Background
The cyclin D-cyclin-dependent kinases 4 and 6 (CDK4/6)-retinoblastoma (RB) pathway is a fundamental component of cell cycle regulation and is implicated in the initiation and progression of various malignancies [
1,
2]. In particular, alterations in the cyclin D-CDK4/6-Rb pathway contribute to both de novo and acquired resistance to endocrine treatment in patients with hormone receptor (HR)-positive breast cancer [
3,
4]. Hence, this pathway is an important target for pharmacologic permutations to overcome endocrine resistance. Concordantly, CDK4/6 inhibitors exhibit robust and durable clinical activity in treating HR-positive, human epidermal growth factor receptor 2 (HER2)-negative advanced breast cancer (ABC) [
5‐
8]. Among CDK4/6 inhibitors, palbociclib demonstrates significant efficacy in combination with letrozole as first-line therapy (PALOMA-2) or fulvestrant as later-line therapy (PALOMA-3) in HR-positive, HER2-negative ABC [
9‐
12]. Although phase III studies consistently show clinical benefits of CDK4/6 inhibitors irrespective of menopausal status, visceral metastasis, or ethnicity, a subset of patients treated with CDK4/6 inhibitors exhibit an early progression, warranting the identification of biomarkers to predict treatment outcome. Unfortunately, previously identified potential biomarkers are inconsistent, have limited clinical feasibility, and do not reflect the dynamic changes induced by CDK4/6 inhibitors [
13,
14].
In addition to the essential role of CDK4/6 in cell cycle progression through RB phosphorylation, which underlies the proliferation of HR-positive breast cancer cells [
15], novel functions have been discovered [
16]. CDK6 is involved in the proliferation and differentiation of hematopoietic cells [
17‐
19]. Consequently, hematologic adverse events, including neutropenia, commonly occur after CDK4/6 inhibitor administration [
9,
20]. Moreover, CDK4/6 inhibitors enhance antitumor immunity through tumor infiltration and activation of T cells via transcriptional reprogramming [
21,
22]. Combining immune checkpoint and CDK4/6 inhibitors synergistically delays tumor growth, providing a rationale for combination strategies comprising CDK4/6 inhibitors and immunotherapies. In this context, a series of clinical trials testing various combinations of CDK4/6 and immune checkpoint inhibitors are currently underway.
Baseline and/or treatment-related lymphopenia is associated with poor clinical outcomes in patients with various malignancies, including metastatic breast cancer [
23‐
25]. Meanwhile, neutrophilia is generally associated with poor outcomes [
26], consistent with preclinical evidence that phenotypic and functional characteristics of neutrophils primarily overlap with those of polymorphonuclear myeloid-derived suppressor cells in cancer [
27,
28]. Accordingly, the neutrophil-to-lymphocyte ratio (NLR) is commonly used as an inflammatory biomarker in cancer patients, and a higher NLR is commonly associated with poor prognosis in patients with breast cancer [
29].
It is well known that CDK4/6 inhibitors modulate the absolute numbers and composition of circulating immune cells [
30]; however, their predictive value has not been determined. We hypothesized that the dynamic changes in the immune cell populations induced by CDK4/6 inhibitors could reflect the resulting antitumor response and may be predictive of patient survival. To test this hypothesis, we analyzed a multicenter retrospective cohort of palbociclib-treated patients with HR-positive, HER2-negative ABC as well as patients of prospective phase III studies (PALOMA-2 and PALOMA-3). The composition and characteristics of immune cells in the peripheral blood of patients being treated with CDK4/6 inhibitors were evaluated using flow cytometry analysis of T cell and myeloid-derived suppressor cell (MDSC) markers to garner further mechanistic insights.
