Alpelisib and fulvestrant in PIK3CA-mutated hormone receptor-positive HER2-negative advanced breast cancer included in the French early access program

SOLAR-1 and BYLieve trials documented the efficacy of the PI3K-inhibitor alpelisib in pre-treated PIK3CA-mutant, hormone receptor-positive, HER2-negative (HR+/HER2-) advanced breast cancer (ABC) patients. We report here real-life data of patients prospectively registered in the French alpelisib early access program (EAP) opened to PIK3CA-mutant HR+/HER2- ABC patients treated with alpelisib and fulvestrant. Primary endpoint was PFS by local investigators using RECIST1.1. Eleven centers provided individual data on 233 consecutive patients. Patients had received a median number of 4 (range: 1–16) prior systemic treatments for ABC, including CDK4/6 inhibitor, chemotherapy, fulvestrant and everolimus in 227 (97.4%), 180 (77.3%), 175 (75.1%) and 131 (56.2%) patients, respectively. After a median follow-up of 7.1 months and 168 events, median PFS was 5.3 months (95% CI: 4.7–6.0). Among 186 evaluable patients, CBR at 6 months was 45.3% (95% CI: 37.8–52.8). In multivariable analysis, characteristics significantly associated with a shorter PFS were age < 60 years (HR = 1.5, 95% CI = 1.1–2.1), >5 lines of prior treatments (HR = 1.4, 95% CI = 1.0–2.0) and the C420R PI3KCA mutation (HR = 4.1, 95% CI = 1.3–13.6). N = 91 (39.1%) patients discontinued alpelisib due to adverse events. To our knowledge, this is the largest real-life assessment of alpelisib efficacy. Despite heavy pre-treatments, patients derived a clinically relevant benefit from alpelisib and fulvestrant.

Alpelisib is an orally bioavailable, α-selective PI3K inhibitor and degrader [5]. SOLAR-1, a phase 3 trial, investigated the efficacy of fulvestrant associated with alpelisib versus placebo [6] in endocrine-resistant HR-positive, HER2-negative ABC. In the cohort of patients with PIK3CA-mutated ABC, progressionfree survival (PFS) was significantly improved in the alpelisibfulvestrant group, as compared with the placebo-fulvestrant group (with 11 and 5.7 months median PFS, respectively). Despite a particular toxicity profile, adverse events were manageable. Of note, only few (5.9%) patients in SOLAR-1 received a prior CDK4/6 inhibitor [7]. The median PFS achieved by fulvestrant-alpelisib after a first line therapy with CDK4/6 inhibitor was later documented by the BYLieve phase 2 trial: the 127 patients pretreated with an aromatase inhibitor and a CDK4/6 inhibitor (cohort A) experienced a median PFS of 7.3 months on alpelisib and fulvestrant [8].
Based on the results of SOLAR-1, FDA approved alpelisibfulvestrant for the treatment of post-menopausal patients with PIK3CA-mutated HR+/HER2− endocrine resistant ABC. In France, a temporary early access program (EAP) was opened on November 26th, 2018 to provide access to alpelisibfulvestrant in PIK3CA-mutated HR+/HER2− ABC patients pretreated with at least two systemic treatments including an aromatase inhibitor and a CDK4/6 inhibitor. Following the decision of the European Medicine Agency to restrict the European label of alpelisib to patients who did not receive a prior CDK4/6 inhibitor, the French EAP was closed to new patients in 10.2020. The purpose of this study is to evaluate the efficacy of alpelisib-fulvestrant through the French EAP, in a population that received treatments corresponding to the current standard therapeutic landscape.

Patients' and tumor characteristics
Out of a total number of 365 patients included in the French EAP (according to the ANSM public report of June 2, 2020) (9), 233 (63.8%) were included in the 11 participating centers that met the study eligibility criteria. Baseline demographic and clinicopathological characteristics of patients are shown in Table 1 Table 2 displays the distribution of PI3KCA mutations, the most common being H1047R (38.6%), E545K (14.5%) and E542K (14.1%).

