Clinical and Genotypic Insights Into Higher Prevalence of Palbociclib Associated Neutropenia in Asian Patients

Abstract

Background

CDK4/6 inhibitors (CDK4/6i) have shown great efficacy in prolonging progression-free survival and is the current standard of care for hormone positive (HR(+)) metastatic breast cancer (mBC). Despite well tolerability and ease of use, the most common side effect of CDK4/6i is myelosuppression, with neutropenia the most prevalent adverse effect. Studies show that the prevalence and severity of neutropenia are more marked in Asian patients, although details remain obscure.

Methods

In this study, we retrospectively analyzed 105 Taiwanese patients who received palbociclib for HR(+) HER2(−) mBC at the Taipei Veterans General Hospital. To investigate a possible genetic association for high prevalence of neutropenia, we queried the Taiwan Biobank with publicly available germline databases (ALFA, gnomAD, ExAC, 1000 Genomes project, HapMap), for the allele frequencies of 4 neutropenia-related SNPs (ABCB1_rs1045642, ABCB1_rs1128503, ERCC1_rs3212986, ERCC1_rs11615) and compared between different ethnicities. In addition, one of the patients was a long-term patient with peritoneal dialysis. We quantified the levels of palbociclib in her serum and peritoneal fluid by liquid chromatography-mass spectrometry (LC-MS).

Results

Interestingly, in our cohort, early neutropenia nadir (occurred within 56 days of start) was associated with worse treatment outcome, while occurrence of grade 3/4 neutropenia was associated with better outcome. We observed an extremely high incidence of neutropenia (96.2% any grade, 70.4% grade 3/4). In the analyzed germline databases, we discovered a higher SNP frequency of the T allele in ABCB1_rs1128503, a lower frequency of T allele in ABCB1_rs1045642, and a higher SNP frequency of G allele in ERCC1_rs11615. We observed that palbociclib levels in peritoneal dialysate ranged from around 20-50 ppb, and serum levels reached 100-110 ppb during drug administration and decreased to <10 ppb during discontinuation.

Conclusion

Our retrospective analysis of real world palbociclib use reveals an association with grade 3/4 neutropenia with better outcome and early neutropenia nadir with worse outcome. Our findings of Asian specific SNPs support a predisposition toward profound and prevalent neutropenia in Asian patients under CDK4/6i. We also report the first pharmacokinetics analysis on a patient with peritoneal dialysis receiving CDK4/6i. In summary, our study provides novel clinical and genotypic insights into CDK4/6i associated neutropenia.

CDK4/6 inhibitors (CDK4/6i) are the current standard of care for HR+ metastatic breast cancer; however, the most common adverse effect is neutropenia. This article reports novel clinical and genotypic insights into CDK4/6i-associated neutropenia.

Implications for Practice.

CDK4/6 inhibitors are well-established treatment modalities for HR(+) advanced breast cancer. Neutropenia is an important adverse effect that leads to dose interruption and modification. This article reports an extremely high prevalence of neutropenia in retrospective clinical analysis of patients receiving palbociclib. Data from the Taiwan Biobank and other public germline databases reveal ethnic differences in allele frequencies of neutropenia associated SNPs. This study also reports serum and peritoneal fluid palbociclib quantification by LC-MS and describes pharmacokinetics in response to dose modification of palbociclib. This study provides novel clinical and genotypic insights into CDK4/6i-associated neutropenia.

Introduction

CDK4/6 inhibitors (CDK4/6i) have shown great efficacy in prolonging progression-free survival (PFS; palbociclib, ribociclib, and abemaciclib) in both first- and second-line setting for hormone positive (HR(+)) metastatic breast cancer (mBC). Currently, ribociclib (MONALEESA-2, -3, -7) and abemaciclib (MONARCH-2) have also demonstrated overall survival benefit. Despite tolerability and ease of use, the most common side effect of CDK4/6i is myelosuppression and especially neutropenia. This is evidenced in PALOMA-2 and MONALEESA-2 (all grade neutropenia 79.5%, grade 3/4 neutropenia 66.5% in PALOMA-2; all grade neutropenia: 74.3%, grade 3/4 55.3% in MONALEESA-2), although febrile neutropenia seldom occurred (less than 3%).^1,2 Abemaciclib had a relatively lower incidence of neutropenia (MONARCH-3: all grade neutropenia: 41.3%, grade 3/4 35.8%), likely owing to its distinct profile of CDK inhibition.^3,4 Subgroup studies from the PALOMA trials suggests that Asian patients receiving palbociclib have higher rates of neutropenia.^5-8 Studies have proposed different genetic factors predisposing to CDK4/6i-induced neutropenia, including Duffy antigen polymorphisms,⁹ ABCB1 and ERCC1 polymorphisms (ABCB1_rs1128503, ABCB1_rs1045642, ERCC1_rs11615, ERCC1_rs3212986),¹⁰ CDK6 polymorphisms,¹¹ and others. Guidelines for CDK4/6i recommend dose adjustments according to neutrophil levels, leading to dosing interruption and complicating effective treatment.

