Immune-mediated herb-induced liver injury: a potential association with herbal artemisinin use as supported by the updated RUCAM

Discussion

Due to the liver’s primary role in metabolism and detoxification, the organ is a significant target of drug-induced toxicity. DILI is a well-recognised precipitant of acute and chronic liver disease and a sizeable number of inciting agents have been described. However, it is clear that DILI is incredibly complex in not just the heterogeneity of suspected hepatotoxins but also the diversity of proposed pathophysiological mechanisms, assortment of concomitant risk factors and unpredictability of patients’ clinical courses. Although individual variance of each drug-specific pathway introduces mechanistic complexity, DILI may be observed in a few generalised clinical classifications to aid in understanding such adverse reactions. Specifically, DILI is often classified into intrinsic and idiosyncratic hepatotoxicity.4 13 Intrinsic hepatotoxicity is defined by a clear dose-dependency and predictable clinical course once a threshold dose has been superseded (eg, acetaminophen). Idiosyncratic hepatotoxicity is more unpredictable and maintains a less obvious dose-dependency. Idiosyncratic hepatoxicity is further subdivided into allergic idiosyncratic hepatotoxicity, defined by the presence of adaptive immune reactions like eosinophilia, rash, and fever, and non-allergic idiosyncratic hepatotoxicity, which demonstrates a lack of the aforementioned clinical signs and symptoms as well as a longer latency period.

Consumption of herbal and dietary supplements is increasingly recognised as a trigger of adverse reactions and hepatotoxicity; the drug-induced liver injury network reported that over 15% of liver injury cases between 2004 and 2013 were due to such products.2 14 Similar to DILI, HILI presents with clinically non-specific and remarkably variable signs and symptoms, including a spectrum of hepatocellular, cholestatic and mixed liver injury. Moreover, the pathogenesis of HILI is often ascribed to an idiosyncratic hepatotoxicity; more than a combined 1000 drugs, herbs and dietary supplements have been associated with idiosyncratic liver injury.15 In a systematic review of HILI by Lin et al, the most frequently reported hepatotoxic herbal products were greater celandine, germander and chaparral.2 The authors notably observed that the majority of HILI cases included female patients, although predilection transitioned towards male patients in the elderly population. Moreover, female patients were significantly more vulnerable to cholestatic HILI. Clearly, there is an interplay between the host, environment and agent in both DILI and HILI.

A subgroup of DILI/HILI cases demonstrate an immune-mediated pattern of livery injury. Features of autoimmunity include the presence of autoantibodies as well as evidence of hepatic infiltration by immunocompetent cells. Some of these characteristics have led to a clinical designation of allergic idiosyncratic hepatotoxicity, while others have termed such cases ‘autoimmune-like’ DILI/HILI.5 Regardless of moniker, immune-mediated DILI/HILI is of interest due to the notable clinical and pathological resemblance with AIH that makes differentiation difficult. AIH is a chronic, fluctuating inflammatory liver disease that presents with hepatocellular injury. Diagnostic criteria include the presence of autoantibodies (ANA and ASMA), selective elevation of IgG and compatible histological profile (eg, interface hepatitis exceeding portal tract borders).11 Although disease-specific antibodies in a minority of AIH cases aid in identifying respective antigens, the pathogenic stimulus for autoimmunity is currently unknown for the majority of patients with AIH. The clinical presentation of AIH is further complicated in the scenario in which consumption of a hepatotoxic agent is believed to trigger AIH. Several reports have detailed such cases, termed drug or herb-induced AIH; commonly implicated agents have included nitrofurantoin or minocycline.16 Presenting features occur in similar proportions in cases of both AIH and drug or herb-induced AIH; clinical deviations generally lie in the absence of underlying cirrhosis in patients with drug or herb-induced AIH and the presence of a favourable temporal sequence between consumption of a hepatotoxic agent and onset of signs and symptoms.17 18 Reactive metabolites of these toxic agents have been proposed to bind hepatocellular proteins to generate immunogenic neoantigens.19 However, in addition to the difficult nature of establishing causality between intake of a possible inciting agent and manifestation of a chronic autoimmune disease like AIH, other studies have indicated a more likely scenario may be that such agents serve as either a trigger or ‘coincidental exposure’ that uncovers or ‘flares’ AIH that is already pre-existent in a patient. Regardless of the pathogenesis, clinical and histological features are currently believed to be indistinguishable and, as such, the same simplified diagnostic criteria are applied in these cases of AIH.

