Introduction

Cyclin-dependent kinases (CDKs) play an important role in cell cycle physiology. In vitro studies of CDK4/6 inhibitors (CDK4/6i) in human breast cancer cell lines showed hormone positive cell lines to be the most sensitive; moreover, there was synergy when combined with tamoxifen [1, 2]. This work led to the phase 2 PALOMA-1 trial, where the CDK4/6i palbociclib improved progression-free survival (PFS, 20.2 vs. 10.2 months) when added to antiestrogen therapy in patients with untreated hormone receptor (HR)-positive, human epidermal growth factor receptor 2 (HER2)-negative metastatic breast cancer (MBC). As a result, the FDA-approved palbociclib in 2015 for use in combination with antiestrogen therapy as first line treatment in HR-positive, HER2-negative MBC [3]. The three currently available CDK4/6is, palbociclib, ribociclib, and abemaciclib, are all FDA-approved in combination with aromatase inhibitors and with the selective estrogen receptor degrader, fulvestrant.

CDK4/6is are well tolerated and adverse events are usually managed with dose modification and supportive care measures. Neutropenia, the dose-limiting toxicity for palbociclib and ribociclib, was the most common adverse event reported in landmark trials [3,4,5,6,7,8,9,10,11,12]. In PALOMA-2, 80% of patients in the palbociclib group developed neutropenia, with 66% of events being grade 3/4 [4]. Similar results were seen in the MONALEESA trials which studied the use of ribociclib, while the MONARCH trials generally reported lower rates of neutropenia with abemaciclib [6, 8,9,10,11]. CDK4/6i-induced neutropenia is rapidly reversible, reflecting a cytostatic effect on bone marrow precursors, and is not associated with an increased risk for infection. Abemaciclib commonly exhibits gastrointestinal toxicities, whereas neutropenia is less evident due to its stronger selectivity for CDK4 compared to CDK6 [13, 14]. CDK6 regulates cytokine expression in hematopoiesis [15], contributes to the “control” of myeloid progenitor expansion, and plays many important roles in myeloid differentiation [16]. Most data about CDK4 and CDK6 are derived from total knockout mouse models. Maurer et al. generated transgenic mice that lack either CDK4 or CDK6 in adult hematopoiesis. Deletion of CDK6 affected all stem cell fractions and led to neutropenia, while deletion of CDK4 resulted in elevated numbers of myeloid progenitors without translating into numeric changes of differentiated myeloid cells [17]. Neutrophils comprise most circulating leukocytes and serve a critical antimicrobial role [18]. Asymptomatic reductions in peripheral blood neutrophil counts are observed in up to 10–30% of individuals of African descent [19]. In the USA, Blacks have lower neutrophil counts than Whites (mean difference 0.83 × 109 cells/L) and higher rates of neutropenia (4.5% vs. 0.79%) [20, 21].

Most patients enrolled in the CDK4/6i landmark trials were White, and data on the toxicity and efficacy in Black patients are limited. The number of Black patients enrolled in the PALOMA-2, PALMOMA-3, MONALEESA-3, and MONALEESA-7 trials were 11 (3.2%), 29 (6%), 5 (1.4%), and 19 (6%), respectively [4, 5, 8, 9, 12]. Given this unmet need, we aimed (1) To assess the impact of CDK4/6i on neutrophil counts in Black and Non-Black patients, and (2) To compare the efficacy measured by PFS, of CDK 4/6i between Black and Non-Black patients with MBC.

Methods

Study design and data collection

We conducted a single-center retrospective study at Montefiore Medical Center. Using the institutional clinical software, “Clinical Looking Glass”, we identified patients with HR-positive, HER2-negative MBC who were prescribed a CDK4/6i as first or subsequent line therapy between January 1st, 2015 and April 28th, 2020. We excluded patients with inaccessible medical records, patients whose race was not available, patients who were lost to follow up, patients who did not take the CDK4/6i, those treated at outside institutions, and those who took the CDK4/6i for less than 14 days. Baseline characteristics including age, race, ethnicity, prior treatment lines for metastatic disease, tumor grade, visceral involvement, and menopausal status were collected through chart review. The Standard Charlson Comorbidity Index was calculated for each patient. Medication compliance and adherence were carefully tracked through diligent review of clinical notes. Patients who were not adherent to CDK4/6is as documented in clinicians’ notes were excluded from the study.

