Abstract
Background & Aims
Cyclin-dependent kinase 4/6 inhibitors (CDK4/6i)—ribociclib, abemaciclib, palbociclib—are the standard of care for metastatic breast cancer. Although hepatotoxicity is a frequent adverse event, data on CDK4/6i drug-induced liver injury (DILI) remain scarce. We aimed to define the incidence, clinical presentation, severity, management, and relapse of CDK4/6i-related DILI in a large real-world cohort.
Methods
Retrospective multicentre study including all patients who developed ≥grade 2 DILI (alanine aminotransferase/aspartate aminotransferase >3 × upper limit of normal) while receiving CDK4/6i between January/2018 and December/2022. Severity was assessed according to Common Terminology Criteria for Adverse Events version 5 (CTCAEv5) and DILI-International Expert Working Group criteria.
Results
Among 2,222 CDK4/6i-treated patients, 100 (4.5%, 95% CI 3.7–5.4%) developed DILI. Incidence was significantly higher for ribociclib (8.3%, 95% CI 6.3–10.7%) and abemaciclib (6.7%, 95% CI 4.6–9.3%) than for palbociclib (1.4%, 95% CI 0.8–2.2%) (Difference, p <0.001). Most cases were grade 2 or 3 (CTCAE) and mild (DILI-International Expert Working Group), although moderate DILI (bilirubin >2 × ULN, n = 12) occurred with all three CDK4/6i. The most frequent pattern was hepatocellular, whereas cholestatic was mainly observed with abemaciclib. Liver biopsy findings (n = 11) were heterogeneous and, together with biochemical results, did not support an immune-mediated mechanism as the primary driver. Corticosteroids (n = 20) did not impact time to grade 1 or alanine aminotransferase normalisation. DILI recurred in 28.2% of CDK4/6i rechallenges (n = 85). Among patients treated with ribociclib, rechallenge with an alternative CDK4/6i was associated with lower risk of recurrence (p = 0.010).
Conclusions
DILI caused by ribociclib and abemaciclib are more common that with palbociclib. The absence of immune-mediated features and the limited benefit of corticosteroids question their routine use. After CDK4/6i reintroduction, DILI recurrence rate was 28%. Standardised management is needed, particularly following the approval of ribociclib and abemaciclib as adjuvant therapy.
Impact and implications
Although hepatotoxicity is a frequent adverse event, real-world data on CDK4/6i DILI remain scarce. DILI associated with ribociclib and abemaciclib is more common than for palbociclib, although moderate DILI (bilirubin >2 × normality) was observed with the three drugs. The absence of immune-mediated features and the limited benefit of corticosteroids question their routine use in CDK4/6i DILI. Relapse of DILI after rechallenge with a CDK4/6i was 28%, with all cases presenting as mild. The increasing use of CDK4/6i (metastatic breast cancer and ribociclib and abemaciclib as adjuvant therapy) highlights the need of standardising the management of CDK4/6i DILI.
Keywords: Breast cancer, Ribociclib, Abemaciclib, Drug-induced liver injury
Graphical abstract
Highlights
-
•
Incidence of liver injury was higher for ribociclib and abemaciclib than for palbociclib.
-
•
The three CDK4/6i were associated with moderate DILI cases.
-
•
Liver biopsy findings were heterogeneous and together with biochemical results, did not support an immune-mediated mechanism.
-
•
Corticosteroid therapy did not impact on time to ALT normalisation.
-
•
Among ribociclib-treated patients, rechallenge with an alternative CDK4/6i was associated with lower risk of recurrence (p = 0.010).
Introduction
Breast cancer (BC) is the most common malignancy among women in Europe, with ∼380,000 new cases every year, accounting for ∼30% of all new cancers diagnosed in women. Furthermore, BC remains the leading cause of cancer-related mortality in women.1 Hormone receptor-positive/human epidermal growth factor receptor 2-negative (HR+/HER2-) is the most frequent BC subtype, representing ∼70% of cases.2 For patients with early HR+ disease, adjuvant endocrine therapy (ET) is recommended across all stages because of its effects on reducing locoregional recurrence, distant metastatic recurrence and improving overall survival (OS).3
In recent years, cyclin-dependent kinase 4/6 inhibitors (CDK4/6i)—ribociclib, abemaciclib, and palbociclib—combined with ET, have become the standard of care for HR+/HER2-metastatic BC. These agents have consistently improved progression-free survival.[4], [5], [6] However, treatment discontinuation rates range from 6% for ribociclib to 19% for abemaciclib, with gastrointestinal toxicity being one of the most common severe adverse events.7 Importantly, hepatotoxicity has emerged as a clinically relevant safety concern associated with CDK4/6i therapy. In pivotal trials, the percentage of women presenting grade 3/4 transaminases increase according to the Common Terminology Criteria for Adverse Events (CTCAE)8 (i.e. >5 × upper limit of normal [ULN]) was 9.3%, 7.0% and 3.2% for ribociclib, abemaciclib and palbociclib, respectively.[4], [5], [6] Notably, in ribociclib registry studies, 1.2% patients with transaminases elevations also developed bilirubin levels over two-fold upper limit of normal.5,9 Furthermore, in the PALOMA-3 trial assessing palbociclib and fulvestrant in advanced BC, two cases of acute liver failure occurred among 521 treated patients, highlighting the need for careful hepatic monitoring.10
In the CDK4/6i clinical trials, characterisation of hepatic adverse events was not homogeneous, and recommendations regarding management or rechallenge were not standardised. To date, real-world data on CDK4/6i-induced liver injury remains scarce, particularly concerning liver injury management. We aimed to characterise the incidence, biochemical patterns, severity, and clinical management of CDK4/6i-induced liver injury in a real-world cohort of patients with HR+/HER2-metastatic BC, with special attention to drug rechallenge safety and the risk of recurrent drug-induced liver injury (DILI) upon re-exposure to CDK4/6i.
