A phase II trial of an alternative schedule of palbociclib and embedded serum TK1 analysis

Palbociclib 3-weeks-on/1-week-off, combined with hormonal therapy, is approved for hormone receptor positive (HR+)/HER2-negative (HER2−) advanced/metastatic breast cancer (MBC). Neutropenia is the most frequent adverse event (AE). We aim to determine whether an alternative 5-days-on/2-days-off weekly schedule reduces grade 3 and above neutropenia (G3 + ANC) incidence. In this single-arm phase II trial, patients with HR+/HER2− MBC received palbociclib 125 mg, 5-days-on/2-days-off, plus letrozole or fulvestrant per physician, on a 28-day cycle (C), as their first- or second-line treatment. The primary endpoint was G3 + ANC in the first 29 days (C1). Secondary endpoints included AEs, efficacy, and serum thymidine kinase 1 (sTK1) activity. At data-cutoff, fifty-four patients received a median of 13 cycles (range 2.6–43.5). The rate of G3 + ANC was 21.3% (95% CI: 11.2–36.1%) without G4 in C1, and 40.7% (95% CI: 27.9–54.9%), including 38.9% G3 and 1.8% G4, in all cycles. The clinical benefit rate was 80.4% (95% CI: 66.5–89.7%). The median progression-free survival (mPFS) (95% CI) was 19.75 (12.11–34.89), 33.5 (17.25–not reached [NR]), and 11.96 (10.43–NR) months, in the overall, endocrine sensitive or resistant population, respectively. High sTK1 at baseline, C1 day 15 (C1D15), and C2D1 were independently prognostic for shorter PFS (p = 9.91 × 10−4, 0.001, 0.007, respectively). sTK1 decreased on C1D15 (p = 4.03 × 10−7), indicating target inhibition. Rise in sTK1 predicted progression, with the median lead time of 59.5 (inter-quartile range: −206.25–0) days. Palbociclib, 5-days-on/2-days-off weekly, met its primary endpoint with reduced G3 + ANC, without compromising efficacy. sTK1 is prognostic and shows promise in monitoring the palbociclib response. ClinicalTrials.gov#: NCT3007979.


INTRODUCTION
Hormone receptor positive (HR+) and human epidermal growth factor receptor 2 negative (HER2−) breast cancer accounts for 70% of breast cancer diagnoses and is a leading cause of cancer death in women 1,2 . The discovery of cyclin D/cyclin-dependent kinase 4/6 (CDK4/6) as a key downstream target of estrogen receptor (ER) and endocrine resistance mechanisms has led to the development of CDK4/6 inhibitors for the treatment of HR+/HER2− breast cancer 3,4 . CDK4/6 inhibitors, including palbociclib, ribociclib, and abemaciclib, have gained FDA approval based on the significant improvement in progression-free survival (PFS) when added to endocrine therapy in patients with advanced disease as front-line therapy [5][6][7][8][9][10][11] or following disease progression on prior endocrine therapy [12][13][14] . Overall survival (OS) benefit has also been observed in several studies [15][16][17] . These agents have now become the standard of care in combination with an endocrine therapy partner for HR+/HER2− metastatic breast cancer (MBC). However, treatment-related neutropenia is a common adverse event (AE), leading to dose interruption, reduction, and at times discontinuation. Palbociclib (Ibrance, Pfizer) is the first CDK4/6 inhibitor approved in combination with an aromatase inhibitor as first-line therapy and with fulvestrant following prior hormonal therapy based on results from PALOMA trials 5,9,10,13 . Neutropenia was the most frequent high grade (G) AE related to palbociclib in PALOMA-2 (79.5% all grades, 56.1% G3, 10.4% G4) and PALOMA-3 (78.8% all grades, 53.3% G3, 8.7% G4) 9,13 . Febrile neutropenia occurred in 1.8%, while dose reduction occurred in over a third of patients who received palbociclib across PALOMA-2 and PALOMA-3 9,13,18 . In addition, with the half-life of palbociclib being~27 h, recovery of Rb phosphorylation and cell proliferation during the off-treatment week is a concern. In a longitudinal biomarker study of palbociclib plus letrozole, recovery of tissue Rb phosphorylation and Ki67 levels to baseline was observed on day 3 and day 4-5 during the off week, respectively 19 . We, therefore, proposed a 5days-on/2-days-off weekly schedule to allow bone marrow recovery during the 2 off days to avoid the week-long break. We hypothesize that this alternative schedule is more tolerable with less frequent high-grade (G3+) neutropenia and dose interruptions/reductions compared to the historical data with the 3-weeks-on/1-week-off schedule. Based on our previous study supporting serum thymidine kinase 1 (sTK1), an E2F-dependent enzyme critical for DNA synthesis, as a pharmacodynamic indicator of CDK4/6 inhibition 20 , we assessed sTK1 dynamics for target inhibition, and examined its potential prognostic value and utility in disease monitoring in this study.  Fig. 1).