Safety
Alpelisib dose reduction occurred in 91 (39.1%) patients. Permanent discontinuation because of treatment toxicity occurred in 84 (36.9%) patients. Most frequent grade 3 or 4 toxicities observed were hyperglycemia in 27 patients (11.6%), rash in 23 (9.9%) patients, fatigue in 10 (4.3%) and diarrhea in 7 (3%) ( Table 3). No toxic death was observed. The management of these toxicities was based on the administration of oral antidiabetics and/or insulin for hyperglycemia, antihistamines and corticosteroids for rash and transit retardants for diarrhea.

DISCUSSION
EAPs offer ethical, compliant, and controlled mechanisms of access to investigational drugs outside of the clinical trial space and before the commercial launch of the drug [9]. The prospective registration of participating patients and the tracability of alpelisib delivery by hospital pharmacies was considered as a unique opportunity to report real-life alpelisib data. Our retrospective study gathered the individual data of almost two-thirds of patients treated as part of the French EAP nationwide and is, to our knowledge, the most significant population-based study on alpelisib and fulvestrant in a post-CDK4/6 inhibitor setting.
In our multicentric real-life cohort of N = 233 patients, the median PFS obtained under alpelisib and fulvestrant (5.3 months) must be weighed against clinical trial results obtained with either fulvestrant and alpelisib combination or single agent fulvestrant, in a post-CDK4/6 inhibitor setting. On the one hand, the median PFS in our study is numerically shorter than that observed in BYLieve trial cohort A (7.3 months, N = 127 patients [8]). We posit that this difference is attributable by a much fitter and less heavily pretreated population in BYLieve: 62% of patients in BYLieve had a performans status (PS) = 0, vs 36% in our population; the median number of prior lines of therapy was one in BYLieve, vs four in our study. In the primary alpelisib pivotal trial, SOLAR-1, the N = 20 patients who received a prior CDK4/6 inhibitor displayed a similar median PFS of 5.5 months under alpelisib and fulvestrant [10]. On the other hand, in the recently reported EMERALD trial a median PFS of 1.9 months was observed in N = 165 HR+/HER2− ABC patients treated with fulvestrant single agent as second-or thirdline therapy after CDK4/6 inhibitor [11]. In addition to these median PFS data, response rates and 6mCBR reported in our study suggests that alpelisib and fulvestrant is an effective therapy for real-life endocrine treatment-resistant PIK3CA-mutant ER+/HER2− ABC patients.
An exploratory analysis of clinicopathological characteristics associated with shorter PFS on alpelisib and fulvestrant retrieved that the most heavily pre-treated patients had shorter PFS, suggesting that the full benefit of alpelisib could be obtained in the earliest lines of therapy for ABC. Patients younger than 60 years of age also experienced significantly shorter median PFS, although we found no difference in baseline characteristics with older patients (data not shown). Lastly, we observed that the 4 patients displaying a C420R PIK3CA mutation had a worse outcome. While the small number of patients precludes any definitive conclusion, mutations located in exon 4 and that impact the C2 domain (C420R and N345K) have been previously reported as associated with shorter OS in the METABRIC dataset (12). In vitro screening experiments also showed that the C420R PIK3CAmutant EFM192A and JIMT-1 cell lines display limited sensitivity to alpelisib (14). Of note, most centers participating to the EAP did not sequence PIK3CA exon 4, explaining the underrepresentation of exon 4 mutations in our cohort: 1.7% (4/233 patients), while C420R and N3345K account for 7.5% of all activating PIK3CA mutations (15). Further research is therefore required to explore the potential impact of these exon 4 mutations on alpelisib efficacy.
As part of the EAP, clinicians were requested to document and report treatment-related adverse events in a prospective manner. Interestingly, the frequency of any grade hyperglycemia (53%) appears very similar to that observed in SOLAR-1 (64%) and BYLIEVE (58%). Similarly, the proportion grade 3 or 4 rashes was consistent with that observed in BYLieve [8]. This may be explained by the application of prophylactic antihistamines recommended for the first 8 weeks of alpelisib treatment [12]. Our real-life study however reports a numerically higher rate of alpelisib discontinuation (37%) than in trials (SOLAR-1: 25%; BYLieve: 21%), which could be explained by the different clinical profile of treated patients and less stringent monitoring. Overall, our data suggest that alpelisib toxicity management seems feasible in real life, notwithstanding the fact that our study was conducted in expert cancer centers, which may have participated to prior alpelisib trials. Yet, regarding the lack of documented significant OS and/or quality of life benefit, as illustrated by the score of 3 on ESMO-magnitude of clinical benefit scale (MCBS), such a tolerance profile makes the actual place of alpelisib in the management of ABC still discussable.
Limitations of this study stem from its retrospective nature, preventing any direct comparison with other studies and trials. Nevertheless, a strength of the EAP is that patients were prospectively registered and no patient was lost to follow-up, allowing robust outcomes analyses. This cohort is, to our knowledge, the largest real-life dataset reported so far, and confirms the efficacy and manageability of alpelisib and fulvestrant in PIK3CA-mutant HR+/HER2− ABC patients.

