Immune‐Mediated Liver Injury From Checkpoint Inhibitor: An Evolving Frontier With Emerging Challenges

ABSTRACT

Over the past decade, immune checkpoint inhibitors (ICIs) have transformed the treatment of cancer, though they come with the risk of immune‐related adverse (irAEs) events such as hepatotoxicity or Immune‐mediated Liver Injury from Checkpoint Inhibitors (ILICI). ILICI is a serious irAE that, when severe, requires cessation of ICI and initiation of immunosuppression. Cytotoxic T Lymphocytes (CTLs) play a central role in ILICI; however, they are just part of the picture as immunotherapy broadly impacts all aspects of the immune microenvironment and can directly and indirectly activate innate and adaptive immune cells. Clinically, as our understanding of this entity grows, we encounter new challenges. The presentation of ILICI is heterogeneous with respect to latency, pattern of injury (hepatitis vs. cholangitis) and severity. This review focuses on our knowledge regarding risk factors, presentation and treatment of ILICI including ILICI refractory to steroids. An emerging topic, the possibility of rechallenge while accepting some risk, in patients who experience ILICI but require immunotherapy, is also discussed. This review provides an update on the current knowns and unknowns in ILICI and highlights several knowledge gaps where studies are needed.

Summary.

• ICI‐induced liver injury is due to a disruption of immune tolerance but the mechanism is still largely unknown.

• Several risk factors have been identified such as female sex, young age, history of ICI therapy and ICI combination therapy.

• ILICI is heterogeneous with respect to the time interval between initiation of ICI and onset of toxicity, biological characteristics (hepatitis vs. cholangitis) and severity.

• Corticosteroid therapy should be tailored to the severity of the hepatitis. In steroid‐refractory hepatitis, MMF is a successful second‐line therapy, while third‐line therapy remains controversial.

• Reintroduction of immunotherapy after ILICI is possible in certain cases and is patient and circumstance dependent and should be discussed by a multidisciplinary team to assess risk vs. benefit.

Abbreviations

AIH

autoimmune hepatitis

ALP

alkaline phosphatase

ALT

alanine transferase

AST

aspartate aminotransferase

CNI

calcineurin inhibitor

CTCAE

Common Terminology Criteria for Adverse Events

CTLA‐4

cytotoxic T‐lymphocyte‐associated protein‐4

CTLs

cytotoxic T‐cells

DILI

drug‐induced liver injury

FGL1

fibrinogen‐like protein‐1

HCC

hepatocellular carcinoma

ICIs

immune checkpoint inhibitors

IFN

interferon

IL

interleukin

ILICI

immunotherapy‐induced liver injury

irAEs

immune‐related Adverse Events

IVIG

intravenous immunoglobulin.

JAK

Janus kinase

LAG‐3

lymphocyte‐activation gene‐3

MALFD

metabolic dysfunction‐associated liver disease

MMF

mycophenolate mofetil

MRCP

magnetic resonance cholangiopancreatography

NK

natural killer

PD‐1

programmed cell death protein‐1

PDL‐1

PD ligand 1

PSC

primary sclerosing cholangitis

TIGIT

T‐cell immunoglobulin—ITIM domain

TIM3

T‐cell immunoglobulin and mucin domain‐containing protein 3

TNF

tumour necrosis factor

UDCA

Ursodeoxycholic Acid

ULN

upper limits of normal

VISTA

V‐domain Ig suppressor of T‐cell activation

VSIG‐1

V‐Set and immunoglobulin

1. Introduction

The liver is an immunotolerant organ. Due to constant exposure to foreign antigens, it has evolved several mechanisms to dampen the immune response. These mechanisms are many fold, involve hepatocytes, innate, adaptive, and non‐parenchymal liver cells, and importantly, they include the immune checkpoints. The immune checkpoints are ligand–receptor pairs that regulate the immune response and exert control over immune cells to prevent their aberrant activation. As such, they have a critical role in maintaining self‐tolerance and preventing autoimmunity. Malignant cells exploit these checkpoints to evade the immune system. The expression of immune checkpoints can lead to T‐cell exhaustion, a decline in T‐cell proliferation and reduced T‐cell function, allowing tumors to grow. Therefore, immune checkpoint inhibitors (ICIs) have been developed to block these immune‐regulatory pathways and augment immune responses. Since the liver relies on these self‐tolerance mechanisms to maintain its state of immune privilege, an unintended side effect of ICIs is immune‐mediated liver injury from checkpoint inhibitors, herein, ILICI.

