Epidermal necrolysis (EN), a severe cutaneous adverse drug reaction characterized by epidermal detachment, encompasses Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN). Stevens-Johnson syndrome affects 10% or less of the body surface area, and there is TEN of 30% or greater with SJS/TEN overlap in between. Epidermal necrolysis is a delayed T cell–emediated hypersensitivity reaction in which mortality reaches 40%, depending on variables in the prognostic Severity-of-Illness Score for Toxic Epidermal Necrolysis tool. Drug-specific HLA alleles, drug exposure, and comorbidities are some determinants of EN epidemiology. In HIV infection, prevalence can be as high as 100-fold; antiretrovirals, sulfonamide antibiotics, and antituberculosis drugs are the common offenders.1 Epidermal necrolysis-associated drug-induced liver injury (DILI) is thought to result from systemic inflammatory response and/or direct drug injury. The current literature suggests that EN-DILI has a prevalence as high as 40%, prolongs hospitalization, and has higher mortality in jaundiced patients.2–4 However, data from African and/or HIV-infected populations is limited.
We aimed to determine the prevalence, clinical features, injury patterns, and outcomes of EN-DILI in the HIV-endemic African setting by retrospectively studying records of EN patients managed by a tertiary dermatology service in Cape Town, South Africa between January 2004 and December 2015. Data extracted included demographics, HIV parameters, risk factors, current and previous markers affecting liver disease, comorbidities, drug exposures, and adverse drug reaction history.
We determined drug causality using the Algorithm for Assessment of Drug Causality in Stevens–Johnson Syndrome and Toxic Epidermal Necrolysis (ALDEN) score4 and graded DILI severity using common terminology criteria for adverse events. Grades 1 to 5 refer to mild, moderate, severe, and life-threatening reactions, and death respectively.5 Hepatocellular injury pattern, using alanine aminotransferase (ALT) and alkaline phosphatase (ALP), was defined as greater than two times the upper limit of normal (ULN) of ALT alone, or R greater than 5, where R = [actual ALT/ALT ULN]/[actual ALP/ALP ULN]. Cholestatic pattern was defined as greater than 2 × ULN alone or R = 2 or less. A mixed pattern referred to both ALT and ALP being greater than 2 × ULN and 2 less than R less than 5.4,6
Our institutional human ethics review committee approved the study. Standard statistical methods were used to evaluate and analyze categorical and continuous data. Univariate and multivariate logistic regression analyses were performed to examine factors associated with EN-DILI. We analyzed data with Stata software (StataCorp. 2017. Stata Statistical Software: Release 15. College Station, TX: StataCorp LLC.); a significance level of P less than .05 was used for all analyses.
The study included 184 patients, median (interquartile range [IQR]) age 33 years (28–38 years). Most were female (129 of 184 [70%) and HIV-positive (142 of 155 of tested patients; 92%). Among HIV-positive patients, median CD4 count was 185 cells/mm3; 49 of 142 patients had WHO grade 3 or 4 HIV/AIDS (35%), and 90 of 155 were receiving antiretroviral therapy (58%). Of 184 patients, Stevens-Johnson syndrome, TEN, and SJS/TEN overlap accounted for 103 (56%), 50 (27%), and 31 (17%), respectively. Tuberculosis was being treated in 38 (21%), and 11 of 38 had disseminated disease (29%).
Developed drug-induced liver injury criteria were fulfilled by 62 of 184 patients (34%), with similar rates in HIV-infected and uninfected patients (48 of 142 [34%] vs five of 13 [38%]; P = .36). Common terminology criteria for adverse events grades 1 through 5 of DILI occurred in 15 of 62 (24%), 17 of 62 (27%), 23 of 62 (37%), seven of 62 (11%), and no patients, respectively. Deaths were attributed to sepsis, renal failure, and pneumonia each. Developed drug-induced liver injury occurred in 22 of 50 patients with TEN (44%), 11 of 31 with SJS/TEN overlap (35%), and 29 of 103 with SJS (28%). Median hospitalization for EN-DILI patients was 14 days (IQR, 9–32 days) versus 11 days (IQR, 8–18 days) for those who did not have EN-DILI (P = .02). Mortality rates were similar in the EN-DILI and non-DILI groups (three of 62 [5%] vs three of 122 [3%]; P = .39) (Table I). Median latency between offending drug initiation and EN symptoms was 19 days (IQR, 14–32 days) for DILI patients versus 20 days (IQR, 10–28 days) for those did not have EN-DILI (P = .51). Heavy alcohol use was self-reported by 12 patients; seven of those (58%) developed DILI. Jaundice developed in three of 61 patients (5%), none of whom died. The international normalized ratio was determined in 25 of 62 DILI patients (40%) and was elevated in four (16%), two of whom died. Apart from one case of gynecomastia, no peripheral signs of chronic liver disease were recorded.
