We believe the most likely cause of death to be an idiosyncratic allergic reaction to sorafenib manifesting as hepatotoxicity with associated renal impairment. This is supported by a RUCAM causality score of 7, consistent with sorafenib being the probable cause [4]. Drug induced liver injury (DILI) is attributable to either direct hepatotoxicity occurring soon after drug exposure, or a delayed idiosyncratic reaction. Idiosyncratic drug reactions are thought secondary to immune response, possibly to drug-protein haptens. They typically take 2–8 weeks to manifest and may demonstrate ongoing progression despite withdrawal of the offending drug [5, 6].
Although sorafenib-induced hepatic dysfunction has been previously described, this is the first description of sorafenib-induced hepato-renal failure and represents the first fatality in an individual otherwise systemically well without underlying liver disease or known metastases. Sorafenib associated liver failure was first described in a phase II trial for advanced thyroid cancer [7] where the patient developed elevated LFTs 8 weeks after starting sorafenib. Despite sorafenib withdrawal, hepatic function worsened, culminating in death secondary to hepatic failure. Sorafenib hepatotoxicity was likewise reported in a 65 year old female with underlying compensated cirrhosis, secondary to non-alcoholic steatohepatitis, who received sorafenib as a third-line therapy for advanced follicular thyroid carcinoma [8]. Treatment was stopped after approximately 3 weeks when she developed a skin rash with associated fever and flu-like symptoms. At this point there was mild derangement of hepatic function which progressed rapidly over the subsequent 3 weeks, resulting in hospitalisation with a hepatocellular picture of markedly raised trans-aminases. A liver biopsy demonstrated drug-induced necrotic hepatitis and the patient responded to treatment with corticosteroids. Two further cases of sorafenib related hepatotoxicity have also been described; in the treatment of HCC [9], and also in an individual who had undergone cadaveric liver transplant for HCC [10]. However, these cases differ in that LFTs became abnormal soon after sorafenib administration (less than 5 days) and hepatic function normalised upon withdrawal of therapy. The second of these cases had a liver biopsy demonstrating hyperallergic features with hepatic inflammatory infiltrate consistent with an immune mediated process [10].
Genome wide association studies of DILI consistently demonstrate susceptibility conferred by major histo-compatibility complex (MHC) polymorphisms, supporting an immune basis to idiosyncratic DILI [11]. Although antibiotics are the most studied DILI culprits showing consistent HLA associations [12–14] similar HLA associated hepatotoxicity has been noted to small molecule kinase inhibitors. MHC associations have been reported with susceptibility to lapatanib and pazopanib hepatotoxicity [15, 16]. Polymorphisms within HFE, the haemachromatosis gene residing within the extended MHC, are associated with pazopanib toxicity; whilst lapatinib toxicity is reproducibly associated with carriage of the class II HLA allele HLA_DQA1*0201.
The case we describe showed a pronounced immunoresponse with elevated CRP, low-grade pyrexia, complement consumption and an immune infiltrate in the liver. We suggest that this, coupled to a typical delay in onset between commencing sorafenib and hepatic failure, would support an immune mediated process. Given the precedents shown with other DILI, a link to HLA subtype may be anticipated. Patterns of DILI tend to be drug specific, with three phenotypes in terms of hepatic dysfunction recognised. These being a hepatocellular pattern, a cholestatic pattern and a mixed picture with derangement in both trans-aminases and cholestasis but neither picture predominating [6, 17]. The pattern of liver panel abnormalities in this case and the other 2 reported cases of delayed DILI to sorafenib would suggest sorafenib can cause DILI with a hepatocellular pattern of LFTs, with an ALT>3 x upper limit of normal (ULN) and ratio of ALT/ULN: AlkP/ULN >5 between 3–7 weeks after initial exposure.
As with many idiosyncratic DILI, sorafenib hepatotoxicity is uncommon. In phase III clinical trials of sorafenib monotherapy the incidence of hepatic failure was similar amongst sorafenib and placebo recipients [18]. Similarly, in two prospective phase III clinical trials of sorafenib in patients with advanced HCC and mild liver impairment (Child-Pugh A), the incidence of hepatic failure and hepatic encephalopathy events were comparable between patients taking placebo (2.4% and 2.1%) and sorafenib (2.5% and 1.8%) [19]. Likewise hepatic failure/encephalopathy events were not observed in a placebo controlled prospective phase III study in advanced RCC (TARGET, n=903) [2]. Analysis of reported toxicities from the SORCE trial supports the rarity of sorafenib hepatotoxicity in the adjuvant setting, with this case representing an idiosyncratic reaction as opposed to a marked example of generalized hepatotoxicity (Meade A., personal communication). Nonetheless, given that a proportion of patients receiving sorafenib may have impaired liver function due to HCC or metastatic disease, it is possible a delayed drug reaction to sorafenib may be mistaken for, or coincide with disease progression. With this in mind it is noteworthy that, including this case, all episodes of sorafenib-associated acute liver failure have been reported outside HCC treatment.
This case is the first documented death directly attributable to sorafenib in an otherwise well individual. Since the patient had no evidence of metastatic disease prior to commencement of treatment or at autopsy, the case raises questions regarding the stratification of adjuvant treatments for which no current evidence exists in an outwardly healthy group. While currently available data suggests such reactions are likely to be extremely rare, we would urge clinicians to be vigilant as to possible sorafenib DILI and ensure suspected cases are reported. We suggest that future trial-wide genetic profiling of patients may potentially expedite identification of markers of idiosyncratic drug reaction to novel oncological agents, including those denoting HLA allele status [20], permitting genotype based risk stratification.
Ethics statement
The SORCE trial is a Medical Research Council sponsored randomised double-blind study with both main UK wide Research Ethics Committee (REC) approval and local ethical approval. It is a National Cancer Institute verified trial and registered with the U.S. National Institutes of Health database ClinicalTrials.gov (Identifier: NCT00492258).
Consent
Written informed consent was obtained from this patient’s next of kin for publication of this Case report and accompanying images. A copy of the written consent is available for review by the Series Editor of this journal.