Patients and methods
Exploratory cohort
Patients from the metastatic breast cancer database registry of two tertiary referral centers (161 patients from Yonsei Cancer Center and 60 from Gangnam Severance Hospital) in the Republic of Korea who received palbociclib and letrozole as first-line treatment for HR-positive, HER2-negative ABC between January 2017 and April 2020 constituted the exploratory cohort. The exploratory cohort included premenopausal and postmenopausal patients. All premenopausal patients underwent bilateral salpingo-oophorectomy before palbociclib and letrozole treatment because the national health insurance policy did not allow the use of luteinizing hormone-releasing hormone agonists with palbociclib and letrozole in premenopausal patients. Baseline patient demographics were collected, including age, initial stage, recurrence type, disease-free interval, metastatic site and number, treatment outcome, and survival data. This study was approved by the institutional review board of each institution (IRB number: 4-2016-0574), and informed consent was obtained from all patients.
Validation cohort
Results from the exploratory cohort were validated using the double-blind multicenter phase III PALOMA-2 study (NCT01740427) [
9]. In this study, women with HR-positive, HER2-negative ABC were randomly assigned to receive 2.5 mg letrozole orally continuously in combination with either palbociclib (125 mg/day orally for 3 weeks followed by 1 week without drug administration in a 4-week cycle) or a matching placebo. Postmenopausal women with no prior systemic therapy for advanced disease, an Eastern Cooperative Oncology Group (ECOG) performance status of 0–2, and adequate organ function were eligible for enrollment. The primary study end-point was investigator-assessed progression-free survival (PFS), defined as the time from randomization to radiologically confirmed disease progression or death during the study. Women with an adequate follow-up period (> 29 days) and available leukocyte and neutrophil counts at cycle 2 day 1 with a window period of 7 days were analyzed (
N = 619; 410 patients from the palbociclib- and letrozole-treated groups and 209 patients from the placebo and letrozole-treated groups). Lymphocyte counts were not collected in the PALOMA-2 study case report. Patients treated with palbociclib (125 mg/day orally for 3 weeks followed by 1 week without drug administration in a 4-week cycle) and fulvestrant (500 mg intramuscularly every 2 weeks for the first three injections and then every 4 weeks) in the PALOMA-3 study (NCT01942135) were also analyzed [
12]. Both PALOMA-2 and PALOMA-3 study data were obtained from Pfizer at the investigators’ request through the Vivli platform (
https://vivli.org).
Peripheral immune cell counts
Leukocyte, neutrophil, and lymphocyte counts in peripheral blood were collected at baseline, cycle 1 day 15, and cycle 2 day 1 of treatment. The NLR was calculated as neutrophil count/lymphocyte count. The derived NLR (dNLR) was calculated as neutrophil count/(leukocyte count—neutrophil count) and was used as an NLR surrogate because the lymphocyte count was unavailable in the PALOMA-2 and PALOMA-3 validation cohort.
Next-generation sequencing (NGS)
Tumor samples were collected and targeted DNA and RNA sequencing were performed using TruSight Tumor 500 (Illumina, San Diego, CA, USA). DNA and RNA from formalin-fixed paraffin-embedded (FFPE) tissue using a Qiagen All Prep DNA/RNA FFPE kit (Qiagen, Hilden, Germany). Library preparation was performed according to the manufacturer’s instructions. After hybridization capture-based target enrichment, pair-ended sequencing (2 × 150 bp) was performed using a NextSeq sequencer (Illumina) according to the manufacturer’s instructions as previously described [
31].