Patients and methods
Study design and patients
This is a retrospective multicentre study involving seven university hospitals in Spain. All patients treated with any CDK4/6i (abemaciclib, palbociclib, or ribociclib) for metastatic BC between January 2018 and December 2022 who developed DILI, defined as a ≥grade 2 transaminase elevation (CTCAE), were included. The decision to use grade 2 as the entry threshold was based on current oncology guidelines, which recommend therapeutic modifications—such as temporary drug interruption—at this level of abnormalities.11 According to the CTCAEv5, liver toxicity was defined as an alanine aminotransferase (ALT) or aspartate aminotransferase (AST) elevation of 3–5 ULN (grade 2), 5–20 × (grade 3), and >20 × ULN (grade 4).8 The incidence of DILI was calculated based on the total number of CDK4/6i-treated patients during the study period.
This study was approved by the Vall d’Hebron Hospital Ethics Committee (Barcelona, Spain; EOM[(AG]054/2023[6199]) and was conducted in compliance with the principles of the Declaration of Helsinki, Good Clinical Practice Guidelines, and applicable local regulations. Informed consent was waived because of the retrospective nature of the study. All data were anonymised.
Methods
Demographic and clinical variables included sex, age, race, and oncological history (BC status, presence of liver metastasis, and prior therapies). Baseline liver data at CDK4/6i initiation included liver biochemical tests, evidence of steatosis on imaging, and type of concomitant ET. Data at DILI onset included routine laboratory tests (platelet count, prothrombin time, total and direct bilirubin, creatinine, ALT, AST, gamma-glutamyltransferase [GGT], alkaline phosphatase [ALP], albumin, and immunoglobulin G [IgG] levels), and autoantibodies—mainly antinuclear antibodies (ANAs). Liver biopsy findings were recorded when available. CDK4/6i-induced liver injury management, rechallenge strategy, date of last follow-up, oncological progression and death were also recorded. Last access to data was June 2025 or at the time of death.
Rechallenge was defined as re-exposure to a drug from the same family,12 that is, the same or another CDK4/6i, at the same or reduced dose, Switch was defined as retreatment with a different CDK4/6i than the one associated with the initial DILI.
All patients were monitored according to the respective Summary of Product Characteristics for these drugs. This means that laboratory tests and clinical visits were performed every 14 days during the first two cycles, and thereafter every 28 days until cycle 6 for palbociclib and ribociclib, and at least until cycle 4 for abemaciclib. From that point onward, laboratory tests and visits were scheduled every 2–3 months.
Causality assessment for CDK4/6i-related hepatotoxicity was performed using the Roussel Uclaf Causality Assessment Method (RUCAM), interpreted as originally described.13 Severity was assessed according to both CTCAEv5 and the DILI-International Expert Working Group (IEWG) criteria, that, briefly, defines the liver enzymes elevation (ALT ≥5 × ULN or ALP ≥2 × ULN) based on the bilirubin values (mild if bilirubin <2 × ULN, moderate if bilirubin ≥2 × ULN, severe if bilirubin ≥2 × ULN plus international normalised ratio >1.5/ascites/hepatic encephalopathy/other organ failure).14 DILI pattern (hepatocellular, mixed, or cholestatic) was classified according to Aithal et al.14
Liver biopsy specimens were evaluated in H&E-stained slides. Necroinflammatory activity was graded using the Ishak modification for Hepatic Activity Index for chronic hepatitis, and fibrosis by the Ishak score (scale of 0–6). The presence of plasma cell infiltrate was assessed by CD138 immunostaining. Additional features including the presence of granulomas, eosinophils, endothelitis, ductal reaction or damage, and suppurative cholangitis were also evaluated.
Statistical analysis
Quantitative variables were expressed as median and interquartile range (IQR) and compared using the Mann–Whitney U test. Categorical variables were expressed as frequencies and percentages, and compared using the Χ2 or Fisher’s exact test when expected frequencies were <5. Statistical significance was defined as p <0.05. Analyses were performed using IBM SPSS, version 26.0 (SPSS Inc., Armonk, NY, USA).
Results
During the 5-year study period (2018–2022), 2,222 patients with metastatic BC received a CDK4/6i, most commonly palbociclib (n = 1,103), followed by ribociclib (n = 640) and abemaciclib (n = 479). Overall, 100 patients developed ≥grade 2 DILI (CTCAEv5), corresponding to an incidence of 4.5% (95% CI 3.7–5.4%). As shown in Fig. 1, the incidence of CDK4/6i-induced liver injury was significantly higher for ribociclib (8.3%, 95% CI 6.3–10.7%) and abemaciclib (6.7%, 95% CI 4.6–9.3%) than for palbociclib (1.4%, 95% CI 0.8–2.2%) (overall difference, p <0.001; ribociclib vs. palbociclib, p <0.001; abemaciclib vs. palbociclib, p <0.001), with similar proportion between ribociclib and abemaciclib (p = 0.317).