The association of neutropenia with CDK4/6i treatment efficacy is still unclear. A recent study reported association between neutropenia and better outcome in 196 patients (61 patients with breast cancer) receiving palbociclib.¹² In the same study, early neutropenia, defined as nadir of absolute neutrophil count (ANC) in the first 56 days of starting palbociclib, was also associated with prolonged PFS.¹² Other studies also link Palbociclib associated neutropenia to improved PFS benefit.^12,13 Currently, it is still unclear whether an “on-target” effect of neutropenia predicts efficacy toward CDK4/6i, as frequently seen with tyrosine kinase inhibitors.¹⁴

CDK4/6is are metabolized primarily by the cytochrome P450, with the majority excreted through feces (>70%).³ Pharmacokinetics of palbociclib in patients with normal renal function has been well studied;^15,16 however, data are lacking in patients with renal failure. Peritoneal or hemodialysis is indicated for patients with irreversible severe kidney failure.¹⁷ The FDA approved package insert for palbociclib recommended no dose modification for patients with renal failure, noting that pharmacokinetics have not been studied in patients with hemodialysis.¹⁸ Package inserts for ribociclib and abemaciclib stated that pharmacokinetics was not affected by mild or moderate renal impairment, while data also lacking in several renal impairment (eGFR < 30 mL/minute/1.73m²).^19,20 As all 3 CDK4/6i are metabolized by CYP3A4, there is a clinical need to accurately determine the pharmacokinetics in patients with dialysis.

In this study, we conducted a retrospective analysis of 105 Taiwanese patients who received palbociclib for HR(+) HER2(−) mBC. We observed an extremely high incidence of any grade neutropenia in our cohort, as well as associated treatment benefit with grade 3/4 neutropenia. To investigate a possible genetic association for the high prevalence of neutropenia, we queried the Taiwan Biobank, a nationwide biospecimen repository with 129 586 healthy individual genome wide DNA sequencing data deposited.^21,22 We investigated the prevalence of the 4 neutropenia associated SNPs previously reported in the PALOMA studies and demonstrate that germline SNP frequency ethnic imbalance exist in the general population.¹⁰ In addition, in our clinical cohort, one of the patients had long-term peritoneal dialysis and was started on palbociclib. By liquid chromatography-mass spectrometry (LC-MS), we quantified the levels of palbociclib in her serum and peritoneal fluid at various time points, while on different dosage use. To our knowledge, this is the first pharmacokinetics analysis on a patient with peritoneal dialysis receiving CDK4/6i. Our study reports several novel insights regarding CDK4/6i associated neutropenia and expands the knowledge breadth for this clinically relevant topic.

Material and Methods

Clinical Analysis of Patients Receiving Palbociclib

Clinical characteristics of adult patients (≥18 years) diagnosed with metastatic breast cancer with pathology proven HR(+) HER2(−) and initiating palbociclib between March 1, 2016 and March 1, 2021 were included. Cases of interest were identified through medical records. The study was approved by the Institutional Review Board (IRB) of Taipei Veterans General Hospital (IRB approval number: 2021-11-005BC). Other detailed parameters are described in Supplementary File S1. Neutropenia was defined per CTCAE (Common Terminology Criteria for Adverse Events) v 4.0, (Absolute neutrophil count (ANC): grade 1: ANC < LLN: 1500/mm³; grade 2: ANC 1000: <1500/mm³; grade 3: ANC 500: <1000/mm³; grade 4: ANC <500/mm³. The normal range for WBC in TVGH was 4180to 9380/µl, neutrophils was 1890 to 6760/µl; therefore, the LLN for neutrophil determination was 1890 for this study.

Genome-Wide SNP Genotyping of the Taiwan Biobank

The Taiwan Biobank is a major community-driven initiative, providing a vital platform for investigating associations between genetics, environment, and human health.^21,23 Further details of the Taiwan Biobank are described in Supplementary File S1.

