key: cord-0977430-6npozfqm
authors: Gao, Xuemei; Tang, Xuhua; Ai, Lu; Gao, Qian; Liao, Qiman; Chen, Mukai; Chen, Xiaohong; Zhou, Hui; Ye, Yanting; Li, Minyi; Han, Jiande; Wang, Fang
title: Acute pancreatic injuries: A complication of Stevens-Johnson syndrome/toxic epidermal necrolysis associated with cytotoxic immune cell activation
date: 2020-06-17
journal: J Am Acad Dermatol
DOI: 10.1016/j.jaad.2020.06.043
sha: 14b6b01da1c309e33a9ee5e7beec7f4e603d5457
doc_id: 977430
cord_uid: 6npozfqm

Abstract Background Complications involving internal organs are usually present in Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN). However, pancreatic complications are rarely reported and studied. Objective To summarize clinical characteristics of SJS/TEN-associated acute pancreatic injuries (API) and to investigate underlying inflammatory mechanisms. Methods Clinical records of 124 inpatients with SJS/TEN were reviewed. Serum levels of tumor necrosis factor-α, interleukin (IL)-6, IL-18, IL-15, IL-12p70, and soluble CD56 were determined in 18 healthy donors and 17 patients with SJS/TEN, including three patients with API. Results API was diagnosed in 7.3% (9/124) of patients in the SJS/TEN cohort. Serum transaminase elevation and hypoalbuminemia occurred more frequently in patients with API compared to those without pancreatic symptoms (P = 0.004, P = 0.0002, respectively). Although API did not alter mortality rate of SJS/TEN, it was associated with longer hospitalization stays (P = 0.008). Within the serum cytokines that were elevated in SJS/TEN, only IL-18 was found to be selectively increased in patients with API compared to those without pancreatic injuries (P = 0.032). Limitations Cohort size is small. Conclusion API is a gastrointestinal complication of SJS/TEN in which hepatotoxicity is more likely to occur. Overexpression of IL-18 might be involved in this unique entity.

Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) are life-76 threatening skin diseases that are often triggered by a drug and are characterized by erythema, 77

A total of 124 patients with SJS/TEN admitted to the First Affiliated Hospital, Sun Yat-sen 111

University from January 2000 to June 2019 were identified according to the diagnostic criteria 112 described by Bastuji-Garin et al, in which SJS/TEN subtypes were mainly classified by the 113 extent of epidermal detachment in addition to skin lesions (widespread erythematous or purpuric 114 macules and flat atypical targets) and involvement of mucous membranes. 14 SJS and SJS/TEN 115 overlap were respectively defined when detachment of the body surface area (BSA) below 10% 116 and between 10% and 30%, while TEN was met with the detachment greater than 30% of 117 BSA. 14 All enrolled patients had clear medical records of drug administration before skin 118 symptoms. We retrospectively documented their disease history, clinical manifestations, 119 laboratory test results including CBC, compressive metabolic panel, serum lipase and amylase 120 activity, autoantibodies, and imaging examination results including transabdominal ultrasound 121 and abdominal CT scan. 122

For diagnosis of API, the Atlanta classification standard was used. 15 The diagnosis of 123 acute pancreatitis (AP) was made when at least two of the following three features were met: (1) 124 typical abdominal pain; (2) biochemical evidence consistent with pancreatitis (serum level of 7 characteristic findings of AP suggested by imaging tomography. Pancreatic hyperenzymemia 127 was defined as a rise in pancreatic enzymes more than three times the upper limit of normal in 128 the absence of symptoms and abdominal CT imaging of pancreatic diseases. 16 129

In January 2017, we started a series of SJS/TEN-related scientific studies, which were 131 approved by the ethics committee of the hospital (IRB number 2019[01]). For cytokine tests, 17 132 patients with SJS/TEN, 3 API included, were enrolled after obtaining consent from each 133 individual from December 2017 to June 2019. Simultaneously, 18 sex-and age-matched 134 healthy controls were recruited (Supplemental Table 1 ). Peripheral blood (5 mL) was obtained 135 from each of 17 patients during the acute stage before treatment as well as from 18 controls. 136

Serum specimens were suspended from the clotted blood centrifuged at 4°C and 2,500 rpm for 137 10 minutes and then stored at -80°C until tested. 