The best studied immune checkpoints include the programmed cell death protein‐1 (PD‐1) and PD ligand 1 (PDL‐1) and the cytotoxic T‐lymphocyte‐associated protein‐4 (CTLA‐4) pathways [1]. There are currently over 10 FDA‐approved ICIs but given resistance and the side effects of these antibodies, there are more than 100 immunotherapy targets under development as next‐generation immunomodulators [2, 3]. Some of these targets include T‐cell immunoglobulin and mucin domain‐containing protein 3 (TIM3), lymphocyte‐ activation gene‐3 (LAG‐3), and T‐cell immunoglobulin and ITIM domain (TIGIT), fibrinogen‐like protein‐1 (FGL1), V‐Set and immunoglobulin domain containing (VSIG‐1), and V‐domain Ig suppressor of T‐cell activation (VISTA), among others.

ICIs are currently critical tools in the oncology toolbox for treatment of various malignancies. It is estimated that over 50% of cancers will be exposed to immunotherapy due to their high efficacy and better tolerability compared to conventional and targeted chemotherapy. Despite their undeniable effectiveness, the consequences and immune‐related Adverse Events (irAEs) such as ILICI are currently inevitable. Therefore, it is critical to understand, prevent, diagnose, and treat these unintended side effects. And to complicate matters further, unlike other forms of drug‐induced liver injury (DILI) where discontinuing the drug is the best solution, in cancer patients where immunotherapy is working, the benefits are significant. Therefore, finding solutions towards safe continuation of the drug is imperative.

2. Mechanism of ILICI

Immunotherapy‐induced liver injury is thought to be due to the activation of cytotoxic T‐cells (CTLs) that inadvertently target the liver. However, immunotherapy can have very broad effects on the immune microenvironment and can directly or indirectly effect cells other than CTLs, such as B cells, FoxP3+ Tregs, T helper cells, and even innate immune cells such as macrophages and dendritic cells [4] (Figure 1). In addition, ICIs modify the tumor microenvironment and circulating chemokines and cytokine levels, in fact, baseline and post‐treatment cytokine levels have been associated with the development of irAEs such as ILICI [5]. Various cytokines have been shown to be upregulated in patients and mice with ILICI, and these include interleukin (IL)‐6, IL1b, interferon (IFN) ‐γ, tumour necrosis factor (TNF)‐α and chemokines including CXCL9, CXCL10, CXCL11 and CXCL13 [5].

FIGURE 1.

Immune Checkpoint Inhibitors (ICIs) alter the immune microenvironment in various ways and by affecting multiple immune cell types.

In some tumors, neoepitope formation and chronic inflammation, break immune tolerance to tissue‐specific self‐antigens leading to serious irAEs [6]. One of the current leading hypotheses for immunotherapy side effects is the concept of epitope spreading. It is hypothesised that the lysis of tumors causes the release of proteins, forming of neoantigens which then drives irAEs [7]. Epitope spreading is the diversification of the initial T‐cell response against these novel epitopes and neoantigens that diverge from the original epitopes targeted. Skin, intestines and the liver are among the most commonly affected organs. Why T‐cells would specifically target neoantigens in certain tissues and organs and not others is not clear, as many neoantigens are present across multiple tissues and cell types. ICIs induce diversification of T‐cell clones and expansion of CTLs, and studies have shown similarities between tumour infiltrating CTLs and tissues affected by irAEs suggesting a mechanistic role for epitope spreading in irAEs [7].

B cells are in constant crosstalk with other cells within the tumor immune microenvironment. CD20+ B cells and their crosstalk with CD8+ T‐cells play a crucial role in the anti‐tumor immune response through costimulatory signalling and by promoting cytotoxic T‐cell survival and proliferation independently of antigen presentation through CD27/CD70 interactions [8]. B cells also respond to immunotherapy, and activation of B cells portends a favourable anti‐tumour response. Blockade of immune checkpoints has been associated with distinct changes in the B cell compartment, increase in CD21^lo B cells and increase in plasmablasts that correlate with the development of irAEs [9, 10]. CTLA4 is a critical checkpoint in B cells, and its loss results in immune dysregulation and autoimmunity [4]. PD‐1 is also expressed in B cells and the blockade of the PD‐1 pathways increases their proliferation, activation and the production of inflammatory cytokines [9, 11]. The physiologic importance of CTLA4 and PD‐1 in keeping B cells in check suggests that the effect of ICIs on B cells likely contributes to the development of irAEs and ILICI. Although small studies have reported autoantibody formation during irAEs, the significance of these and the role of B cells during these events requires further study [12].