Quantification of Palbociclib Concentration in Peritoneal Fluid and Serum

Chemicals and Reagents

Detailed LC-MS based sample collection, preparation, and quantification methods are described in Supplementary File S1.

Statistical Analysis

Kaplan-Meier curves were generated with GraphPad Prism (version 10.0.2 for Windows 64 bit, GraphPad). P-values were calculated using either the log rank method, Gehan-Breslow-Wilcoxon test, or Fisher’s exact test. A P-value of less than .05 was considered significant.

Results

1. Clinical analysis of a HR(+)HER2(-) metastatic breast cancer patient cohort receiving palbociclib

We retrospectively analyzed a cohort of 105 patients who received palbociclib from 2017-2021 at the Taipei Veterans General Hospital. The clinical profile of the cohort is shown in Table 1. All patients were of female gender with radiologically proven metastatic HR (+) HER2(−) breast cancer. In this cohort, the median age was 56.9 years old, with 88.6% postmenopausal status. All premenopausal patients received concurrent ovarian function suppression (OFS) with GnRH agonists or oophorectomy. The cohort had a median follow up time of 968 days (32.3 months). 66.7% and 47.6% had lung or liver metastases, respectively.

Table 1.

Clinical characteristics of patients receiving palbociclib for mBC at Taipei Veterans General Hospital.

Characteristics	Number	Percentage
Patients, n	105
Median age, y	56.9
Female, %	100
Premenopausal	12	11.43%
Median BMI	23.72
Median height (cm)	155.8
Median weight (kg)	57.5
Lung metastasis	70	66.7%
Liver metastasis	50	47.6%
Median PFS	503 (17.9 months)
Median OS	968 (32.27 months)
Death events	53	50.4%
1L	57	54.28%
2L	15	14.29%
≥3L	33	31.43%
Response
CR	0
PR	39	37.8%
SD	54	52.4%
PD	10	9.7%
ORR	39/103	37.8%
In first-line patients
Median PFS	612 (20.4 months)
Median OS	Not reached
Progression/death events	36	63.2%
Death events	19	33.3%
PFS of second-line treatment	176 (5.87 months)
PFS2	883 (29.4 months)
ORR		47.4%
In second-line patients
Median follow-up time	870 days (29 months)
Median PFS	318 days (10.6 months)
Median OS	Not reached
Progression/death events	13	86.67%
Death events	5	33.3%
PFS2	562 (17.7 months)
ORR		33.3%
In third-line patients
Median PFS	181 days (6.0 months)
Median OS	710 days (23.7 months)
Progression/death events	26	76.5%%
Death events	23	67.6%
PFS2	413 (13.8 months)
ORR	21.2%	CR 0, PR 7, SD 18, PD 8
Dose reduction	30	28.6%
Patients reached neutrophil nadir <56 days	56	55.4%
Entire cohort	Early neutropenia	Delayed neutropenia	P-value
Patient number	56	55
Median age	60	57.3	.718
Dose reductions	17	13
% of dose reductions	30.3%	28.9%	1.00
Liver metastases (%)	64.3	60	.56
Lung metastases (%)	44.6	41.8	1.00
Bone metastases (%)	73.2	70.1	.49

In this cohort, 57 patients (54.3%) received palbociclib as first-line treatment, 15 (14.3%) as second line, and the remaining (31.4%) as third- or later line. In the 103 patients with evaluable response, 39 patients had best response as CR or PR, 54 had SD, and 10 had progressive disease, with an objective response rate (ORR) of 37.8% and disease control rate (DCR) of 90.3%.

In the 57 patients who received first-line treatment, PFS was 20.4 months, median OS not reached (19 death events, 33% dead). (Fig. 1A and 1B). PFS2 (progression-free survival from starting palbociclib to end of subsequent treatment) was 29.4 months. As expected, palbociclib first-line PFS was significantly longer than later lines (P = .0123) (Fig. 1A).

Figure 1.

(A) Progression-free survival (PFS) curves by lines receiving palbociclib. (B) Overall survival curves by lines receiving palbociclib. (C) PFS curves of patients receiving chemotherapy or non-chemotherapy as second line following progression on palbociclib. (D) Pie chart of treatment selection following progression on Palbociclib. (E) Endocrine partner shown according to treatment line. (F) PFS curves of patients with or without liver metastases (mets). (G) PFS curves of patients with liver or lung metastases, compared to patients without either. (H) PFS curves of patients with or without lung metastases (mets). (I) PFS curves of patients with Ki-67 higher or lower than 20%. Abbreviations: TAM: tamoxifen; FULV: fulvestrant; AI: aromatase inhibitors; GnRH: Gonadotropin hormone-releasing hormone (leuprorelin or goserelin).