Fig. 1.
Rate of CDK4/6i-induced liver injury ≥grade 2 (CTCAE) according to the inhibitor used.
Level of significance: p <0.001 (Χ2 test). CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; CTCAE, Common Terminology Criteria for Adverse Events.
Baseline characteristics and pre-existing liver conditions of patients who developed CDK4/6i-induced liver injury are summarised in Table 1. All patients were women, with a median age of 59 years; 63% had recurrent BC and 22% had liver metastasis. Among the 100 DILI cases, 53 occurred on ribociclib, 32 on abemaciclib, and 15 on palbociclib. Type of metastatic onset (recurrent vs. de novo) and the presence of liver metastasis were similar across the three CDK4/6i. Patients treated with abemaciclib had higher baseline ALT, with six (18.8%) presenting abnormal ALT or AST at therapy initiation. Alcohol consumption was reported in six individuals, but none had evidence of advanced liver disease. Liver steatosis was present in 18 patients, with similar frequency for the three inhibitors (p = 0.640).
Table 1.
Baseline and Liver Injury characteristics of patients with CDK4/6i-related DILI.
| Overall (N = 100) | Ribociclib (n = 53) | Palbociclib (n = 15) | Abemaciclib (n = 32) | p value | |
|---|---|---|---|---|---|
| At baseline | |||||
| Age, years, median (IQR) | 59 (51–69) | 58 (50–68) | 60 (53–74) | 60 (51–71) | 0.973 |
| Recurrent metastatic disease, n (%) | 63 (63.0) | 32 (60.4) | 8 (50.0) | 22 (84.6) | 0.399 |
| Liver metastasis, n (%) | 22 (22.0) | 12 (22.6) | 3 (20.0) | 7 (21.9) | 0.976 |
| Previous chemotherapy, n (%) | 59 (59.0) | 28 (52.8) | 8 (53.3) | 23 (71.9) | 0.199 |
| AST, IU/ml, median (IQR) | 26 (18–33) | 21 (17–29) | 26 (18–37) | 42 (28–60) | <0.001 |
| Elevated AST, n (%) | 11 (11.0) | 3 (5.7) | 2 (13.3) | 6 (18.8) | 0.166 |
| GGT, IU/ml, median (IQR) | 24 (17–52) | 21 (16–31) | 21 (19–51) | 38 (19–92) | 0.266 |
| At CDK4/6i-induced liver injury | |||||
| Time from CDK4/6i start, months, median (IQR) | 3.0 (1.4–6.0) | 3.0 (1.4–5.3) | 3.7 (2.3–6.4) | 2.0 (0.9–8.3) | 0.991 |
| Endocrine therapy, n (%) Aromatase inhibitor Fulvestrant |
73 (73.0) 27 (27.0) |
39 (73.9) 14 (26.4) |
10 (66.7) 5 (33.3) |
24 (75.0) 8 (25.0) |
0.827 |
| AST, IU/ml, median (IQR) | 143 (91–254) | 144 (98–308) | 141 (74–215) | 135 (73–226) | 0.346 |
| ALT, IU/ml, median (IQR) | 220 (154–365) | 248 (160–482) | 223 (140–315) | 189 (83–331) | 0.126 |
| Total bilirubin, mg/dl, median (IQR) Bilirubin >ULN, n (%) Bilirubin >2.5 mg/dl, n (%) |
0.63 (0.44–0.90) 14 (14.0) 6 (6.0) |
0.7 (0.5–1.0) 8 (15.4) 1 (1.9) |
0.6 (0.4–1.1) 2 (13.3) 2 (13.3) |
0.5 (0.4–0.9) 4 (12.5) 3 (9.4) |
0.829 0.930 0.167 |
| ALP, IU/ml, median (IQR)∗ | 106 (87–154) | 108 (89–132) | 91 (64–166) | 107 (83–331) | 0.025 |
| GGT, IU/ml, median (IQR)∗ | 62 (31–184) | 55 (31–96) | 56 (28–345) | 157 (44–412) | 0.071 |
| IgG >ULN, n (%) | 0/22 (0) | 0 (0) | 0 (0) | 0 (0) | – |
| ANAs ≥1/80, n (%) | 5/28 (17.9) | 2/14 (14.3) | 0/3 (0) | 3/11 (27.3) | 0.587 |
| R value, median (IQR) | 5.0 (3.0–8.2) | 5.9 (3.5–10.7) | 5.7 (2.5–11.4) | 3.9 (2.2–5.0) | 0.075 |
| Admission to hospital, n (%) | 7 (7.0) | 3 (5.7) | 1 (6.7) | 3 (9.4) | 0.808 |
| Discontinuation of endocrine therapy, n (%) | 17 (17.0) | 9 (17.0) | 2 (13.3) | 18.8) | 0.899 |
Variables are expressed as n (%) or median (IQR). ∗Data were missing in five patients. ALP, alkaline phosphatase; ALT, alanine aminotransferase; ANAs, antinuclear antibodies; AST, aspartate aminotransferase; CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; CTCAE, Common Terminology Criteria for Adverse Events; GGT, gamma-glutamyltransferase; IEWG, International Expert Working Group; IgG, immunoglobulin G; ULN, upper limit of normal.