Race was defined as Black, White, Asian, or Other. Complete blood count data were collected at cycle 1, day 1 (C1D1); cycle 1, day 14 (C1D14); cycle 2, day 1 (C2D1); cycle 2, day 14 (C2D14); cycle 3, day 1 (C3D1); cycle 4, day 1 (C4D1); cycle 5, day 1 (C5D1); and cycle 6, day 1 (C6D1). All relevant oncology clinic and inpatient notes were reviewed and scrutinized for information regarding treatment response and adverse events to CDK4/6i.

Statistical analysis

Baseline characteristics were summarized using descriptive statistics. To assess associations between two categorical variables, we used the Pearson’s Chi-square test or Fisher’s exact test. The Wilcoxon rank sum test was used to compare neutrophil counts at each time point between Black vs. Non-Black patients. Change in neutrophil count over time was compared using the Wilcoxon signed rank sum test.

Toxicities were recorded and graded using the Common Terminology Criteria for Adverse Events (CTCAE) version 4.0. The number of dose reductions/delays were also recorded. The Pearson’s Chi-square test was used to compare adverse event rates between Black vs. Non-Black patients.

PFS was calculated as the time in days between the CKD4/6i initiation date and the date of disease progression or death. Data was censored at the date each patient was last seen at our institution. The Kaplan–Meier method was used to estimate PFS, and the log-rank test was used to compare PFS between groups. To assess the effect of neutropenia and dose reduction on PFS, a landmark analysis was conducted excluding patients who died or had disease progression within 28 days of CDK4/6i initiation. Multivariable Cox Proportional-Hazards model, adjusting for age, visceral disease, menopausal status, prior endocrine therapy (ET) lines, prior chemotherapy lines, and neutropenia, was used to compare the PFS between Black and Non-Black patients. Statistical analysis was conducted using STATA version 14.0 and R statistical software. This study was approved by the Albert Einstein Institutional Board Review.

When accounting for missing data, we started by excluding patients who had missing data for race, since this was our main variable of interest. We conducted detailed chart review to obtain most values for collected variables. For the evaluation of ANC over time in Black vs. Non-Black patients, we encountered 10–15% missing data for some ANC values. It was deemed that values were missing completely at random and there was no indication that any missing data were systematic. Therefore, we removed records with missing data to conduct statistical analysis of ANC at certain time points. To assess disease progression, we obtained information from imaging and clinical notes; only 1 patient had missing data for PFS analysis.

Results

Baseline characteristics

A total of 233 patients were identified. Of those, 45 were excluded for the following reasons: 11 were treated at outside institutions, 4 were lost to follow up, 13 never took the prescribed CDK4/6i, 2 had HER2 + disease, 6 had inaccessible medical records, 8 were treated with CDK4/6i for less than 14 days, and 1 patient did not have stage IV disease. Of the 188 remaining evaluable patients, an additional 6 were excluded as there were no data available on their race (Fig. 1). Our final cohort included 182 patients with a median age of 64 years [interquartile range (IQR), 54–72]. There were 61 Hispanic (34%) and 121 non-Hispanic (66%) patients. Most patients (77%) were postmenopausal and 67% had visceral metastases. Palbociclib, ribociclib, and abemaciclib were prescribed in 153 (84%), 9 (5%) and 20 (11%) patients, respectively, and this was similar between racial groups. In most circumstances, CDK4/6is were prescribed at the FDA-approved starting dose. Letrozole, fulvestrant, and anastrozole were prescribed in 88 (48%), 68 (37%), and 19 (10%) patients, respectively. The most common reason for CDK4/6i discontinuation was progression of disease (94 patients, 82%). Neutropenia was responsible for drug discontinuation in only 2 patients (2%), (Table 1).

Fig. 1
figure 1

Consort diagram: A total of 233 patients were initially identified and 51 patients were excluded, leaving 182 patients in the final analysis

Full size image
Table 1 Baseline characteristics
Full size table

In our cohort, 83 patients (46%) were Black, 39 (21%), 10 (5%), and 50 (27%) patients were White, Asian, and Other race, respectively. For the purpose of comparison, we consolidated all the Non-Black patients into one group. Baseline characteristics were similar between Black and Non-Black: median age (64 vs. 63 years, p = 0.47), postmenopausal status (75% vs. 80%, p = 0.67), and visceral metastases (67% vs 67%, p > 0.99). CDK4/6is were used in the first line for metastatic disease in 48% of patients in both groups. The endocrine therapy backbones were also similar between Black and Non-Black (p = 0.67), (Table 1).

