CDK4/6 抑制剂所致肝毒性及其消退：7 例研究Cyclin-dependent kinase 4/6 (CDK4/6) inhibitor-induced hepatotoxicity and its resolution: a study of seven cases

Scope: open-access full text (CC BY 4.0) from Histopathology — abstract, introduction, material and methods, results and discussion, with 5 figures and 3 tables; the abbreviations list, author contributions, funding, conflict of interest, data availability statement, supporting information (Tables S1–S2) and references are not included. Superscript numbers are the original reference numbers.

Abstract

Aims

Hepatotoxicity is an increasingly recognised and potentially dose-limiting adverse effect of CDK4/6 inhibitors, but their histopathological spectrum remains incompletely characterised. We aimed to describe the clinicopathological features of histologically characterised CDK4/6 inhibitor-induced liver injury.

Methods and results

We retrospectively identified patients with metastatic breast cancer who developed liver enzyme elevation and underwent liver biopsy while receiving a CDK4/6 inhibitor at a single comprehensive cancer centre (2020–2024). Inclusion required a RUCAM score ≥6 and exclusion of alternative aetiologies. Seven female patients met the criteria (ribociclib, n = 6; palbociclib, n = 1). The mean interval to first enzyme elevation was 60 days (range 20–114) and the mean RUCAM score was 8. Antinuclear antibody (ANA) positivity was noted in three patients, with negative extended autoimmune panels. Liver biopsy demonstrated severe acute hepatitis in five patients (71.4%), mild lobular hepatitis in one (14.3%) and subacute hepatitis with early fibrosis in one (14.3%). All acute biopsies showed perivenular (zone 3) confluent necrosis, with additional bridging necrosis in three biopsies. Portal inflammation was identified in all biopsies and lymphocytic cholangitis in five (71.4%) biopsies. No ANA-positive patient showed histological features of autoimmune hepatitis. All patients achieved biochemical normalisation after drug withdrawal (mean 63 days from peak; range 30–129), and six were successfully switched to an alternative agent.

Conclusions

CDK4/6 inhibitor-induced liver injury predominantly manifests as acute hepatocellular injury with zone 3 necrosis and lymphocytic cholangitis. Despite severe histological abnormalities, it resolves with drug withdrawal, and limited cross-reactivity between agents supports individualised therapeutic strategies.

Introduction

Cyclin-dependent kinase 4 and 6 (CDK4/6) inhibitors combined with endocrine therapy have become the standard treatment for hormone receptor-positive, human epidermal growth factor receptor 2 (HER2) negative breast cancer.¹ Ribociclib, palbociclib and abemaciclib are FDA-approved orally bioavailable small molecules that act by binding to the ATP cleft of CDK4 and/or CDK6.² They significantly increase progression-free survival (PFS) when added to endocrine therapy. Although these drugs have shown similar efficacy, hepatotoxicity emerges as a dose-limiting adverse effect of CDK4/6 inhibitors, occurring in 14%–25% of patients across clinical trials.^{3, 4} Ribociclib received FDA approval in March 2017 in combination with letrozole for the first-line treatment of hormone receptor-positive, HER2-negative (metastatic or locally advanced) breast carcinoma.¹ The choice between the drugs, in practice, is influenced by patient-specific factors including menopausal status, prior treatment history, comorbidities and tolerability profile.^{2, 5, 6}

Ribociclib is among the most widely prescribed CDK4/6 inhibitors and confers a clear survival benefit, but its adverse-event burden is substantial; in the MONALEESA trials, nearly all treated patients experienced at least one treatment-emergent adverse event, and grade III–IV events occurred in up to 84% of patients.⁶ Common toxicities of ribociclib include fever, fatigue, diarrhoea and thrombocytopaenia, while severe reactions include QTc prolongation and severe neutropaenia.¹ Palbociclib is often preferred when minimising hepatotoxicity risk is a clinical priority, given its consistently more favourable hepatic safety profile.^{5, 7} Abemaciclib occupies an intermediate position with respect to hepatic risk.^{3, 5} Hepatotoxicity caused by ribociclib is echoed in real-world series and in isolated case reports of grade IV hepatotoxicity and fulminant hepatitis.^8-10

Despite its growing clinical relevance, histopathological characterisation of CDK4/6 inhibitor-induced liver injury remains inadequately described, posing diagnostic challenges for pathologists.