Diagnosis of acutely evident autoimmune liver disease, especially AIH, is also complicated by the intersection of clinical presentation with alternative autoimmune diseases that involve liver injury without a direct hepatotoxic trigger, such as primary biliary cholangitis. Primary biliary cholangitis is a chronic liver disease typified by immune-mediated bile duct injury liver injury; elevated ALP levels and prominent cholestasis generally favours the diagnosis. However, rarer cholestatic presentations of AIH may obfuscate differentiation. Serology can support a particular autoimmune process; however, considerable overlap exists. For example, AMA serology, typically indicative of primary biliary cholangitis, may be positive in 5%–12% of isolated type 1 AIH cases.20 Notably, reported cases of AIH-primary biliary cholangitis overlap syndrome can further complicate diagnoses; however, the diagnostic approach to and significance of this presentation is poorly established.21

As opposed to AIH, immune-mediated DILI/HILI is an alternative type of hepatotoxin-induced autoimmune liver disease in which exposure to a hepatotoxic agent triggers an adverse acute or chronic immune response. Neoantigen formation and haptenisation may be involved in pathogenesis. Patients will more often present with fevers, eosinophilia, lymphadenopathy and rash—features indicating a component of hypersensitivity reaction. Histology can show portal inflammation, infiltrates of mononuclear cells, lymphocytes, and eosinophils, cholestasis, and lack of bile ducts.22 Interface hepatitis and focal necrosis may be evident but is more often associated with AIH. Whereas AIH has been described as a self-perpetuating autoimmune liver disease incited by a hepatotoxic trigger, immune-mediated DILI/HILI is defined by more limited duration directly related to the degree of exposure to the insulting agent with a tendency towards resolution following cessation of the drug.3 Moreover, steroids are the most well-studied treatment for both disorders. However, patients with AIH will often require permanent immunosuppression therapy due to relapse on withdrawal, while patients with immune-mediated DILI/HILI more often exhibit complete response to treatment with no remission.

Contemporary research in DILI/HILI has frequently emphasised the identification of potential hepatotoxins and the discovery of underlying pathophysiological mechanisms. However, effective diagnostic tools in the clinical assessment of DILI/HILI have lagged. Because most cases of DILI/HILI are idiosyncratic in nature, the presence of toxic metabolites is generally unreliable as a diagnostic biomarker. Similarly, immunologic or genetic parameters are not yet validated. The Roussel Uclaf Causality Assessment Method (RUCAM) is an established, hepatotoxicity-specific diagnostic tool in common use to quantitatively assess a causality association in cases of suspected DILI/HILI.23 The updated RUCAM considers additional criteria such as temporality dependent on the pattern of liver injury as well as a diverse set of alternative causes to be excluded.24 Of note, because the simplified criteria used to assess AIH does not include the exclusion of DILI/HILI, the use of validated tools such as the RUCAM in addition to clinical acumen is vital when investigating immune-mediated hepatic pathologies.

A growing spectrum of agents are considered to have idiosyncratic hepatotoxicity; each adverse reaction presents an agent-specific pathway that, by extension, complicates mechanistic classifications. A review by Russmann et al emphasises mechanistic ‘upstream’ events in DILI/HILI and further simplifies these events into a three-step model of principle mechanisms.4 Initial mechanisms of toxicity are classified as direct cell stress, direct mitochondrial impairment and specific immune reactions. An agent-specific pathway is likely a consequence of some combination of these initial insults. When considering immune-mediated DILI/HILI, the principal mechanism of specific immune reactions is naturally of great interest. Suspected inciting agents are often substrates of phase I and phase II metabolising enzymes; these agents may generate neoantigens via modified cytochrome isoenzymes, yielding targets of autoimmunity.22 Additional hypotheses of pathogenesis include drug metabolites serving as a hapten that covalently binds with other macromolecule carriers. These hapten-carrier complexes may be internalised by antigen-presenting cells, after which antigens are presented to naïve T cells via major histocompatibility complex interactions to ultimately produce the immunogenic response observed in immune-mediated DILI/HILI.

It is important to note that no agent has been definitively recognised as a trigger for immune-mediated liver disease. In patients like ours, who lacked a history of alcohol abuse or previous hepatotoxic reactions to drugs, causality association is supported by identifying a compatible toxic agent in the setting of an appropriate temporal sequence. The patient reported no significant recent changes in diet. Her only relevant outpatient medication consisted of opiates, which have limited links to liver toxicity in the setting of overdose. Regarding the patient’s supplement therapy (sodium butyrate and artemisinin/piperine), existing evidence interestingly supports sodium butyrate’s hepatoprotective properties.25