Neutrophil and white blood cell count at baseline and over time

The baseline (C1D1) median absolute neutrophil count (ANC) was lower in Black vs. Non-Black (3.0 vs. 4.0 × 109/L, p = 0.001), with similar results at C3D1(1.40 vs. 1.75 × 109/L, p = 0.03). However, there was no significant difference in median ANC between Black and Non-Black at C1D14 (1.15 vs. 1.30 × 109/L, p = 0.07), C2D1 (1.2 vs. 1.3 × 109/L, p = 0.40), and C2D14 (1.65 vs. 1.5 × 109/L, p = 0.38) (Table 2). When the change in ANC from baseline was calculated (delta-ANC), Black patients showed a smaller decrease in median ANC over time compared to Non-Black patients. Delta-ANC at C2D1 was − 1.5 vs. − 2.6 × 109/L (p = 0.001), for Black vs. Non-Black, respectively, and delta-ANC at C3D1 was − 1.5 vs. − 2.4 × 109/L (p = 0.007) for Black vs. Non-Black, respectively (Table 2; Fig. 2). White blood cell count (WBC) showed a similar pattern. The baseline WBC (C1D1) was lower in Black vs. Non-Black (5.5 vs. 6.6 × 109/L, p = 0.001). Black patients also experienced a smaller decrease in WBC over time compared with Non-Black patients. Delta-WBC at C2D1 was -1.95 vs. − 3.3 × 109/L (p = 0.003) and delta-WBC at C3D1 was − 1.9 vs. − 2.9 × 109/L (p = 0.005) for Black vs. Non-Black, respectively (Table 4; Figure 3 see appendix).

Table 2 Neutrophil count over time in black vs. non-black cohorts
Full size table
Fig. 2
figure 2

Neutrophil count over time while on treatment with CDK4/6 inhibitors in Black vs. Non-Black patients. Abbreviations: C1D1: Cycle 1 Day 1; C1D14: Cycle 1 Day 14; C2D1: Cycle 2 Day 1; C2D14: Cycle 2 Day 14; C3D1: Cycle 3 Day 1; C4D1: Cycle 4 Day 1; C5D1: Cycle 5 Day 1; C6D1: Cycle 6 Day 1

Full size image

Neutropenia

Most patients (86%) experienced neutropenia, with grade 1, 2, 3, and 4 neutropenia reported in 10%, 24%, 43%, and 8% of patients, respectively (Table 5; see appendix). The rates of grade 1, 2, 3, and 4 neutropenia were 11%, 17%, 51%, 12% and 10%, 29%, 37%, 5%, in Black vs. Non-Black, respectively (Table 5; see appendix). There was a statistically significant difference in neutropenia rates between Black vs. Non-Black (p = 0.04).

Infections

Infections were seen in 32 patients (18%), with no difference in the rate of infection between Black vs. Non-Black (13% vs. 21%, p = 0.22) (Table 5; see appendix). Of note, 3 Black patients (4%) and 7 Non-Black patients (7%) experienced more than one infection. Among Black patients, there were a total of 14 infections [pneumonia (n = 4), urinary tract infection (n = 2), bacteremia (n = 2), febrile neutropenia (n = 2), and other (n = 4)]. In Non-Black patients, there were a total of 28 infections [urinary tract infection (n = 6), upper respiratory infection (n = 5), pneumonia (n = 3), osteomyelitis (n = 3), skin infection (n = 6), and other (n = 5)]. Grade 3/4 infections were present in 10 Black patients and 11 Non-Black patients (Table 6; see appendix).

Other toxicities

There were no differences in the rates of any thrombocytopenia (45% vs. 49%) or grade 3/4 thrombocytopenia (4% in each group), between racial groups (p = 0.42; Table 5; see appendix). There was also no statistically significant difference in the rate of anemia for Black vs. Non-Black (48% vs. 56%, p = 0.25). Grade 3 anemia was present in 11% vs. 5% in Black vs. Non-Black, (Table 5; see appendix). Commonly reported gastrointestinal toxicities were nausea, diarrhea, and decreased appetite. Grade 3 fatigue was seen in 2% of Black and 1% of Non-Black patients (p = 0.44), and grade 3 diarrhea was seen in 1% of both cohorts (p = 0.19) (Table 5 see appendix).

Dose reductions and delays

At least one dose reduction was required in 53 patients (29%) (Table 7; see appendix). Black patients required more dose reductions, but this difference was not statistically significant (34% vs. 25%, p = 0.27). One dose reduction was required in 22 Black and 17 Non-Black patients, whereas more than one dose reduction was required in 6 Black and 8 Non-Black patients. Dose delay was required in 61 patients (34%); 29 patients required one dose delay, while 32 patients required more than 1 dose delay. There was no difference in the rate of dose delay between Black vs. Non-Black (36% vs. 31%, p = 0.59). Both a dose reduction and delay were required in 26% of patients, with no difference between Black and Non-Black (33% vs. 21%, p = 0.11). In our entire cohort, 36% of patients required either a dose reduction or dose delay, with similar rates in Black vs. Non-Black (37% vs. 35%, p = 0.90) (Table 7; see appendix).

