A comprehensive literature search was conducted in the Cochrane, PubMed, Embase, and Web of Science databases from their inception until August 1st. The objective was to gather literature pertaining to the use of CDK4/6 inhibitors in combination with ET for breast cancer. Both subject terms and free terms were employed in each database. Search phrases included “breast cancer," “Cyclin-Dependent Kinases 4 and 6 Inhibitors,” “Systematic review,” “Meta-analysis,” and the searches were limited to English (Supplementary Data S1).

Inclusion criteria were based on the PICOS (participant, intervention, comparison, outcome, and study type) framework (P) Participants were breast cancer patients of any race, age, or disease stage (I/C) the trial group received CDK4/6 inhibitors combined with ET, while the control group received ET alone or with placebo (O) outcomes measured included AEs, with data such as risk difference (RD), relative risk (RR), odds ratio (OR), hazard ratio (HR) (S) Study types were meta-analyses (MAs) and systematic reviews (SRs) comprising exclusively RCTs.

Exclusion criteria were: (a) duplicate publications; (b) articles that did not report the necessary data; (c) systematic review reevaluation plans, conference abstracts, etc.

Two reviewers (WY and XDB) independently conducted the screening and data extraction process. Initially, duplicate titles were removed, followed by screening of titles and abstracts, and then full-text evaluation based on the inclusion and exclusion criteria. Data extracted included first author, publication year, country, study population, sample size, interventions/control measures, outcome measures, and quality assessment methods. Any disagreements were resolved through the consensus of a third reviewer (CHH).

Following the Joanna Briggs Institute guidelines for Umbrella Reviews, we conducted a descriptive analysis of AEs for CDK4/6 inhibitors in combination with ET for breast cancer, without reanalyzing data from RCTs or MAs/SRs (Aromataris et al., 2015). We summarized indicators for CDK4/6 inhibitors, including RD, RR, OR, HR, 95% confidence interval (95% CI), and p-value. I² was utilized to assess study heterogeneity, with statistical significance set at a p-value <0.05.

Two independent evaluators (WY and XDB) performed assessments using the Assessment of Multiple Systematic Reviews 2 (AMSTAR-2) scale, PRISMA statement, and the Grades of Recommendation, Assessment, Development, and Evaluation (GRADE) instrument and resolved disagreements by consensus of a third-party reviewer (Supplementary Data S2). First, the AMSTAR-2 scale, a systematic evaluative methodological quality assessment tool, was used to assess the study’s methodological quality (Shea et al., 2009; Shea et al., 2017). The AMSTAR-2 scale, a tool for assessing methodological quality, includes 16 items scored as “yes,” “no,” or “partially yes,” based on the criteria fulfillment. The PRISMA 2020 checklist was employed for assessing reporting quality, comprising 27 items (42 sub-item levels) (Hutton et al., 2015). Each item was scored as 1 for complete reporting, 0.5 for partial, and 0 for non-reporting, with a total possible score of 42. Scores of 33–42 indicated relatively complete reporting, 25–32 indicated some deficiencies, and scores below 25 signified serious information deficiencies (Page et al., 2021). The GRADE tool evaluated evidence quality for each outcome by examining risk of bias, inconsistency, indirectness, precision, and publication bias, categorizing outcome indicators as high, moderate, low, or very low based on downgrades (Schunemann et al., 2020; Zeng et al., 2021).

From four database searches, 425 pieces of literature were initially retrieved. After removing duplicates, screening titles and abstracts, and evaluating full texts, 24 studies met the inclusion criteria. The literature screening process and results are depicted (Figure 1). Publications spanned from 2017 to 2021, with study numbers ranging from 3 to 9 and sample sizes from 1,352 to 12,647. Regarding the included population, two studies focused on HR+/HER2- EBC(Agostinetto et al., 2021; Gao et al., 2021), with the remainder addressing HR+/HER2- ABC. In terms of therapeutic interventions, one study (Ramos-Esquivel et al., 2020) compared CDK4/6 inhibitor combined with fulvestrant versus fulvestrant alone, two studies (Ramos-Esquivel et al., 2018; Shimoi et al., 2020) compared CDK4/6 inhibitor with AI versus AI alone, and the rest involved CDK4/6 inhibitor combined with ET versus ET treatment. For study outcomes, 11 studies reported pooled AE outcomes, such as all-grade AEs, grade 3/4 AEs, and grades 3–5 AEs (Deng et al., 2018; Martel et al., 2018; Messina et al., 2018; Ramos-Esquivel et al., 2018; Ramos-Esquivel et al., 2020; Shimoi et al., 2020; Zheng et al., 2020; Agostinetto et al., 2021; Gao et al., 2021; Li et al., 2021; Tian et al., 2021). The other studies reported specific AE outcomes without pooled AE data (Table 1).

