Clinical and demographic characteristics and treatment patterns of patients with HR+ HER2– locally advanced or metastatic breast cancer receiving CDK4/6 inhibitors in real-world clinical practice in Moscow: data from the ICEDORA study

Introduction

Breast cancer (BC) is the most common cancer among women in Russia. In 2023, 74,008 new cases of breast cancer were diagnosed, accounting for 19.1% of all detected malignancies. According to published data, approximately 7.4% of women with newly diagnosed breast cancer have distant metastases, and about 17% of breast cancer cases are diagnosed at stage III [1].

Randomized clinical trials (RCTs) have demonstrated significant benefits of using CDK4/6 inhibitors in combination with endocrine therapy (ET) — fulvestrant or nonsteroidal aromatase inhibitors (NSAIs) — as first and subsequent lines of therapy in patients with hormone receptor-positive (HR+), human epidermal growth factor receptor 2-negative (HER2–) advanced breast cancer (ABC) in terms of progression-free survival (PFS) (ribociclib, palbociclib, and abemaciclib) and overall survival (OS) (ribociclib and abemaciclib) [2–12].

Data from RCTs provide compelling evidence of the clinical efficacy of CDK4/6 inhibitors. When conducting retrospective studies based on real-world clinical practice (RCP) data, multiple challenges arise, including mismatches between patients' clinical and demographic characteristics and RCT inclusion criteria, as well as imbalance in comparison groups. However, RCP studies are highly valuable because they provide data on a broader and more diverse patient population, including patients with comorbidities typically excluded from RCTs. These studies demonstrate the drug’s effectiveness in less controlled settings, i.e., the same real-world clinical environment in which nearly all patients receive treatment across different countries [13–19, 30–31].

Analysis of retrospective data from the ICEDORA study provided, for the first time in Russia, extensive information on the use of combination ET in RCP in a large patient population, and also allowed analysis of patient characteristics, treatment regimens, and clinical outcomes.

Materials and Methods

Study goals and objectives. ICEDORA is a non‑interventional, retrospective study of patient characteristics, treatment approaches, and clinical outcomes in patients with HR+ HER2– locally advanced breast cancer or metastatic breast cancer (mBC) receiving ribociclib, palbociclib, or abemaciclib in real‑world clinical practice in Moscow.

The primary objective of the study was to describe the clinical‑demographic profile and pharmacotherapy approaches in patients receiving CDK4/6 inhibitors in routine clinical practice. The key secondary objective was to retrospectively assess the effectiveness of ribociclib in patients with HR+ HER2– advanced or metastatic breast cancer in RCP. The analysis did not compare the efficacy of treatment options due to incomparability of patient groups.

Study design. The ICEDORA study included patients with HR+ HER2– ABC and mBC who received CDK4/6 inhibitors from January 2020 through the end of December 2022. Because the study was observational and retrospective, it did not include a treatment protocol, diagnostic or therapeutic interventions, a visit schedule, or the collection of written informed consent from patients. The study included all patients from the Moscow Oncology Society registry with HR+ HER2– advanced breast cancer who received CDK4/6 inhibitor therapy during the index period; the total study population size was 2,051 patients. In this study, the index date is the date of initiation of the study treatment.

During the study, available data on key demographic characteristics, treatment responses, treatment sequence, and treatment duration were collected. The following data were collected: age at diagnosis of ABC or mBC, sensitivity or resistance to ET, ECOG performance status at the start and end of CDK4/6 inhibitor therapy, primary tumor characteristics (progesterone receptor (PR) and estrogen receptor (ER) levels, HER2/neu, Ki‑67), TNM stage at diagnosis, location and number of distant metastases, presence of comorbid diagnoses, prior treatment options (surgery, radiotherapy (RT), chemotherapy (CT)), concomitant ET options, time from diagnosis to initiation of CDK4/6 inhibitor therapy, lines of therapy, duration of CDK4/6 inhibitor therapy, time to progression (including while on CDK4/6 inhibitors), time from diagnosis to confirmation of first metastases, death, and causes of death. Adverse event data were not collected.