In this study, we present detailed histopathological and biochemical findings from seven well-characterised patients with advanced-stage breast carcinoma, who developed clinically apparent liver injury following treatment with ribociclib (n = 6) and palbociclib (n = 1), and in whom other causes of liver injury were excluded, confirming drug-induced liver injury (DILI) due to CDK4/6 inhibitors.

Material and methods

Study Design

After obtaining requisite approval from the Institute Ethics Committee, a retrospective review of breast cancer patients receiving CDK4/6 inhibitor therapy who had elevated liver enzymes and underwent liver biopsy at a single comprehensive cancer centre from January 2020 through December 2024 was performed. Histopathological review was performed independently and remotely by co-authors based at other institutions (R.S., S.M.); no additional cases were contributed by any other centre.

Clinical and laboratory data were collected, including demographic characteristics, breast cancer subtype, specific CDK4/6 inhibitor used and dosing schedule, duration from therapy initiation to liver enzyme elevation, biochemical parameters and underlying liver disease. All liver biopsies underwent detailed histopathological evaluation to characterise the liver injury pattern and assess for concurrent hepatic pathology. Liver function tests were monitored at a median interval of 15 days (range 1–30) prior to the first recorded abnormality, and at a median interval of 7 days thereafter. No interval between successive measurements before the first abnormality exceeded 30 days in any patient (Table S1).

Patient Selection

Patients were included if they met these criteria: (1) Histologically characterised DILI with Roussel Uclaf Causality Assessment Method (RUCAM) score ≥6 (revised electronic RUCAM, 2022)¹¹; (2) Exclusion of alternative aetiologies, including viral hepatitis (negative HCV RNA, HBsAg, HBV DNA), autoimmune hepatitis (AIH) (negative extended autoimmune liver disease panel, anti-LKM <1:40, IgG ≤16 g/L), biliary obstruction (normal magnetic resonance cholangiopancreatography [MRCP]) and hepatic metastatic disease; (3) Absence of competing causes of liver injury, including concurrent use of hepatotoxic medications within 30 days, significant alcohol consumption (more than 20 g/day) and pre-existing chronic liver disease.

Time to first liver enzyme elevation was defined as the interval from initiation of the CDK4/6 inhibitor to the first ALT (alanine aminotransferase) or AST (aspartate aminotransferase) value above the laboratory reference range. Clinically significant hepatotoxicity, the threshold at which the drug was withheld and further evaluation undertaken, was defined as an ALT or AST value greater than three times the upper limit of normal (ULN), corresponding to grade ≥2 toxicity by the Common Terminology Criteria for Adverse Events (CTCAE) version 5.0.¹² All patients had documented normal baseline liver biochemistry before starting therapy, and chronic liver disease, including metabolic dysfunction-associated steatotic liver disease (MASLD), alcohol-related liver disease and MetALD, was excluded on pretreatment metabolic risk-factor assessment, alcohol history and imaging. Biopsy was performed when the cause of the enzyme elevation remained unclear after non-invasive investigation, or when the abnormality persisted despite drug withdrawal.

Data Collection

Clinical parameters for all eligible patients were extracted from the electronic medical records. Primary breast tumour characteristics included histological tumour type and grade, as well as hormone receptor (oestrogen receptor [ER] and progesterone receptor [PgR]) and HER2 status. Laboratory parameters evaluated consisted of liver function tests (LFTs; ALT, AST, alkaline phosphatase [ALP], and total bilirubin), viral serologies (hepatitis B surface antigen, hepatitis C antibody, HBV DNA, HCV RNA), and autoimmune markers (antinuclear antibody [ANA], anti-liver–kidney microsomal antibody [ALKMA], anti-smooth muscle antibodies [ASMA], anti-mitochondrial antibodies [AMA-M2], and IgG levels).

