Amoxicillin-associated Stevens-Johnson syndrome or toxic epidermal necrolysis: systematic review

Abstract Our aim was to explore and summarize available cases of Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) suspected to be associated with amoxicillin reported in the literature. Electronic searches were conducted in several databases. Fifty-one publications describing a total of 64 patients who satisfied inclusion criteria were included in the review. The age of the patients ranged from 1.5-80 years (median: 24.5 years). TEN, SJS and SJS/TEN overlap were diagnosed in 30 (46.9%), 28 (43.8%) and 1 (1.6%) patients, respectively. SJS/TEN may occur promptly after administration of amoxicillin, but it could also be a delayed adverse effect. The total length of hospital stay ranged from 3-70 days (median: 16 days). Amoxicillin-induced SJS/TEN is accompanied by frequent occurrence of serious complications, long-term ocular and skin sequelae and high mortality rate. Clinicians should be aware that amoxicillin alone or combined with clavulanic acid can cause SJS/TEN in patients of all ages.


Introduction
Stevens-Johnson syndrome (SJS), toxic epidermal necrolysis (TEN, formally known as Lyell syndrome) and SJS/TEN overlap are rare, acute dermatoses characterized by extensive epidermal detachment, erosion of mucosae and failure of the internal organs [1,2]. Incidence of those syndromes varies in the literature, from 0.4-1.45 per million patients for TEN to 3.96-5.3 cases per million patients for SJS [3,4]. SJS, TEN and SJS/TEN overlap have similar prodromal symptoms (fever, headache, sore throat, weakness, malaise, or cough) and disease-specific symptoms (detachment of mucous membranes, skin erosions) [2], but differ by the degree of body surface area (BSA) involvement. While SJS is defined as skin involvement of less than 10%, TEN is established when it is up to 30% and the overlap of both is between 10-30% of skin involvement [5]. Those reactions are mostly caused by drugs, being in this case considered delayed-type hypersensitivity reaction [2]. The drugs that frequently cause SJS/TEN are anticonvulsants, nonsteroidal anti-inflammatory drugs, allopurinol, nevirapine and antibiotics [2]. Although sulfonamides are responsible for the highest proportion of antibiotic-induced cases of SJS/TEN, numerous cases of SJS/TEN induced by beta-lactams antibiotics, especially amoxicillin, can also be found in the literature [6].
Amoxicillin is one of the most commonly used antibiotics in primary care setting [7]. It is a semisynthetic amino-penicillin with bactericidal activity toward a wide variety of gram-positive bacteria and some gram-negative organisms [7]. Amoxicillin is used for the treatment of tonsillitis, pharyngitis, otitis media, lower respiratory tract infections, skin and skin structure infections, urinary tract infections and for the eradication of Helicobacter pylori [8]. It can be administered alone or in combination with clavulanic acid, a beta-lactamase inhibitor that provides the efficacy of amoxicillin against resistant strains of bacteria [7]. Both amoxicillin and amoxicillin/clavulanate are generally well-tolerated [7,9]. Their use may occasionally be accompanied by mild gastrointestinal side effects, such as diarrhea, nausea, vomiting, while serious side effects such as hepatotoxicity and hypersensitivity reactions are rare [7,9]. SJS/TEN is one of very rare adverse drug reactions associated with the use of amoxicillin [10].
Although more than 100 drugs were associated with SJS and TEN, at least in singular case reports [11,12], strength of these associations remains questionable, especially for antibiotics that are used characteristically for a short time period. Therefore, the aim of this review was to investigate and summarize information about the characteristics of published cases of SJS/TEN suspected to be associated with the use of amoxicillin to estimate degree of the causal relationship between amoxicillin therapy and SJS/TEN, as well as to analyze outcomes of amoxicillin-associated SJS/TEN.