Tregs suppress anti‐tumour immunity, and therefore checkpoint blockade helps overcome this by attenuating Treg function and decreasing the Treg/T helper (Th) ratio [13]. This suppression of T reg numbers and function promotes autoimmunity, and in preclinical models of irAEs there is a negative correlation between Treg numbers and side effects. In concert with this, ICIs promote increased inflammatory CD4+ Th17 cells and Th1 proliferation resulting in increased IL‐2, IFN, and TNF production to activate CTLs, as well as innate immune cells such as macrophages and natural killer (NK) cells [7, 13].

An increase in the Th17 population and IL‐17 has also been reported post ICI therapy and with certain irAEs [14, 15]. These proinflammatory cytokines can subsequently activate an innate immune response by recruiting NK cells and macrophages contributing to liver injury. Macrophages play a role in ILICI, and this has been shown in both humans and mice models [16, 17]. Liver biopsies from patients with ILICI revealed the presence of both CD8+ CTLs and CCR2 + CD68+ macrophages in proximity to areas of injury, suggesting that immune cells work in concert and the adaptive innate crosstalk is important in the pathogenesis of ILICI [16]. This crosstalk was confirmed in a mouse model of ILICI where macrophage and T‐cell interactions leading to NLRP3 inflammasome activation were shown to drive hepatocyte apoptosis [17]. Indeed, while the effector cells causing liver injury seem to be the CD8+ T‐cells, as depletion of these cells significantly dampens injury, there still exists a macrophage infiltrate in the liver. This suggests that macrophage activation and recruitment to the liver are independent of CD8+ T‐cell infiltration [17]. There are still many unknowns in the mechanism of ILICI, and it is important to decipher the contribution of the various cell types and signalling pathways to move us towards finding therapeutic solutions.

3. Risk Factors for ICI Toxicity

The identification of risk factors for irAEs presents a significant challenge due to the high variability of these events. The use of two ICIs increases the risk of ILICI [1]. It is well known that females are more susceptible to autoimmune disease. Indeed male sex portends a more favourable response to immunotherapy, and female sex has been associated with a higher incidence of irAEs [18, 19, 20]. Analysis of over 23 000 patients from 202 trials experiencing adverse events revealed a 49% increased risk among women receiving immunotherapy (OR 1.49; 95% CI,1.24 to 1.78; p = 0.001) [18]. This finding was also reported in a subgroup of patients treated with immunosuppressive therapy for an autoimmune disease who received ICI. The results indicated that female patients were more likely to develop irAEs, although in this small study the risk of exacerbation of existing autoimmune disease was lower [21]. While larger studies have shown this in diverse cohorts, some small studies have not been able to replicate this finding [22]. Not surprisingly, female sex has been shown to be a predictor of ILICI in large cohorts and real‐world studies [23, 24]. Given the well‐known effect of sex differences in immune‐oncology and the higher incidence of autoimmune disease in females, the hypothesis of a role of sex as a risk factor for toxicity appears plausible [20]. However, much of this data is retrospectively extrapolated from randomized controlled trials and cohort studies, highlighting the need for further prospectively collected data to reach definitive conclusions.

Age has also been suggested as a risk factor for irAEs. A meta‐analysis of 13 studies found that younger age was significantly associated with the risk of any grade and of grade ≥ 3 ILICI [25]. Other identified risk factors were the use of two ICIs (dual checkpoint blockade), a history of prior ICI exposure, higher aspartate aminotransferase (AST), higher baseline alanine transferase (ALT) and lower baseline alkaline phosphatase (ALP), malignant melanoma, and high lymphocyte count [24, 25, 26].

Regarding the association between underlying cancer and incidence of irAEs, it is important to note that in large meta‐analyses, the highest mean all‐grade adverse events incidence was observed in melanoma. However, the mean incidences of all‐grade and grade 3 or higher adverse events were similar across different cancer types [27]. While this holds true for different cancer types, it likely does not apply to patients with hepatocellular carcinoma (HCC) and, consequently, underlying liver disease. Indeed, HCC and pre‐existing hepatic disease appear to be associated with a higher rate of ILICI compared to the treatment of non‐hepatic cancers [28, 29]. Furthermore, other studies on patients with underlying liver disease have indicated that chronic liver diseases such as hepatitis B and metabolic dysfunction‐associated liver disease (MALFD) may also be risk factors for ILICI [30, 31].