Overall, in this cohort the median duration of CDK4/6i therapy was 29.1 months (IQR 10.4–41.9, range 0.6–91.7). At last follow-up, 34% patients were still undergoing CDK4/6i, whereas 33% had died.
Characteristics and chronology of CDK4/6i-induced liver injury
Analytical findings at DILI presentation are summarised in Table 1. AST elevations were more prominent in ribociclib-related DILI, whereas ALP and GGT levels were higher in those exposed to abemaciclib. ANAs and IgG were available in 28 and 22 patients, respectively. ANA titres ≥1/80 were observed in five (17.9%) cases, and none had elevated IgG.
Regarding severity, most DILI cases were grade 2–3 (CTCAE) or mild according to DILI-IEWG (Fig. 2A -and B). Grade 4 DILI (transaminases >20 × ULN) occurred only with ribociclib and abemaciclib, with 10 (18.9%) and two (6.3%) cases, respectively (p = 0.145). Importantly, moderate DILI (total bilirubin >2 × ULN) was observed in 12 patients across all three inhibitors (one of 15 palbociclib, seven of 53 ribociclib, four of 32 abemaciclib, p = 0.785). Among these patients, six (50%) presented grade 4 hepatitis, four (33.3%) grade 3, and two (16.7%) grade 2 (p <0.001) according to CTCAEv5. No liver related deaths or cases of liver failure were reported.
Fig. 2.
Overall rate of CDK4/6i-induced liver injury according to the CTCAEv5 criteria (A) and the DILI-IEWG score (B).
The CTCAE defines severity of DILI based on transaminases values: grade 2 for AST and/or ALT three-to five-fold over ULN, grade 3 ranging from five-to 20-fold and grade 4 for AST and/or ALT over 20-fold ULN. The DILI-IEWG defines DILI as mild for patients with ALT ≥5 × ULN or ALP ≥2 × ULN and total bilirubin <2 × ULN and moderate when total bilirubin is two-fold ULN. ALT, alanine aminotransferase; ALP, alkaline phosphatase; CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; CTCAE, Common Terminology Criteria for Adverse Events; DILI, drug-induced liver injury; IEWG, International Expert Working Group; ULN, upper limit of normal.
Moderate DILI (DILI-IEWG) was more frequent in patients with oncologic disease progression at the time of DILI (33.3% vs. 9.7%, p = 0.030) and in those previously exposed to chemotherapy (16.7% vs. 5.0%, p = 0.070). Elevated transaminases at CDK4/6i initiation (p = 0.125), liver metastasis (p = 0.519), steatosis (p = 0.376) or the type of ET (aromatase inhibitors vs. fulvestrant, p = 0.587) were not associated with moderate DILI. Regarding causality, 11 patients scored highly likely on RUCAM, 37 scored probable, 36 scored possible, and six scored unlikely.
Overall, the most frequent DILI pattern was hepatocellular (50.5%), followed by mixed (37.9%), and cholestatic (11.6%). Hepatocellular pattern predominated with ribociclib and palbociclib, whereas the mixed pattern dominated in those on abemaciclib (Fig. 3). Concerning cholestatic pattern, incidence was significantly lower for ribociclib than for the other two CDK4/6i (p = 0.002): 3.9% (95% CI 0.5–13.5%) for ribociclib, 21.4% (95% CI 4.7–50.8%) for palbociclib, and 20.0% (95% CI 7.7–38.6%) for abemaciclib.
Fig. 3.
DILI pattern associated with each CDK inhibitor.
The biochemical pattern was defined according to the classical classification by the DILI-IEWG as hepatocellular, mixed or cholestatic.14 Level of significance: p = 0.002 (Χ2 test). CDK, cyclin-dependent kinase; DILI, drug-induced liver injury; IEWG, International Expert Working Group.
Time to DILI onset and liver histology
Half of DILI cases (51.0%) occurred within the first 3 months of therapy, and most (75.0%) within the first 6 months (Fig. S1). Time to DILI onset was not associated with severity (DILI-IEWG, p = 0.714), type of CDK4/6i (p = 0.655), presence or absence of liver metastasis (p = 0.232), type of ET (p = 0.123) or biochemical pattern (p = 0.623). However, patients with elevated AST or ALT at CDK4/6i initiation were more likely to develop DILI within the 3 months (81.8% vs. 47.2%, p = 0.030), with no cases occurring beyond month 6.
Liver biopsy was performed in 11 patients (six receiving ribociclib, four abemaciclib and one palbociclib). Three presented moderate DILI (bilirubin >2 × ULN), and biochemical patterns were hepatocellular (seven, 63.6%) or mixed (four, 36.4%). Median time from the start of CDK4/6i therapy to DILI in these patients was 56 (IQR 33–138) days and from DILI to liver biopsy 29 (IQR 15–48). Only one (9.1%) of the 11 patients had received corticosteroids before the biopsy.