Data with normal distributions were described as mean ± standard deviation. For data 145 not following a normal distribution, median (interquartile range) was used. Differences of 146 incidence rate between each two groups were compared by Fisher's exact test. Differences of 147 cytokine levels were assessed with the two-tailed unpaired Student's t test when they were 148 normally distributed. For data that were not normally distributed, the Wilcoxon-Mann-Whitney 149 nonparametric test was performed to detect significant differences. All tests were performed 150 using GraphPad Prism software, version 8.0 (GraphPad Software, Inc, USA). P values lower

In the entire cohort, SJS, SJS/TEN overlap, and TEN were diagnosed in 91, 4, and 29 156 patients, respectively. API was documented in 9 of these patients. Among them, 8 patients met 157 the diagnosis of AP, while the other 1 patient received the diagnosis of hyperamylasemia. 158 Therefore, the morbidity rate of API was 7.3%. Within the patients with API, there were 3 with 159 SJS, 1 with SJS/TEN overlap, and 5 with TEN ( Fig 1A) . Accordingly, the API incidence rate was 160 3.3% (3/91) in SJS, 25.0% (1/4) in SJS/TEN overlap, and 17.2% (5/29) in TEN. The API 161 morbidity rate in TEN was much higher compared to that in SJS (P = 0.020; Fig 1B) . 162

The API cohort consisted of 5 females and 4 males with the age range from 12 years to 164 69 years (median, 31 [19-56] years). In the other 115 patients without pancreatic symptoms, the 165 sex ratio of female: male was 1:1.05, and the ages ranged from 5 years to 93 years (median, 46 166

[26-60] years). The statistical analyses of sex distribution and age revealed no significant 167 difference between the two groups (P = 0.742, P = 0.148, respectively). 168

Similar to classical AP, abdominal pain occurred in 7 (7/9, 77.8%) patients in the initial 170 stage (from day 0 to day 8) of SJS/TEN (Table Ⅰ) . However, common triggers in classical AP 171 appear dispensable for SJS/TEN-associated API because gallstones history and alcohol abuse 172 were found in only 1 patient each ( Table Ⅰ) . Neither of the incidence rates showed significant 173 difference between API and NP groups (P = 0.204, P = 0.263, respectively; Figs 2A and 2B). 9 found in 77.8% (7/9) and 100% (9/9) patients with API. Both were significantly higher than those 178 in NP patients (31/115, 27.0%, P = 0.004 for comparison of transaminase elevation; 42/115, 179 36.5%, P = 0.0002 for hypoalbuminemia; Fig 2C) . In contrast to liver dysfunction, no difference 180 in the incidence of kidney dysfunction was observed (P = 0.648; Fig 2D) . Results of 181 autoantibody tests were available in 7 patients with API and 56 individuals in the NP cohort. The 182 percentage of positive autoantibodies was 3/9 (33.3%) in patients with API, while 15 (15/115, 183 13.0%) patients without pancreatic injuries exhibited positive serum autoantibodies. The 184 difference between the two groups was not statistically significant (P = 0.120; Fig 2E) . According 185 to the imaging examinations, pancreas abnormality presenting as diffuse enlargement of the 186 pancreas was found in 4 (4/9, 44.4%) patients. Detailed auxiliary test results are listed in Table  187 Ⅱ. 188