DISCUSSION
CDK 4/6 inhibitors represent a major advance in the treatment of HR+/HER2− MBC. However, a one-week break following 3 weeks of administration is required for both palbociclib and ribociclib due to treatment-induced neutropenia 22,23 . To improve the tolerability and to avoid the 1-week break, which has been shown to led to the recovery of target inhibition in previous studies 19  improved tolerability of palbociclib administered in a 5-days-on/2days-off weekly schedule compared to historical data from palbociclib trials 9,13,18 .
To assess target inhibition, we examined serial sTK1 activity as a surrogate pharmacodynamics marker in this trial, and demonstrated a significant reduction of sTK1 to an undetectable level in 78.2% of patients at C1D15. This result mirrors the impressive reduction in sTK1 observed 2 weeks after initiation of palbociclib in our prior study of patients with early stage HR+/HER2− breast cancer receiving neoadjuvant palbociclib and anastrozole 20 . The significant reduction of sTK1 at C1D15 and C2D1 indicates that the 5-days-on/2-days-off schedule achieved effective target inhibition and provides reassurance with the 2-day break each week.
This alternative schedule offers an advantage in avoiding the 1-week break with a 3-weeks-on/1-week-off schedule, which potentially compromises the efficacy of CDK4/6 inhibitors because of their short terminal half-life (~26 h for palbociclib 23 ; 32.6 h for ribociclib 22 ). In a biomarker study that assessed the longitudinal effect of palbociclib following a single dose or 3-week administration on Ki67 and pRb expression in skin biopsies and sTK1 in blood samples collected from 26 patients with HR+/HER2− MBC, recovery of pRb, Ki67, and sTK1 was observed during the offtreatment week, back to baseline on day 2-3 for pRb, day 3-4 for Ki67 and sTK1 19 . The rebound of sTK1 at the end of C1 therapy with the 3-weeks-on/1-week-off schedule was also demonstrated Similarly, the mPFS of 11.96 (10.43~NR) months observed in the endocrine-resistant population was comparable to that reported in PALOMA-3 (mPFS 11 months in the updated analysis) 25 .
In addition, our study demonstrated the potential utility of sTK1 activity at baseline and on-treatment as a prognostic marker and in monitoring disease status for patients with MBC receiving a CDK4/6 inhibitor. High sTK1 at baseline or early on-treatment time point (C1D15 or C2D1), especially at C1D15, had high degrees of accuracy in predicting progression within 6 months (84% accuracy for C1D15). In addition, we demonstrated that a rise in sTK1 predicted subsequent clinical/RECIST progression.
Our data are consistent with previous studies demonstrating that a higher baseline sTK1 is associated with a shorter time to progression in patients with advanced HR+/HER2− breast cancer receiving endocrine therapy and a rise in sTK1 on-therapy from baseline was associated with treatment resistance [26][27][28][29] . Few studies have evaluated sTK1 activity in patients receiving standard dosing palbociclib in combination with endocrine therapy. McCartney et al analyzed serial plasma TK1 activity at baseline (T0), after 1 cycle (T1), and at progression (T2) in 46 patients with HR + MBC treated with palbociclib within the TREnd trial 30 . The median TK1 activity was 75 Du/L at T0, decreased to 35 Du/L at T1, and increased to 251 Du/L at progression 30 . Patients with increasing TK1 at T1 correlated with a worse outcome than those with decreased/stable TK1 (n = 33; mPFS 3.0 vs 9.0 months; p = 0002), similar to our study 30 . Although TK1 above the median at T2 was associated with worse outcomes on post-study therapy, baseline TK1 was not prognostic 30 . In addition, in vitro studies demonstrated that TK1 reduction occurred in palbociclib sensitive but not resistant cells 30 . Cabel et al. 31 reported a study that assessed the plasma TK1 activity at baseline and a 4-week time point in 103 patients with ER+/HER2− MBC treated with palbociclib and endocrine therapy, which demonstrated that baseline TK1 activity (using median value as cutoff) was an independent prognostic factor for both PFS and OS, but adding TK1 activity at 4 weeks did not further increase survival prediction. More recently, Malorni et al. assessed TK1 at pre-treatment, C1D15, and D28 in patients with endocrine-resistant luminal MBC receiving palbociclib and fulvestrant. In this study, TK1 was significantly suppressed on C1D15, with 90/108 (83%) patients to <20 Du/L, similar to our observation with the 5-days-on/2-days-off schedule. However, on D28, a TK1 rebound was observed in most patients in the Malorni study, with TK1 < 20 Du/L in only 29% patients, while 28 of 44 (63.6%) had TK1 < 20 Du/L on C2D1 in our study. Similar to our study, at each time point, higher TK1 was significantly and consistently associated with shorter PFS, with C1D15 being most prognostic. However, none of the prior studies assessed longitudinal TK1 activity assessment until progression. Our study is the first to show that increases in TK1 occur earlier than clinical progression, indicating detection of subclinical progression.
Our study is limited by the small sample size and the nonrandomized single-arm trial design. The data regarding sTK1, at BL and early on-treatment time points, in predicting response on CDK4/6 inhibitors is intriguing as there are no predictive biomarkers currently available for CDK4/6 inhibitors. Future prospective trials are warranted to confirm our results on the safety and efficacy of the 5-days-on/2-days-off weekly schedule and to validate the clinical utility of sTK1 in guiding the management of patients with HR + /HER2-MBC.
In conclusion, this single-arm phase II trial of palbociclib administered at an alternative 5-days-on/2-days-off weekly schedule, in combination with either letrozole or fulvestrant, met the predefined primary endpoint in reducing the incidence of high-grade neutropenia, without compromising efficacy. While randomized trials are  needed to confirm this finding, this alternative schedule provides an option for patients having difficulty tolerating the standard 3-weekson/1-week-off schedule and to potentially avoid drug discontinuation due to neutropenia. Our data demonstrate sTK1 activity a promising biomarker of prognosis and disease monitoring in patients receiving CDK4/6 inhibitors.     20,32 for all patients with available samples at baseline, C1D15, C2D1, and C4D1, as well as every 3 cycles up to progression in those who progressed by the time of data cutoff. TK1 activity was determined using a refined ELISA-based method (DiviTum ® ) according to the manufacturer's instruction (www.biovica.com) and was performed at the Biovica laboratory in Uppsala, Sweden, with laboratory investigators blinded to patient data.