Histological findings were heterogeneous, with acute cholestasis, steatosis, and interphase hepatitis as the most common features (Table S1). Only one patient had moderate fibrosis (F3, Ishak score). ANA positivity was not associated with interphase hepatitis (p = 0.708) or significant lymphoplasmacytic infiltrate (p = 0.545). No associations were observed between histology and type of CDK4/6i, ET, or biochemical pattern.
Management of CDK4/6i-induced liver injury
Overall, seven (7.0%) patients were admitted to hospital because of CDK4/6i-induced liver injury, five of whom presented with moderate DILI (bilirubin two-fold ULN). Among the whole cohort, 20 (20.0%) patients received corticosteroids for DILI. Corticosteroid use was more frequent in ribociclib-treated patients (p = 0.025), those with grade 4 hepatitis (p <0.001) or moderate DILI (p = 0.013) (Table 2). Median prednisone dose was 30 mg/day (IQR 5–50), with a median duration of 42 days (IQR 29-57). Median cumulative dose was 550 mg (IQR 160–900). Five patients (25.0%) experienced steroid-related adverse events, including two cases of oedema, one respiratory infection, one oesophageal candidiasis, and one steroid myopathy, none requiring admission to hospital.
Table 2.
Comparison of characteristics and clinical outcomes of patients treated with and without corticosteroids.
| Corticosteroids (n = 20) | No corticosteroids (n = 80) | p values | |
|---|---|---|---|
| Ribociclib, n (%) | 16 (80.0) | 37 (46.3) | 0.025 |
| Moderate DILI (IEWG), n (%) | 6 (30.0) | 6 (7.5) | 0.013 |
| Grade 4 DILI (CTCAE), n (%) | 9 (45.0) | 3 (3.8) | <0.001 |
| Hepatocellular pattern, n (%) | 13 (65.0) | 34 (46.7) | 0.299 |
| Liver biopsy, n (%) | 7 (35.0) | 4 (5.0) | 0.001 |
| ≤3 months from CDK4/6i start, n (%) | 12 (60) | 39 (48.8) | 0.258 |
| Total bilirubin, mg/dl, median (IQR) | 0.9 (0.5–1.4) | 0.6 (0.4–0.9) | 0.085 |
| ALT, IU/ml, median (IQR) | 463 (264–866) | 197 (142–272) | <0.001 |
| Time to grade 1 ALT, days, median (IQR) | 70 (22–102) | 27 (14–64) | 0.033 |
| Rate of ALT normalisation, n (%) | 20 (100) | 66 (83.5) | 0.043 |
| Time to ALT normalisation, days, median (IQR) | 113 (62–139) | 52 (28–129) | 0.838 |
| Rechallenge with a CDK4/6i, n (%) | 16 (80.0) | 69 (86.3) | 0.347 |
ALT, alanine aminotransferase; CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; CTCAE, Common Terminology Criteria for Adverse Events; IEWG, International Expert Working Group.
Time to ALT normalisation or grade 1 was longer in corticosteroid-treated patients, likely attributable to higher baseline ALT (median 723 vs. 280 IU/ml, p <0.001). After stratifying grade 3 hepatitis into 3a (5–8 × ULN) and 3b (8–20 × ULN) using an FDA-recommended framework for DILI in premarketing drugs,15 time of ALT normalisation was similar within these subgroups. Among 3b cases (n = 24), peak ALT levels (median 528 vs. 378 IU/ml, p = 0.253) and time to ALT normalisation or to grade 1 were similar with and without corticosteroid therapy (time to grade 1 ALT: 45 vs. 48 days, p = 0.831; time to ALT normalisation: 91 vs. 86 days, p = 0.613). Comparable results were seen in the 3a cohort (time to grade 1 ALT: 51 vs. 27 days, p = 0.593; time to ALT normalisation: 118 vs. 47 days, p = 823), although corticosteroid use was more limited in this subgroup (3/32 vs. 9/32). Despite nearly 50% of patients presenting with mixed or cholestatic DILI, ursodeoxycholic acid was used in only one case.
In addition to CDK4/6i, ET was interrupted in 17 (17.0%) cases. This interruption was more frequent in patients with grade 4 hepatitis (33.3% vs. 21.4% grade 3 and 3.1% grade 2, p = 0.025) and in those with moderate DILI (41.7% vs. 13.6%, p = 0.029), but was similar according to the type of ET (19.2% aromatase inhibitors vs. 11.1% fulvestrant, p = 0.263). There were no differences in terms to time to ALT normalisation or to grade 1 according to ET maintenance (p = 0.456 and p = 0.483), even when only grade 3a or 3b were considered.
Rechallenge with CDK4/6i and risk of recurrence
Overall, 85 patients (85.0%) were rechallenged with a CDK4/6i after a median of 30 days (IQR 15–74). As shown in Fig. 4A, rechallenge rates did not differ by CDK4/6i (p = 0.198), although rechallenge strategies varied. Switching to a different CDK4/6i was more common after ribociclib-related severe DILI (p = 0.003). Dose reduction at rechallenge was similar across the three CDK4/6i (p = 0.940).
Fig. 4.
