Drug-induced Liver Injury With Skin Reactions

Drugs and Host Risk Factors, Clinical Phenotypes and Prognosis

Harshad Devarbhavi; Sujata Raj


Liver International. 2019;39(5):802-811. 

In This Article

Diagnostic Criteria

Drug Reaction With Eosinophilia and Systemic Symptoms (DRESS)

Drug reaction with eosinophilia and systemic symptoms (DRESS) is defined according to the International Registry of Severe Cutaneous Adverse Reactions (RegiSCAR) based on the presence of at least three or more criteria of the following criteria: fever >38°C, acute skin rash, lymphadenopathy, internal organ involvement, blood count abnormalities and hospitalization.[50]

Associated liver dysfunction in DRESS is defined as per RegiScar criteria as serum transaminase levels >2 × ULN (upper limit of normal) or bilirubin at >2 × ULN on two successive dates or aspartate aminotransferase, gamma-glutamyl transpeptidase and alkaline phosphatase levels >2 × ULN at least once (see appendix S1 in Ref. 11).

Stevens-Johnson Syndrome (SJS) and Toxic Epidermal Necrolysis (TEN)

Although rarer than DRESS, SJS/TEN are most often drug-induced. Drugs which cause DRESS can also cause SJS/TEN. SJS/TEN is distinguished by severe skin reactions including blisters, erosions and peeling of skin and mucous membranes. Severity is determined by the extent of skin detachment related to the body surface area (BSA): SJS <10%, TEN >30% and SJS/TEN overlap 10%-30% of BSA.[51] Unlike DRESS, liver involvement is uncommon in SJS/TEN

Drug-induced liver injury (DILI). The diagnosis of DILI as per the international DILI expert working group[52] includes any one of the three following criteria: (a) More than or equal to five-fold elevation above the upper limit of normal (ULN) for alanine aminotransferase (ALT),[52] or (b) more than or equal to two-fold elevation above the ULN for alkaline phosphatase (ALP),[52] or (c) more than or equal to three-fold elevation in serum ALT concentration and simultaneous elevation of serum bilirubin concentration exceeding 2 × ULN.[52]

An additional criterion is any symptomatic hepatitis and three-fold elevation in transaminases.[13,37] Symptoms include skin rash and itching and others such as fatigue, right upper quadrant abdominal pain, nausea/vomiting, weakness, anorexia and weight loss.[13,37,52] Symptoms if attributable to DILI qualifies to be included under severity assessment.[52]

Since DILI is a diagnosis of exclusion, other causes of liver injury must be excluded by appropriate tests. These include hepatitis A-E, alcoholic liver disease, autoimmune liver disease, biliary tract disease and ischaemic hepatitis.[52]

There are minor differences in liver injury definitions by RegiSCAR[50] and DILI international consensus group.[53] The presence of obvious skin injury with two times elevation of bilirubin or ALP and or ALT is akin to an earlier definition used for DILI.[53]

Causality Assessment

There are no biomarkers or specific tests for diagnosing DILI or DISI. Attributing causality to an implicated drug is challenging and should take into account the temporal relationship, clinical and laboratory features, drug's hepatotoxicity profile, resolution with cessation of drug and recurrence in the event of re-challenge. A host of alternative causes such as viruses, infections and biliary obstructive disease shave to be excluded. Often a drug will have its clinical and laboratory signature which along with together with exclusion of common causes of liver or skin injury determines the likelihood of diagnosis.

For DILI, the widely employed causality assessment tool is the Roussel-Uclaff Causality Assessment Method.[54] Others include Maria and Victorino[55] clinical diagnostic scale which gives credence to extrahepatic manifestations and expert opinion causality determination by the DILIN group, which cannot be used in solo practice setting and needs experts in liver disease to adjudicate causality.[56]

With regard to DRESS causality assessment, the well-known and widely used tools are the RegiSCAR assessment[11] and the Japanese Consensus Diagnostic system.[57]

The algorithm for drug causality for epidermal necrolysis (ALDEN) is specific for SJS/TEN and is particularly useful in discriminating the "culprit" from "innocent" drugs when patients are on multiple drugs.[58] Points allocated to different elements in the scoring systems are tallied to yield a score that aids the determination of the likelihood of diagnosis into possible, probable, definite (certain) or unlikely causes.

Drug Reaction With Eosinophilia and Systemic Symptoms (DRESS)

Drug reaction with eosinophilia and systemic symptoms is the original phenotype amongst SCAR in terms of clinical and biological characteristics, drugs, timeliness and clinical course.[11,59] Females are more commonly affected[11] although not all studies support this observation.[60]

Internal organ involvement is a defining feature of DRESS and can involve any organ system.[2] It plays a very important role in prognosis as well. Liver involvement is the most common; it varies from 75%[11] to 94% of cases.[60] In some series, liver involvement occurs in 100% of cases.[61,62] Other organs that are involved include kidneys (8%-37%), lung (5%-32%), heart (2%-13%), pancreas (4%) and others (11%).[11,60] Reactivation of herpes family virus (human herpes virus 6 [HHV-6]) or HHV-7 or Epstein-Barr virus (EBV) is seen in up to 76% of patients with DRESS in some series.[62]