Thirty-six patients progressed on first-line palbociclib. In these patients, 31 patients received second-line treatment, including 12 patients with chemotherapy, 19 patients with non-chemotherapy (Fig. 1C). Interestingly, patients who received chemotherapy had a trend of better PFS compared to patients with non-chemotherapy (P = .183). The most used second-line treatment was everolimus plus exemestane (8 patients, 22.2% of progressors) followed by capecitabine (5 patients as monotherapy, 2 patients capecitabine + cyclophosphamide, total 19.44%) (Fig. 1D).

In the 15 patients who received palbociclib as second-line setting, PFS was 318 days (10.6 months) while OS was 1119 days (37.3 months). PFS2 was 17.7 months, and ORR was 33.3%. In the ≥third-line setting with 33 patients, median PFS and OS were 181 days (6.0 months) and 710 days (23.7 months), respectively. PFS2 was 13.8 months, and ORR was 21.2%.

We also documented the concomitant endocrine partner for palbociclib according to treatment lines (Fig. 1E, Supplementary Table S1). In each line, aromatase inhibitors (AI) were the majority endocrine partner (91% in first line, 60% in second line, 64% in third line). Fulvestrant was rarely used in first line (4% of patients) but more frequently in later lines (27%, 27% in second and third lines, respectively). This reflected the adherence to PALOMA-2 regimen in the real-world setting.

A total of 84 patients (80% of the entire cohort) had bone metastasis upon starting palbociclib, 70 patients (66.7%) had lung metastases, 50 patients (47.6%) had liver metastases. Eleven patients presented with bone only metastases (BOM). Since presence of liver metastases has been well characterized as a poor prognostic factor in CDK4/6 inhibitors,⁴ we examined the impact of liver metastases in our cohort. Consistent with prior studies, presence of liver metastases or coexisting lung/liver metastases were associated with worse survival (P = .0001,.0393, respectively), while lung metastases alone did not show significantly worse survival (P = .66) (Fig. 1F-1H). Since in the MONARCHe study, Ki-67 was proven to be an important prognostic factor, but not a predictive factor of CDK4/6 inhibitor efficacy, we wondered if we could observe this phenomenon in our cohort. However, in the 64 patients with Ki-67 data available, we did not observe a PFS difference in Ki-67 high (≥20%) or low patients (P = .38), consistent with the hypothesis that Ki-67 may not be a predictor for CDK4/6 inhibitor efficacy (Fig. 1I).

1. Palbociclib is associated with very high incidence of neutropenia in Taiwanese cohort

We assessed the prevalence and features of palbociclib associated neutropenia in our cohort. Overall, a total of 96.2% of patients (101 patients) presented with any grade of neutropenia, while 70.4% of patients developed grade 3/4 neutropenia (Table 1). In the 4 patients who did not develop neutropenia, 2 received Palbociclib as first line and 2 as ≥third line. Only 1 neutropenic fever episode occurred. A total of 30 patients had dose reduction (28.6%), including 2 patients who had 2 reductions (to 75 mg) and 28 patients with 1 reduction (to 100 mg). Interestingly, patients who developed grade 3/4 neutropenia had a better outcome (median PFS 23 months vs. 10 months, P = .0025) (Fig. 2A).

Figure 2.

(A) PFS curves of patients with or without grade 3/4 neutropenia. (B) PFS curves of patients who developed neutropenia nadir before or after 56 days since initiation of palbociclib. (C) PFS curves of patients who developed neutropenia nadir before or after 56 days since initiation of palbociclib as first-line.

In the 101 patients who developed neutropenia on palbociclib, 56 patients (55.4%) developed the lowest ANC level during the first 2 cycles (defined as time to reaching neutrophil nadir in the first 56 days), and 55 patients reached ANC nadir after day 56 (median time to nadir was 25 vs. 215 days, respectively). Interestingly, patients who reached ANC nadir during the first 2 cycles had a worse outcome compared to patients who reached ANC nadir later (median PFS: 592 days vs. 892 days, P = .011). (Fig. 2B). There was a comparable percentage of patients who had dose reduction in the early neutropenia cohort (early vs. delayed: 30.3% vs. 28.9%, Fisher’s exact test P = 1.00). We did not observe vast differences in baseline characteristics between early and delayed neutropenia regarding age, percentage of liver, lung, bone metastases, suggesting disease severity may not predispose to the difference in outcome (Table 1).

