The identification of clinical, pathological and molecular features predicting the benefit of treatments is a major goal for medical oncologists.
Fulvestrant has been proven a valuable option for treatment of HR+ABC. First-line fulvestrant was associated with a mPFS of 16.6 months which extended to 22 months in patients with bone-only metastases in a population of mostly treatment-naïve patients (5). In the cohort of patients treated as 1st line in the MonaLEEsa-3 study mPFS in the Fulvestrant /placebo arm was 19 months, while in the control arms of 2nd/3rd line studies with CDK 4/6i and Fulvestrant, mPFS decreased to 9 and 4.6 months, respectively (15-17). Similar data have been reported in patients treated with Fulvestrant in real-world series (18-20).
Identification of pathological and molecular features predicting benefit from treatment in tumor samples of patients treated with fulvestrant has been attempted previously (21,22).The TransCONFIRM, a translational analysis within the CONFIRM trial aimed to identify clinical-pathological features and molecular signatures in the primary tumors predicting response to fulvestrant, showed in the 112 samples analyzed that only PgR and HER2 expression and a signature of 37 genes were independently associated with PFS, (21). Of note, no correlation with ESR1 mutations on primary tumors was found (22). Christensen et al. investigated in mRNA extracted from 226 tumor samples of patients treated with fulvestrant the predictive value of a mathematical algorithm based on the expression of multiple genes (DRP) (22). The DRP was associated although not significantly with outcome in patients treated in earlier lines and unexposed to previous adjuvant endocrine therapies (23).
Our clinical data are comparable with those mentioned above, showing a mPFS of 10.6 months in a heavily pretreated cohort, since 46% of our patients were treated in 3rd or later lines. Only the occurrence of visceral metastases was associated with a worse outcome, although we cannot exclude that the correlations between HR and PI3KCA mutation status and outcome were not significant only because of the limited sample size. Similarly, only the absence of visceral metastases was significantly associated with a 4-fold higher likelihood of long-term benefit from fulvestrant,
Notably, patients starting Fulvestrant after progressing to other therapies experienced a similar benefit as patients receiving treatment as maintenance after chemotherapy (mPFS=9.9 months vs 10.8 months, respectively).
The role of PI3KCA mutations in the mechanisms of resistance to fulvestrant is not clear. Results of the fulvestrant/placebo arms in studies with PI3K inhibitors are inconsistent (23). While in the BELLE-2 study mPFS in patients with PI3KCA wild-type tumors doubled that of patients with PI3KCA mutant tumors, no difference was observed in the same comparisons within the BELLE-3 and SOLAR-1 studies (23). In a recent analysis on ctDNA of patients included in the PALOMA 3 trial PI3KCA mutations while being among the most common mutations at baseline did not increase after progression and did not correlate with PFS (24).
In our study PI3KCA mutations were found in about 36% of patients, consistently with what expected in Luminal breast cancers (25) and were associated with a numerically shorter PFS but not with a lower likelihood of being progression-free at 18 months. These findings are only partially consistent with the analysis of the long-term benefit in the fulvestrant/placebo arm of the PALOMA 3 trial since in this study only 6% of patients harboring PI3KCA mutations vs 39% of patients with PI3KCA wild type tumors were treated for ≥ 18 months (10).
Similarly to what observed in the PALOMA 3 study, the lack of PgR expression was associated with a lower likelihood of long-term benefit from Fulvestrant (10).
The rate of ESR1 mutations was lower than expected, despite more than 50% of patients had previously received a NSAI Notably ESR1 mutations were observed only in post-progression biopsy in 2 patients, one of whom had lost at the same time PI3KCA mutation, suggesting that a different mechanism of resistance to fulvestrant was developed.