Conversely, while a PubMed search for ‘artemisinin’ and ‘immune-mediated liver injury’ does not yield relevant data, search for ‘artemisinin’ and ‘hepatoxicity’, ‘drug-induced liver injury’ or ‘herb-induced liver injury’ shows several reports that cite liver toxicity secondary to artemisinin use.8–10 26 27 A review and case reports by Efferth et al detailed the role of cancer combination therapies with artemisinin-type drugs.8 26 27 In one case, the use of combination dichloroacetate and artesunate therapy in addition to standard radiochemotherapy for glioblastoma multiforme precipitated liver injury and bone marrow toxicity. Assessment with the updated RUCAM yielded a grading score of 6, indicating probable causality association. Another patient with glioblastoma multiforme also demonstrated acute liver injury after use of standard chemotherapy, the antimalarial drug artesunate, and a variety of Chinese herbs; however, RUCAM causality assessment was not performed in this instance. A case report by Kumar details a patient presenting with cholestatic liver injury following consumption of oral herbal artemisinin (125 mg approximately two to three times per day) for 6 weeks.10 The associated updated RUCAM score was 7, suggesting a probable causality association. This case also emphasises that chronic use of low-dose agents may still incite an idiosyncratic hepatoxic reaction. Indeed, while dose-dependency in idiosyncratic hepatotoxicity is not as strong as in intrinsic reactions, the risk of DILI/HILI has been shown to be increased if a minimal dosage threshold is superseded.4 5 Although the mechanism of injury is not completely understood, volatile oils of artemisinin are thought to be metabolised by the liver, chiefly cytochrome P450 3A4, into the active metabolite artenimol (figure 3).28 Moreover, complex interactions may also produce hapten-protein conjugates that activate T-cells and generate the immunogenic response. Additionally, piperine is a black pepper derivative shown to suppress hepatic metabolism to increase bioavailability of drugs.29

Our patient consumed three times the recommended daily dosage of artemisinin/piperine for 4 weeks with cessation 1 day prior to admission. She then presented with clinical findings and laboratory results indicative of an autoimmune liver disease. Appropriate features yielded only four points on diagnostic criteria for AIH, making this aetiology unlikely. Histology was inconsistent with primary biliary cholangitis and markedly positive response to steroid therapy, which is of uncertain benefit in primary biliary cholangitis, lent credence to an alternative diagnosis. Positive autoantibodies, cholestatic liver disease and a history of concomitant fevers and rash along with compatible histology was supportive of immune-mediated HILI. However, it must be noted the pruritic rash could also be the result of excoriation due to marked mixed hyperbilirubinaemia. The association between artemisinin use and subsequent liver injury was ‘probable’ in the updated RUCAM scale (score of 6). Thus, with compatible clinical features and temporal association, the patient was diagnosed with immune-mediated HILI, likely associated with artemisinin use.

We emphasise the limitations of this singular case presentation with regard to the eventual diagnosis of immune-mediated HILI associated with artemisinin use. The clinical history and presentation obtained from the patient, diagnostic approach used during evaluation, validated tools employed during assessment (ie, the updated RUCAM) and additional cases of artemisinin toxicity discovered during research support the suggestion of an association between artemisinin use and manifestation of immune-mediated HILI. However, a direct causative relationship cannot be firmly concluded, and more robust data collection and analyses of complications following artemisinin use are warranted to better characterise these observations. Additionally, as is the case in idiosyncratic hepatotoxicity, the dose-response relationship in this patient’s presentation is not well understood. We cannot appropriately discern whether the patient’s immune-mediated pathology was triggered from a singular initial consumption or gradually evolved after continuous use. Figure 3 only summarises pathogenic mechanisms of artemisinin toxicity as detailed in the existing literature; the specific underlying mechanism prompting the patient’s clinical presentation cannot be identified and is likely subject to additional environmental and immunogenetic factors. Thus, while we cannot faithfully support a causative relationship between artemisinin use and immune-mediated HILI, we present a comprehensive diagnostic evaluation used to arrive at such an association in our patient. Moreover, appropriate medical therapy (avoiding re-introduction of the suspected offending agent and administering immunosuppressive medication) resulted in a positive response, lending credence to the presumed diagnosis.

Although there is no literature regarding efficacy of artemisinin-based therapies in the treatment of pelvic sarcomas as sought for by the patient, recent works have underlined the potential of artemisinin-type drugs as radiosensitisers and tumour growth inhibitors. One study investigated glioma cells treated with varying concentrations of dihydroartemisinin with radiation, evaluating the effectiveness of radiosensitisation with clonogenic survival assays.30 Findings suggested that dihydroartemisinin precipitates the formation of reactive oxygen species and inhibition of antioxidant enzymes, chiefly glutathione-S-transferase, leading to therapeutic radiosensitisation of human glioma cells. Another investigation evaluating the effects of artemisinin and a novel derivative, artemisone, demonstrated antiproliferative effects in human breast, colon, pancreatic and melanoma cancer cells via reductions of cyclin D1, CDK4 and pRb with consequent cell cycle arrest.31 While there is a great deal of potential benefit of artemisinin-based therapies, it is clear that there is also a great deal of unrecognised sequelae. This is true for many alternative therapies, which is of particular importance for patients with cancer where 20%–50% use herbal medicine.32 33 While thoroughly investigating their underlying mechanisms is infeasible, this case stresses the importance of clinician awareness of patient’s usage of and education on common therapies to enhance appreciation of unanticipated side effects.