The time to onset of liver enzyme elevation (ALT, AST) as well as peak enzyme values were recorded for each patient. Peak values were defined as the single highest recorded value at any timepoint during follow-up, verified against the complete serial monitoring dataset (Table S1). Time to biochemical normalisation was measured from the day of peak transaminase elevation to the first day on which both ALT and AST returned within the reference range. The R value was calculated as R = (ALT/ULN) ÷ (ALP/ULN), and liver injury classified as hepatocellular (R ≥ 5), cholestatic (R ≤ 2) or mixed (R = 2–5).¹³

Histopathological Evaluation

All liver biopsies were independently reviewed by three pathologists (R.M., R.S., S.M.), each blinded to the clinical history. Haematoxylin and eosin (H&E), Masson's trichrome and cytokeratin 7 (CK7) stained sections were examined in every case, and each biopsy was classified as hepatitic, cholestatic or mixed. Portal inflammation (graded as minimal, mild, moderate or severe), interface activity (0–4), lobular necroinflammatory activity (1–4) and fibrosis (0–6) were graded using the modified Ishak system;¹⁴ the equivalent METAVIR fibrosis stage (0–4) is also reported.¹⁵ Periportal inflammatory spill-over without destruction of the limiting plate (Ishak interface score 1) was recorded separately from definite interface hepatitis (score ≥2), as the former commonly accompanies acute lobular injury. The inflammatory infiltrate was assessed semi-quantitatively as predominantly lymphocytic, plasma cell-rich or eosinophil-rich. Bile duct injury, including lymphocytic cholangitis and epithelial degenerative change, was documented, and ductopenia defined as loss of interlobular bile ducts in more than 50% of portal tracts on H&E and CK7. The number of portal tracts was recorded in each biopsy as a measure of adequacy. Confluent, perivenular and bridging necrosis were recorded, with the proportion of parenchyma involved by confluent necrosis estimated on H&E as none or <5%, 5%–10%, 10%–30%, or >30%. Steatosis, granulomas, cholestasis and bilirubinostasis (canalicular or hepatocellular bile plugs) were noted when present.

Results

Clinical Findings

Nine female patients with advanced breast cancer receiving CDK4/6 inhibitor therapy underwent liver biopsy for abnormal liver tests; two were excluded because of concurrent complementary-alternative medication use with potential hepatotoxicity. The remaining seven (ribociclib, n = 6; palbociclib, n = 1) met inclusion criteria for CDK4/6 inhibitor-induced DILI, with RUCAM scores ≥6 (mean 8; range 7–9). Ages ranged from 36 to 72 years (mean 50.4; median 48) and all had hormone receptor-positive carcinoma (Table 1). Ribociclib was given at 600 mg and palbociclib at 125 mg once daily, both on a 3-weeks-on/1-week-off schedule; all patients were on standard doses when hepatotoxicity was identified and none had undergone a prior dose reduction (Table S2).

Note: 21/7 = 3 weeks on, 1 week off. In every patient, the CDK4/6 inhibitor was withdrawn on identification of grade ≥2 transaminase elevation at the next scheduled monitoring visit; concurrent endocrine therapy was continued throughout. All patients were receiving standard label doses at the time hepatotoxicity was identified and none had undergone a prior dose reduction.

Abbreviations: DILI, drug-induced liver injury; ER, oestrogen receptor; G, grade; HER2, human epidermal growth factor receptor 2; IDC, invasive ductal carcinoma of no special type; IHC, immunohistochemistry; ILC, invasive lobular carcinoma; LN, lymph node; PgR, progesterone receptor.