TABLE I.
Patient characteristics of epidermal necrolysis admissions during 12-y study period (n = 184)
| Variable | Patients with missing data, n (%) | All (n = 184) | Liver injury (n = 62) | No liver injury (n = 122) |
|---|---|---|---|---|
| Age, y (median [IQR]) | 2 (1) | 33 (28–38) | 34 (28–41) | 32 (28–38) |
| Female, n (%) | — | 129 (70) | 37 (60) | 92 (75) |
| Pregnant, n (%) | — | 21/129 (16) | 5/37 (14) | 16/92 (17) |
| African race* | 41 (22) | 130/143 (91) | 44/51 (86) | 86/92 (93) |
| Phenotype, n (%) | ||||
| Stevens-Johnson syndrome | — | 103 (56) | 29 (47) | 74 (61) |
| Stevens-Johnson syndrome/toxic epidermal necrolysis overlap | — | 31 (17) | 11 (18) | 20 (16) |
| Toxic epidermal necrolysis | — | 50 (27) | 22 (35) | 28 (23) |
| HIV status | ||||
| Positive, n (%) | 29 (16) | 142/155 (92) | 48/53 (91) | 94/102 (92) |
| World Health Organization clinical stages 3 and 4, n (%) | 8 (6) | 49/134 (37) | 16/46 (35) | 33/88 (38) |
| Median CD4 cell count (IQR) | 12/142 (8) | 185 (97–264) | 200 (97–272) | 180 (95–251) |
| Receiving antiretroviral therapy, n (%) | 2/142 (1) | 90/140 (64) | 26/46 (57) | 64/94 (68) |
| Previous adverse drug reaction, n (%) | 170 (92) | 4/14 (29) | 3/6 (50) | 1/8 (13) |
| Tuberculosis | ||||
| Current treatment for tuberculosis | 38 (21) | 38/146 (26) | 18/52 (35) | 20/94 (21) |
| Disseminated tuberculosis | 1 (3) | 11/37 (30) | 3/18 (17) | 8/19 (42) |
| Comorbidity | ||||
| Hypertension | 2 (1) | 15/182 (8) | 4/61 (7) | 11/121 (9) |
| Diabetes | 1 (1) | 7/183 (4) | 1/61 (2) | 6/122 (5) |
| Renal impairment (grade) | 13 (7) | |||
| 1 or 2 (mild or moderate) | — | 26/171 (15) | 13 (21) | 13/109 (12) |
| 3 or 4 (severe or life threatening) | — | 3/171 (1) | 1 (1) | 2/109 (1) |
| Hospital length of stay, d (median [IQR]) | 10 (5) | 13 (8–23) | 14 (9–32) | 11 (8–18) |
| Mortality | 0 | 6 (3) | 3 (5) | 3 (2) |
IQR, interquartile range.
Self-reported ethnicity
Hepatocellular injury overall was the most common pattern (32 of 62 patients; 52%); it accounted for 25 of 32 HIV-infected patients (78%). Compared with non-hepatocellular patterns, it was strongly associated with severe (grade 3 or 4) reactions (22 of 30 patients; 73%; P < .001). A mixed pattern occurred in 28 of 62 patients (45%): eight of 30 had severe reactions (27%) and 21 of 28 were HIV-infected (75%). Only two of 62 HIV-infected patients had a cholestatic pattern, both mild (3%).