Blood-based immune monitoring with flow cytometry
We prospectively collected peripheral blood at baseline, cycle 1 day 15, and cycle 2 day 1 in an independent prospective cohort of 20 patients treated with CDK4/6 inhibitors for ABC from September 2020 to April 2021 at the Yonsei Cancer Center. Immune cells were isolated as previously described [
32]. Single-cell suspensions were incubated with the LIVE/DEAD Fixable Near-IR Dead Cell Stain kit (Thermo Fisher Scientific) and then stained with fluorochrome-conjugated antibodies for 15 min at room temperature. For intracellular staining, cells were permeabilized and fixed with the FoxP3 staining buffer kit (Thermo Fisher Scientific) and further stained with fluorochrome-conjugated antibodies. Flow cytometry analysis was conducted using an LSR Fortessa system (BD Biosciences) and FlowJo software version 10.4.0 (Tree Star Inc.). To characterize CD8
+ T cell exhaustion, the relative frequency of PD-1
+ cells and mean fluorescent intensities of TOX, NFATc1, and CTLA-4 in PD-1
+CD8
+ T cells were measured based on previously published articles [
33,
34]. Relative ratios of MDSCs and T cells, numbers of monocytic MDSCs (M-MDSC; CD14
+CD11b
+CD15
−HLA-DR
−CD3
−CD19
− cells), or polymorphonuclear MDSCs (PMN-MDSC; CD15
+CD11b
+CD14
−HLA
−DR
−CD3
−CD19
− cells) were obtained by dividing those of T cells (CD3
+CD19
−CD14
− cells) based on previously defined gating strategies [
35]. This study was approved by the institutional review board of Severance Hospital (IRB number: 4–2020-0472). Informed consent was obtained from patients. The reagents used for flow cytometry are described in Additional file
1: Table S1.
Statistical analysis
Statistical analyses were performed using SPSS version 25 (IBM), R version 3.5.3. (R Foundation for Statistical Computing), and GraphPad Prism (GraphPad Software). Chi-squared and Mann–Whitney U tests were used to compare patient characteristics between two groups. Paired t-tests were used for pair-wise comparisons before and after treatment. Mean values with standard deviations or ranges are presented for continuous variables. Correlations between variables were analyzed based on a linear regression model. The log-rank maximization method was used to find the best dNLR cut-off values for PFS in the exploratory cohort, the PALOMA-2 validation cohort treated with letrozole plus placebo, and the PALOMA-3 exploratory cohort using the MaxStat R package. Clinical benefit was defined as complete response, partial response, or stable disease lasting more than 24 weeks. PFS was defined as the time interval from the initiation of palbociclib treatment to radiologically confirmed disease progression or patient death, and overall survival (OS) was defined as the time interval from the initiation of palbociclib treatment to patient death in the exploratory cohort. Patient survival was monitored until November 3, 2020. Survival differences among the patient groups were compared using log-rank tests and depicted using Kaplan–Meier plots. A Cox proportional hazards regression model was used to estimate the hazard ratio, and multivariate analysis was conducted to adjust variables associated with patient outcomes. In the PALOMA-2 validation cohort, multiple log-rank tests for PFS comparing letrozole plus palbociclib therapy with letrozole monotherapy were conducted in palbociclib-treated patients with increasing dNLR cut-offs to investigate the correlation between palbociclib-driven PFS benefit and on-treatment dNLR.
Discussion
CDK4/6 inhibitors have been established as a standard treatment for HR-positive, HER2-negative ABC, demonstrating clinical efficacy and safety in a series of clinical trials. Previous studies proposed that CDK4/6 inhibitors elicit robust antitumor immunity and that the response to CDK4/6 inhibitors essentially relies on their immunologic effects [
40]; however, the role of dynamic immune biomarkers in predicting CDK4/6 inhibitor benefits had not been addressed. In this study, we demonstrated that the on-treatment (d)NLR was significantly reduced by CDK4/6 inhibitors and specifically predicted the benefits of letrozole plus palbociclib treatment. Mechanistically, we showed that pharmacologic inhibition of CDK4/6 promoted antitumor immunity by attenuating T cell exhaustion and MDSC abundance. This study sheds new light on the value of immune monitoring during treatment with CDK 4/6 inhibitors in patients with HR-positive, HER2-negative ABC.
