Circulating tumor DNA as a dynamic biomarker of response to palbociclib and fulvestrant in metastatic breast cancer patients

Abstract

Background: Following the PALOMA-3 study results, the combination of palbociclib, a CDK4/6 inhibitor, with fulvestrant, a selective estrogen receptor degrader, has become a standard therapy in women with estrogen receptor-positive (ER+) HER2-negative (HER2-) metastatic breast cancer (MBC). Palbociclib has been shown to increase the progression-free survival (PFS) overall but no predictive biomarker of palbociclib efficacy has been validated so far. We thus evaluated whether early changes of circulating tumor DNA (ctDNA) levels are associated with palbociclib plus fulvestrant efficiency.

Methods: ER+ HER2- MBC patients were included in a prospective observational cohort before treatment initiation. Tumor response was assessed by radiological evaluation (RECIST v1.1) every 3 months. Plasma samples were collected before treatment (baseline), at day 15 (D15), at day 30 (D30), and at disease progression. We searched for somatic mutations from archived tumor tissues by targeted deep sequencing. For patients with somatic mutations identified, circulating tumor DNA (ctDNA) was tracked using digital droplet PCR. Ratios of ctDNA levels ([D15/baseline] and [D30/baseline]) were then correlated with prospectively registered patient characteristics and outcomes.

Results: Twenty-five of the 61 patients enrolled had a somatic mutation testable in plasma (N_PIK3CA = 21, N_TP53 = 2, N_AKT1 = 2). At baseline, 84% of patients had detectable ctDNA levels but ctDNA levels had no prognostic impact on PFS (p = 0.10). Among those patients, ctDNA was still detected in 82% at D15 and 68% at D30. ctDNA clearance observed at day 30 was associated with longer PFS (HR = 7.2, 95% CI = 1.5-32.6, p = 0.004). On the contrary, a [D30/baseline] ctDNA ratio > 1 was associated with a shorter PFS (HR = 5.1, 95% CI = 1.4-18.3, p = 0.02) and all 5 patients with increased ctDNA levels at D30 showed disease progression after 3 months under palbociclib-fulvestrant. Finally, at the time of radiological tumor progression, ctDNA was detected in all patients tested.

Conclusion: Our study demonstrates that the efficiency of palbociclib and fulvestrant can be monitored by serial analyses of ctDNA before radiological evaluation and that early ctDNA variation is a prognostic factor of PFS.

Trial registration: ClinicalTrials.gov NCT02866149.

Keywords: Breast cancer; Circulating tumor DNA; Palbociclib-fulvestrant; Precision medicine; Treatment follow-up.

Fig. 1

Identification of trackable somatic mutations. Study flow chart. N = number of patients

Fig. 2

ctDNA detection at baseline. cfcDNA and ctDNA concentrations, prior palbociclib-fulvestrant treatment initiation, are displayed in copies per ml of plasma extracted. Patients are ranked by decreasing ctDNA concentrations. Gray bars represent cfcDNA concentrations; pink bars represent ctDNA concentrations

Fig. 3

Early ctDNA dynamics during palbociclib-fulvestrant treatment and impact on survival. cfcDNA (a) or ctDNA (b) quantifications in copies per ml of plasma at 4 different time points during therapy. Black lines refer to median values. cfcDNA (c) or ctDNA (d) variations along 4 sampling time points during therapy. Pink triangles highlight samples with increased ctDNA level compared to the previous assessable time point. Wilcoxon tests were used to compare DNA levels. Bsl, baseline; D15, day 15; D30, day 30; Progr., disease progression. Progression free survival of patients split by ctDNA detected or not, at day 15 (e) and day 30 (f)

Fig. 4

ctDNA dynamics correlation with treatment response. a ctDNA concentrations at the 4 sampling time points in patients with non-progressive disease (green) versus patients with progressive disease (red) after 3 months of palbociclib-fulvestrant. Black lines refer to median values. ctDNA variation in patients with non-progressive disease (b) or progressive disease (c) during treatment follow-up. Each patient is highlighted with a unique color. Bsl, baseline; D15, day 15; D30, day 30; Progr., disease progression; PD, progressive disease

Fig. 5

ctDNA ratios, treatment response, and impact on survival. ctDNA ratios for patients with non-progressive disease (green) vs progressive disease (red) after 15 days (a) or 30 days (b) of treatment. P values were calculated using Mann–Whitney tests. Median is highlighted by a solid black line. Patients highlighted with a star display increased ctDNA levels, which predict progression at 3 months. c–e Progression free survival of patients under palbociclib-fulvestrant therapy split by median ctDNA ratio at D15 (c, median D15 = 0.16), median ctDNA ratio at D30 (d, median D30 = 0.09), ctDNA ratio > 1 or ≤ 1 at D30 (e)