To control SJS/TEN, a combination of systemic corticosteroids and intravenous 190 immunoglobulin therapy was administered. To treat pancreatic injuries, 7 patients with 191 abdominal pain were given a short period of fasting, while the other 2 patients were fed with a 192 low-fat solid diet. Simultaneously, all 9 patients were given intravenous injection of proton pump 193 inhibitors and anti-secretory agents. The therapeutics were effective and the survival rate for API 194 cohort was 100%. Although 3 out of the 115 patients without pancreatic symptoms were 195 deceased due to SJS/TEN, no significant difference in mortality rate was found between the two 196 groups (P > 0.999; Fig 2F) . However, the hospitalization stay of patients with API ranged from 9 197 days to 53 days (average, 26.3 ± 13.9 days), which was significantly longer than that of NP 198 patients (average, 14.9 ± 4.9 days, P = 0.008; Fig 2G) . 199

In patients who had serum cytokine tests, the SJS/TEN group exhibited elevated serum 201 levels of TNF-α, IL-6, IL-18, IL-15, and IL-12p70 compared to healthy controls (Supplemental 10   Table 2 ). Conversely, soluble CD56 levels were significantly lower in SJS/TEN group than those 203 in heathy controls (Supplemental Table 2 ). In the SJS/TEN cohort, we compared cytokine 204 results between groups classified by the existence of pancreatic injuries (Supplemental Table 1 damage in the process of SJS/TEN. 19, 20 In the current study, we found 9 patients with API in a 217 cohort of 124 individuals. These epidemiology data indicate that API should be identified as a 218 complication of SJS/TEN, although the incidence rate may be variable among studies. 219

In the present API cohort, we found that more patients developed liver dysfunction 220 compared to those without pancreatic symptoms. Thus, we suspect that API is a gastrointestinal 221 complication that can parallel with injuries in other gastrointestinal organs. Notably, patients with 222 API did not show more evidence of kidney toxicity or positive autoantibodies, suggesting that 223 SJS/TEN may have preference and subtypes regarding toxicity in diverse internal organs. 224 Moreover, we found that patients with API on average had longer hospitalization stays, implying 225 increased financial and medical burdens. Therefore, screening and early monitoring for API are 226 required via abdominal symptom inquiry, physical examination of abdomen, and routine 227 laboratory tests for lipase and amylase activity. 11 It has been well recognized that the classical AP is usually triggered by gallstones or 229 alcohol abuse. 21 Occasionally, drugs such as acetaminophen and valproic acid could also 230 contribute to the etiology. 22 Surprisingly, in the present cohort, neither biliary disorders nor 231 alcohol abuse were present as triggers for SJS/TEN-associated API. Additionally, the culprit 232 drug spectrum is broader than the one that is regularly related to AP. Thus, we hypothesize that 233 API in the context of SJS/TEN is more likely to be mediated by the systemic inflammation rather 234 than common mechanical obstruction or dysfunction in pancreatic ducts. 235

Among the cytokines that we examined, IL-6 is commonly used in classical AP as an 236 early marker to predict disease severity. 23 A recent study showed that the assessment of serum 237 IL-6 increases the accuracy of systemic inflammatory response syndrome in severe AP. [24] [25] [26] In 238 an AP rat model, an increase of serum IL-6 precedes severe pancreatic edema and necrosis. 27 239

However, in our present study, IL-6 did not show significant elevation in SJS/TEN patients with 240

API. This negative finding indicates that API related to SJS/TEN may have a distinct cytokine 241 profile from classical AP. 242

Previously, only one study reported that serum IL-18 levels were elevated in patients 243 with classical AP and might aggravate liver injuries. 28 In an obese murine model, injection of IL-244 18 combined with IL-12 is sufficient to cause pancreatic necrolysis. 29 In another pancreatitis 245 murine model, TNF-α directly mediates protease activation and subsequent necrosis of 246 pancreatic acinar cells. 30 In the present API cohort, in addition to the significant elevation of 247 serum IL-18, IL-12p70 and TNF-α also exhibited an increasing trend. These results indicate that 248 cytotoxic cytokine dysregulation might be a cause for API in the setting of SJS/TEN. However, 249

due to the small cohort size and the lack of pancreas histopathology, the precise role cytokines 250 play in API mechanisms and whether they have distinct function in mediating other internal 251 organ toxicities associated with SJS/TEN will be an interesting area of further inquiry. 252