When we limited our analysis to patients receiving palbociclib as first line and developed any grade neutropenia (total 55 patients), 29 patients (55.2%) developed nadir before the first 2 cycles compared to 26 patients who did not. There was a trend for patients with earlier neutropenia associated with worse outcome (median PFS: 702 days vs. 1007 days, P = .085) (Fig. 2C). In summary, our data suggests an inferior outcome in patients with early neutropenia in patients with breast cancer.

1. Neutropenia associated SNP allele frequency in Taiwan Biobank

We investigated the Taiwan Biobank to assess the allele frequency of the following 4 SNPs: ABCB1_rs1045642, ABCB1_rs1128503, ERCC1_rs3212986, ERCC1_rs11615,¹² in a sample size of 108 801 to 108 829 subjects.

For ethnic comparison, we investigated several SNP database cohorts downloaded from the dbSNP database, including ALFA SNP database (version 20200227123210),²⁴ Genome Aggregation Database (gnomAD),²⁵ Exome Aggregation Consortium (ExAC),²⁵ the 1000 Genomes project,²⁶ and the HapMap project.²⁷ For the ABCB1_rs1045642 SNP, the T allele frequency was ~0.4 in Asian subjects, mildly lower compared to European and American subjects, while extremely low (~0.2) in African subjects (Fig. 3A). For the ABCB1_rs1128503, the T allele frequency in the Asian cohort was >0.6, substantially higher than American (~0.45) and European (~0.4) patients and was extremely low (~0.2) in African subjects (Fig. 3B).

Figure 3.

(A-D) Allele frequency by ethnicity in SNPs analyzed across databases.

For ERCC1_rs11615, the A allele frequency was substantially higher in European subjects (~0.6) and American subjects (~0.4), low in Asian (~0.25) and African (~0.1) subjects (Fig. 3C). However, for ERCC1_rs3212986, the allele frequency of the reference allele C was not substantially different across different ethnic cohorts (~0.7) (Fig. 3D). SNP frequencies from the Taiwan Biobank was largely consistent with data from Iwata et al¹⁰ (Table 2).

Table 2.

Comparison of SNP frequencies from the PALOMA-2/3 studies and the Taiwan Biobank. Note: The PALOMA-2/3 studies SNP are compiled from data from Fig. 1 in Iwata et al,¹⁰ with permission from the authors.

	SNP		TT	TC	CC	T allele (%)	C allele (%)
PALOMA-2/3	rs1045642	Non-Asian	25	52	22	51	49
		Asian	15	47	39	38.5	61.5
Taiwan BioBank		Taiwanese	14	46	40	37	63
			TT	TC	CC	T allele (%)	C allele (%)
PALOMA-2/3	rs1128503	Non-Asian	19	47	34	42.5	57.5
		Asian	43	45	12	65.5	34.5
Taiwan Biobank		Taiwanese	42	45	12	65	35
			AA	AG	GG	G allele (%)	A allele (%)
PALOMA-2/3	rs11615	Non-Asian	36	48	16	60	40
		Asian	7	43	49	70.5	29.5
Taiwan Biobank		Taiwanese	7	38	55	74	26
			AA	AC	CC	A allele (%)	C allele (%)
PALOMA-2/3	rs3212986	Non-Asian	7	38	55	26	74
		Asian	7	41	52	27.5	72.5
Taiwan Biobank		Taiwanese	11	44	45	33	67

In summary, of the 4 SNP examined from germline databases with >100 000 subjects, we discovered a germline ethnic difference in the allele frequency of the reference alleles in ABCB1_rs1128503, ABCB1_rs1045642, and ERCC1_rs11615 in the Asian population, consistent with the findings from PALOMA-2/3.

1. Dynamic monitoring of serum and dialysate palbociclib in a patient with peritoneal dialysis who developed neutropenia on palbociclib

As previously mentioned, data regarding pharmacokinetics of CDK4/6i in patients with dialysis are lacking. We hereby report a female patient from our cohort who had started receiving peritoneal dialysis 10 years before the diagnosis of breast cancer. She was diagnosed with early luminal breast cancer in 2007 and received breast conserving surgery followed by adjuvant radiotherapy. Twelve years later, disease recurrence in lung, liver, bone, chest wall, and mediastinal lymph nodes were discovered, and rebiopsy confirmed HR(+) HER2(−) mBC.

