Immune checkpoint inhibitors (ICIs) have changed the landscape of systemic anticancer therapy (SACT) over the past decade [1]. Their use has been associated with improved survival in a wide variety of human solid tumours and is increasingly extending beyond the treatment of metastatic disease into adjuvant, and even neoadjuvant settings [2, 3]. The most widely-used ICIs target the programmed death (ligand) 1 and cytotoxic T lymphocyte 4 axes, and may be utilised as monotherapy or in combination. The toxicity profile of ICIs is distinct from that of cytotoxic chemotherapy, insofar as toxicity is less predictable and the range of potential ICI-induced immune-related adverse events (irAEs) is far broader, with the potential to affect almost any organ system. In the absence of higher-quality data, clinical guidance for some of the rarer irAEs remains based on case reports and expert opinion [4]. Grading of irAEs is as per the Common Terminology Criteria for Adverse Events version 5.0 [5].

Any-grade ICI-induced hepatotoxicity (ICI-H) is seen among approximately 5–10% of patients treated with ICI monotherapy (Grade 3: 1–2%) [6, 7]. A subset of patients with more severe ICI-H develop ICI-induced cholangitis (ICI-C), in which liver function test (LFT) abnormalities are predominantly of γ-glutamyltransferase and alkaline phosphatase as opposed to the transaminases, aspartate aminotransferase and alanine aminotransferase [8]. Patients who develop ICI-C typically have ICI-H identified initially, and ICI-C is subsequently diagnosed by ultrasound or cholangiopancreatography, the latter of which may be either endoscopic or magnetic resonance-based (ERCP, MRCP). Three subtypes of ICI-C are described, being typically defined by imaging and in some cases correlated histologically—large duct cholangitis (often termed sclerosing cholangitis), small duct cholangitis, and mixed-type [8]. Treatment of ICI-C is typically with ursodeoxycholic acid (UDCA) to facilitate biliary drainage, and immunosuppression to arrest the ductal inflammation. Corticosteroids are consistently utilised, while additional agents including mycophenolate mofetil (MMF), azathioprine, tacrolimus, and tocilizumab show greater case-to-case variability [8]. We sought to describe three adult cases of ICI-C managed at our centre, including biochemical and radiological findings.

The cases described here illustrate some of the heterogeneity seen in cases of ICI-C, in terms of severity, biochemical onset, radiological findings, requirements for treatment and clinical outcome. Many cases reported in the literature report immunosuppressant medication use as present/absent, whereas here specific doses of medications received are specified, as is the chronology of their use which is useful, especially given the relapsing and remitting nature of many cases of ICI-C [8, 10]. The inclusion of radiologic images that can be aligned with pharmacological management is also a strength of the present series [8, 10].

Diagnosis of ICI-H can be challenging in patients with known liver metastases, or concurrently receiving other potentially hepatotoxic SACT. The temporal pattern of a patient’s LFTs is likely to be useful: chemotherapy, unlike ICIs, is likely to produce a transient derangement which resolves on holding the offending agent; case 2 in the present series illustrates this, with baseline liver metastases and treatment with ICI-chemotherapy. Case 2 also illustrates the potential severity of ICI-C, with rapidly progressive radiological appearances (Figure 2C–E) alongside gross hyperbilirubinaemia (peak value 453 μmol/L) and coagulopathy (1.4-fold increase in prothrombin time versus baseline) contributing to subarachnoid haemorrhage, and subsequently death.

Case 1 is notable for its severity and refractoriness to multiple treatments, with apparent stabilisation following infliximab therapy; this finding warrants subsequent exploration particularly as guidance from the European Society of Medical Oncology (ESMO) advises avoiding infliximab in ICI-H due to a possible association with idiosyncratic liver failure, though this has not been observed in the context of irAEs [4]. Indeed, ESMO guidance only specifically recommends UDCA and corticosteroids in ICI-C [4]; the American Gastroenterological Association advises considering infliximab for severe ICI-H on a case-by-case basis with no specific mention of ICI-C, while American Society of Clinical Oncology guidelines state infliximab is contraindicated in ICI-H [11, 12]. These cases also illustrate our institutional experience that for patients whose ICI-C is treated with initial success, the course is typically prolonged, requires multiple immunosuppressant medications, and attempting to taper these prematurely commonly leads to flaring of toxicity. We include our centre’s local guidance for diagnosis and management of ICI-H as supplementary materials.

Regarding the radiological findings of ICI-C, we note that in several of the cases described, rather than requiring MR to detect any signs of ICI-C, relevant abnormalities were detected using CT imaging in the first instance, which prompted MR for a more complete evaluation. Specifically, CT imaging displayed CBD enhancement in case 1, as well as enhancement of the CBD and gallbladder in case 3 (Figure 2E). Previous comparisons of imaging in pancreatobiliary disease have demonstrated the added value of MR for diagnosis, particularly in the evaluation of the extrahepatic bile ducts [13]. As radiologists often only have access to the clinical information provided on imaging requests rather than a full clinical history, clinicians should be aware to include details of ICI treatment and possible cholangitis on requests for CT, as well as MR imaging, if the clinical picture is consistent. Increased awareness of ICI-C among radiologists, particularly among those who do not specialise in hepatobiliary radiology, will further help to optimise the diagnostic process. We report that partially due to the cases described herein, our centre’s protocols for MR liver scans in patients who have received ICI were altered to include some sequences previously reserved for MRCP, to improve diagnosis of ICI-C at our centre. In the future, radiomics may assist in the rapid assessment of tumour response which could facilitate early cessation of therapies in patients who are not benefiting, which may help to prevent treatment-related toxicities in the metastatic and neoadjuvant settings [14, 15].

In this series, none of the patients underwent liver biopsy during their ICI-C management. Biopsy findings of cases reported elsewhere [8] have been demonstrated to vary depending on subtype—large-duct versus small-duct. There is currently no evidence for managing these subgroups differently, and so a biopsy is not routinely pursued at our centre. However, developing more optimal management approaches is likely to depend on a deeper understanding of the biochemical, radiological, and histological subtypes of ICI-C, and for this reason, prospective clinical trials including liver biopsy in a research context should be advocated for.

In summary, ICI-C is rare and severe toxicity that can present with varying tempo, provides characteristic findings on imaging, and often requires a complex immunosuppressive approach with tumour necrosis factor alpha-inhibiting agents illustrating therapeutic benefit.