The diagnosis of pre‐existing autoimmune disease is associated with an increased risk for disease flares and irAEs occurrence, although this is highly variable according to the type of the disease and the ICI administered [32]. In general, the greater the immunosuppressive therapy received prior to the introduction of ICIs, the greater the risk of irAEs [32]. The risk of ILICI in patients with autoimmune disease has not been extensively studied. In a small European cohort of 22 patients with hepatic autoimmune diseases, treated with immunotherapy for various malignancies, 36% (8/22) experienced grades 1–2 irAEs, three of which were ILICI, and none experienced ≥ grade 3 injury [33]. In addition, a recent systematic review of 22 studies in patients with Child‐Pugh A and Child‐Pugh B cirrhosis found a similar incidence of treatment‐related adverse events regardless of Child‐Pugh classification [34].

As ICIs break tolerance, it is also feasible to hypothesize that they can predispose patients to idiosyncratic DILI from concomitant meds. Indeed, in one study, a combination of ipilimumab plus dacarbazine resulted in significantly more irAEs compared to ipilimumab alone (56.3% vs. 27.5%, p < 0.001) [35]. In addition to dacarbazine recently multiple reports of DILI following the addition of a small‐molecule inhibitor such as enzyme inhibitors and tyrosine kinase inhibitors to ICIs have been published. Although these are early reports, they point to the complexity of determining causality when multiple drugs are used concomitantly with or even following ICI therapy [36, 37, 38, 39, 40].

Recently, the Drug‐Induced Liver Injury Network (DILIN) published its experience on 57 cases of ILICI, adjudicated by an expert panel of hepatologists as high likelihood [41]. In this cohort, younger age was associated with a hepatocellular pattern of injury. In terms of genetic risk factors, the HLA alleles for autoimmune hepatitis (AIH) were not overrepresented in the ILICI patients. However, two host immune response genes (EDIL3 and SAMA5A) and three other genes (GABRP, SMAD3 and SLCO1B1) were associated with ILICI (OR 2.08–2.4, p < 0.01) [41].

Lastly, the identification of risk factors should not preclude the use of these drugs but rather guide the clinician in identifying higher‐risk patients for hepatotoxicity events and allow for personalized monitoring during therapy.

4. Clinical Presentation

The presentation of ICI‐induced liver injury is heterogeneous in terms of the time interval between initiation of therapy and onset of toxicity, in terms of clinical and biological features (hepatitis vs. cholangitis), and in terms of severity. An accurate specific assessment, along with the causality assessment for all DILI, is paramount for diagnosis and management (Table 1). Hepatic toxicity may occur after a single injection or several weeks after discontinuation of therapy, although the latter is very rare [42, 43]. The median interval between ICI and liver injury is between 3 and 14 weeks, with some variation depending on the drug used, according to some studies: shorter in patients treated with anti‐CTLA4 compared to those treated with anti‐PD1/anti‐PDL1 [44, 45, 46, 47]. The median latency of DILI in patients with hepatocellular injury in the well‐characterised DILIN cohort was 2–3 months, while latency was closer to 6 months in patients with cholestatic ILICI [41]. Although it is possible for ILICI to manifest after 6 months and even one year of therapy, the likelihood is much lower [24]. The late onset is explained by the prolonged activity of these drugs which exceeds their half‐life [48].

TABLE 1.

Key clinical considerations in immune‐mediated liver injury from checkpoint inhibitors (ILICI).

Key points	Comments
Rule out common causes of liver injury	Similar to other forms of DILI, viral, metabolic, alcohol, stone disease should be ruled out [72, 73]
Concomitant medications and recent pharmacological history	Previous exposure to ICI increases the risk of toxicity and the delay between last ICI dose, and new ICI introduction is of great importance due to the potential prolonged activity of the first drug [49]
Pre‐existing liver disease	Increased risk of toxicity in the setting of HCC and/or cirrhosis [31, 32, 33, 34]
Type of ICI	Association with different liver injury patterns such as cholestatic pattern being more common with anti‐PD1s such as pembrolizumab [54, 55, 56, 57, 58, 59, 60, 61]
Time interval between ICI injection and liver injury onset	Late onset (even after ICI discontinuation) does not exclude ICI given long half‐life and receptor occupancy [44]
ICI in conjunction with traditional antineoplastic agents	ICIs break immune tolerance and can predispose to DILI from other drugs such as dacarbazine [42, 43]
ICI in conjunction with small‐molecule inhibitors (SMI)	Concomitant or subsequent treatment with small‐molecule inhibitors such as Tyrosine Kinase Inhibitors (TKI) is associated with hepatotoxicity. Whether this is due to a break in immune tolerance predisposing to TKI/SMI hepatotoxicity or ILICI or an additive or synergistic effect due to dual therapy is not yet understood [38, 39, 40, 41, 42]
Signs and symptoms	Clinical symptoms may be associated with a more severe disease (e.g. jaundice, fever, oedema/ascites) and require different managements. Furthermore, they are associated with a higher risk of steroid‐refractory hepatitis [87]
Hepatocellular versus cholestatic pattern of DILI	Distinction between hepatocellular and cholestatic patterns can lead to consideration of different options. Namely UDCA has been used in some centres in Europe for cholestatic ILICI although this is not recommended by most guidelines [53]
Severity of injury	Relevant for the introduction, timing and dose of immunosuppressive therapy, as well as consideration of ICI reintroduction or rechallenge [1, 47, 75, 76, 77]
Imaging	Can avoid misdiagnosis by detecting disease progression, tumoural hepatic thrombosis and biliary obstruction. Also can help in follow‐up [62, 63]