Patient age at the time of breast cancer diagnosis was recorded. To describe PR/ER levels on tumor cells, a scoring method was used with the following scoring scheme: 0 (negative), 1–3 (low level, may range up to intermediate values, partially positive), 4–8 (high level, positive).

Functional status dynamics were assessed using the ECOG scale: 0 (fully active, able to carry on all pre‑disease activities without restriction), 1 (restricted in physically strenuous activity but ambulatory and able to carry out work of a light or sedentary nature), 2 (ambulatory and capable of all self‑care but unable to carry out any work activities; up and about more than 50% of waking hours), 3 (capable of only limited self‑care, confined to bed or chair more than 50% of waking hours), 4 (completely disabled, cannot carry on any self‑care, totally confined to bed or chair).

Statistical analysis. Treatment duration and overall survival (OS) were analyzed using the Kaplan–Meier product‑limit method. Survival curves were compared using the log‑rank test. No stratification factors were included in the analysis. Median OS was calculated from the date of initial breast cancer diagnosis to death from any cause. The date of diagnosis was the date recorded in the patient's medical record, and the date of death was the date registered with the civil registry office. Results were presented as mean survival in years, median survival in years (the time by which 50% of patients in the respective treatment group had survived), and 95% confidence interval (95% CI) for the median survival.

Comparison of frequency values of quantitative variables between groups was performed using the chi‑square test or the Fisher–Freeman–Halton test, depending on the number of observations in the contingency table. Quantitative data were compared between groups using the Kruskal–Wallis test.

Ethics approval. The study design was approved by the Local Ethics Committee of the Loginov Moscow Clinical Scientific Center of the Moscow Department of Health (extract from Protocol No. 5/2023 dated April 26, 2023).

Results

Clinical and demographic characteristics of patients. The clinical and demographic characteristics and data on the number of study participants receiving CDK4/6 inhibitors (ribociclib, palbociclib, abemaciclib) are presented in Table 1. The majority of patients received palbociclib (1,204; 58.7%) or ribociclib (712; 34.7%). Abemaciclib was used in the treatment regimen in only 135 patients (6.6%). This imbalance in treatment group sizes reflects Russian RCP and is due to the fact that palbociclib was approved by the Ministry of Health of the Russian Federation for use in Russia earlier than the other CDK4/6 inhibitors. It is important to note that from a methodological and analytical perspective, the nine‑fold difference in the size of the palbociclib and abemaciclib groups, as well as the more than five‑fold difference between the ribociclib and abemaciclib groups, lead to different reliability of estimates for these groups, complicating the statistical and clinical interpretation of the results.

The median age of patients at the time of breast cancer stage determination was 58 years (Table 1). The age of patients in the ribociclib and palbociclib groups was almost identical: the median was 58 years for both groups. In the abemaciclib group, the median age was higher at 62 years.

For the vast majority of patients, the stage of the disease at diagnosis was known (Table 1). The overall population was dominated by patients with stage IV (42.1%), approximately equal numbers of patients had stage II or III (26.1% and 23.8%, respectively), and 160 patients (7.8%) were diagnosed with stage I breast cancer. Thus, 42% of patients were diagnosed with de novo metastatic breast cancer, while the remainder progressed to mBC after initial detection of early‑stage breast cancer. Intergroup differences in stage distribution did not reach statistical significance (p=0.078, chi‑square test), although it should be noted that the abemaciclib group had a predominance of later stages. Regarding distant metastases (M) identified in de novo breast cancer or as a result of disease recurrence at an earlier stage, the differences between groups reached statistical significance (p=0.021, chi‑square test) due to a higher frequency of de novo metastatic breast cancer in the abemaciclib group (51.9%) compared to the ribociclib (43.5%) and palbociclib (40.1%) groups. This result may reflect prescribing patterns in which abemaciclib is more often chosen for patients with de novo mBC.

Overall, the stage distribution and observed trends reflect the pattern characteristic of the disease in general.

In the majority of patients (Table 1), negative human epidermal growth factor receptor 2 (HER2/neu) status was recorded (83.5% in the overall population), for 308 patients (15.0%) information on HER2/neu expression was missing, and 31 (1.5%) had a positive status. The ribociclib, palbociclib, and abemaciclib groups were homogeneous with respect to HER2/neu status (p=0.592, Fisher's exact test).