Due to good performance status and no visceral crisis, we opted to start CDK4/6i with ET for this patient. With no clear guidance for patients with peritoneal dialysis CDK4/6i use, we decided to start the patient on 125 mg palbociclib daily. To closely monitor the pharmacokinetics of palbociclib in this patient, we developed and implemented a quantification method of palbociclib in body fluids using LC-MS/MS (see Methods section for details). We collected peritoneal dialysate at baseline (day 0, before starting palbociclib) and at various time points in the first week after starting treatment.

The patient developed grade 3 neutropenia on the 17th day after starting palbociclib 125 mg daily (WBC 2100, ANC 693). The decision was made to temporarily discontinue palbociclib. On the 10th day of discontinuing palbociclib, profound neutropenia was still noted (WBC 1200), and the drug was still discontinued. The neutrophil count normalized (WBC 5400) on day 15 after discontinuation, and palbociclib 100 mg was restarted (Fig. 4A). The patient tolerated this dose well and achieved partial response on palbociclib (Fig 4B). She continued to receive palbociclib until a sudden onset of dialysis-related infection occurred, which resulted in clinical deterioration and ultimately death. Since she expired outside of our hospital, we could not locate the exact neutrophil levels at time of her death, but records from our hospital in the past 5 months (2021 Jan-May) before her death (2021-June) revealed relatively steady neutrophil levels (last ANC count: 2156) and no infection episodes (Supplementary Fig. S1). Her PFS was 360 days since initiation of palbociclib.

Figure 4.

(A) changes in peritoneal fluid palbociclib concentrations (y-axis, ppb) in respect to time (x-axis). Squares are values from peritoneal dialysate fluid (PDF), triangles are values from serum. (B) Chest computed tomography (CT) of the patient showing changes in pulmonary metastases. Left: taken on April 14, 2020, prior to initiation of palbociclib. Right: taken on August 27, 2020.

By analyzing palbociclib concentration at various time points, several observations could be made: (1) Palbociclib levels in peritoneal dialysate ranged from around 20-50 ppb during administration of palbociclib and did not vary much during treatment of 100 mg or 125 mg. The drug levels decreased to undetectable state (<0.1 ppb, the detection limit of the quantification method) during discontinuation of palbociclib. (2) Serum levels of palbociclib reached 100-110 ppb during drug administration and decreased to <10 ppb during discontinuation. The serum levels did not differ significantly during treatment of 100 mg or 125 mg.

Discussion

In our study, we report a cohort of patients receiving palbociclib for metastatic, HR(+) breast cancer. Efficacy data in our cohort were consistent with landmark trials of palbociclib. Our study specifically examined the second-line PFS data following progression of palbociclib. Patients who directly received chemotherapy had a numerically longer PFS compared with non-chemotherapy. The optimal treatment regimen post CDK4/6i progression is currently under intense research. Recent results from the MAINTAIN,²⁸ PACE,²⁹ CAPITELLO-291,³⁰ and PALMIRA³¹ studies collectively show that fulvestrant monotherapy in the post CDK4/6i setting tend to have relatively short PFS, ranging from 3-5 months. MAINTAIN study demonstrated the advantage of CDK4/6i plus fulvestrant in this setting.²⁸ Interestingly, while currently no phase III trial have confirmed using CDK4/6i post-progression of CDK4/6i, this practice is already well in real world use.^32,33 In a large real-world study of post-CDK4/6i treatment patterns, Sawaki et al³² reported subsequent chemotherapy had relatively shorter time to failure (TTF) when compared to subsequent ET + CDK4/6i (7.2 m vs. 10.9 m). However, in the same study, chemotherapy has longer TTF than ET alone (4.4 m) or ET + everolimus (6.1 m). The RIGHT Choice study³⁴ demonstrated PFS benefit of CDK4/6i over chemotherapy in the first-line setting patients with visceral tumor burden, but no such data exists in later line setting, especially post CDK4/6i progression. The above evidence suggests that when post-CDK4/6i, switching to another CDK4/6i and another ET (such as fulvestrant) is superior to non-CDK4/6i regimens, and may also be superior to chemotherapy. We anticipate that the near future, 2 questions will be crucial in shaping the optimal treatment sequence: (1) whether CDK4/6i post CDK4/6i progression has solid phase III evidence (postMONARCH study³⁵), and whether CDK4/6i post CDK4/6i progression is superior to chemotherapy.