Patients may be asymptomatic or present with non‐specific symptoms such as fever, nausea/vomiting, abdominal/back pain, arthralgia/myalgia and rash [47, 49], sometimes with jaundice and less commonly with signs of portal hypertension such as edema and/or ascites. This may be due to vascular changes in the liver parenchyma, nodular regenerative hyperplasia [50] or sinusoidal obstructive syndrome (formerly known as veno‐occlusive disease) [51] with clear features of portal hypertension alone, or generalized edema associated with polyserositis [52].

Like other types of DILI, the pattern can be hepatocellular, cholestatic or mixed; however, the hepatocellular pattern of injury is much more common (Table 2) [1, 53].

TABLE 2.

Hepatocellular versus cholestatic pattern of liver injury in ILICI.

Pattern of liver injury (R value = (ALT/ULN)/(ALP/ULN))
Hepatocellular (cytolytic) (R ≥ 5)	Radiology Normal liver Hepatomegaly Portal hypertension Periportal lymphadenopathy	Histology Lobular hepatitis with polymorphic infiltration Spotty or confluent necrosis predominantly in the centrilobular zone Endothelialitis Granulomatous hepatitis Fibrin ring granuloma
Cholestatic (R ≤ 2)
	Normal liver	Cholangitis with mixed inflammatory infiltrate
	Dilatation and strictures segmental or diffuse (sclerosing cholangitis aspect)	Lymphocytic cholangitis Bile duct dystrophy and proliferation
	Thickening of the bile ducts and gallbladder wall	Concentric fibrosis and complete obliteration of the bile duct
		Vanishing bile duct syndrome
		Granulomatous cholangitis
		Fibrin ring granuloma
		Cholestasis
Mixed (2 < R < 5)	All features mentioned above can be present	All features mentioned above can be present

Secondary Sclerosing Cholangitis (SSC) is a manifestation of ILICI and may involve small and/or large bile ducts or both and can present with periductal fibrosis on liver biopsy, cholangitis and intraepithelial inflammatory injury, and even vanishing bile ducts [54]. In laboratories, these patients have a very elevated ALP often in the thousands. There is hypertrophy of the bile duct wall and an absence of obstructive pathology [55, 56, 57]. Biliary strictures and areas of dilation are seen intra‐ and extra hepatically on imaging studies such as magnetic resonance cholangiopancreatography (MRCP) resembling primary sclerosing cholangitis (PSC) [54, 55, 56, 57, 58, 59, 60, 61]. This cholestatic pattern of DILI which mimicks PSC has been associated in particular with anti‐PD1 therapy, such as pembrolizumab and nivolumab, and responds poorly to immune suppression (discussed below) [54, 61, 62]. The hepatocellular pattern of liver injury appears to be more frequently associated with combination therapy (dual blockade) and anti‐CTLA4 agents [53]. The distinction seems important for differential diagnosis and management. Patients may experience mild to severe abnormalities in liver tests. In addition, jaundice and hyperbilirubinemia are seen in up to a quarter of patients with SSC which is associated with a more severe outcome [63].

Mortality from ILICI has been reported although this is generally rare [64, 65, 66]. Systematic reviews estimate the incidence of death from fulminant hepatitis to be between 0.5% and 0.07% [67, 68]. Wang et al. found 613 fatal ICI toxic events from 2009 through January 2018 in VigiLyze internationally, 22% of which were attributed to hepatitis [68]. In another study utilising the World Health Organisation's large VigiBase database of individual safety case reports, death from fatal hepatitis occurred in 19% (94/490) of cases, with only age ≥ 65 years being a risk factor [44]. Indeed, some patients exhibited evidence of multiple organ toxicity. Therefore, in the setting of underlying cancer, it is not readily apparent whether the ultimate cause of death in these studies was fulminant hepatitis and liver‐related mortality.