AMSTAR-2, an extensive critical appraisal tool, facilitates rapid and reproducible assessments of systematic reviews of randomized controlled trials (RCTs) for interventions, thereby identifying high-quality evaluations for decision-makers (Shea et al., 2017). We utilized the AMSTAR-2 scale to appraise the methodological quality of the included studies. Two studies were evaluated as low quality (Agostinetto et al., 2021; Yang et al., 2021), while the remainder were deemed very low quality. For specific AMSTAR-2 criteria, entries 1, 3, and 11 achieved 100% compliance, indicating all included PICO, specified the types of literature included, and employed appropriate statistical methods for combined result analysis. Conversely, entries 7, 8, 10, and 14 had compliance rates of 0%, 13%, 8%, and 17%, respectively, highlighting significant gaps in justifying the inclusion of study types, listing and explaining excluded literature, detailing study characteristics, reporting funding sources, and elucidating result heterogeneity (Table 1; Figure 2, and Supplementary Data S3).

PRISMA 2020 serves as a reporting guideline for systematic reviews of health intervention studies, regardless of study design. Its comprehensive reporting allows readers to evaluate the methodological soundness and credibility of study results (Page et al., 2021). We applied the PRISMA checklist to assess the reporting quality of the 24 included papers, which scored between 23.5 and 36. Two reports, scoring from 33 to 42, were relatively complete (Agostinetto et al., 2021; Gao et al., 2021). Nineteen reports, scoring from 25 to 32, exhibited some deficiencies. Three reports (Lasheen et al., 2017; Lee et al., 2019; Li et al., 2021) scoring less than 25 had serious deficiencies. Common omissions included item 7 (detailing the search strategy for databases), 13f (conducting sensitivity analysis in synthesis methods), 15 (assessing certainty), 16b (explaining literature exclusion reasons in results), 20d (providing sensitivity analysis in synthesis results), 22 (providing evidence certainty in results), 23c (discussing any limitations in the review process), and 24abc (providing registration and protocol-related information) (Figure 3; Table 1 and Supplementary Data S4).

Patients treated with CDK4/6 inhibitors, compared to ET alone, experienced a higher incidence of AEs. The main AE categories were distributed across nine areas (n = number of MAs): summary AEs (n = 14), hematological AEs (n = 15), gastrointestinal AEs (n = 13), systemic AEs (n = 11), liver AEs (n = 2), circulatory AEs s (n = 2), skin AEs (n = 5), pain-related AEs (n = 5) and other types AEs (n = 2). Notably, hematological and gastrointestinal AEs were more prevalent, with frequent reports of leucopenia, neutropenia, thrombocytopenia, anemia, diarrhea, nausea, and vomiting. Among systemic AEs, reports of fatigue were substantial (Figure 4).

GRADE provides a framework for authors of systematic reviews and health technology assessments to rate the certainty of their evidence (Zeng et al., 2021). This study summarized 13 AE outcomes for two EBCs and 158 AE outcomes for 22 ABC studies. The evidence quality for EBC was predominantly low (4 items, about 30.77%) and very low (9 items, about 69.23%), as assessed using GRADE. The highest level of evidence in ABC was for the venous thromboembolism (VTE) outcome (Thein et al., 2020), Moderate quality evidence was found in five studies (Deng et al., 2018), Tian Q (Tian et al., 2021), Xu ZH (Xu et al., 2020), Li J (Li et al., 2020), Toss A (Toss et al., 2019) addressing various AEs including all-grade AEs, grade 3/4 AEs, leukopenia, grade 3/4 leukopenia, neutropenia, grade 3/4 neutropenia, anemia, grade 3/4 anemia, grade 3/4 diarrhea, and nausea, among others (20 items, about 12.66%). The remaining evidence was categorized as low quality (70 items, about 44.30%) and very low quality (67 items, about 42.40%). Overall, the reduced quality of evidence was largely attributed to inconsistency, risk of bias, and publication bias (Supplementary Data S5).

Our review identified 24 MAs and SRs evaluating AEs in 13 EBC cases and 158 ABC cases. Our findings indicated that AEs associated with CDK4/6 inhibitors spanned nine disease areas, including summary AEs, hematologic AEs, gastrointestinal AEs, systemic AEs, liver AEs, circulatory AEs, skin AEs, pain-related AEs, and other AEs. In HR+/HER2- EBC patients, CDK4/6 inhibitor addition significantly correlated with all-grade AEs and grade 3/4 AEs, and notably increased early treatment discontinuation risk. Subgroup analysis showed that EBC AEs were independent of whether palbociclib or abemaciclib was used. For HR+/HER2- ABC patients, combined therapy significantly elevated AEs across all grades, including grade 3/4 leukopenia, neutropenia, anemia, diarrhea, nausea, constipation, fatigue, pyrexia, VTE, abdominal pain, and cough. However, their safety profile was considered acceptable, with evidence quality mostly high to medium. Subgroup analysis of CDK4/6 inhibitor types in ABC revealed that palbociclib was associated with greater hematologic toxicity, especially grade 3/4 neutropenia, anemia, and thrombocytopenia, while ribociclib was linked mainly to grade 3/4 neutropenia, QT interval prolongation, and alopecia. Abemaciclib is closely associated with diarrhea and elevated blood creatinine.