It is tantalizing to hypothesize that CDK4/6i as first line, while significantly extending first-line PFS and overall survival,³⁶ changes the endocrine sensitivity landscape. Intriguingly, the antibody drug conjugate Trastuzumab Deruxtecan (T-Dxd) reported substantial PFS benefit in the DESTINY-Breast 04 trial³⁷ with majority of patients refractory to prior CDK4/6 inhibitors. Since DESTINY-Breast 04 mandates 1-2 lines of prior chemotherapy, whether T-Dxd is efficacious in chemo-naïve patients is addressed in the ongoing phase III DESTINY-Breast 06 trial³⁸ (notably, DESTINY-6 compares between T-Dxd and chemotherapy, not T-Dxd versus ET or CDK4/6i). Besides chemotherapy, novel therapeutics including oral SERDs, the oral PROTAC ER degrader ARV-471,³⁹ the novel SERM lasofoxifene (phase III ELAINE-3 study⁴⁰) are now actively undergoing clinical investigation. The results are eagerly anticipated.

Neutropenia is an important aspect of CDK4/6i and plays a factor in routine clinical use. Pooled studies from the PALOMA trials reported a total of 35.5% patients who required dose reduction from 125 mg to 100 mg.⁸ Real world studies have shown comparable results.⁴¹ Interestingly, multiple studies now collectively show that reduction in dose did not compromise treatment efficacy.^8,42 Data from pooled MONALEESA studies (MONALEESA -2, -3, -7) reported 41.8% patients had ≥ 1 dose reduction,⁴³ mostly due to neutropenia, and had no impact on efficacy.^43,44 In comparison, abemaciclib has a lower dose reduction rate with 10%, 12.8% in MONARCH-2 and MONARCH-3, respectively.⁴⁵ Currently, the body of evidence suggests that discontinuation due to neutropenia does not ostensibly impact clinical outcomes.

In our study, we reported a high incidence of neutropenia (96.2% all grade, 70.4% grade 3/4). Although our cohort consisted about 46% of patients receiving Palbociclib in second line or later, prior treatment resulting in bone marrow impairment may not be sufficient to explain the high neutropenia incidence, as our first-line patients had equally high prevalence of neutropenia (55/57, 96.4%). Consistent with this hypothesis, neutropenia rates in PALOMA-2 and PALOMA-3 did not differ significantly (grade 3/4 neutropenia 66.5% in PALOMA-2, 65% in PALOMA-3^1,46). Our findings are consistent with the previous reports showing a relatively higher incidence of neutropenia to CDK4/6i in Asian patients.^5-7 The recently reported PALOMA-4, a phase III study in Asian patients with identical trial design to PALOMA-2,⁴⁷ reported all grade neutropenia occurred in 98.2%. The PALOMA-4 data (168 patients) are nearly identical to our cohort (105 patients), again confirming the high incidence of neutropenia in Asian patients receiving palbociclib. The propensity of Asians for higher incidence of neutropenia has also been proposed in studies utilizing chemotherapy.^48-50 Iwata et al reported a pooled analysis of the PALOMA-2/3 on 4 SNPs encoding ABCB1 and ERCC1 polymorphisms, which inspired our study.¹⁰ It must be emphasized that in the PALOMA studies, germline SNP analysis was done in the 652 trial patients, whereas our study provides germline data from the presumably cancer free general population with a far larger sample size (129 586 participants in the Taiwan Biobank, 60 706 from ExAC database). Observations from our analysis reveal that SNP differences regarding the ABCB1 and ERCC1 genes exist as a disequilibrium across ethnicities, providing a genetic predisposition basis for higher neutropenia in Asians.

In contrary to McAndrew et al,¹² early neutropenia (development of neutropenia in the first 2 cycles) was associated with poorer outcomes in our cohort. The McAndrew study consisted of only 61 patients with breast cancer, in which 88% had ≥3 prior lines treatment before palbociclib, 91.8% (56/61) developed early neutropenia,¹² and PFS in patients with early neutropenia was 17.3 weeks (~4 months). This may reflect the bone marrow accumulated toxicity in heavily pretreated status.⁵¹ In our cohort, 54 patients (53.5%) developed early neutropenia. The lower incidence of early neutropenia in our cohort (53.5%) compared to the McAndrew study (91.8%) suggested a difference in bone marrow reserve that could partially be explained by prior treatment lines. A mechanistic explanation for the impact of neutropenia on efficacy regarding either timing or severity remains elusive. Studies have proposed that CDK6 may be the main culprit for CDK4/6i-induced neutropenia, as mouse models suggest deletion of CDK6, but not CDK4, results in neutropenia.⁵² In vitro data suggest ribociclib and abemaciclib,^53,54 but not Palbociclib,^55,56 to be more selective of CDK4 over CDK6. One study proposed that different CDK4/6is demonstrate variable efficacy in cell lines with CDK4 or CDK6 dependence,⁵⁷ raising the tantalizing idea that CDK4, and CDK6 are not “equal” as determinants to efficacy. We hypothesize that the ability to inhibit CDK6, as well as the timing and strength of inhibition, relates to “on-target” efficacy that could translate into clinical benefit; however, further research is needed to confirm this idea.