Variance in definition of liver injury across disciplines makes obtaining of accurate data gathering challenging. Liver involvement can range from mild often asymptomatic elevation in liver biochemical tests to severe disease including acute liver injury and acute liver failure. Cholestatic or mixed hepatitis is more common than hepatocellular pattern of liver injury, but this is dependent on the implicated drug, timing of presentation and liver biochemistry tests.[8] The presence of jaundice, any grade of hepatic encephalopathy or coagulopathy is a sign of poor prognosis.[63] Such patients should be considered for liver transplantation. Reports of liver transplantation are rare. See Table 3 for a summary of published reports of liver transplantation for DRESS. In one multicenter French study, all of the nine patients without hepatic encephalopathy improved without a liver transplant.[63] Five patients with encephalopathy received transplantation; 3-month survival rate after transplantation was 80%, and the actuarial survival rate was 60% which is lower than other common causes for transplantation.[63] Recurrence of DRESS without cutaneous rash can occur complicating post-liver transplantation management.[63,64]

There is an inconsistent relationship between the severity of skin injury and liver; Lin et al[65] found no significant relationship between skin histopathological findings and liver injury, whereas Walsh et al[61] from Kings College Hospital found patients with DRESS presenting with an atypical erythema multiforme (EM)-like eruption (purpura and atypical targets) to be associated with severe hepatic involvement such as significantly marked elevation of serum aminotransferases (AST 2758 IU/L) compared with other types of skin pathology (AST 292; P = 0.01). Further, two out of eight patients with EM-like DRESS died in contrast to one mortality out of 19 cases in the non-EM-like DRESS cohort. All three patients who died had severe liver involvement including one who died following a failed liver transplantation.[61] These inconsistent results could be accounted for by patient severity and referral patterns. Patients reported by Lin et al study[65] had less severe disease whereas several patients reported in the Walsh et al[61] study had hepatic injury and failure.

The association of eosinophilia and its relationship with liver injury and outcome in DRESS is interesting. The presence of eosinophilia is variable. In a literature review, eosinophilia was seen in 66% (114/172)[60] while it was as high as 95% (108/114) in a prospectively studied patients from the RegiSCAR group.[11] The association of eosinophilia is inconsistent, with one study not showing a correlation between peripheral eosinophilia and liver dysfunction severity,[61] whereas Lin et al (in 62 of the 72 patients) demonstrated eosinophils in the dermis in patients with less severe liver disease (30.8% vs 88.9%; P = 0.002).[8] Incidentally, eosinophilia in patients with liver injury is associated with better survival[6,66,67] which has been attributed to the inclusion of immuno-allergic response. It has to be noted that both skin rashes and eosinophilia individually need not to be present separately in a case with DRESS syndrome.

Drug reaction with eosinophilia and systemic symptoms is a multisystem disease. Therefore, a multidisciplinary approach with inputs from dermatologists, hepatologists, nephrologists, pulmonologists and immunologists may be required. Often it is not the skin involvement, but the presence of hepatitis that that influences prognosis.[60,68]

Stevens-Johnson Syndrome/Toxic Epidermolysis Necrosis

Internal organ involvement is less common in SJS/TEN compared to in DRESS. Furthermore, unlike DRESS, the association of the extent of skin injury with outcome is strong with SJS/TEN and less dependent on internal organ involvement. This is assessed by SCORTEN, a prognostic score for in-hospital mortality, in which a higher score portends increased mortality.[69]

The mortality at 6 weeks varies by severity of skin involvement with a mortality of 12% for SJS, 29% for SJS/TEN overlap and 46% for TEN.[70] The mortality can approach 49% in 1 year mostly due to the presence of other comorbidities. Liver involvement in SJS/TEN is less well studied but is generally mild and reversible in the form of raised liver biochemical tests. However severe hepatitis including acute liver failure and mortality has been reported.[13] When skin lesions resolve quickly, long-term complications from prolonged jaundice such as vanishing bile duct syndrome may occur from shared immune-related mechanisms of both diseases.[71] A report of a case in an infant and additional six cases showed complete recovery of VBDS in five of seven cases mostly treated with immunosuppressive drugs such as steroids and calcineurin inhibitors.[71]

Whereas the frequency of SJS/TEN in large registries varies from 1% in the DILIN series[4] to 4.8% in India.[6] Mortality in the presence of DILI is even higher (36%) but has been reported to be higher (44%-46%) in the presence of jaundice.[4,6] Often the skin injury resolves quickly, but liver disease can progress resulting in death or liver transplantation has been reported in some cases.[72,73] Mortality is less in children and in HIV infected individuals,[13] although HIV/AIDS-infected individuals have a several fold higher risk of developing SJS/TEN.[74]

Similar to DRESS, a small number or class of drugs produce a majority of the cases of SJS/TEN, including newer anti-epileptic drugs such as lamotrigine.[73]

Advanced age and the presence of underlying liver disease are risk factors associated with mortality. Extra cutaneous and extra hepatic complications like the ophthalmic injury and sepsis overshadow those of liver. Recovery is prolonged and is complicated by infection.