With the increasing use of CDK4/6 inhibitors combined with ET in first and second-line clinical settings, anticipating AEs' risk, timely diagnosis, and management are pivotal in enhancing patients' quality of life (Martel et al., 2018). Despite a high incidence of hematologic, gastrointestinal, and systemic AEs in both EBC and ABC, most were reversible and manageable. Neutropenia was the most common hematologic toxicity, particularly with palbociclib and ribociclib, but it led to febrile neutropenia or infection at much lower rates than chemotherapy. The mechanism of CDK4/6 inhibitors involves reversible blocking of neutrophil precursor cycles, inhibiting proliferation, which is reversible upon discontinuation and tends to lessen over time (Kassem et al., 2018). Thus, clinical applications require monitoring of complete blood counts and timely intervention through discontinuation, dose adjustment, or symptomatic treatment based on individual safety and tolerability (Thill and Schmidt, 2018). Gastrointestinal AEs, such as diarrhea and nausea, especially with abemaciclib, did not significantly increase serious gastrointestinal risk. This may be attributed to abemaciclib’s CDK9 inhibitory effect (Marra and Curigliano, 2019). As diarrhea is typically short-lived and of low severity, it can be managed through dose adjustment or antidiarrheal medications like loperamide. Fatigue, significantly increased in any grade and grade 3/4 AEs, poses a challenge in systemic AEs due to its vague and multidimensional nature, making identification of contributing factors and mitigation measures difficult (Lasheen et al., 2017).

Significantly, the findings related to nausea and vomiting associated with CDK4/6 inhibitors were inconsistent across different AE categories. This inconsistency may stem from variations in the types of CDK4/6 inhibitors used in clinical trials, their modes of administration, differences in study subjects, and the range of clinical trials included in the SRs. Consequently, further research is necessary to reach definitive conclusions. Additionally, we noted substantial heterogeneity in some outcomes. Fundamental aspects of the included studies, such as study population, design, intervention/control measures (dose), follow-up activities, analysis procedures, and outcome indicators, were not adequately detailed, and limited raw data might have contributed to this heterogeneity.

The clinical selection of the three CDK4/6 inhibitors presents a challenging issue, primarily due to the absence of distinct clinical case characteristics, predictive biomarkers, and comparable clinical benefits in PFS and OS(Agostinetto et al., 2021; Gao et al., 2021; Munzone et al., 2021; Falato et al., 2023). A subgroup analysis focusing on CDK4/6 inhibitor types in ABC aimed to address this issue. Palbociclib was associated with increased neutropenia, anemia, and thrombocytopenia, while ribociclib was linked mainly to neutropenia, prolonged QT interval, and alopecia. Abemaciclib was strongly correlated with diarrhea and elevated blood creatinine. This suggests that patients with ABC who have hematological disorders or tendencies may be better suited for abemaciclib, while those with gastrointestinal disorders and renal dysfunction or tendencies might be more appropriate for ribociclib or palbociclib. Furthermore, patients with cardiac disorders or tendencies might be more suitable for ribociclib or abemaciclib. This could be a significant opportunity to improve patient quality of life, advance clinical decision-making accuracy, and achieve personalized precision medicine.

To our knowledge, this is the first to use an umbrella review of studies to summarize CDK4/6 inhibitors combined with ET for HR+/HER2-breast cancer AEs. This approach mitigates the bias inherent in individual MAs/SRs and enhances the accuracy of study outcomes. Second, we included MAs/SRs that used only RCT study types and excluded literature that used single-arm studies, and non-randomized prospective, retrospective, and observational studies as study types, which further reduced the interference of subjective factors and improved the scientific validity of the findings. In Additionally, this review employed the recently published PRISMA 2020 guidelines for reporting quality assessment (Sohrabi et al., 2021), a significant update over the PRISMA 2009 guidelines typically referenced in most MAs/SRs (Shea et al., 2009). The PRISMA 2020 guidelines offer improvements in areas like data items, data synthesis methods, study outcome selection, and results synthesis. These updated guidelines facilitate the generation of high-quality evidence that can support clinical decision-making and practice (Page et al., 2021).

This review, however, is not without limitations. Firstly, despite employing the latest versions of AMSTAR-2, PRISMA 2020, and GRADE for quality assessment of the selected literature, it is important to note that all three scales are inherently subjective, which could introduce bias to our findings. To mitigate this, two researchers independently conducted the assessments, with any differences resolved through consensus by a third-party reviewer (CHH), thereby aiming to maximize the accuracy of our evaluation. Moreover, the MAs/SRs included in this study were based on global multicenter RCTs, potentially leading to publication bias and a diminished quality of evidence due to the limited number of RCTs.