Nevirapine and cotrimoxazole were the two most common drugs that ALDEN scored as probable or very probable causes of EN. Similarly, they accounted for the highest number of DILI cases (see Table E1 in this article’s Online Repository at www.jaci-inpractice.org). No clear offending drug (scoring probable or very probable with ALDEN) was more likely to cause EN-DILI. Nevirapine was the most common drug causing both hepatocellular injury (11 of 30 patients; 37%) and a mixed pattern (six of 25 patients; 24%), and cotrimoxazole was the second most common cause in both hepatocellular and mixed-injury patterns (Table II). Univariate analysis found male sex (P = .03), current tuberculosis (P = .04), and body surface area greater than 30% (P = .05) to be associated with EN-DILI; however, none remained significant upon multivariate analysis (see Table E2 in this article’s Online Repository at www.jaci-inpractice.org).
TABLE II.
Characteristics of drug-induced liver injury cases by offending drug
| Drug name | CD4 cell count (median [IQR]) | International normalized ratio (median [IQR]) | Latency, d (median [IQR])* | Algorithm for Assessment of Drug Causality in Stevens–Johnson Syndrome and Toxic Epidermal Necrolysis probable or very probable | |||
|---|---|---|---|---|---|---|---|
| Hepatocellular, n (%) (n = 30) | Mixed, n (%) (n = 25) | Cholestatic, n (%) (n = 1) | Severe or life-threatening drug-induced liver injury, n (%) (n = 28) | ||||
| Nevirapine | 211 (167–280) | 1.4 (1.2–1.7) | 20 (14–25) | 11 (37) | 6 (24) | 0 | 11 (39) |
| Antituberculosis drugs | 138 (39–226) | 1.2 (1.0–1.4) | — | 3 (10) | 3 (12) | 1 (100) | 5 (18) |
| Rifampicin | 362 (113–952) | 1.3 (1.2–1.4) | 18 (14–34) | 0 | 3 (12) | 0 | 2 (7) |
| Isoniazid | 186 (134–336) | 1.2 (1.2) | 44 (18–73) | 1 (3) | 0 | 1 (100) | 1 (4) |
| Pyrazinamide | 108 (32–232) | 1.1 (1.0–1.2) | 15 (11–35) | 2 (7) | 1 (4) | 0 | 3 (11) |
| Ethambutol | 181 (76–226) | 1.0 (1.0) | 28 (15–39) | 1 (3) | 0 | 0 | 2 (7) |
| Anticonvulsants | 266 (219–335) | 1.1 (1.0–1.2) | — | 3 (10) | 3 (12) | 0 | 2 (7) |
| Phenytoin | 301 (232–369) | 1.1 (1.0–1.2) | 27 (20–53) | 0 | 2 (8) | 0 | 0 |
| Carbamazepine | 300 (300) | — | 17 (11–21) | 2 (7) | 1 (4) | 0 | 1 (4) |
| Phenobarbital | — | — | 11 (7–15) | 0 | 0 | 0 | 4 (14) |
| Lamotrigine | 205 (205) | — | — | 1 (3) | 0 | 0 | 1 (4) |
| Antibiotics | 114 (43–220) | 1.3 (1.1–1.5) | 3 (10) | 6 (24) | 0 | 4 (14) | |
| Cotrimoxazole† | 127 (43–220) | 1.3 (1.1–1.4) | 23 (6–46) | 3 (10) | 5 (20) | 0 | 4 (14) |
| Amoxicillin | 390 (93–686) | 1.5 (1.5) | — | 0 | 0 | 0 | 0 |
| Streptomycin | 107 (32–134) | — | 19 (11–35) | 0 | 2 (8) | 0 | 0 |
| Allopurinol | — | 1.2 (1.1–1.2) | 218 (51–555) | 0 | 0 | 0 | 0 |
IQR, interquartile range.
Interval between initiation of drug and development of first signs of reaction.
Cotrimoxazole was identified as a possible offender by Algorithm for Assessment of Drug Causality in Stevens–Johnson Syndrome and Toxic Epidermal Necrolysis algorithm in six of 30 hepatocellular (20%), two of 25 mixed (8%), and no cholestatic patterns of drug-induced liver injury.