The diagnosis of ILICI, like other forms of hepatotoxicity, is a diagnosis of exclusion and requires ruling out other causes. Causality assessment also involves considering the clinical presentation, latency and careful study of other medications administered and is beyond the scope of this review [69, 70, 71, 72]. Importantly, the very long half‐life of these drugs and the fact that active drug levels persist in patients’ weeks after their last dose should be taken into account during DILI causality assessment.

5. Treatment

Treatment of ILICI depends on the severity of the presentation. Severity is categorized by the degree of elevation over upper limits of normal (ULN) and is most commonly categorized from 1 to 4 according to the Common Terminology Criteria for Adverse Events (CTCAE) classification, with severe ILICI encompassing grades 3–4 [1, 73]. Although this classification has recently been challenged, the utilisation of an updated Roussel Uclaf Causality Assessment Method has been proposed to show a lower incidence of severe cases, resulting in less steroid usage [74]. Mild transaminase elevation and low‐grade ILICI require discontinuation of the ICI and strict follow‐up, but not necessarily high‐dose corticosteroids, as spontaneous improvement has been observed [45, 47, 75]. Despite the fact that a precise cut‐off of liver tests for prognosis has not yet been identified, the kinetics of liver test evolution should guide treatment decisions.

Indeed, high doses of steroids are required in cases of hyperbilirubinemia or impending liver failure. Although there has been some recent controversy as a recent retrospective study on 215 patients with Grade ≥ 3, ILICI showed that there was no significant benefit from doses above 1.5 mg/kg/day [76]. The benefits of high‐dose steroids must be balanced against the risk of side effects such as infections and glycemic control [76, 77]. In general, steroids for irAEs are associated with improved overall survival, however, the long‐term impact of high doses of steroids on tumor progression is still controversial [78]. Early steroid use, within 60 days of initiating ICI, has been shown to be detrimental and is independently associated with lower survival benefit regardless of ICI continuation or discontinuation [78, 79]. This could be interpreted as evidence for the fact that systemic immunosuppression counteracts the beneficial effect of immune activation by immunotherapy. However, when interpreting these findings, variables such as steroid dosage and cancer stage should be considered. To avoid high doses of corticosteroids, budesonide may appear to be an attractive option [80]; however, its use in patients with acute flares of autoimmune hepatitis is inferior to prednisone [81] and cohort studies regarding its efficacy for ILICI are lacking.

For cases of mild cholestatic ILICI, Ursodeoxycholic Acid (UDCA) has been used and may help in improving serum liver tests, although still controversial and not routinely recommended [63, 71, 73, 82, 83]. The use of UDCA has not been established as a treatment for cholestatic DILI due to the lack of adequate evidence. However, it does not appear to be harmful [71, 84]. In general, UDCA is well tolerated, and there is minimal risk associated with its use in ILICI; however, large, randomized controlled trials are needed to determine its efficacy, optimal dose and duration of therapy [46, 53, 71]. Due to the risk of SSC, an MRCP is recommended for those with severe cholestatic liver injury or progressive jaundice to look for these rare cases that have recently been recognized as an important clinical entity due to the therapeutic challenges they pose. Pembrolizumab and nivolumab in particular have been implicated in this form of cholestatic ILICI, although other ICIs, and in particular PD‐1 inhibitors, can cause SSC [54, 55, 56, 57]. Corticosteroids in combination with UDCA have been used to treat cholestatic ILICI although the results are mixed, and many patients with ICI‐mediated cholangitis do not respond to steroids [54].

Risk factors for steroid‐refractory cases of hepatitis include previous liver comorbidity, hyperbilirubinemia, and general symptoms associated with hepatitis such as fever, jaundice, ascites and/or encephalopathy [85]. There are no randomized controlled trials or high‐quality studies with level‐one evidence, or FDA‐approved medications indicated for management of steroid‐refractory ILICI. The information summarized herein are a compilation of published case reports, case series, or retrospective experiences from various centres (Table 3).

TABLE 3.