A huge number of miRNAs have been associated with breast cancer with alternative and not always consistent suppressive or oncogenic properties for each miRNA and neither meta-analyses have been able to define an unequivocal expression and ro le even for more frequently expressed miRNAs (9,26-29). Discrepancies among studies may be attributed to several factors as differences in patient populations, biological samples (fresh tissues, paraffin-embedded tissues and blood) and methodological procedures (qRT- PCR, NGS, microarray) but may also be related to different functions of each miRNA according to the tumor microenvironment (9, 26-29).
In our study we identified 9 miRNAs which were significantly and differently associated with 18-month PFS. Literature data provide evidence about 3 of these miRNAs (let-7c, miR-520d-3p, miR-181a) to be involved in HR+ABC (30-35). In particular miR-520d-3p was among miRNAs which were downregulated upon estradiol stimulation in BCCL (30). On the other hand, miR-520d-3p has also been proposed to suppress ESR1 expression and to be involved in endocrine resistance (31,32).
A larger amount of consistent evidence is available on the let-7 miRNA family which was demonstrated to target ER-a and negatively affect its function in ER-positive BCCL (33). In particular, let-7c targets ESR1 and is less expressed in metastatic tissue than in primary tumor and normal tissue (32). Upon analysis of clinical data from The Cancer Genome Atlas, it was suggested that low expression of let-7c in addition to other miRNAs (miR-99a and miR-125b) was associated with worse overall survival compared with patients who had high expression of these miRNAs (34).
Our findings showing that let-7c-5p was significantly and positively correlated with a higher likelihood of being progression-free at 18 months and let-7f-5p significantly directly correlated with PFS confirmed the suppressive properties for the members of the let-7 family.
In addition, we found a not significantly higher expression of miR-125b in patients achieving 18-month PFS.
Contradictory evidence is available on miR-181a-5p, which belongs to a family of largely expressed miRNAs (29). It was upregulated and associated with poor survival in metastatic breast cancer patients particularly in those with TNBC (29), but, on the other hand, it is among the most potent miRNAs repressing cell growth and counteracting estradiol-dependent cell proliferation (30). In our study a greater expression of miR-181a was positively associated with long-term fulvestrant benefit.
Also on miR-16-5p literature data are somehow inconsistent since preclinical data support tumor suppressive properties but an increased expression in TNBC as compared to normal tissues has been found as well as either a down- or upregulation in the serum of breast cancer patients as compared to healthy controls (33,35-37). In our study miR-16-5p was associated with a decreased probability of being progression-free at 18 months.
In silico analysis of the targets of miRNAs inversely associated with long-term outcome showed that oncogene-induced cell senescence was the most affected biological processes. Analysis of predicted pathways suggested that miR-520d-3p and miR-548g-3p decrease might induce Hippo and Wnt signaling. On the other hand miR- 603 and miR-181a-5p and miR-199a-miR-199b-3p were predicted to suppress endocrine resistance.
The miRNAs we found significantly associated with long-term benefit or resistance to fulvestrant were not among miRNAs differently expressed in fulvestrant-resistant cell lines as compared with parental MCF-7 cells (38,39). Moreover, in our series miR-221 and mir-222 which had been associated with fulvestrant resistance in BCCL were detected only in 2 and 8 samples, respectively and therefore were not further correlated with outcome; at the same time miR-21, despite being confirmed among the most highly expressed miRNAs, did not show any association with outcome (40,41).
We acknowledge that our study has some limitations. First of all, the limited number of patients with available metastatic tissue which resulted in a small sample considering the large number of putative predictive variables examined especially for miRNA analyses; then the retrospective design which resulted in missing clinical information and which may have not allowed to meet with a strict application of RECIST 1.1 criteria for all patients. Moreover, we did not perform in vitro assays in BCCL to better elucidate the correlation between the miRNA profile associated with the in vivo response to fulvestrant and gene expression analyses to identify the related target genes involved. Finally, since the current use of fulvestrant monotherapy is recommended only after CDK 4/6i, we cannot rule out that this treatment can induce molecular patterns different from those observed in our CDK 4/6i naïve population.