Peak ALT ranged from 243 to 1588 U/L (mean 674) and peak AST from 106 to 1059 U/L (mean 480) (Table 2, Figure 1). Total bilirubin exceeded the reference range in two patients, markedly in Patient 3 (9.6 mg/dL) and mildly in Patient 5 (2.8 mg/dL). ALP was normal or only mildly elevated (mean 190 U/L). Where ALP exceeded the reference range (Patients 4–7), it peaked concomitantly with or within 1–2 weeks after the transaminase peak, consistent with a hepatocellular-predominant injury with secondary biliary involvement. ANA titres ≥1:160 were found in three patients, none of whom had other markers of autoimmune liver disease; one patient had isolated low-titre anti-dsDNA without clinical SLE. The mean interval to first enzyme elevation was 60 days (median 45; range 20–114) and to peak elevation 81 days (median 60; range 20–188). Four patients (1, 5, 6 and 7) reached both within approximately 2 months; onset was later in the others; Day 68 in Patient 3, Day 100 in Patient 2 and Day 114 in Patient 4, with peaks at Days 104, 120 and 188. R value analysis indicated a hepatocellular pattern (R ≥ 5) in 5/7 (71.4%) and a mixed pattern in 2/7 (28.6%), ranging from 2.6 to 33.7.

Note: Values in parentheses denote the laboratory reference (normal) range. Peak values represent the single highest recorded value at any timepoint during follow-up, verified against the complete serial monitoring dataset (Table S1). ‘Days to peak’ refers to peak transaminase (ALT/AST) elevation. ‘Days to normalisation’ is measured from the day of peak transaminase elevation to the first day on which both ALT and AST returned within the reference range. R value = (ALT/ULN) ÷ (ALP/ULN); hepatocellular R ≥ 5, mixed R 2–5, cholestatic R ≤ 2.

Abbreviations: ALP, alkaline phosphatase; ALT, alanine transaminase; ANA, antinuclear antibody; AST, aspartate transaminase; Bil(T), total bilirubin; dsDNA, double-stranded DNA; NA, not available; RUCAM, Roussel Uclaf Causality Assessment Method.

The CDK4/6 inhibitor was discontinued in all patients on identification of grade ≥2 transaminase elevation at the next scheduled visit while concurrent endocrine therapy was continued throughout. Corticosteroids were given in three patients for persistent transaminase elevation and in one (Patient 3) as an empirical therapeutic trial. Steroids began a median of 32 days after withdrawal (range 5–45), as fixed-dose oral prednisolone 50–100 mg/day (Patient 1 received intravenous methylprednisolone first; Table S2). Enzymes normalised in all patients, 30–129 days after peak (mean 63; median 61). Six patients (85.7%) were switched to an alternative CDK4/6 inhibitor, four to abemaciclib and two to palbociclib, without recurrence over a median follow-up of 18.5 months (range 5–34). Patient 6 was not switched and was lost to follow-up after 9 months.

Histological Findings

Biopsies were adequate in all cases, with a mean of 17 portal tracts (range 12–27); findings are summarised in Table 3. The predominant pattern was acute hepatitis in five patients (71.4%), with mild lobular hepatitis and subacute hepatitis with early fibrosis in one each (14.3%). All five acute cases showed confluent perivenular (zone 3) necrosis and three also showed bridging necrosis (Figure 2). Confluent necrosis involved more than 30% of the parenchyma in Patient 1, 10%–30% in Patients 3, 5 and 6, and 5%–10% in Patient 7; none was identified in Patients 2 or 4. The necrotic zones were consistently accompanied by a lymphomononuclear infiltrate, confirming the necroinflammatory nature of the injury. The patient with mild lobular hepatitis showed occasional lobular inflammation and resolving features, including perivenular ceroid-laden macrophages (Figure 3).

Note: Reference ranges: ALT 7–56 U/L, AST 15–46 U/L, ALP 38–126 U/L. Biopsy interval = days from first LFT abnormality to biopsy. All confluent necrosis was perivenular (zone 3). Definite interface hepatitis (Ishak interface score ≥2) was present only in Patient 4; Patient 5 showed periportal spill-over (score 1) without true interface hepatitis. No biopsy showed significant steatosis, granulomas, or a cholestatic (biliary-dominant) pattern. Canalicular bilirubinostasis was absent in all biopsies except Patient 3, which showed focal centrizonal canalicular bile plugs.

Abbreviations: ALT, alanine transaminase; ALP, alkaline phosphatase; AST, aspartate transaminase; F, METAVIR fibrosis stage; LFT, liver function test; Ishak, modified Ishak fibrosis stage.

