Clinical management of metastatic hormone receptor-positive, HER2-negative breast cancer (MBC) after CDK 4/6 inhibitors: a retrospective single-institution study

Cyclin-dependent kinase 4/6 inhibitors (CDK4/6is), in combination with endocrine therapy (ET), are standard either in the first (1L) or second-line (2L) setting for the treatment of hormone receptor (HR) positive, HER2-negative metastatic breast cancer (MBC). However, the optimal sequencing of treatments after progression on CDK4/6i remains unknown. We performed a single-institution analysis to identify treatments and outcomes after progression on a CDK4/6i. We identified patients with HR-positive, HER2-negative MBC prescribed a CDK4/6i in the 1L or 2L settings from December 2014 to February 2018 at Mayo Clinic in Rochester, Minnesota. Outcomes were collected through September 30, 2020. Palbociclib, in combination with letrozole or fulvestrant, was the most prescribed CDK4/6i. The 1L and 2L CDK4/6i cohorts exhibited comparable overall survival (OS), but progression-free survival (PFS) was longer in the 1L than the 2L cohort [28.2 months (95% CI 19.6–34.9) vs 19.8 months (95% CI 15.7–29.6)]. The most common post-CDK4/6i treatments were PI3K/mTOR inhibitors (PI3K/mTORi), single-agent ET, or chemotherapy. PFS in the 1L CDK4/6i cohort following PI3K/mTORi was 8.5 months (95% CI 5.5 months—NE), single-agent ET was 6.0 months (95% CI 3.3–14.0 months), and chemotherapy PFS was 5.4 months (95% CI 3.3 months—NE). Following progression on a CDK 4/6i, mPFS was short, with similar PFS times comparing chemotherapy and ET, with slightly longer PFS for targeted strategies (PI3K/mTOR). These results highlight a major need to better understand the mechanisms of CDK4/6i resistance and identify new therapeutic strategies for these patients.


Introduction
Approximately 60-70% of the 150,000 women currently living with metastatic breast cancer (MBC) in the United States have tumors that are hormone receptor 1 3 (HR)-positive and HER2-negative [1][2][3]. While endocrine therapy (ET) remains an essential therapeutic option for this population [4], intrinsic or acquired resistance inevitably emerges. Preventing and reversing resistance to ET necessitates the use of effective and optimally sequenced therapies.
While CDK4/6i have become standard of care as 1L or 2L treatment of HR-positive MBC, the optimal treatment strategy after progression on a CDK4/6i remains uncertain. To determine the prognosis of patients following progression on CDK4/6i based on the type of post-CDK4/6i therapy used, we performed a single-institution retrospective review of patients with HR-positive MBC receiving 1L or 2L CDK4/6i with the goal of describing the prescribing patterns and clinical responses following progression on CDK4/6i-based treatment.

Patient selection
As part of an Institutional Review Board approved protocol, we utilized the electronic medical record to identify individuals with MBC prescribed palbociclib, ribociclib, or abemaciclib by a medical oncologist in Mayo Clinic (Rochester, MN) between December 2014 and February 2018. We utilized ICD-10 codes and record review to focus on those prescribed 1L or 2L CDK4/6i who received at least 30 days of therapy and attended more than two clinic visits at our institution during treatment. In-class switching of ET (i.e., changing from one aromatase inhibitor to another for toxicity) was not considered a new line of therapy. If a patient switched to a different CDK4/6i (e.g., palbociclib to abemaciclib) or a different class of ET (e.g., letrozole to fulvestrant while remaining on the same CDK4/6i), this was considered a subsequent line of therapy.

Clinical outcomes
Endocrine therapy resistance was classified as primary and secondary, per the 4th ESO-ESMO International Consensus Guidelines for Advanced Breast Cancer [17]. Based on this definition, our 1L cohort was classified as primary ET resistance when metastatic relapse occurred within the first two years of adjuvant ET. Secondary ET resistance was defined as (a) relapse while on adjuvant ET, but after the first 2 years, (b) relapse within 12 months of completing adjuvant ET. In the 2L cohort, primary ET resistance was defined as progression within the first six months of ET for MBC.
Patient data were abstracted from December 1, 2014 through September 30, 2020, and three authors (GMC, KVG, SL) participated in data abstraction. We separately evaluated 1L and 2L cohorts to determine the PFS during and after CDK4/6i therapy. We defined PFS as the time from the date of initiation of CDK4/6i to the date of disease progression that was determined clinically, radiographically, or pathologically. Patients who stopped treatment due to toxicity prior to disease progression were censored at time of therapy discontinuation. We measured OS from the start of CDK4/6i-based therapy until time of death or last followup where patients were censored if they were alive or lost to follow-up.