In this large African study in which greater than 90% of EN patients were HIV-infected, major findings were that (1) EN-DILI was common, occurring in a third of patients; (2) half of EN-DILI cases were mild to moderate; (3) multiple drugs caused EN-DILI, none of which was associated with higher prevalence or severity; (4) hepatocellular and mixed injury patterns were the most common, and the pattern was usually associated with the most frequent offenders, nevirapine and cotrimoxazole; (5) HIV infection and AIDS stage did not influence the prevalence and severity of EN-DILI; (6) EN-DILI prolonged EN hospitalization; (7) EN-DILI did not increase EN-associated mortality; (8) the hepatocellular injury pattern was most likely to be severe; and (9) we found no clear predictive factors for EN-DILI in this cohort. The major offending drugs reflected the HIV/AIDs prescribing patterns, which were antiretrovirals or prophylaxis or treatment for opportunistic infections.
This cohort’s 30% prevalence of EN-DILI is consistent with reports in other ethnic populations with low or negligible HIV prevalence, as is the association of nevirapine with hepatocellular injury.7 However, we did not note higher CD4 counts predis-posing to nevirapine-induced DILI and cotrimoxazole mainly causing cholestatic injuries, perhaps suggesting distinct immunopathogeneses.8,9 Limitations of this study include its retrospective design, missing data, and a relatively small sample size that restricted the power to detect small effect sizes. Nevertheless, we were largely able to restrict missing data to less than 10% for all variables considered.
This study’s findings add to the current body of knowledge on EN-DILI by describing it in an African HIV-endemic population.
Supplementary Material
Clinical Implications.
Among an African epidermal necrolysis patient cohort (>90% HIV infected), one third developed drug-induced liver injury (DILI), most of which was of mild to moderate severity. Multiple drugs caused EN-DILI, predominantly hepatocellular and mixed injury pattens and DILI prolonged hospitalization without increasing mortality.
Acknowledgments
R. Lehloenya was supported by the South African Medical Research Council and received nonrated researcher support from the South African National Research Foundation. J.G. Peter is supported by National Institutes of Health Grant K43TW011178-01, EDCTP2 programme supported by the European Union (grant number TMA2017SF-1981) and rated-researcher support from the South African National Research Foundation.
Footnotes
Conflicts of interest: The authors declare that they have no relevant conflicts of interest.
REFERENCES
- 1.Peter J, Choshi P, Lehloenya RJ. Drug hypersensitivity in HIV infection. Curr Opin Allergy Clin Immunol 2019;19:272–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Jee A, Sernoskie SC, Uetrecht J. Idiosyncratic drug-induced liver injury: mechanistic and clinical challenges. Int J Mol Sci 2021;22:2954. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Devarbhavi H, Raj S. Drug-induced liver injury with skin reactions: drugs and host risk factors, clinical phenotypes and prognosis. Liver Int 2019;39:802–11. [DOI] [PubMed] [Google Scholar]
- 4.Devarbhavi H, Raj S, Aradya VH, Rangegowda VT, Veeranna GP, Singh R, et al. Drug-induced liver injury associated with Stevens-Johnson syndrome/toxic epidermal necrolysis: patient characteristics, causes, and outcome in 36 cases. Hepatology 2016;63:993–9. [DOI] [PubMed] [Google Scholar]
- 5.Common Terminology Criteria for Adverse Events (CTCAE) Version 4.032010. 2011. http://evs.nci.nih.gov/ftp1/CTCAE/CTCAE_4.03_2010-06-14_QuickReference_5×7.pdf. Accessed March 14, 2019.
- 6.Benichou C. Standardization of definitions and criteria of causality assessment of adverse drug reactions. Drug-induced liver disorders: report of an international consensus meeting. Int J Clin Pharmacol Ther Toxicol 1990;28:317–22. [PubMed] [Google Scholar]
- 7.Yang L, Shou YH, Li F, Zhu XH, Yang YS, Xu JH. Retrospective study of 213 cases of Stevens-Johnson syndrome and toxic epidermal necrolysis from China. Burns 2020;46:959–69. [DOI] [PubMed] [Google Scholar]
- 8.Zhang C, Wang W, Zhou M, Han Y, Xie J, Qiu Z, et al. The interaction of CD4 T-cell count and nevirapine hepatotoxicity in China: a change in national treatment guidelines may be warranted. J Acquir Immune Defic Syndr 2013;62:540–5. [DOI] [PubMed] [Google Scholar]
- 9.Prakash M, Poreddy V, Tiyyagura L, Bonacini M. Jaundice and hepatocellular damage associated with nevirapine therapy. Am J Gastroenterol 2001;96:1571–4. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.