(A) Rate of rechallenge with the same or an alternative CDK4/6i and the preferred strategy according to the initial CDK4/6i Level of significance: p = 0.198; p = 0.003, and p = 0.980, respectively, Χ2 test. (B) Rate of DILI recurrence according to both the initial CDK4/6i (X axis) and the CDK4/6i used at rechallenge (coloured columns). Level of significance: p = 0.010, p = 0.389, and p = 0.5032 for initial DILI associated with ribociclib, palbociclib, and abemaciclib, respectively (Χ2 test). CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; DILI, drug-induced liver injury.
Rechallenge was less frequent in patients with moderate DILI (58.3% vs. 88.6%, p = 0.016), but similar according to CTCAE grade (grade 4: 75.0%; grade 3: 83.9%; grade 2: 90.6%, p = 0.409) or biochemical pattern (hepatocellular: 87.5%; mixed: 86.1%; cholestatic: 63.6%, p = 0.136). Switching the CDK4/6i was more frequent after moderate DILI (85.7% vs 32.1%, p = 0.009) or grade 4 hepatitis (100% vs 36.2% for grade 3, p <0.001).
Overall, DILI recurrence occurred in 24 of 85 patients (28.2%) after a median of 18 days (IQR, 13–45). Recurrence severity by CTCAE was grade 1 (8.3%), grade 2 (37.5%), grade 3 (50.0%), and grade 4 (4.2%). All except one were mild by DILI-IEWG.
Recurrence rates did not differ by initial DILI grade (p = 0.167), initial CDK4/6i (ribociclib: 25%; palbociclib: 36.4%; abemaciclib 30.8%, p = 0.709) or time to DILI onset (p = 0.957). However, the choice of CDK4/6i at rechallenge influenced recurrence risk (Fig. 4B). Among patients initially receiving ribociclib, switching to a different CDK4/6i was associated with a lower relapse rate (p = 0.010). Higher ALT or GGT at rechallenge were associated to greater recurrence risk (p = 0.077 and p = 0.005, respectively). Dose reduction (75% vs. 60.4%, p = 0.303) or concomitant corticosteroids (zero of 24 vs. four of 55, p = 0.318) did not prevent recurrence.
Discussion
This study provides several novel and clinically relevant insights into CDK4/6i-induced liver injury from a hepatology perspective. First, in this large real-world multicentre cohort, we demonstrated that CDK4/6i-related DILI is heterogeneous and exhibits drug-specific phenotypes, with ribociclib and abemaciclib conferring a higher risk of clinically relevant liver injury than palbociclib. Second, we show that moderate DILI—defined by more than two-fold bilirubin elevation—can occur with all three agents, including palbociclib, and may be underestimated when severity assessment relies solely on ALT values according to CTCAE. This observation is noteworthy, because in liver toxicity associated with other oncological agents such as immune checkpoint inhibitors, DILI-IEWG severity grading correlates with outcome, whereas CTCAE does not.16
Third, DILI most frequently developed during the first months of therapy. In our cohort, half of cases occurred within the first 3 months and up to 75% within the first 6 months. Close monitoring of liver function is therefore particularly warranted during this initial period, as also highlighted in pivotal trials and CDK4/6i product labels, which recommend intensive biochemical surveillance at treatment initiation, especially among patients with elevated baseline transaminases.
Fourth, we provide a detailed real-world assessment of CDK4/6i rechallenge, showing that recurrence of DILI is not universal. Notably, in the case of ribociclib-related hepatotoxicity, switching to an alternative CDK4/6i reduced the risk of relapse. Together, these findings have direct and practical implications for the management of CDK4/6i-induced liver injury.
In line with a previously published cohort,17 the hepatocellular pattern was the most frequent in our study. However, more than half of patients presented with mixed or cholestatic pattern, particularly those treated with abemaciclib, underscoring the need for systematic assessment of all liver biochemical parameters rather than transaminases alone.
From a mechanistic standpoint, 17.9% of patients had ANAs titres ≥1/80 but none had elevated IgG levels. Liver histology was available in 11 patients and revealed heterogeneous findings, including interphase hepatitis and moderate lymphoplasmacytic infiltrate in three and two patients, respectively. Similarly, in the cohort reported by from Meunier et al.,17 no histological features suggestive of autoimmune-like hepatitis were observed, with only five patients exhibiting ANA ≥1/80 and one elevated IgG values.17 These data should be interpreted cautiously because of the limited number of biopsied cases.