Portal inflammation was moderate in two biopsies (28.6%), mild in four (57.1%) and minimal in one (14.3%). Lymphocytic cholangitis and bile duct injury were present in five (71.4%) (Figure 4) with mild to moderate ductular reaction, both corroborated by CK7; no biopsy showed ductopenia. Periportal spill-over of inflammation was seen in five biopsies (71.4%), but definite interface hepatitis (Ishak score ≥2) in only one (Patient 4), which also showed periportal fibrosis (Ishak stage 1; METAVIR F1) (Figure 5). No other biopsy showed significant fibrosis. There was no significant steatosis or granulomas, and no cholestatic (biliary-dominant) pattern. Canalicular bilirubinostasis was absent in six biopsies; in Patient 3, a few bile plugs were present in centrizonal (zone 3) areas; this patient also had the highest total bilirubin (9.6 mg/dL).

Individual Case Descriptions

Two patients are described separately. Hepatic injury in Patient 6 (the only patient treated with palbociclib) was morphologically indistinguishable from that seen with ribociclib: a hepatocellular pattern (R = 7.7) with confluent zone 3 necrosis and lymphocytic cholangitis, but no bridging necrosis or fibrosis. She recovered fully on drug withdrawal alone.

The time course and morphology differed from the rest in Patient 4. Her baseline metabolic evaluation was normal, with no features of MASLD and negligible alcohol intake. Serial liver tests were normal to Day 85, with a first mild elevation at Day 114 (ALT 69 U/L) and a marked rise at Day 121; no gap between measurements exceeded 30 days. Biopsy showed moderate portal lymphoplasmacytic inflammation with definite interface activity, bile duct injury and periportal fibrosis, without confluent or bridging necrosis. Given the normal baseline and absence of pre-existing liver disease, we consider ‘chronic hepatitis’ inappropriate here and have re-designated this pattern as subacute hepatitis with early fibrosis. Notably, this biopsy was obtained 121 days after the first abnormality and 47 days after the peak, by which time corticosteroids had been started and liver tests were improving; both the late sampling and prior immunosuppression may have contributed to its fibrosing rather than acutely necrotic appearance.

Clinicopathological Correlations

Among patients with an acute hepatitic pattern, the magnitude of transaminase elevation broadly paralleled the extent of necrosis, the highest peak ALT (Patient 1, 1588 U/L) corresponding to the most extensive confluent necrosis (>30%). Yet, the correspondence was not exact: bridging necrosis was present in Patient 5 at a peak ALT of 367 U/L but absent in Patient 6 at 783 U/L, and Patient 4 showed marked enzyme elevation without confluent necrosis.

Corticosteroid use did not track with the extent of necrosis: only two of the four steroid-treated patients showed confluent or bridging necrosis, whereas all three who recovered on withdrawal alone did so despite confluent necrosis being present. The determinant was the biochemical trajectory, failure of transaminases to fall after withdrawal, rather than any histological feature. Extensive necrosis did not by itself mandate immunosuppression, nor did a relatively bland biopsy preclude it. ANA positivity likewise showed no relationship to histological severity: the highest titre (Patient 2, >1:2560) accompanied the mildest biopsy, while the most extensive necrosis occurred in an ANA-negative patient. Lymphocytic cholangitis occurred predominantly in patients with a hepatocellular R value and did not track with isolated ALP elevation, suggesting that the biliary injury is a secondary component of a predominantly hepatocellular process.

Discussion

CDK4/6 inhibitors have transformed the treatment landscape for hormone receptor-positive, HER2-negative breast cancer.³ Hepatotoxicity is an important dose-limiting adverse event of this class.^{1, 3, 4} Pharmacovigilance data from the FDA Adverse Event Reporting System (FAERS) demonstrate significant disproportionality signals for DILI, based on the reporting odds ratio (ROR), for ribociclib (ROR = 2.60; 95% CI: 2.48–2.72) and abemaciclib (ROR = 2.37; 95% CI: 2.18–2.58), whereas no signal is observed for palbociclib (ROR = 0.70; 95% CI: 0.67–0.73).¹⁶ In a recent review, Wong et al. concluded that ribociclib carries the highest risk of hepatotoxicity, palbociclib the lowest and abemaciclib an intermediate risk profile.⁵