Immunosuppressive therapy reported in the literature for ILICI. ^a

Line of therapy	Drug	Comments
First line	Corticosteroids	Steroids remain first line; however, there is no consensus regarding time of administration (according to severity), route, doses and time of escalating doses. In general, once liver enzymes decreased to less than 5× ULN, a slow taper over 30 days with close lab follow‐up is recommended [45, 74, 77, 83, 84]
	UDCA	Either alone or with corticosteroids for cholestatic or mixed liver injury pattern recommended by the European Society of Medical Oncology (ESMO), not in the US. There is no consensus in its use, but it is considered a safe drug [45, 83, 94]
Second line (recommended as second line by most guidelines)	MMF	AZA has been used [88, 90] but there are more data with MMF [24, 52, 85, 86, 87, 88, 89, 90, 91]
Third line (several options exist and have been reported in the literature, but experience is limited)	Tacrolimus	Calcineurin inhibitor. Several case reports showed its efficacy. Cyclosporine is used in other irAEs, not much data in ILICI [92, 93, 94, 95, 96, 97, 98]
	ATG	T‐cell‐depleting agent. Successfully used in several case reports [103, 104, 105]
	Tocilizumab	Interleukin‐6 (IL‐6) receptor blockers. Emerging strategy with several successful case reports [106, 107, 108, 109, 110]
	Plasma exchange +/− IVIG	Technique to remove circulating autoantibodies and other humoral factors +/− polyclonal serum IgG, anti‐inflammatory and immunomodulatory treatment. Emerging strategy with several successful case reports [84, 112, 113, 114, 115]
	Tofacitinib	Janus kinase (JAK) inhibitor. Just one report. More frequently used for other irAEs [102]

^a Many of the reports reviewed here are single cases or small case series where an agent was used to treat ILICI. There are no large prospective studies on this topic. Clinicians should use their own judgement when treating patients, and risk–benefit analysis should be undertaken in a multidisciplinary fashion before the off‐label use of these medications.

Patients who are refractory to corticosteroids should be treated with second‐line immunosuppressive therapy, the most commonly used drug being mycophenolate mofetil (MMF), which is successful in 82% of patients with hepatocellular injury [54, 85, 86, 87]. A recent meta‐analysis of 30 studies that included 1120 patients showed that the pooled ILICI response rate was 79% (95% CI 0.73–0.84) for treatment with corticosteroids and 93% (95% CI 0.79–1.0) for treatment with MMF [87].

Azathioprine has been reported as a successful therapy for steroid‐resistant ILICI presenting as hepatitis [88] but is less commonly reported in the literature in comparison to MMF. Several third‐line therapies have been described in case reports with some differences depending on the organ affected by the toxicity [89]. Calcineurin inhibitors (CNIs), tacrolimus and cyclosporine are commonly used in the liver transplant setting and refractory autoimmune hepatitis (AIH) and have the advantage of being familiar and easy to use for hepatologists. CNIs inhibit the expression of IL‐2 and prevent the differentiation and maturation of T‐cells. Tacrolimus has been shown to be safe and effective in managing steroid‐refractory ILICI in several case reports [90, 91, 92, 93, 94]. Cyclosporine has been effectively used in case reports of ILICI and several non‐liver irAEs including ICI‐induced dermatitis and anaemia [95, 96, 97]. Anti‐TNF antibodies that are commonly used on colitis from ICIs have also been used for the treatment of ILICI [98, 99], but due to potential hepatotoxicity and fear of liver decompensation, they are currently avoided in ILICI until further safety studies can be conducted. The Janus kinase (JAK) inhibitor, tofacitinib, which has been used for ICI‐induced colitis and myocarditis, has been used successfully in a few cases of steroid‐refractory hepatitis [100]. The use of T‐cell‐depleting agent, antithymocyte globulin, ATG has also been reported in patients with steroid side effects (steroid‐related psychosis), or those refractory to both steroids and MMF or insufficient response to MMF [101, 102, 103, 104]. Please note that while these case reports indicate these drugs are successful in controlling severe liver inflammation in some cases, the long‐term consequences of these therapies on the underlying cancer require study.

Two additional strategies seem to be emerging more recently as third‐line therapy: the use of tocilizumab or plasma exchange +/− intravenous immunoglobulin (IVIG). Tocilizumab, a monoclonal antibody that blocks the IL‐6 receptor, has been reported as effective in controlling hepatitis in several case reports [105, 106, 107, 108]. The safety and efficacy of tocilizumab which is given at a dose of 4 mg/kg IV over one hour have also been reported for the treatment of other organ irAEs or for the prevention of flares in patients with rheumatic and autoimmune diseases [109]. Another third‐line option that has been used successfully in steroid and MMF‐refractory irAEs is plasma exchange +/− IVIG [110]. A few case reports have described the successful use of this treatment in acute liver failure [111, 112, 113]. It is important to consider that there may be an element of publication bias where the unsuccessful use of these drugs is less commonly reported, and it is hard to know the true percentage of success of third‐line immunosuppressive therapies in the absence of controlled studies. Given the lack of data and unknown long‐term safety of these interventions and the effect they may have on underlying cancer, prospective randomized studies are needed to make firm recommendations as to the best option for steroid‐refractory disease [114]. We urge our colleagues to consider these options in steroid‐refractory disease, in a multidisciplinary manner, and in conversation with patients.