In a patient receiving peritoneal dialysis, we quantified the drug level of palbociclib in both peritoneal fluid and serum. To our knowledge, this is the first quantification of palbociclib concentration in peritoneal fluid. We discovered that palbociclib concentration in peritoneal fluid is approximately 30%-50% of the serum levels. Other small-molecule drugs in peritoneal fluid have been quantified, including sorafenib to be ~40% of plasma levels,⁵⁸ and antibiotics drugs ranging 45% to 107% of serum levels.^59,60 Our data suggest that small-molecule drugs such as Palbociclib can reach a certain stable level in the peritoneal fluid. A recent study investigating palbociclib in patients with non-breast cancer with renal impairment⁶¹ concluded that total drug exposure was largely similar between patients with varying degree of renal failure. In summary, our observations suggest that palbociclib can be used safely in patients with dialysis, but judicious monitoring of neutropenia should be carried out.

The neutropenia in this patient took 15 days for neutrophil count to normalize. Interestingly, during discontinuation of Palbociclib, both serum and peritoneal fluid levels were near undetectable, and when restarted at a lower dose, levels of palbociclib resumed to similar levels as before. This was quite unexpected, and several questions, therefore, emerge: Would this finding (similar serum palbociclib levels in different dosage) be reproducible in a large, non-dialysis patient cohort? Would upfront dose reduction (starting at 100 mg or 75 mg of palbociclib) still reproduce the same outcome as full dose? Is this phenomenon exclusive to palbociclib or does it include other CDK4/6i? These are all questions that would be most suitably answered in a randomized controlled trial, which could potentially change future practice guidelines of CDK4/6i.

Our study is not without limits, including a retrospective nature and unblinded treatment status. Our study only reported data on Palbociclib, without data from the other 2 CDK4/6i; however, since ribociclib has a very similar neutropenia pattern to Palbociclib, we speculate that ribociclib would have a similar neutropenia presentation in Asian/Taiwanese cohort. Our serial palbociclib quantification was limited to only one patient with peritoneal dialysis, thus the applicability of the findings would be in question until further confirmed in a large-scale cohort study. However, we believe that despite these shortcomings, our study has provided several crucial insights: confirming high neutropenia incidence in Asian patients, providing SNP allele disequilibrium in Asian patients from large-scale genomic databases, and quantification of peritoneal fluid and serum palbociclib levels. Our findings provide important data and insights for the use of CDK4/6i in breast cancer.

Supplementary Material

Supplementary material is available at The Oncologist online.

Funding

This study was partially funded by the grants to J.I.L.: MOST 110-2314-B-A49A-549, NSTC 112-2314-B-A49-054 (National Science and Technology Council, Taiwan), the Taiwan Clinical Oncology Research Foundation, the Melissa Lee Cancer Foundation, and 112DHA0100294 (Taipei Veterans General Hospital internal grant).

Conflict of interest

The authors indicated no financial relationships.

Author Contributions

Conception/design: J.I.L., T.H.K., L.M.X.C. Provision of study material or patients: J.I.L., C.Y.L., Y.F.T., C.C.H., L.M.T., T.C.C. Collection and/or assembly of data: K.J.H., J.I.L., L.M.X.C., T.H.K., C.C.H. Data analysis and interpretation: K.J.H., J.I.L., L.M.X.C., T.H.K., C.C.H. Manuscript writing: K.J.H., J.I.L., L.M.X.C., T.H.K., C.C.H. Final approval of manuscript: All authors.

Data Availability

The original contributions presented in the study are included in the article. Clinical data are analyzed from medical records in TVGH. SNP studies were conducted using publicly available genome databases and data can be accessed per each database. Further inquiries can be directed to the corresponding authors.

Ethical Approval

The study was approved by the Institutional Review Board (IRB) of Taipei Veterans General Hospital (IRB approval number: 2021-11-005BC)