MATERIALS AND METHODS Patients and treatment
The French EAP was accessible to all patients and centers in France. To be eligible, the following criteria were mandatory: women aged 18 years and more; treated for a histologically proven HR + HER2-ABC (HR-positivity being defined by ≥10% tumor cells positive for either estrogen and/or progesterone receptor using immunohistochemistry); having received at least 2 prior lines of systemic therapy, including an aromatase inhibitor and a CDK4/6 inhibitor; activating PI3KCA mutation, per local sequencing of the primary and/or metastatic tumors and/or liquid biopsy. Patients with visceral crisis or inflammatory breast cancer were excluded. Patients had to have a fasting plasma glucose below 140 mg/dl (7.7 mmol/l) and glycated hemoglobin A 1c below 6.4%. All patients treated as part of the EAP were prospectively registered at their site in a national registry. Alpelisib was delivered to patients by their hospital pharmacy every month, at a starting dose of 300 mg daily. Fulvestrant was used at standard doses (500 mg intramuscularly on day 1 of each 28-day cycle and on day 15 of cycle 1). Luteinizing hormone-releasing hormone (LH-RH) agonists were allowed per standard of care and investigator's opinion. Treatment and dose reductions were managed by oncologists, following drug labels.

Data collection
This study was approved by the Institut Curie review board; a waiver of informed consent was granted due to the retrospective nature of the work. A synopsis of this retrospective study was sent to French cancer centers, known for having significantly contributed to the alpelisib EAP. Eleven of these centers agreed to participate in the study: Institut Curie Paris and Saint Cloud Hospitals, Institut Paoli Calmettes, Center Eugene Marquis, Institut de Cancérologie de l'Ouest, Gustave Roussy, Center Antoine Lacassagne, Tenon Hospital, Brest University Hospital Center, Center Georges-François Leclerc, American Hospital of Paris, Institut Sainte-Catherine. Medical oncologists retrospectively collected data using electronic medical records for all patients enrolled in the EAP at their sites. Participating sites had to grade toxicities according to Common Terminology Criteria for Adverse Events version 5.0 (CTCAE v5.0), whereas tumor assessment by imaging had to be reported using Response Evaluation Criteria in Solid Tumor version 1.1 (RECIST v1.1) [13]. Data cutoff was February 28th 2021. Pseudonymized individual data were further manually reviewed for quality and coherence at Institut Curie, and queries were issued whenever needed.

Endpoints and statistics
The primary endpoint was PFS. Secondary endpoints were overall survival (OS), objective response rate (ORR), 6 months clinical benefit rate (6mCBR) defined as the percentage of patients who have achieved a complete or partial response and/or experienced a stable disease for at least 6 months, and safety. The following prognostic factors for PFS were explored: median age at alpelisib initiation, performance status, histological subtypes, hormone receptors expression, disease stage at diagnosis, presence of visceral disease, number of metastatic sites, number and type of previous systemic treatments administered for metastatic disease and type of PIK3CA mutation-with no correction for multiple testing.
Descriptive statistics were used to summarize patients' characteristics. Progression-free survival was defined as the time from treatment initiation to disease progression or death, whichever came first. Patients discontinuing both alpelisib and fulvestrant for other reason than a PFS event were censored at time of discontinuation. Survival curves for PFS, median PFS and its 95% confidence interval (95% CI) were generated using the Kaplan-Meier method. Multivariate Cox proportional hazards models were constructed on the general population using a backward step-by-step manual selection procedure to identify independent prognosis factors. All factors significant at a conservative 10% level in univariate analysis were included in multivariate analysis. The final model was reached when including only factors at a p = 0.05 significance level. All analyses were performed using R version 3.3.2. Statistical significance was defined by a two-tailed p < 0.05.