In our study, corticosteroids were used in 20% of patients, particularly those with grade 4 or moderate DILI. However, after stratifying grade 3 hepatitis into 3a and 3b according to an FDA-recommended framework for DILI severity, time to ALT normalisation or grade 1 ALT (i.e. the recommended threshold for CDK4/6i reintroduction) was similar regardless of the steroid use. Moreover, five of the 20 steroid-treated patients developed steroid-related adverse events. These results are consistent with current recommendations for DILI management, in which corticosteroid therapy is generally discouraged except in cases of suspected drug-induced autoimmune-like hepatitis or acute liver failure.18,19 However, our data, together with those from the Meunier et al.17 series, do not support an immune-mediated mechanism, which might explain the limited effect of corticosteroids on recovery time. Indeed, although the European Society for Medical Oncology and American Society of Clinical Oncology guidelines recommend steroids for the management of grade 3 immune-mediated hepatitis (e.g. that induced by immune checkpoint inhibitors), real-world data suggest that >65% of these patients, including those with grade 3 or 4 hepatitis, can be successfully managed without steroids.20 In our cohort, ET was discontinued because of liver toxicity in 17 patients (17.0%), mainly those presenting with moderate or grade 4 DILI. Although time to ALT normalisation or to grade 1 was similar regardless of ET discontinuation, from a mechanistic perspective early discontinuation of both drugs (the CDK4/6i and ET) may still be reasonable in order to achieve faster transaminase improvement. Nevertheless, this hypothesis remains speculative and should be tested. Prospectively some of the most relevant findings relate to rechallenge outcomes. In our cohort, 85.0% of patients underwent rechallenge after resolution of DILI, with a recurrence rate of 28.2% after a median of 18 days. Most recurrent DILI episodes occurred within the first 8 weeks of reintroduction, and half cases were grade 1 or 2 according to CTCAE. Dose reduction did not reduce recurrence risk, consistent with the AMALEE trial, a phase II randomised study in which ribociclib at 400 mg did not mitigate hepatotoxicity compared the full dose with.21,22 By contrast, switching to an alternative CDK4/6i reduced the risk of recurrence in ribociclib-treated patients: among 48 patients initially exposed to ribociclib, recurrence was 43.5% (10/23) of those rechallenged with the same agent, compared with 0% (zero of 14) and 18.2% (two of 11) in those switched to palbociclib and abemaciclib, respectively.
Our study has limitations inherent to its retrospective design, including potential selection bias, incomplete biochemical and immunological data in some cases and the possibility of the inclusion of patients with an alternative diagnosis as shown by those with just ‘possible’ causality according to the RUCAM score. Besides, the decision to administer corticosteroids and rechallenge strategies was at the discretion of treating physicians, resulting in heterogeneity of management. Moreover, inclusion of patients based on CTCAE criteria may have introduced inaccuracies. Strengths include the multicentre design, the large number of well-characterised DILI cases, and the use of multiple severity frameworks, including subclassification of grade 3 hepatitis, which reinforce the robustness of our observations.
In summary, CDK4/6i-induced liver injury is not uncommon, particularly within the first 3–6 months of therapy with ribociclib and abemaciclib, although moderate DILI (bilirubin >2 × ULN) can occur with any of the three approved compounds. Corticosteroids appear to provide limited benefit, and rechallenge is feasible, although switching CDK4/6i should be considered in cases of ribociclib-related DILI. Standardised management strategies are urgently needed, particularly for ribociclib and abemaciclib, especially because of their expanding use in the adjuvant setting.
Abbreviations
ALP, alkaline phosphatase; ALT, alanine aminotransferase; ANAs, antinuclear antibodies; AST, aspartate aminotransferase; BC, breast cancer; CDK4/6i, cyclin-dependent kinase 4/6 inhibitors; CTCAE, Common Terminology Criteria for Adverse Events; DILI, drug-induced liver injury; ET, endocrine therapy; GGT, gamma-glutamyltransferase; IEWG, International Expert Working Group; IgG, immunoglobulin G; OS, overall survival; RUCAM, Roussel Uclaf Causality Assessment Method; ULN, upper limit of normal.
Authors’ contributions
Guarantor of the article: MR-B. Concept and design: MR-B, MBellet. Acquisition, analysis, or interpretation of data: MR-B, PE, KSV-C, BM, ELM, PJ, MM, SAL, MS, XF, I G-F, OM-S, JP-B, MG-C, NR, ÁRN, DM-A, CAR, RL, RD-G, MC, JCR-C, AV, EZ. Drafting of the manuscript: MR-B, EZ, MButi, MBellet. Approved the final version of the article, including the authorship list: all authors.
Data availability
The data that support the findings of this study are available from the corresponding author, MR-B, upon request.
Financial support
This study has been funded by Instituto de Salud Carlos III (ISCIII) through the project ‘PI25/01602’ and co-funded by the European Union.
Conflicts of interest
The authors declare no conflicts of interest.
Please refer to the accompanying ICMJE disclosure forms for further details.
Acknowledgements
The graphical abstract was created with the support of BioRender (https://BioRender.com/78sj2q4). English writing support was provided by Maura O’Sullivan.
Footnotes
Author names in bold designate shared co-first authorship
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jhepr.2026.101873.