The benefit of CDK4/6 inhibitors is consistent across patients with various clinical characteristics [
9,
11,
20,
41]. In the PALOMA-3 study, cyclin E1 expression was associated with resistance to palbociclib combined with fulvestrant and with diminished antiproliferative effects after palbociclib treatment [
13]. In the PALOMA-2 study, expression of receptor tyrosine kinases (e.g., FGFR2 or HER3) predicted palbociclib benefit [
14]. In addition, various alterations (e.g., loss of
RB1, estrogen receptor expression, and activating mutations in
AKT1,
RAS,
AURKA,
CCNE2,
ERBB2, and
FGFR2) mediate both intrinsic and acquired resistance to palbociclib [
42]. Mutations in
RB1,
PIK3CA, and
ESR1 resulted in acquired resistance to palbociclib in the PALOMA-3 study [
43]; however, these biomarkers were heterogeneous, explaining only a small portion of resistance mechanisms without a clear delineation between prognostic and predictive implications. In addition, next-generation sequencing was employed to explore the value of these biomarkers, which largely limited their usefulness and expandability in routine clinical practice. Therefore, apart from hormone receptor positivity, there are no clinically available biomarkers to identify patients who would benefit from CDK4/6 inhibitor treatment despite a clear clinical need.
In this study, we showed that palbociclib induced changes in immune cell composition, leading to a reduction in the dNLR after treatment. In addition, on-treatment neutrophil and lymphocyte counts were associated with palbociclib benefit in the opposite direction, necessarily giving predictive significance to the on-treatment (d)NLR in the exploratory cohort. The versatility of dNLRs to predict clinical outcomes was validated using data from letrozole plus palbociclib-treated patients enrolled in the PALOMA-2 study. Importantly, the PFS of patients treated with letrozole plus placebo was independent of the on-treatment dNLR value, indicating that the on-treatment dNLR is a predictive factor for palbociclib outcome rather than a simple prognostic factor. We also obtained concordant results in patients treated with fulvestrant and palbociclib in the PALOMA-3 study, although the on-treatment dNLR cut-off value was different due to marginally different on-treatment dNLR value between the patients enrolled in each study (Additional file
11: Figure S10A), probably can be attributed to patients’ disease status and exposure to previous therapies [
44]. Additionally, our finding that CDK4/6 inhibitors modified the characteristics and numbers of lymphocytes and myeloid cells further supports the notion that response to CDK4/6 inhibitors depends on immune-modulating effects. Taken together, our on-treatment biomarkers showed strengths in general with independent external validation, were convenient and less invasive than current methods, and exhibited biological validity that reflects systemic immune responses. In line with our results, Zattarin et al. also reported that on-treatment NLRs and platelet-to-lymphocyte ratios after three cycles of treatment were associated with PFS [
45].
In triple-negative and HER2-positive breast cancer, a high degree of tumor-infiltrating lymphocytes (TILs) is correlated with a low risk of relapse and with the effectiveness of chemotherapy [
46‐
48]. However, increased TIL density is associated with shorter OS in HR-positive breast cancer patients [
49], indicating a different underlying mechanism. Although HR-positive breast cancer is considered less immunogenic [
50,
51], CDK4/6 inhibitors magnify immune recognition and the reaction of host immune cells toward tumor cells [
52]. Of note, CDK4/6 inhibitors regulate PD-L1 stability in tumor cells [
53], augment cytotoxic T cell activation [
21], and attenuate suppressive functions of regulatory T cells [
22], all of which contribute to robust antitumor immunity with reinforced host immune surveillance. These studies suggest that the mechanism of action of CDK4/6 inhibitors may involve mediation by host immune responses toward tumors. Our findings that CDK4/6 inhibitors prevented T cell exhaustion phenotypes via reducing PD-1, NFATc1, TOX, and CTLA-4 expression levels and MDSC numbers demonstrate that high dNLRs reflect active immune suppression with progressive T cell exhaustion in the tumor microenvironment, which is largely alleviated by CDK4/6 inhibitor treatment. A prospective study with larger cohorts of patients would help to evaluate further whether the on-treatment dNLR is a useful biomarker for predicting patient response to CDK4/6 inhibitor treatment. The utility of on-treatment dNLRs as a tool to measure systemic antitumor immune responses will be an interesting area of future study [
54].