CDK4/6 inhibitor discontinuation

CDK4/6i was discontinued in 114 patients, and the most common reason was progression of disease (n = 94). Among Black patients, the most common reasons for CDK 4/6i discontinuation were progression of disease (n = 47), other (n = 8), infection (n = 1), and neutropenia (n = 1). Among Non-Black patients, the most common reasons for discontinuation were progression of disease (n = 47), other (n = 8), infection (n = 1), and neutropenia (n = 1) (Table 1).

Progression-free survival

There was no difference in median PFS in Black vs. Non-Black (316 vs. 407 days, p = 0.51) (Table 3, Figure 4; see appendix). Among patients who received CDK4/6i as first line treatment, the median PFS for Black vs. Non-Black was 390 and 518 days (p = 0.48), respectively (Table 3, Figure 7; see appendix). We stratified breast cancer outcomes by neutropenia, to evaluate whether the drug’s effects on the bone marrow could predict disease response. Among patients who developed neutropenia, the median PFS was 336 vs. 421 days (p = 0.45) for Black and Non-Black, respectively (Table 3, Figure 5; see appendix). In patients who did not develop neutropenia, the median PFS was 141 vs. 259 days (p = 0.41) for Black and Non-Black, respectively (Table 3, Figure 8; see appendix).

Next, we assessed the effect of dose reductions on breast cancer outcomes by racial groups. Among patients who required a dose reduction, the median PFS was 312 vs. 627 days (p = 0.26) for Black vs. Non-Black (Table 3, Figure 6; see appendix). While in those without dose reduction, the median PFS was 346 vs. 377 days (p = 0.91) for Black vs. Non-Black, respectively (Table 3, Figure 9; see appendix).

In a univariate analysis, the use of prior endocrine therapy and prior chemotherapy for metastatic disease (p = 0.01 for both) and presence of visceral disease (p = 0.04) were statistically significantly associated with PFS. Black race (p = 0.51), neutropenia (p = 0.09), and age (p = 0.27) were not associated with PFS (Table 8; see appendix). We then created a multivariable analysis using Cox Proportional-Hazards model including Black race, neutropenia, age, visceral status, prior endocrine therapy for metastatic disease, and prior chemotherapy for metastatic disease. In this model, the use of prior endocrine therapy for metastatic disease was inversely associated with PFS (HR 1.87, 95% CI 1.22–2.87, p < 0.01), and the presence of visceral metastasis was inversely associated with PFS, with a trend towards statistical significance (HR 1.57, 95% CI 0.95–2.60, p = 0.08) (Table 8; see appendix).

Table 3 Median PFS in Black vs. Non-Black Cohorts
Full size table

Discussion

Our study included 182 patients with HR-positive/HER2-negative MBC treated with CDK4/6i in a racially diverse cohort (Black = 83, Non-Black = 99). We report that Black patients have a lower baseline ANC (3.0 vs. 4.0 × 109/L, p = 0.001), and smaller decreases in ANC over time compared to Non-Black patients (delta-ANC at C2D1: − 1.5 vs. − 2.6 × 109/L, p = 0.001) while on CDK4/6is. A similar trend was observed for the total WBC count. Most patients did develop neutropenia while on treatment, and although there was a statistically significant difference in neutropenia rates between Black vs. Non-Black, there were no differences in the infection rates between racial groups. Further, the rates of dose reductions and delays were similar between groups. We found no difference in the median PFS between Black vs. Non-Black (316 vs. 407 days, p = 0.51), and this was not affected by the presence or absence of neutropenia (Figure 4, Table 3; see appendix). Of note, the PFS results should be interpreted with caution, given the number of patients included in each analysis. Further, our multivariable analysis did not show an association between PFS and Black race or neutropenia. These data support the notion that the lower baseline ANC seen in Black patients does not negatively impact survival or toxicity outcomes during CDK4/6i treatment. Our results also suggest that baseline ANC alone may be a poor indicator of infection risk during CDK4/6i therapy.