Data on progesterone receptor (PR) and estrogen receptor (ER) levels were unavailable for more than half of the patients in the overall population (63.5% and 61.4%, respectively). Among the remaining patients, 22% had low PR levels, 10.1% had high PR levels, and 4.4% of patients had a negative result. No intergroup differences were found for progesterone receptor levels (p=0.340, chi‑square test). For estrogen receptors (ER), 22.2% of patients had low levels, 14.9% had high levels, and 1.5% had a negative result. Statistically significant differences were observed between the study drug groups (p=0.005, Fisher–Freeman–Halton test) due to a significantly higher proportion of patients with high ER levels in the abemaciclib group (27.4%) compared to the ribociclib and palbociclib groups (16.4% and 12.5%, respectively). However, due to limited information on ER and PR levels, extrapolation of these data to the overall sample should be done with caution.

Data for assessing the proliferation marker Ki‑67 were available for 1,471 out of 2,051 patients. In the overall population, the mean Ki‑67 value was 33.8%, corresponding to a high proliferation level, and the median was 28% (threshold for moderate proliferation). In the abemaciclib group, the Ki‑67 index (mean 39.2%, median 35%) was statistically significantly higher (p=0.029, Kruskal–Wallis test) than in the ribociclib group (mean 33.7%, median 29%) and the palbociclib group (mean 33.2%, median 27%).

Among the overall population of patients with recorded metastases, in the vast majority of cases they were found in one (18.4%), two (13.8%), three (9.0%), or four (4.9%) locations. Metastases in 5–8 locations were found in less than 5% of women. When comparing the study drug groups, statistically significant differences in the number of metastases between groups were recorded (p=0.021, Fisher–Freeman–Halton test), primarily due to a significantly higher proportion of patients with one metastasis (31.9%) in the abemaciclib group compared to the ribociclib (19.0%) and palbociclib (16.6%) groups. The most common locations of distant metastases were bone (41.1%), lung (17.7%), lymph nodes (16.8%), and liver (14.1%) (Table 1). Statistically significant differences between the study drug groups were found only for peritoneal metastases: in the palbociclib group, the frequency of such metastases was almost two times lower (1.7%) than in the ribociclib and abemaciclib groups (3.2% and 3.7%, respectively; p=0.04, Fisher–Freeman–Halton test). In addition, there was a trend toward a higher frequency of distant lymph node metastases in the abemaciclib group (23.7%) compared to the ribociclib and palbociclib groups (17.3% and 15.8%, respectively).

The majority of patients included in the study had comorbidities (Table 1). Overall, comorbidity was recorded in almost 90% of patients, with no statistically significant differences between groups.

Table 1. Clinical and demographic characteristics of patients at the time of initial diagnosis of breast cancer

Notes: ^a 0 points – no expression (negative result); 1-3 points – low expression, may range from 0 to 30% (partially positive result); 4-8 points – high expression (positive result); ^b – at diagnosis; ^c – treatment received prior to initiation of study drug therapy (CDK4/6 inhibitors).

* Patients may have had multiple comorbidities, so the overall proportion exceeds 100%.
** In 6 women, pregnancy was confirmed after initiation of therapy and was registered according to standard procedures.

Dynamics of ECOG performance status. Analysis of ECOG data was performed only for patients whose functional status was known both before starting CDK4/6 inhibitors and during or after taking the study drugs. As a result, the samples for the analysis of ECOG status dynamics included 129 patients from the ribociclib group, 190 patients from the palbociclib group, and 30 patients from the abemaciclib group. While receiving CDK4/6 inhibitors, a slight deterioration in functional status was observed (p <0.05, Fisher's exact test), mainly due to patients moving from ECOG 0 to ECOG 1 (Fig. 1), which is directly related to disease progression. In none of the groups was there a statistically significant increase in the proportion of patients with significant functional status limitations. There were no differences between the ribociclib, palbociclib, and abemaciclib groups in ECOG functional status either before starting the drugs (p=0.585, Fisher's exact test) or during/after treatment (p=0.114, Fisher's exact test).