Our study provides a framework for future clinical trials and important clues for the interpreting several pivotal clinical trials. The PFS on palbociclib and letrozole therapy was superior to letrozole monotherapy regardless of on-treatment dNLR in our PALOMA-2 analysis, suggesting that (d)NLR status is not a determinant to select endocrine monotherapy or endocrine plus CDK4/6 inhibitor therapy. However, we think on-treatment dNLR can serve as important biomarker for drug selection and patient stratification for various clinical situations. In this regard, whether the (d)NLR can be used as a predictive marker for the benefit of adjuvant treatment with CDK4/6 inhibitors should be evaluated further. Recently, adjuvant trials with two different CDK4/6 inhibitors (NCT03155997 [monarchE] and NCT02513394 [PALLAS]) have shown conflicting results, highlighting the importance of identifying subgroups of patients that derive clinical benefits from CDK4/6 inhibitors. Our study revealed that patients with extremely high on-treatment dNLRs (≥ 2.00) did not derive benefit from adding palbociclib to letrozole treatment (Additional file
8: Figure S7A-S7B), although the number of these patients was small (17 of 410 patients treated with palbociclib and letrozole, 4.1%). Thus, our on-treatment biomarker may be essential tool for identifying these patient subgroups. Second, whether dose modification of CDK4/6 inhibitors has any effect on survival outcomes should be investigated. Dose reduction of CDK4/6 inhibitors is recommended in grade 3/4 hematologic adverse events, making it challenging to identify the dose–response relationship. Complicating infection accompanied by febrile neutropenia is rarely seen during treatment with CDK4/6 inhibitors, and continuing palbociclib treatment is safe in patients experiencing grade 3 neutropenia [
55]. Thus, whether maintaining or increasing the dose of CDK4/6 inhibitors confers clinical benefits requires further investigation in future clinical trials because the NLR on cycle 2 day 1 may reflect the pharmacodynamic effects of palbociclib on bone marrow hematopoietic stem cells in addition to its immune-modulating effects. The third aspect of interest to future research is examining whether cytokines for T cell expansion (such as recombinant IL-7 or IL-15) restore lymphopenia and improve treatment responses to CDK4/6 inhibitors, because CDK4/6 inhibitor-related lymphopenia was associated with poor outcomes in our study. In addition, whether immune checkpoint inhibitor plus CDK4/6 inhibitor combinations elicit meaningful treatment responses should be investigated, especially in patients with HR-positive ABC, for whom several early phase studies are underway (NCT02778685, NCT02779751, and NCT02791334). We anticipate that these studies will show the clinical efficacy of these combined therapies, highlighting the immunomodulatory capacity of CDK4/6 inhibitors. Accordingly, we expect that on-treatment dNLR may guide optimal treatment sequence in HR-positive, HER2-negative ABC. Patients with high on-treatment (d)NLR is predicted to show poor response to CDK4/6 inhibitors, and they would be better candidates for early switching to cytotoxic chemotherapy and other novel therapies including targeted agents, antibody drug conjugates, or immunotherapy combined with CDK4/6 inhibitors. Lastly, the potential predictive value of on-treatment (d)NLR after CDK4/6 inhibitor therapy in terms of long-term OS needs to be explored. Considering that the results from long-term OS in three phase III studies (PALOMA-2, MONALEESA-2, and MONARCH-3) have been recently reported to be inconsistent, the influence of on-treatment (d)NLR on long-term OS outcome would give insights for more precise stratification of the patients.
The strengths of this study include the exploration and validation of clinically accessible biomarker for HR-positive, HER-2-negative ABC patients treated with CDK4/6 inhibitors, which was not observed in patients treated with only letrozole. In addition, immune correlates were explored along with translational relevance. Limitations to this study include the imbalances of baseline characteristics of the exploratory and validation cohort. Absence of lymphocyte counts in the validation cohort led us to use dNLR instead of NLR. Relatively small number of patients with immunologic analysis should be validated in a large scale with prospective manner. The statistical significance of on-treatment dNLR in the validation cohort was not as pronounced compared to that witnessed in the exploratory cohort.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.