCBM [16–21]. Difficulties in accessing intra-cranial tissue samples have hampered development of systemic treatments that could have an effect in the treatment of patients with BMs. Our cohort, consists of 52 analyzed cases, and 37 cases with matched PT, all from one region in Sweden. The difficulty in obtaining tissue from the brain results in few tumors which limits the study when comparing BC subgroups or in survival analyses. Although the tissue material is partly old, with the first samples collected in 1994, sequencing results by NGS were successfully obtained in 90% of the BMs. At least one mutation, among 50 cancer driver genes, was present in 62% of the analyzed samples which is in the similar range as reported in previous publications [16–21]. Concordant with these previous studies, we found TP53 and PIK3CA as the most commonly mutated driver genes, both in the BMs and PTs. Of these two mutations, PIK3CA are considered targetable, whilst TP53 mutations can be indirectly attacked through restoration of the transcriptional activity resulting in a functional wild-type TP53 protein [22, 23].
The drug alpelisib, is a PI3Kα-specific inhibitor available for patients with recurrent PIK3CA mutated Luminal/HER2 negative BC [23]. Previous published data show that approximately 40% of ER positive PT harbor a PIK3CA mutation [24]. Two studies that compared PIK3CA mutations in PTs and BC metastases reported mutations in 33% of PTs and 30% of metastases and 45% of PTs and 53% in metastases respectively [18, 25]. There were few or no patients with BMs in the above mentioned studies. Our figures in BCBM were lower with a PIK3CA mutation frequency of 20% in PTs and 23% in the BMs. A recently published systematic review compared 164 BMs with its matched PT in 126 patients extracted from 13 studies and found PIK3CA mutations in 22% of the patients with BMs which in line with our results [26]. It might be hypothesized that PIK3CA mutations is lower in BMs compared to other metastatic sites, however, additional studies are required to adequately answer this question. As expected, we found the highest proportion of PIK3CA mutations in Luminal BC. Of note, in our material two out of eight pathogenic PIK3CA mutations and five TP53 mutations were found in the BMs only and not in the PT underlining the need for re-evaluation of metastatic tissue or possibly by analysis of cell free DNA (cfDNA) [27].
We found PIK3CA mutations as the second most common mutation after TP53 in matched pairs of TNBC with 18% PIK3CA mutations all concordant in PTs and BMs. This is in line with previous published data in which PIK3CA mutations was the second most common mutation after TP53, especially in basal-like and luminal androgen receptor subtypes of TNBC [28–30]. The combined treatment of alpelisib and nab-paclitaxel is currently under investigation in pre-treated TNBC with either loss of PTEN expression or a PIK3CA mutation (NCT04251533).
The prognostic role of PIK3CA mutations in HER2 + BC has been extensively investigated. Mutations in PIK3CA has been associated with less effect of HER2 blocking therapy, both in the neo-adjuvant setting, and for recurrent BC in terms of lower pCR rates and shorter survival respectively [31, 32]. We found PIK3CA mutations in only 10% of the HER2 + cases, which is lower than the previously reported frequency of 20–40%. [24]. The limited number of HER2 + patients in our cohort is considered to be the cause of this discrepancy. The effect of alpelisib in HER2 + breast cancer is under investigation in one ongoing and one completed clinical trial (NCT02038010; NCT04208178).
Almost 50% of the PTs in the present study population that have a high proportion of TNBC harbored a TP53 mutation, in contrast to the 20–35% mutation prevalence reported in unselected primary BC [33]. The Cancer Genome Atlas reveals an enrichment of TP53 mutations in basal-like and HER2 enriched BC [24]. Interestingly, previous data reveal that a high proportion of patients with a TP53 mutation in the primary BC developed BM [34, 35]. We lack data on the molecular subtypes in our material, but find an increased number of TP53 mutations in TNBC and HER2/ER- BC. Enrichment of TP53 mutations in BMs was not seen in the HER2 amplified subgroup, a finding that must be interpreted with caution due to the very small sample size of the HER2 group. Other detected mutations in our series, CDH1, EGFR, HRAS, RB1, CDKN2A and PTEN were rare, in general found in single or both samples from one patient.
Roughly 25% of the matched pairs changed IHC based BC subtype in the BMs. The most common change was from Luminal A in PT to other subtypes, in most cases Luminal B in BMs. This is in accordance with previous results in which 219 patients showed a 36% overall discordance with the most common change in form of loss of PgR [36]. In a review pooling a total of 3384 matched pairs of BC and metastases from all organs, BM showed a discordant median rate of 22% compared to 45% in liver metastases and 16% in lymph node metastasis [37]. TNBC was the most stable group with less than 10% showing a gain in ER (one case) and HER2 (one case).
A limitation of our study is that genetic profiling was performed using a relatively small NGS panel of 50 genes. Even though the panel gives broad coverage of important cancer driver genes, more comprehensive sequencing might have provided further details in the landscape of mutation discordancy between PTs and BMs. Still, we believe that the relatively large number of matched PT and BM add to the knowledge about the biology of BMs, how the metastatic process affects actionable genes.
In conclusion, we confirm mutations in TP53 and PIK3CA to be common in both primary breast tumors and BMs but the proportion varied depending on the subgroup. Mutation pattern, as well as IHC based subtypes were discordant in approximately 25% of the patients underlining the need for re-biopsy at disease progression. In this context, analysis of cfDNA may be a fruitful avenue in patients with BMs, placing hope in promising results with liquid biopsy from CSF [27, 38, 39].