Table 2 shows the distribution of patients by direction of change in ECOG status (no change, deterioration, improvement), as well as by presence or absence of status change regardless of direction. No statistically significant differences were found either in the direction of change (p=0.796, Fisher's exact test) or in its presence or absence (p=0.725, chi-square test). On average, 45% of patients had no change in ECOG status, another 45% of patients showed deterioration in functional status, and the remaining 10% showed improvement. The median change (Table 2) in all groups was +1 point, which corresponds to a change toward increased functional limitations and is consistent with the data that the dynamics in all groups are mainly associated with the transition of patients from ECOG category 0 to ECOG category 1.

Fig. 1. Dynamics of ECOG performance status in the ICEDORA study

Table 2. Change in ECOG performance status in the ICEDORA study

Notes: ^# Difference in scores in patients with change (negative values indicate a decrease in ECOG score, i.e., improvement in functional status).

Treatment before CDK4/6 inhibitors. Adjuvant radiotherapy was prescribed with approximately equal frequency in all groups (Table 1). Although the frequency of radiotherapy use in the abemaciclib group was numerically lower, no statistically significant differences from the other groups were observed (p=0.172, chi-square test). For surgical treatment performed before study entry, on the contrary, statistically significant differences were observed between groups (p=0.024, chi-square test), due to a lower rate of prior surgical treatment in the abemaciclib group (43.7%). This fact is related to the higher frequency of de novo metastatic breast cancer in the abemaciclib group, which precluded surgical treatment. The frequency of surgical treatment in the ribociclib and palbociclib groups was comparable: 52.8% and 55.7%, respectively.

The most common components of chemotherapy regimens were cyclophosphamide (21.3%), doxorubicin (18.2%), and paclitaxel (15.7%) (Table 3), reflecting current clinical practice guidelines for chemotherapy of primary metastatic or progressive breast cancer [20-24]. No significant differences were found between groups or drug positions. It can be noted that the proportion of patients receiving chemotherapy or combination therapy was somewhat lower in the abemaciclib group (51.9%) than in the ribociclib (56.9%) and palbociclib (56.2%) groups, which may be associated with earlier initiation of CDK4/6 inhibitor therapy (due to more patients with de novo mBC).

Table 3. Breast cancer medications used before starting CDK4/6 inhibitors in the ICEDORA study

Notes: N is the total number of patients in the group; NOE is the number of drug prescription episodes; n is the number of patients who had at least one episode; % is (n/N) × 100.

Hormonal therapy combined with CDK4/6 inhibitor treatment. Table 4 shows the frequency of prescription of aromatase inhibitors, antiestrogens, and gonadotropin agonists as part of combination therapy with CDK4/6 inhibitors. Since each drug could be prescribed to a patient in multiple courses, the table presents both the number of patients who received the corresponding prescription at least once (n) and the total number of episodes (NOE) of prescription of the corresponding drug in the group. Due to this data structure, direct statistical comparison between groups is difficult, so the results are presented descriptively. Among female sex hormone antagonists, anastrozole (55.0% in the overall population) and fulvestrant (54.4%) were prescribed most often, while letrozole was prescribed somewhat less frequently (40.5%). Among gonadotropin‑releasing hormone agonists, patients most often received goserelin (12.8%) and much less frequently triptorelin (4.5%). Compared to the ribociclib and palbociclib groups, the abemaciclib group received anastrozole (66.7%) and fulvestrant (31.1%) much less frequently. Overall, 98.4% of all patients received hormone therapy concomitant with CDK4/6 inhibitor treatment, reflecting the requirements of the prescribing information for CDK4/6 inhibitors.

Table 4. Hormonal therapy drugs concomitant with CDK4/6 inhibitors in the ICEDORA study

Notes: N is the total number of patients in the group; NOE is the number of drug prescription episodes; n is the number of patients who had at least one episode; % is (n/N) × 100.