Ribociclib-associated hepatotoxicity may involve inhibition of the bile salt export pump (BSEP) and extensive cytochrome P450 3A4 (CYP3A4)-mediated metabolism, leading to the generation of reactive intermediates.¹⁷ Network analysis of drug–gene interactions has identified key genes such as STAT3, HSP90AA1 and EP300 as potential mediators of CDK4/6 inhibitor hepatotoxicity.¹⁶

Although palbociclib is considered to have the most favourable hepatic safety profile among CDK4/6 inhibitors, severe cases of hepatotoxicity have been rarely reported. In the PALOMA-3 trial, one patient in the palbociclib arm developed fatal hepatic failure in the setting of progressive liver metastases, and isolated reports describe severe acute hepatitis attributed to palbociclib.^18-21

Despite growing clinical and pharmacokinetic insights into CDK4/6 inhibitor-associated hepatotoxicity, comprehensive histopathological characterisation remains limited. Only a few reports describing ribociclib-related liver injury have been published to date.^{6, 8, 9, 22} To the best of our knowledge, this series of seven patients represents the first systematic clinicopathological study of histologically characterised CDK4/6 inhibitor-induced liver injury.

Our histopathological analysis demonstrates a strikingly consistent pattern of severe acute hepatocellular injury, marked by prominent lobular inflammation and confluent perivenular (zone 3) necrosis in most biopsies (71.4%), with three cases also showing bridging necrosis. These findings closely parallel previously reported descriptions of ribociclib-associated acute hepatitis.^{8, 9, 22} Patients exhibiting this acute hepatitic pattern uniformly showed early onset of hepatotoxicity (20–68 days after therapy initiation) and more rapid biochemical recovery (30–72 days after peak). In contrast, the single patient with a subacute pattern with early fibrosis (Patient 4) had no pre-existing liver disease and experienced a markedly delayed onset of LFT elevation (114 days), peak enzyme levels at 188 days (ALT 612 U/L; AST 504 U/L), and the longest recovery interval (129 days) following drug withdrawal and steroid administration.

A notable finding in our series was the high incidence of bile duct injury (71.4%), characterised by lymphocytic cholangitis, which is supported by emerging evidence implicating cholangiocyte vulnerability in DILI.²³ Cholangiocytes possess limited detoxification capacity owing to lower cytochrome P450 activity and reduced glutathione reserves, which may predispose them to toxic injury.²³ The accompanying ductular reaction likely represents a compensatory regenerative response to epithelial damage.²³

Elevated ANA titres in three patients raised the possibility of drug-induced autoimmune hepatitis (DI-AIH), as described in recent ribociclib case reports.^{24, 25} However, none of the biopsies from the ANA-positive patients demonstrated definite interface hepatitis in the context of acute injury, plasma cell-rich infiltrates or other features of true AIH. Conversely, the only biopsy showing definite interface activity with a plasma cell-rich infiltrate (Patient 4) came from an ANA-negative patient with a normal serum IgG. Taken together, these observations indicate that ANA positivity alone should not be over-interpreted without corroborative histological findings, and that histological features suggestive of autoimmunity may occur in its absence.

The histopathological features observed in our cohort share some similarities with immune checkpoint inhibitor (ICI)-induced hepatotoxicity.^{26, 27} Both demonstrate a predominantly lobular pattern of hepatocellular damage, but important distinctions emerge in severity and distribution. In our cohort lobular inflammation was markedly more severe, frequently accompanied by confluent zone 3 necrosis and, in several cases, bridging necrosis, whereas the ICI series described by Zen et al. and Cohen et al. typically exhibited scattered foci of lobular necrosis without extensive zone 3 confluent necrosis.^{26, 27} Biliary lesions in ICI-related hepatotoxicity are more common than was once appreciated, particularly with second-generation checkpoint inhibitors.^{28, 29} Lymphocytic cholangitis was present in 71.4% of our biopsies, which, together with the conspicuous ductular reaction, may reflect a particular vulnerability of cholangiocytes to BSEP inhibition and reactive metabolites. In every case, the bile duct damage accompanied a predominantly hepatocellular injury. Importantly, this bile duct injury occurred as a secondary component of a predominantly hepatocellular process; we did not observe a cholangiopathic (biliary-dominant) pattern of injury analogous to primary biliary cholangitis in any case. Steatohepatitis and bland cholestasis, described in ICI-associated injury, were likewise absent.^{26, 27}

