Patient characteristics
From May 1, 2016, to November 30, 2019, a total of 186 patients were enrolled in this study (Figure 1). Patients were from 89 cities in 18 provinces in China. The greatest number of patients come from five regions: Beijing (74, 42.5%), Hebei (28, 15.1%), Shandong (15, 8.1%), Inner Mongolia (13, 7.0%), and Heilongjiang (11, 5.9%). The median age was 54 years (range 28–90), and 29 patients (15.6%) were younger than 40 years. Thirty-three patients (17.7%) had de novo metastatic breast cancer. Hundred thirty-eight (74.2%) patients had visceral disease when they started the palbociclib therapy, while 16 (8.6%) patients had bone-only metastasis. Hundred forty-four out of 186 patients (77.4%) showed sensitivity to prior endocrine therapy. Patients received distinct endocrine therapy combined with palbociclib: 96 with fulvestrant, 45 with letrozole, 20 with exemestane, 14 with anastrozole, nine with toremifene, and two with medroxyprogesterone. Only 48 patients (25.8%) received palbociclib as first-line treatment for their metastatic disease, while 82 patients (44.1%) had previously undergone at least three lines of systemic treatment. Thirty-four patients (18.3%) had received everolimus for their metastatic disease before the prescription of palbociclib. The basic clinical and pathological characteristics are detailed in Table 1.
Efficacy profile of palbociclib in the Han population
As of February 2020, the median follow-up time was 6.5 months (range 0.9–40.0), and 88 patients had disease progression. Univariate analysis revealed that patients who were younger than 40 years (P=0.022), those with metastasis other than bone metastasis (P=0.069), and those who had received previous lines of systemic treatment (P<0.001) had significantly worse PFS, while various treatment combinations (P=0.763), number of metastatic sites (P=0.508), and disease stage at diagnosis (P=0.620) did not influence PFS. Multivariate analysis (Table 2) showed that age and previous lines of systemic treatment before palbociclib-based therapy were both independent factors for PFS (P=0.027 and P<0.001, respectively).
The DCR of palbociclib-based therapy varied significantly among different age groups (P=0.034), i.e., the DCR was 60.0% (20/29) in patients younger than 40 years, 77.1% (108/140) in those aged 40–70 years, and 82.3% (14/17) in patients older than 70 years. Accordingly, patients younger than 40 years had a worse PFS (4.0 months, 95% confidence interval [CI] 0–8.5) than patients older than 40 years (8.4 months in those aged 40–70 years, and 6.2 months in those older than 70 years, P=0.189, Figure 2A). The DCR after palbociclib-based treatment was promising in patients with bone-only metastasis (93.7% vs. 74.7%%, P=0.087), and these patients also had significantly prolonged PFS compared with the other patients (8.8 vs. 7.8 months, P=0.023, Figure 2B).
The efficacy of palbociclib-based treatment changed in accordance with increasing numbers of previous lines of systemic treatment (Figure 2C). As the number of previous lines of systemic treatment increased, the DCR gradually decreased (P=0.001), i.e., the DCR was 93.7% (45/48) in the setting of first-line treatment, 80.3% (45/56) in patients with one to two prior regimens, 65.9% (29/44) in patients with three to four prior regimens, and 60.5% (23/38) in patients who had been treated with more than five regimens. The median PFS was also longer in patients who had received palbociclib-based therapy as first-line treatment (14.0 months, 95% CI 11.4–16.6) than in those who received subsequent lines of treatment (P<0.001), i.e., 10.0 months (95% CI 7.1–12.9) in the second/third-line group, 6.2 months (95% CI 3.0–9.5) in the fourth/fifth-line group, and 3.4 months (95% CI 0.8–6.1) in patients who had received more than five lines of systemic therapy.
The DCR did not differ significantly among patients who received diverse palbociclib-based treatment (P=0.403), i.e., the DCR was 86.7% (39/45) in the letrozole-combined group, 74.0% (71/96) in the fulvestrant-combined group, 71.4% (10/14) in the anastrozole-combined group, 70.0% (14/20) in the exemestane-combined group, and 66.7% in the toremifene-combined group (6/9). The median PFS was also similar among different palbociclib-combined groups (P=0.566, Figure 2D), i.e., 10.0 months (95% CI 3.8–16.1) in the exemestane plus palbociclib group, 9.7 months (95% CI 6.3–13.1) in the letrozole plus palbociclib group, 7.8 months (95% CI 5.5–10.2) in the fulvestrant plus palbociclib group, 7.2 months (95% CI 3.2–11.3) in the toremifene plus palbociclib group, and 6.1 months (95% CI 1.2–11.0) in the anastrozole plus palbociclib group. One of the two patients receiving medroxyprogesterone plus palbociclib had stable disease for more than 12 months; however, palbociclib treatment was interrupted for financial reasons. The other patient experienced disease progression after 2.5 months of treatment with medroxyprogesterone plus palbociclib.
Efficacy in patients with prior everolimus treatment
Regarding the previous treatment before palbociclib-based therapy, everolimus was mostly combined with exemestane (15/34, 44.1%), followed by fulvestrant (7/34, 20.1%), and toremifene (6/34, 17.6%).
The DCR was significantly lower in patients who had received previous everolimus (50.0%, 17/34) than in the everolimus-naïve group (82.2%, 125/152, P<0.001). Consistently, the Kaplan-Meier estimates indicated that patients pretreated with everolimus had significantly worse PFS (3.4 months, 0.7–6.1) than patients in the everolimus-naïve group (8.8 months, 95% CI 6.6–11.0, P=0.001, Figure 3A). Further, propensity score matching was used to match patients with or without prior everolimus treatment. Propensity score matching resulted in 30 patients in the previous-everolimus group and 30 patients in the everolimus-naïve group (Table 1). After propensity score matching, no significant differences in the clinical characteristics were observed between the two groups. Palbociclib-based therapy resulted in a worse DCR of 53.3% (16/30) in the previous-everolimus group compared with 76.7% (23/30) in the everolimus-naïve group (P=0.058). Consistently, patients pretreated with everolimus had inferior PFS (4.4 months, 95% CI 0.5–8.2) than everolimus-naïve patients (6.1 months, 95% CI 4.7–7.5, P=0.439, Figure 3B).