CDK4/6 inhibitor therapy. The mean time from breast cancer detection to initiation of CDK4/6 inhibitor therapy in the overall population was 4.5 years (median 2.2 years) (Table 5). The Kruskal‑Wallis test revealed statistically significant differences between groups (p <0.001). In the abemaciclib group, the mean time was 3.0 years (median 0.5 years), which is significantly shorter than in the ribociclib (mean 4.3 years, median 2.1 years) and palbociclib (mean 4.7 years, median 2.8 years) groups. The substantially lower median values indicate a predominance in the overall population of patients receiving CDK4/6 inhibitor treatment within the first three years after breast cancer diagnosis. In more than half of the cases (56.2%) in the overall sample, CDK4/6 inhibitors were used in the first line of therapy (Table 5). Figure 2 shows the duration of CDK4/6 inhibitor therapy. The median durations of ribociclib and palbociclib administration were comparable at 17.6 and 15.9 months, respectively, whereas for abemaciclib it was 7.7 months (Table 5).

Table 5. Key patterns of CDK4/6 inhibitor therapy and its duration in the ICEDORA study

Notes: *kw – Kruskal‑Wallis, chisq – chi‑square.

Fig. 2. Time to discontinuation of CDK4/6 inhibitor therapy (Kaplan‑Meier method)

Progression and mortality. Breast cancer progression was recorded in 28.7% (n=588) of the overall population: in 60.7% (n=82) of patients in the abemaciclib group, and in 29.4% (n=209) and 24.7% (n=297) of patients in the ribociclib and palbociclib groups, respectively (Table 6). Since the study population included patients with initially diagnosed early, advanced, and metastatic breast cancer (Table 1), these disease progression data represent a composite measure influenced by various factors (average disease severity, treatment strategy, etc.). These disease progression data in different treatment groups cannot be attributed to the studied CDK4/6 inhibitors and cannot be compared with each other.

While on CDK4/6 inhibitors, progression occurred in 13.3% (n=160) of the overall population: 6.8% (n=5) in the abemaciclib group, 12.5% (n=54) in the ribociclib group, and 14.5% (n=101) in the palbociclib group, with no statistically significant differences between drug groups (p=0.146) (Table 6).

Table 6. Progression and mortality in the ICEDORA study

Notes: * Tests: kw – Kruskal‑Wallis, chisq – chi‑square, f – Fisher's exact test. N/A – not applicable (see explanation in text).

In total, out of 2,051 patients, 823 (40.1%) had died at the time of analysis. The lowest mortality rate was recorded in the abemaciclib group (23.0%), followed by the ribociclib group (37.4%), and the highest rate was in the palbociclib group (43.7%). Breast cancer as the cause of death (in patients' discharge summaries after death) was indicated in 33.2% of deceased patients in the overall population. The retrospective nature of the study does not allow reliable determination of the nature of disease progression to death. Mortality rates differed between treatment groups: 31.6% of patients in the ribociclib group, 35.6% in the palbociclib group, and 20% in the abemaciclib group (Table 6). Given the previously identified differences in age, frequency of metastases, and receptor levels upon histological tumor typing, these differences may be related to the characteristics of the respective groups. Further research using multivariate analysis is required to confirm this result, especially given the divergent nature of the data obtained.

Median OS, calculated from the time of breast cancer diagnosis to patient death, was 11.9 years (95% CI: 9.5–13.6) in the ribociclib group, 10.1 years (95% CI: 9.4–12.2) in the palbociclib group, and 12.6 years (95% CI: 8.4–not reached) in the abemaciclib group. It is important to note that the OS measure in this case cannot reflect the therapeutic effect of the study drugs, and direct comparison between treatment groups is not possible for several reasons. First, the groups were not balanced, so the influence of confounding factors cannot be excluded. Second, calculating OS from the time of diagnosis rather than from the start of CDK4/6 inhibitor therapy does not allow comparison of the results with RCT data or assessment of the direct contribution of the study drugs to survival, since this measure is largely dependent on disease history and prior treatment. Furthermore, survival data in the abemaciclib group should be interpreted with particular caution, as there is a high probability of random errors due to the small sample size.

Discussion

The advent of CDK4/6 inhibitors has radically changed the approach to the treatment of advanced and metastatic HR+ HER2– breast cancer, becoming the backbone of first‑line therapy [20–24]. Although RCTs [2–12; 25–28] have convincingly demonstrated their efficacy, direct comparisons of these drugs in real‑world clinical practice are still limited. Therefore, the choice of a specific CDK4/6 inhibitor in everyday medical practice is often determined by factors such as drug availability, side effect profile, comorbidities, and individual patient tolerance. Thus, the clinical and demographic characteristics of the patient play a key role in the choice of CDK4/6 inhibitor.