Benign ethnic neutropenia (BEN), an entity where an ANC < 1.5 × 109/L is seen in certain populations, is most common in people of African descent [22]. Although the etiology has not been elucidated, studies have identified a chromosome 1q22 polymorphism containing the Duffy antigen and receptor for chemokine (DARC) gene which strongly influences leukocyte counts in African Americans and is one potential mechanism [23]. BEN is not associated with an increased risk for infection or febrile neutropenia. Despite this, a proportion of patients with BEN may be inadvertently excluded from oncology clinical trials (which typically have an eligibility criteria of ANC ≥ 1.5 × 109/L for inclusion), or have unnecessary dose interruptions/reductions, due to their lower than normal baseline ANC [22, 24]. Our study is the first to evaluate racial disparities in toxicities and breast cancer outcomes for patients treated with CDK4/6i for MBC, and has the largest proportion of Black patients (46%; N = 83) reported in the literature in this clinical scenario.

Prior studies have evaluated racial disparities in neutrophil counts in breast cancer patients receiving chemotherapy. Smith et al. found no racial differences in the frequency of febrile neutropenia in breast cancer patients receiving chemotherapy, despite a lower baseline ANC in Black patients [25]. A retrospective analysis by Hershman et al. found that Black patients had a lower baseline WBC prior to starting adjuvant chemotherapy but a similar mean percentage decline in WBC from baseline to treatment completion compared to White patients [26]. In another study analyzing differences in WBC counts and dose delays/discontinuations between Black and White breast cancer patients receiving adjuvant chemotherapy from two clinical trials, the rates of neutropenic fever were similar between racial groups despite a lower baseline WBC/ANC in Black patients [27]. These studies call into question the clinical relevance of a lower baseline WBC and ANC that may be seen in Black patients, and whether clinical trial inclusion criteria should more frequently be liberalized. Vastola et al. explored the ANC inclusion criteria of prostate cancer clinical trials and found that 41.4% of trials excluded patients with BEN, arguing that Black patients may be disproportionately excluded from cancer clinical trials due to benign racial laboratory variations [28]. Hsieh et al. argues that rigid eligibility criteria for WBC in cancer clinical trials may cause undue exclusion of Black patients from participation [22]. Our work provides additional evidence that clinical trial inclusion criteria for ANC should be reexamined, to provide equitable and evidence-based opportunities for participation.

Inclusion criteria for most CDK4/6i landmark trials required an ANC ≥ 1.5 × 109/L, which may have impacted the underrepresentation of Black patients in these studies [4, 7, 11, 12, 29]. The phase 2 single arm PALINA trial analyzed the hematologic safety of palbociclib in combination with letrozole or fulvestrant in self-reported African American, African, or Black women with HR-positive/HER2-negative MBC [29]. The ANC inclusion cutoff was 1.0 × 109/L. The median baseline ANC was 3.1 × 109/L, and 35 patients were enrolled. No patients experienced febrile neutropenia or required drug discontinuation due to neutropenia. Lower baseline ANC (2.4 vs 4.3 × 109/L, p = 0.006), grade 3 neutropenia (66.7% vs. 23%, p = 0.029), and dose reductions (55.6% vs. 7.7%, p = 0.008) were more common in patients found to have the Duffy null polymorphism [29]. Interestingly, our study reports similar conclusions as the PALINA trial, that CDK4/6is may be safely administered in Black patients even if they have a lower baseline ANC.

Our study has certain limitations. Given its retrospective nature, information on medication compliance was limited to review of clinical notes. We encountered missing data points, which were subsequently excluded from statistical analysis and potentially limited the magnitude of the results. Despite careful attempts to reduce missing data, it is possible that systematic bias was introduced by inclusion of only those without missing data. There is also the possibility of selection bias, as perhaps only the compliant patients were included in this retrospective analysis. Our sample size is indeed smaller than those studied in landmark CDK4/6i clinical trials. However, although small, our study includes the highest proportion of Black patients for analysis of toxicities and PFS of CDK4/6i ever reported in the literature.

This is the first study to elucidate that despite lower baseline neutrophil counts seen in Black patients, there are no differences between racial groups in the frequency of infections, dose reductions, or dose delays seen with CDK4/6i use in MBC. Based on these results, the lower baseline neutrophil counts that may be seen in Black patients should not cause concern or hesitancy when initiating CDK4/6i, as they may not be clinically relevant.

Conclusion

In our cohort, Black patients had a lower baseline ANC compared to Non-Black patients, but they experienced less of a decline in ANC from baseline (delta-ANC) during treatment with CDK4/6i. Our results are the first to examine racial disparities in toxicities of patients receiving CDK4/6i. Our data suggest that CDK4/6is can be safely administered to patients who may have BEN, since their baseline ANC does not necessarily translate into increased rates of infections, dose reductions, or dose delays.