Statistical analyses
Continuous variables were summarized as median and standard deviation (SD) reported with 95% confidence intervals (CI) or 25th/75th percentile (Q1, Q3). In subgroup analyses of the 2L and third-line (3L) setting, samples were too small to calculate an accurate 95% CI and were subsequently denoted as not evaluable (NE), where applicable. All statistical analyses, including Kaplan-Meier survival curves were performed using BlueSky statistics (BlueSky Statistics LLC, Chicago, IL USA).

Patient characteristics
We identified 136 patients who started a CDK4/6i as 1L (n = 91) or 2L (n = 45) therapy for HR-positive MBC. A total 37 patients (27.2%) from both cohorts had a diagnosis of de novo metastatic disease. Of the 99 patients who relapsed following an initial diagnosis of early-stage disease, all underwent previous surgical resection, 21 patients (21.2%) received neoadjuvant chemotherapy, and 52 patients (52.5%) received adjuvant chemotherapy. In patients with relapsed disease, 90 patients (90.9%) had received adjuvant ET with a median treatment duration of 45 months (Q1-Q3 26-60 months) in the 1L cohort and 60 months (Q1-Q3 27.5-60 months) in the 2L cohort.
The most common 2L treatment options after stopping 1L CDK4/6i was PI3K/mTOR inhibitors (PI3K/mTORi) combined with ET (n = 22, 33.3%), single-agent ET (n = 22, 33.3%), and single-agent chemotherapy (n = 11, 16.7%) Table 1 Patient characteristics in first-line (1L) and second-line (2L) CDK4/6i cohorts a One patient underwent lumpectomy for bone-only metastatic disease after near complete response on ET and another patient underwent mastectomy and found to have metastatic disease shortly after surgery b ET resistance was higher in the 2L CDK4/6i cohort due to the majority of patients having previously been on ET as 1L treatment for metastatic breast cancer  (Fig. 1C).