Consistent with previous studies, 12, 13, 31, 32 we found a universal elevation of multiple 253 cytokines in SJS/TEN. In addition to TNF-α, IL-6, IL-15, IL-18, and IL-12, other elevated 12 cytokines we previously found in SJS/TEN include IL-10, IFN-γ, and IFN-γ-inducible 255 chemokines CXCL9 and CXCL10. 33, 34 Such phenomena lead us to speculate that drug-induced 256 SJS/TEN could be another form of "cytokine storm" syndrome, 35 the condition that is triggered 257 by a wide variety of infections like COVID-19 36, 37 or noninfectious diseases like graft-versus-258 host disease. 38 Interestingly, in contrast to the broad cytokine elevation, we found a significant 259 decrease of serum soluble CD56 in SJS/TEN. CD56 (also called neural cell adhesion molecule 260 1), is an adhesion molecule expressed on the surface of NK and some innate lymphoid cells 261 and is a well-known phenotypic marker for NK cells. 39 Notwithstanding this, future studies will be required to clearly define cytokine diversity and their 266 specific function in promoting SJS/TEN process. 267 268

In conclusion, API is identified as a visceral complication associated with SJS/TEN. The 270 specific monitoring may help reduce medical burdens. Given that more symptoms in internal 271 organs are identified in SJS/TEN, specific subclassification will be needed. The precise role 272 cytokines play in internal organ injuries related to SJS/TEN is an outstanding question for further 273 studies. 274 275

We thank all patients participating in this study and physicians from Department of Dermatology 277 in the First Affiliated Hospital of Sun Yat-sen University for their clinical practice. We also would 278 like to express our deepest appreciation to Dr. Madison Mack for the language editing work. 

Stevens-Johnson Syndrome and Toxic Epidermal Necrolysis

Years of Progress but Still Much to Be Done

Cytotoxic proteins and therapeutic targets in severe cutaneous adverse 309 reactions

Successful treatment of interstitial lung disease related to 311

syndrome/toxic epidermal necrolysis overlap with etanercept: A case report 312 and published work review

Clinical classification of cases of toxic epidermal 314

Classification of acute pancreatitis--2012: revision of 317

Pancreatic hyperenzymemia is associated with bacterial 319 culture positivity, more severe and right-sided colitis

What is the real function of the liver 'function' tests?

Johnson syndrome/toxic epidermal necrolysis

Chemokine expression in diverse nonimmediate drug 359 hypersensitivity reactions: focus on thymus activation-regulated chemokine, cutaneous T-cell-360 attracting chemokine, and interleukin-10

Serum IFN-gamma-inducible chemokines CXCL9 and CXCL10 362 are elevated in non-immediate drug hypersensitivity reactions

Into the eye of the cytokine storm

COVID-19 and immunomodulator/immunosuppressant 367 use in dermatology

COVID-19: consider cytokine storm syndromes and 369 immunosuppression

Cytokine Dysregulation as a Mechanism of Graft-Versus-Host Disease

Signatures of human NK cell development and 373 terminal differentiation

Neural cell adhesion molecule: structure

immunoglobulin-like domains, cell surface modulation, and alternative RNA splicing

Supplemental Table 1. Demographic data of healthy controls and patients with Stevens-401

Johnson syndrome/toxic epidermal necrolysis from whom blood samples were collected. In the 402 group of 18 health controls, the sex ration of female: male was 1:1.25, and the ages ranged 403 from 14 years to 78 years (median

including 9 females and 8 males, were enrolled. The age ranges spread from 17 years to 75

In addition to results of culprit drugs collected from patients 406 with SJS/TEN, HLA gene tests of HLA-B*58:01 and HLA-B*15:02 were conducted when the 407 culprits were suspicious for allopurinol and carbamazepine