6. Rechallenge

Given the vast survival benefit of immunotherapy and the association between certain irAEs and tumour response to ICIs, oncologists often deem reintroduction of immune checkpoint inhibitors necessary after hepatitis resolves. The question of whether to resume ICIs after a bout of severe ILICI is patient and situation dependent and should be addressed in a multidisciplinary manner. The resumption of an ICI after experiencing toxicity, termed ‘rechallenge’, is a difficult decision, as solid biomarkers for relapse of toxicity have yet to be identified. What is clear, however, is that some form of rechallenge with alternating agent, dose or regimen is a feasible option, even in patients with grade 3 or above liver injury. In most studies, rechallenge is associated with a 20 to 35% ILICI recurrence rate, the majority of which are non‐severe [24, 53, 115, 116]. In the context of reintroducing therapy, two clinically relevant questions emerge: first, which ICI can be reintroduced, and second, whether prophylaxis with steroids is indicated to prevent relapse of liver injury. The use of an anti‐CTLA4 following DILI from an anti‐PD1/PDL1 or combination therapy appears to be associated with an elevated risk of recurrence for hepatitis and other irAEs [113, 117, 118, 119]. In general, anti‐CTLA4 inhibition is more potent and associated with more irAEs than anti‐PD1. However, the cholestatic phenotype of ILICI is more commonly seen in PD1/PDL‐1 inhibitions; therefore, the choice of ICI to rechallenge with will depend on many factors. The underlying liver disease, the presence of underlying autoimmune disorder, previous biliary injury, as well as progression of cancer, have been identified as risk factors for hepatitis recurrence [53, 115]. The prophylactic use of budesonide during rechallenge has been proposed [120]; however, there is no evidence that pre‐emptive corticosteroids have any impact on the recurrence rate of irAEs and as such cannot be recommended [53, 115].

7. Conclusion

We are just scratching the surface of how the immune system can be utilized to neutralize cancer. The future of cancer immunotherapy is poised to revolutionize how we think about cancer care, treatment, and prognosis. Novel preventive and therapeutic vaccines, cutting‐edge cell therapies, novel inhibitors that target new checkpoints, and a new era of treatment combinations will pave the future [3]. Importantly, most if not all of these treatments will alter liver immune homeostasis by affecting multiple types of immune cells and will have the unintended side effect of hepatotoxicity. Therefore, understanding ILICI mechanisms, prompt diagnosis, and novel treatment approaches focused on the liver are imperative in mitigating side effects and allowing patients to stay on their life‐saving immunotherapy.

Corticosteroids are reported to be successful in the majority of patients with ILICI, and MMF is the most commonly used second‐line therapy in steroid‐refractory hepatitis. Although there are no high‐quality prospective studies to demonstrate MMF's efficacy, it has been the most studied in retrospective series, likely due to the comfort and experience of hepatologists with its use. There are few case series and isolated case reports using third‐line agents such as tacrolimus, cyclosporine, ATG, tocilizumab, plasma exchange, and tofacitinib successfully. Data regarding the use of these medications are limited, and none are FDA approved for this indication. In addition, all of these therapeutics alter the immune balance towards immune suppression, and their long‐term use can have potential consequences on the underlying cancer, which has not been studied. More recently, ILICI with a cholestatic phenotype resembling secondary sclerosing cholangitis on imaging and biopsy has been described. This phenotype of ILICI which is more common with anti‐PD1 therapy and has been classically described with nivolumab and pembrolizumab is more resistant to steroids.

Finally, ICIs are rapidly becoming standard of care for cancer therapy. It is estimated that more than half of all tumours will at some point be given ICIs. Given their undeniable impact on improving overall survival in cancer, there is a legitimate debate about balancing risk and benefit with potential rechallenge with these medications. Consideration of rechallenge should be on a case‐by‐case basis, in a multidisciplinary manner, with alteration of regimen, dose (or both), close monitoring and follow‐up, and after careful consideration of the potential benefits and risks.

Author Contributions

L.D. and E.D.M. equally contributed in conceptualisation, drafting and critical revision of the manuscript. All authors approved the final version of the manuscript.

Conflicts of Interest

The authors declare no conflicts of interest.