Discussion
The use of CDK4/6i in combination with ET as 1L treatment for the management of HR-positive HER2-negative MBC has now become standard of care [4]. However, optimal treatment strategies following progression on CDK4/6i are not well defined. Current options include single-agent ET [4], drugs targeting the PI3K/mTOR pathway such as alpelisib + ET for patients with somatic PIK3CA mutations [18] and everolimus + exemestane [19] or chemotherapy [4].
Our study was a single-institution retrospective cohort study designed to investigate the types of therapies being used in the post-CDK 4/6i setting, as well as the clinical outcomes for patients who progress after 1L or 2L CDK4/6i. Though our study was limited by small sample sizes in subgroup analysis where statistical significance could not be determined for clinical outcomes, we observed that 1L CDK4/6i cohort had improved PFS and OS outcomes compared to 2L CDK4/6i cohort. We reassuringly found that median PFS times in our analysis were quite similar in the 1L setting [ Following progression on a CDK4/6i, we found that PI3K/mTORi in combination with ET were the most commonly prescribed 2L or 3L option, followed by single-agent ET and chemotherapy. These data somewhat contrast to a recent large population-based study looking at administrative claims for CDK4/6i-based therapy for treatment of HR-positive HER2-negative MBC, demonstrating that single-agent ET and chemotherapy were the most commonly prescribed regimens following progression on a CDK4/6i with repeat CDK4/6i or everolimus-based regimens less commonly prescribed [20]. This discrepancy is likely due to differing prescribing practices at Mayo Clinic Rochester compared to the community as a large tertiary care center. Overall, we found that the PFS times following progression on a CDK4/6i were short, and ranged from 2.0 to 14.0 months, with shorter PFS times for patients that progressed after receipt of a 2L CDK4/6i. These data did not demonstrate an obvious advantage for one specific post-CDK4/6i-based regimen over another; however, the retrospective nature of this cohort and small sample sizes limits our ability to make comparisons across regimens.
Currently, there are several clinical trials that have evaluated the single-agent antitumor activity of ET after CDK4/6i progression. The Veronica trial was a phase II randomized trial investigating the BCL-2 inhibitor venetoclax combined with fulvestrant compared to fulvestrant monotherapy as 2L or 3L treatment after CDK4/6i progression [21]. There was no statistically significant improvement with the addition of venetoclax, but this study highlighted that single-agent fulvestrant resulted in poor PFS benefit (approximately 1.94 months) after CDK4/6i. Recent data from the EMER-ALD trial, a phase III randomized control trial of elacestrant, an oral selective estrogen receptor degrader (SERD), compared to fulvestrant or an aromatase inhibitor in patients who had previously progressed on CDK4/6i + ET noted an identical PFS benefit with fulvestrant alone of 1.94 months [22], similar to the Veronica trial.
It has been repeatedly observed that some of the mechanisms that drive resistance to CDK4/6i include mutations in ESR1 [23] and RB1 [24], amplification of CDK4 or CDK6 [25], and Cyclin E [26] and decreased ER expression [27,28]. Of these mechanisms, the mTOR/AKT/PI3K pathway may contribute to resistance to CDK4/6i and therefore may be a reasonable target after progression on a CDK4/6i [27,[29][30][31][32]. In a retrospective cohort study of 23 patients with MBC, those previously treated with prior everolimus in the metastatic setting were significantly less likely to receive a clinical benefit with subsequent treatment with palbociclib + letrozole with a median PFS of 4.2 months [33]. In contrast, Cook et al. [34] demonstrated that patients who received everolimus + exemestane after CDK4/6i or without prior CDK4/6i had similar PFS (3.6 vs 4.2 months) and OS (15.6 vs 11.3 months), inferring that prior use of CDK4/6i does not prevent future benefit to everolimus + exemestane. This is further supported by a larger retrospective study of 622 patients who had received everolimus + exemestane for treatment of HR-positive MBC, wherein 54 patients had received everolimus + exemestane as 2L therapy after progression on a CDK4/6i with median time-to-next treatment being 5.5 months vs 8.3 months in patients who received 1L everolimus + exemestane [35]. However, OS was longer in the cohort of patients who had received CDK4/6i than those who had received 1L ET (OS 59.2 months vs 40.8 months p < 0.01). Additionally, cohort A of the BYLieve trial investigating alpelisib + fulvestrant in patients with PIK3CA mutations who had progressed on CDK4/6i + an aromatase inhibitor reported a median PFS 7.3 months (95% CI 5.6-8.3 months) and median OS was 17.3 months (95% CI 17.2-20.7), with 14% discontinuing treatment due to adverse events [36]. Our data further support the use of PI3K/mTOR inhibition after progression on CDK4/6is, given similar PFS time compared to the BYLieve results. These data provide further support for the use of drugs that target the PI3K/mTOR pathway following progression on CDK4/6i and suggest that further prospective studies should evaluate the antitumor activity of everolimus containing regimens post-CDK4/6i.
There are several limitations to our study including its retrospective design, small sample size, and possible differences in restaging frequency depending on the clinical scenario, which may alter PFS times. Further, we were unable to stratify patients in the PI3K/mTORi cohort based on activating PIK3CA mutations, as there were too few patients to allow for adequate subgroup analysis for those treated with alpelisib + fulvestrant. Additionally, our population primarily received palbociclib-based regimens, which is reflective of prescribing practices at the time at Mayo Clinic Rochester and there are few data regarding the benefit of treatment strategies following abemaciclib or ribociclib. Lastly, the true clinical benefit of PI3K/mTOR is in the 2L setting is likely limited by associated toxicities (e.g., hyperglycemia, diarrhea) and poor tolerance, as demonstrated in our population where just over 20% of patients discontinued due to side effects, suggesting that better management of these toxicities is needed to improve clinical outcomes with these drugs.
In conclusion, our study is one of the largest to date evaluating the prognosis of patients following progression on CDK4/6is. Our data confirm that despite the choice of therapy, the PFS times are modest and generally below 6 months in both the 1L and 2L settings. These sobering data clearly illustrate the need for a better understanding of the mechanisms that drive resistance to CDK4/6i as well as for new targeted therapies. Our findings demonstrating numerically longer PFS times for patients receiving PI3K/mTORi are consistent with the prospective data examining PI3K inhibitors in this setting. Prospective studies will help determine the optimal treatment strategy after CDK4/6i failure, several of which are currently ongoing [30,[37][38][39].