To our knowledge, no previous systematic reviews and meta-analyses have evaluated the available evidence regarding the efficacy and safety of CDK4/6i treatment in MBC patients previously treated with another CDK4/6i. Therefore, a specific protocol for a new systematic review and meta-analysis was developed. The present protocol has been registered in the PROSPERO database (registration number: CRD42022330355) and is reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses Protocols (PRISMA-P) statement (11).
Ethics
No ethical approval is required for the performance of this work.
Study identification/information sources
PubMed and Web of Science will be searched from their inception date to 15 July 2022. In addition, the reviewers will perform a manual search of gray literature, including the San Antonio Breast Cancer Conference, American Society of Clinical Oncology annual conference, European Society of Medical Oncology annual conference, and European Society of Medical Oncology Breast annual conference. Finally, the reviewers will examine the citation lists of the reviewed documents to identify possible additional articles of relevance.
Search strategy
Our PubMed search strategy is the following:
(abemaciclib OR ribociclib OR palbociclib OR cyclin-dependent kinase inhibitor OR cyclin-dependent kinase inhibitor OR iCDK OR CDKi OR CDK inhibitor OR CDK4/6i OR CDK4/6 inhibitor) AND (Rechallenge OR previous OR prior OR after OR retreatment OR readministration OR restart OR resume OR reinduction OR reintroduction OR continuing) AND (breast cancer OR breast malignancy OR breast neoplasm).
The reviewers will adapt this search strategy for the Web of Science database.
Study selection
Studies will be included in this systematic review if they meet the following eligibility criteria:
- Human studies.
- ER+/HER2- MBC patients previously treated with CDK4/6i now undergoing CDK4/6 inhibition treatment with another CDKi.
- Any observational retrospective or prospective study and any RCT will be eligible for inclusion.
- Reporting of a hazard ratio (HR) for OS, PFS, and/or survival curves, allowing estimation of the HR for OS or PFS and/or reporting of ORR, DCR, and/or AEs.
- English language publication.
- Other criteria: there will be no limitations on the year of publication.
Studies will be excluded in this systematic review if they meet the following criteria:
- Age group below 18
- Letter to editors, review articles, case studies, non-human studies.
- Studies without full accessibility.
- Duplicate studies.
- Articles written in other languages, not English.
Six reviewers (JBA, RS, MA, CF, ALL, CT) will independently screen the titles and abstracts of these studies for potential inclusion. Potentially eligible studies will be confirmed by an evaluation of the full text. A third-party investigator (FM or JA) will resolve any uncertainty or discrepancy. References of all articles included or excluded at the full-text review stage will be entered into EndNote 20.2. Details of the study screening and selection process will be shown in Figure 1.
Data Extraction
A standardised data extraction template for the study will be developed using Microsoft Excel. Six reviewers will apply data extraction separately (JBA, RS, MA, CF, ALL, and CT). Whenever possible, the original authors will be contacted for any missing data. The following information will be retrieved:
- Studies: publication year, name of the first author, study design, sample size, country, whether single-center or multicentre, follow-up length, follow-up method, and quality of the study.
- Participants: age, sex, ethnicity, performance status, menopausal status, tumour pathology characteristics (histological type, hormone receptor, HER2 status, Ki67), extent of disease, presence of visceral metastases, previous lines of treatment with start and finish date, previous chemotherapy use, time until CDK4/6i use since prior CDK4/6i, PFS obtained with prior CDK4/6i, best response achieved with prior CDK4/6i, and previous AEs with these treatments.
- Interventions: CDK4/6i treatment start and finish date and dosage.
- Outcome measures: HR for OS, PFS, and/or survival curves, allowing estimation of the HR for OS and/or PFS, ORR, DCR, and AEs.
The reviewers will also collect other information regarding the data extraction process (e.g. reviewer, date of data extraction, record number, and missing data). The capacity of this procedure to collect the intended data will be tested in a few studies before performing complete data extraction of all included studies.
Once data extraction is completed, reviewers will compare their results, and a third author (FM or JG) will mediate in case of disagreement.
Risk of Bias Assessment
- For non-randomised studies, the risk of bias tool in non-randomised studies of interventions (ROBINS-I) will be used if non-randomised studies are included (12). Seven domains will be evaluated: confounding, selection of study participants, classification of interventions, bias from deviations of intended interventions, missing data, measurement of outcomes, and selection of the reported outcome. The overall bias will be estimated for these seven domains. Each individual study will be assessed as having a low, moderate, severe, and critical risk of bias. If crucial information for determining the risk of bias is missing, such studies will be considered non-informative.
- For randomised studies, the risk-of-bias tool from the Cochrane Handbook V.5.1.0 will also be used if random controlled trials are included (13). Six domains of the risk of bias will be evaluated: random sequence generation, allocation concealment, blinding, incomplete outcome data, selective reporting, and other bias. Eligible studies will be judged to have a low or high risk of bias.
The risk of bias in all included studies will be evaluated by six reviewers independently (JBA, RS, MA, CF, ALL, and CT). Disagreements will be reported and resolved by a third reviewer (FM or JG).
The Grading of Recommendations Assessment, Development, and Evaluation (GRADE) will be used to rate the quality of the body of evidence retrieved in this systematic review, assessed initially by two reviewers (JB, AL) (14). The overall evidence for each outcome will be rated as very low, low, moderate, or high quality, based on the study design and limitations, consistency of effect, and the directness or generalizability of evidence [15].
Data synthesis and statistical analysis
Statistical analysis will be conducted using Stata software version 16.1 (Stata Corp., College Station, Texas). The Kolmogorov-Smirnov test will be used to check the normality of the extracted data. Descriptive statistics will be used to describe the characteristics of all eligible studies. Normally distributed data will be presented as mean (standard deviation), and non-parametric data will be expressed as median (interquartile range). Categorical data will be presented as proportions (%).
If the studies retrieved report quantitative data that can be combined, the extracted data will be aggregated into a meta-analysis. Heterogeneity will be assessed using the Q and I2 statistics. A p-value <0.10 or an I2 >50% suggests that statistical heterogeneity may exist (16). Publication bias will be assessed using the funnel plot and Begg’s and Egger’s tests (17, 18). A fixed-effects model will be used when the effects are assumed to be homogenous (p>0.05, I2≤50%), and a random-effects model will be used when they are heterogeneous (p<0.05, I2>50%). Confidence intervals will be set at 95%, and two-sided p-values <0.05 will be considered statistically significant. If heterogeneity is detected, a subgroup analysis will be conducted to judge the source of heterogeneity. The criteria for subgroup analysis will potentially include age, ethnicity, performance status, extent of disease, presence of visceral metastases, previous lines of treatment, previous chemotherapy use, time until CDK4/6i use since prior CDK4/6i, PFS obtained with prior CDK4/6i, and best response achieved with prior CDK4/6i.
Pearson’s Chi-square and Fisher’s exact tests will be used for categorical variables, while the t-test or Mann–Whitney tests will be applied for continuous variables, if applicable. OS and PFS will be estimated by Kaplan–Meier analysis. Log-rank tests will be employed to compare differences in OS and PFS between subgroups.