The shared morphological and immunological features between CDK4/6 inhibitor and ICI hepatotoxicity suggest overlapping pathways of immune dysregulation. However, the clinical course appears to differ: ICI-induced liver injury resolves spontaneously in approximately one-third to one-half of patients without corticosteroid therapy, whereas 57% of patients in our series required corticosteroids.^{25, 26} This finding parallels the French REFHEPS multicenter study, in which 41% of CDK4/6 inhibitor-associated DILI cases (predominantly ribociclib) required immunosuppression with corticosteroids.⁴ Consistent with systematic reviews and pharmacovigilance analyses, our observations support the need for close monitoring and early consideration of corticosteroid therapy in patients with severe ribociclib-induced liver injury, particularly when transaminase elevation fails to improve after withdrawal of the drug.^{3, 4, 16}

Our finding that 85.7% of patients were successfully switched to an alternative CDK4/6 inhibitor without recurrence of hepatotoxicity, over a median follow-up of 18.5 months, aligns with recent reports and reviews describing the safety and feasibility of switching to palbociclib or abemaciclib following ribociclib-induced grade III–IV hepatotoxicity.^{5, 30, 31} These observations suggest limited cross-reactivity among CDK4/6 inhibitors and support the concept that hepatotoxicity may be agent-specific rather than class-wide.³⁰ This inference should nonetheless be tempered: the three agents share several pharmacological properties, including CYP3A4-mediated metabolism and biliary elimination pathways, so class-level mechanisms may still contribute to susceptibility in individual patients.

The detailed histopathological characterisation provided by our study may aid pathologists in recognising the distinctive injury patterns associated with CDK4/6 inhibitor-related DILI, thereby improving diagnostic confidence in clinically complex cases.³¹ Baseline liver function evaluation and scheduled monitoring during treatment remain prudent, with drug discontinuation guided by the severity and trajectory of liver test abnormalities. While the precise criteria predicting benefit from corticosteroid therapy in CDK4/6 inhibitor-induced liver injury remain to be established in prospective studies, our observations and those from the REFHEPS cohort suggest that early consideration of immunosuppression may be warranted in patients whose transaminases fail to improve after drug withdrawal; in our series this biochemical trajectory, rather than the extent of necrosis on biopsy, was what prompted treatment.

Limitations of the study include the retrospective design, small cohort size and variability in monitoring intervals and biopsy timing. In particular, the biopsy in Patient 4 was obtained late in the clinical course and after corticosteroid therapy had begun, so the fibrosing appearance of that biopsy may reflect the stage of evolution and prior immunosuppression as much as an intrinsically distinct pattern of injury; this single case should therefore be interpreted with caution. Prospective studies with larger, standardised cohorts are needed to validate these findings and identify predictive biomarkers of CDK4/6 inhibitor-related hepatotoxicity.

To conclude, CDK4/6 inhibitor-induced liver injury represents an important adverse event with a characteristic pattern of predominantly acute hepatocellular injury, predominantly involving zone 3, which may be severe in some cases. Despite the intensity of biochemical and histological injury, most patients experience full recovery with drug discontinuation and, when indicated, corticosteroid therapy. The low incidence of recurrent hepatotoxicity on switching to an alternative CDK4/6 inhibitor underscores the limited cross-reactivity among agents and supports continued use of this drug class through individualised therapeutic adjustments. Characterising the histopathological spectrum of CDK4/6 inhibitor hepatotoxicity can enhance diagnostic confidence and guide clinical management in patients receiving these therapies for metastatic breast cancer.