In the ICEDORA study, the overall population was dominated by patients with stage IV disease (de novo mBC) (about 42.1%), whereas the proportion of patients with relapse of stage I (at initial breast cancer diagnosis) was only 7.8%. Within each group, the stage distribution was similar, with a significant predominance of later stages, highlighting the importance of analyzing the effectiveness of CDK4/6 inhibitors in patients with ABC in real‑world clinical practice in Russia. The majority of patients had comorbidities such as diabetes mellitus, arterial hypertension, circulatory system diseases, and others. This indicates a significant number of comorbid conditions among study participants. If the disease progressed from earlier stages, patients underwent various types of treatment before receiving CDK4/6 inhibitors, including surgery, neoadjuvant and adjuvant chemotherapy. Radiotherapy was used relatively uniformly across all groups.

The greatest similarity was found between the groups of patients receiving ribociclib and palbociclib, while patients receiving abemaciclib differed significantly in their clinical and demographic characteristics (Table 1). For example, patients in the abemaciclib group were statistically significantly older than patients in the other two groups, had a greater number of patients with later stages of the disease, and were more often diagnosed with de novo metastatic breast cancer. Consequently, the frequency of prior surgical treatment in this group was significantly lower than in the ribociclib and palbociclib groups. In addition, the abemaciclib group had the highest levels of ER and Ki‑67, the largest number of patients with distant metastases, and the shortest time from diagnosis to initiation of CDK4/6 inhibitor therapy. The proportion of patients receiving chemotherapy was lower than in the ribociclib and palbociclib groups, which may be related to the larger number of patients with de novo mBC and earlier initiation of CDK4/6 inhibitor therapy. The duration of CDK4/6 inhibitor therapy in the abemaciclib group was significantly shorter than in the ribociclib and palbociclib groups (Table 5). In addition, aromatase inhibitors (anastrozole) were prescribed more often in the abemaciclib group, and SERD (fulvestrant) less often (Table 4). This difference may be related to clinical practice patterns or individual patient characteristics. Thus, the aforementioned features form the profile of patients in the abemaciclib group, which is clearly different from the other two groups.

Another significant difference between treatment groups that complicates statistical analysis and clinical comparison of the results is the number of patients in the treatment groups. The size of the palbociclib group (N=1,204) was almost twice the size of the ribociclib group (N=712) and almost nine times the size of the abemaciclib group (N=135). The difference in sample size is related to the different registration dates of CDK4/6 inhibitors for use in the Russian Federation. Palbociclib was approved in October 2016, ribociclib in January 2018, and abemaciclib in August 2019 [29].

Strengths of this analysis include the diversity of patient groups represented and the completeness of data on patients receiving CDK4/6 inhibitors in real‑world clinical practice in Moscow.

Study Limitations

This study has several limitations. It was a retrospective database analysis with the potential risk of treatment selection bias and the presentation of inaccurate or incomplete data. No causal relationships were established. The definition of OS used in this study differed from that used in RCTs and most RCP studies, resulting in the inability to compare the obtained data with RCT results. In addition, some subgroups may be limited by insufficient sample size. Finally, the results of this study may not be applicable to patient groups whose clinical and demographic characteristics differ from those presented in this study.

Conclusion

The ICEDORA study is the largest Russian analysis of clinical and demographic characteristics, treatment patterns, and clinical outcomes in patients with HR+ HER2– ABC and mBC receiving CDK4/6 inhibitors in real‑world clinical practice in Moscow.

This analysis emphasizes that if a comparison of the three CDK4/6 inhibitors is necessary, well‑planned prospective studies that can ensure homogeneity of patient groups and sufficient statistical power are advisable. However, given that no controlled clinical trials are currently being conducted to directly compare the three CDK4/6 inhibitors, the results of this analysis may be useful in the broad population of patients with HR+ HER2– ABC and mBC receiving CDK4/6 inhibitors in real‑world clinical practice in Russia.