Outcomes
The primary endpoint was the rate of G3 + ANC between C1D1 and C2D1 (C1D1-29). The sample size of 47 provided 90% power, based on a onesample binomial exact test at alpha = 5%, to test the one-sided null hypothesis of G3 + ANC rate >62%, an estimate based on incidences from prior phase III trials of palbociclib 10,13 and that neutropenia occur early in the course of therapy 33 , versus the alternative of <40%. If G3 + ANC was observed in ≤23 patients, the 5-days-on/2-days-off schedule will be deemed as having less neutropenia than the standard schedule. As a subsequent pooled analysis of safety data from three randomized trials (PALOMA-1, 2 and 3) indicates, the rate of G3 + neutropenia in C1 in the palbociclib arm was 44.7% 18 , lower than what we originally expected. A post hoc power calculation was performed on testing against the null hypothesis H0: G3 + neutropenia rate >44.7% versus the observed 21.3% (10 out of 47, including the occurrences on C2D1 beyond C1 D1 to D28) in this trial. With N = 47, the post hoc power is 94.04% based on a one-sided Binomial exact test at 5% alpha level. Secondary endpoints include the rate of G3 + ANC in all cycles, palbociclib dose intensity/reduction/interruption/ discontinuation, AEs, PFS for the overall population and for the endocrine sensitive or resistant population as defined by ESMO guideline 21 , objective response rate (ORR: CR + PR (complete and partial responses)) and clinical benefit rate (CBR: CR + PR + Stable disease (SD) ≥ 24 weeks by RECIST 1.1). Other endpoints include sTK1 at baseline, C1D15, C2D1, and progression, in relation to PFS and CBR, as well as, the lead time from sTK1 rise during therapy to disease progression defined by RECIST.

Statistical analysis
Patient characteristics and AEs were summarized by descriptive statistics. AE rate, ORR, and CBR were estimated accompanied with 95% confidence interval (CI). PFS was defined from the date on treatment to off-study date (due to radiographic progression, clinical deterioration, investigator decision, AE) or date of death and to the date of the last imaging scan demonstrating no progression if patients had no events. Survival endpoints were analyzed by Kaplan-Meier (KM) method and survival difference was compared between patient groups of interests by log-rank test. Cox proportional hazard model was applied to estimate the hazard ratio (HR) with 95% CI. sTK1 ≤ 20 Du/L was deemed undetectable and was replaced with 19 Du/L for analysis. sTK1 was compared between time points by Wilcoxon signed-rank test with p values corrected by the Benjamini-Hochberg method to control false discovery rate (FDR). Baseline and on-treatment sTK1 was compared between patients who had the clinical benefit (CB) or progressive disease (PD) versus not by the Wilcoxon rank-sum test. sTK1 was dichotomized to high versus low by a pre-defined cutoff of 200 Du/L at baseline 28,29 and 20 Du/L at on-treatment time points 24 and was then analyzed in relation to PFS by KM method and logrank test. Diagnostic test operating characteristics including specificity, sensitivity, negative and PPV of the dichotomized sTK1 at different time points were derived for the binary outcomes (CB vs non-CB; PD vs non-PD).

Reporting summary
Further information on research design is available in the Nature Research Reporting Summary linked to this article.

DATA AVAILABILITY
The datasets used and analyzed during the current study are available from the corresponding authors on reasonable request.