Supplementary data
The following are the Supplementary data to this article:
References
- 1.System E-ECI . 2025. ECIS - European cancer information system.https://ecis.jrc.ec.europa.eu/ [Google Scholar]
- 2.Loibl S., André F., Bachelot T., et al. Early breast cancer: ESMO Clinical Practice Guideline for diagnosis, treatment and follow-up. Ann Oncol. 2024;35:159–182. doi: 10.1016/j.annonc.2023.11.016. [DOI] [PubMed] [Google Scholar]
- 3.Early Breast Cancer Trialists’ Collaborative Group (EBCTCG) Aromatase inhibitors versus tamoxifen in premenopausal women with oestrogen receptor-positive early-stage breast cancer treated with ovarian suppression: a patient-level meta-analysis of 7030 women from four randomised trials. Lancet Oncol. 2022;23:382–392. doi: 10.1016/S1470-2045(21)00758-0. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Rugo H.S., Finn R.S., Diéras V., et al. Palbociclib plus letrozole as first-line therapy in estrogen receptor-positive/human epidermal growth factor receptor 2-negative advanced breast cancer with extended follow-up. Breast Cancer Res Treat. 2019;174:719–729. doi: 10.1007/s10549-018-05125-4. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Hortobagyi G.N., Stemmer S.M., Burris H.A., et al. Overall survival with ribociclib plus letrozole in advanced breast cancer. New Engl J Med. 2022;386:942–950. doi: 10.1056/NEJMoa2114663. [DOI] [PubMed] [Google Scholar]
- 6.Goetz M.P., Toi M., Campone M., et al. MONARCH 3: abemaciclib as initial therapy for advanced breast cancer. J Clin Oncol. 2017;35:3638–3646. doi: 10.1200/JCO.2017.75.6155. [DOI] [PubMed] [Google Scholar]
- 7.Farooq F., Cohen J.A. Efficacy and toxicity of CDK 4/6 inhibitors in breast cancer: systematic review and meta-analysis of the phase III clinical trials. J Clin Oncol. 2019;37 [Google Scholar]
- 8.Common Terminology criteria for adverse events (CTCAE) version 5.0. 2017. [Google Scholar]
- 9.Slamon D.J., Neven P., Chia S., et al. Phase III randomized study of ribociclib and fulvestrant in hormone receptor-positive, human epidermal growth factor receptor 2-negative advanced breast cancer: MONALEESA-3. J Clin Oncol. 2018;36:2465–2472. doi: 10.1200/JCO.2018.78.9909. [DOI] [PubMed] [Google Scholar]
- 10.Turner N.C., Slamon D.J., Ro J., et al. Overall survival with palbociclib and fulvestrant in advanced breast cancer. New Engl J Med. 2018;379:1926–1936. doi: 10.1056/NEJMoa1810527. [DOI] [PubMed] [Google Scholar]
- 11.Gennari A., André F., Barrios C.H., et al. ESMO Clinical Practice Guideline for the diagnosis, staging and treatment of patients with metastatic breast cancer. Ann Oncol. 2021;32:1475–1495. doi: 10.1016/j.annonc.2021.09.019. [DOI] [PubMed] [Google Scholar]
- 12.Hunt C.M., Papay J.I., Stanulovic V., et al. Drug rechallenge following drug-induced liver injury. Hepatology. 2017;66:646–654. doi: 10.1002/hep.29152. [DOI] [PubMed] [Google Scholar]
- 13.Danan G., Benichou C. Causality assessment of adverse reactions to drugs – I. A novel method based on the conclusions of international consensus meetings: application to drug-induced liver injuries. J Clin Epidemiol. 1993;46:1323–1330. doi: 10.1016/0895-4356(93)90101-6. [DOI] [PubMed] [Google Scholar]
- 14.Aithal G.P., Watkins P.B., Andrade R.J., et al. Case definition and phenotype standardization in drug-induced liver injury. Clin Pharmacol Therapeut. 2011;89:806–815. doi: 10.1038/clpt.2011.58. [DOI] [PubMed] [Google Scholar]
- 15.FDA. Drug-Induced Liver Injury . 2009. Premarketing clinical evaluation.https://www.fda.gov/regulatory-information/search-fda-guidance-documents/drug-induced-liver-injury-premarketing-clinical-evaluation [Google Scholar]
- 16.Hountondji L., Faure S., Palassin P., et al. Time to use the right classification to predict the severity of checkpoint inhibitor-induced liver injury, as assessed for causality using the updated RUCAM. Aliment Pharmacol Ther. 2024;60:1561–1572. doi: 10.1111/apt.18276. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Meunier L., De Martin E., Delire B., et al. CDK4/6 inhibitor-induced liver injury: clinical phenotypes and role of corticosteroid treatment. JHEP Rep. 2024;6 doi: 10.1016/j.jhepr.2024.101098. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.European Association for the Study of the Liver . Vol. 70. 2019. pp. 1222–1261. (EASL clinical Practice guidelines: drug-induced liver injury). [DOI] [PubMed] [Google Scholar]
- 19.Andrade R.J., Aithal G.P., de Boer Y.S., et al. Nomenclature, diagnosis and management of drug-induced autoimmune-like hepatitis (DI-ALH): an expert opinion meeting report. J Hepatol. 2023;79:853–866. doi: 10.1016/j.jhep.2023.04.033. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Riveiro-Barciela M., Barreira-Díaz A., Salcedo M.T., et al. An algorithm based on immunotherapy discontinuation and liver biopsy spares corticosteroids in two thirds of cases of severe checkpoint inhibitor-induced liver injury. Aliment Pharmacol Therapeut. 2024;59:865–876. doi: 10.1111/apt.17898. [DOI] [PubMed] [Google Scholar]
- 21.Slamon D., Lipatov O., Nowecki Z., et al. Ribociclib plus endocrine therapy in early breast cancer. New Engl J Med. 2024;390:1080–1091. doi: 10.1056/NEJMoa2305488. [DOI] [PubMed] [Google Scholar]
- 22.Cardoso F., Jacot W., Kuemmel S., et al. 600- vs 400-mg first-line ribociclib in hormone receptor-positive/ERBB2-negative advanced breast cancer: the AMALEE randomized clinical trial. JAMA Oncol. 2025;11:1356–1363. doi: 10.1001/jamaoncol.2025.3687. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author, MR-B, upon request.