Materials and methods
Our systematic review was registered in the international prospective register of systematic reviews (PROSPERO) under the registration number CRD42021266253 before initiation of the research. Electronic searches were conducted by two authors independently (DF, JB) in MEDLINE (PubMed), EBSCO, SCIndeks (Serbian citation index), Scopus, Google Scholar, Cochrane Central Register of Controlled Trials -CENTRAL (Cochrane Reviews and Trials), and ClinicalTrials.gov without any language or date restriction from the beginning of indexing up to November 09, 2021. A detailed search strategy for each database is provided in Table 1. Inclusion criteria were as follows: patients of any race, age and both sex (1) receiving amoxicillin or amoxicillin/clavulanic acid therapy, at any dose regimen, for any indication, (2) having diagnosis of SJS, TEN or SJS/TEN overlap, (3) with onset of the abovementioned syndromes after initiation of amoxicillin or amoxicillin/clavulanic acid therapy in which one of these drugs was considered as the only possible cause of the syndrome, (4) reported from clinical trials, observational studies or individual cases including case series. Exclusion criteria were as follows: (1) cases of SJS, TEN or SJS/TEN overlap in patients taking amoxicillin or amoxicillin/clavulanic acid mentioned in review articles, without detailed description; (2) cases of SJS, TEN or SJS/TEN overlap in non-human species; (3) published cases with incomplete data; (4) published cases in which there were other possible causes of SJS, TEN or SJS/TEN overlap or in which causality was not or could not be determined. The reference lists of the retrieved articles were also searched for additional relevant case reports and series.
Initially, the eligibility of retrieved publications was screened based on the title and abstract by two authors independently (DF, JB). In cases in which it was not possible to assess whether the publication fully corresponds to the research topic based on the title and information provided in the abstract, the full text of the publication was retrieved and analyzed. The article was included in review if 5 authors (SJ, DF, JB, MM, AP) agreed that eligibility criteria were met. In case that the reviewers had different opinions about the eligibility of a study for inclusion, the matter was resolved by the senior author (SJ).
A data collection sheet was created and included studies were assessed for: publication ID; report ID; review author initials; citation and contact details; eligibility for review; study design; total study duration; risk of bias, if applicable (randomization if any, sequence generation, allocation sequence concealment, blinding, selection bias, selector bias, extractor bias, selective reporting bias, follow-up time bias, attrition bias, other concerns about bias); total number of patients; age of patients; sex of patients; medical history; used medication (amoxicillin, amoxicillin/clavulanic acid or both); amoxicillin or amoxicillin/clavulanic acid dosing regimen; duration of amoxicillin treatment and time to onset of the reaction (from the first dose of the drug to the first symptoms of the reaction, including prodromal (ARTAK: ((Stevens-Johnson syndrome) OR (toxic epidermal necrolysis)) AND (amoxicillin)) Scopus TITLE-ABS-KEY (((stevens-johnson AND syndrome) OR (toxic AND epidermal AND necrolysis)) AND (amoxicillin)) Google Scholar ((Stevens-Johnson syndrome) OR (toxic epidermal necrolysis)) and amoxicillin and ('adverse drug reaction') CENTRAL ((Stevens-Johnson syndrome) OR (toxic epidermal necrolysis)) AND (amoxicillin) in All Text ClinicalTrials.gov Condition or disease: ((Stevens-Johnson syndrome) OR (toxic epidermal necrolysis)) Others terms: amoxicillin symptoms); indication for amoxicillin or amoxicillin/ clavulanic acid treatment; setting of antibiotic treatment; country; diagnosis of the reaction (SJS, TEN, SJS/ TEN overlap); severity-of-illness score for toxic epidermal necrolysis (SCORTEN); symptoms and signs of the reaction; results of a skin biopsy; Tumor Necrosis Factor (TNF) alpha levels in serum; laboratory parameters of systemic inflammation or myelosuppression; serum level of interleukin-2 receptor; serum level of interleukin-6; assessment of liver and renal function; blood culture result; skin culture result; mucosal lesions confirmed by bronchoscopy, esophagogastroscopy or colonoscopy; pneumonitis on chest X-ray; signs of local infection of skin lesions; testing for drug hypersensitivity (e.g., patch test, intradermal test, lymphocyte transformation test [LTT], et cetera); information about causality assessment by the authors (e.g., algorithm for assessment of drug causality for epidermal necrolysis [ALDEN], Naranjo score, et cetera); method of treating (with complete list of medication); length of hospital stay; outcomes of treatment (cure rate, death rate, functional outcomes); presence of clinical signs of systemic infection (body temperature, heart rate, blood pressure, respiratory rate); antibiotic regimen used for treatment; complete medication list the patient was taking before onset; rate of adverse events related to treatments; any residual functional disability after completion of the treatment; any skin scarring up to one year after onset of syndrome. The data were extracted by two investigators independently (AP, MM) and then collating of the two tables was done by the senior author (SJ), who produced the final extraction table.

Results
Supplementary Table 1 provides an overview of each case, while Table 2 summarizes basic characteristics of included cases. Cases were reported in all age groups, while both genders were almost equally represented. Twenty-seven patients (42.2%) were pediatric patients (<18 years). The largest number of cases occurred in the United States of America (n ¼ 13; 20.3%).
TEN was the most frequently reported diagnosis (n ¼ 30; 46.9%). SCORTEN was reported for 12 patients (18.8%) and values of this score are reported in Supplementary Table 1. Time to onset of the reaction was specified in 34 cases (53.1%) and it ranged from 6 hours to 38 days (median: 3.0 days). Three patients (4.7%) continued to receive amoxicillin or amoxicillin/clavulanic acid after the onset of the reaction which was associated with the worsening of the condition. Naranjo score rating was reported as probable and possible in 8 (12.5%) and 3 cases (4.7%),  Abbreviations: nnumber, SJS -Stevens-Johnson syndrome; TENtoxic epidermal necrolysis.
respectively. In one of the cases (1.6%) ALDEN score was 4, while in another case (1.6%) SJS was rated as probable/likely caused by amoxicillin according to the World Health Organization Causality Scale. In two cases (3.1%) determination of drug culpability was estimated according to the timing method proposed by the French group on TEN. Occurrence of prodromal symptoms (general malaise, sore throat, cough) was specified in less than one-third of the patients (29.7%). Description of skin manifestations was provided in 49 cases (76.6%) and involved presence of erythematous macules, maculopapular lesions, atypical target lesions, targetoid lesions, vesicular lesions, purpuric lesions, ulcers, bullous lesions and blisters with epidermal detachment, necrotic lesions, pruritus, and skin pain. Localizations of skin lesions included scalp, face, neck, trunk, extremities, hands, palms, feet and soles. Positive Nikolsky's sign was reported in 11 patients (17.2%). Skin manifestations were not present in 2 patients (3.1%). Fever was reported in 26 patients (40.6%). Some form of mucosal involvement was specified in 52 patients (81.3%). Oral involvement included manifestations like oral lesions (e.g., blisters, ulcers, vesicles, aphthae, erosions) some of which were associated with bleeding, purulent discharge and crusting, increased salivation, gingival detachment, papules on tongue, blisters, erosions and crusting on the lips, edema, pain sensation in oral mucosa, difficulty in opening mouth, stomatitis, cheilitis, and mucositis. Ocular manifestations were reported as conjunctival hyperemia, erosions or injection, conjunctivitis, keratoconjunctivitis, purulent or watery discharge, eye redness, pseudomembrane formation, symblepharon, vesicular lesions, periorbital edema, reduced visual acuity, photophobia, dryness in eyes, and detachment of the inferior conjunctival border. Involvement of nasal mucosa was reported in the form of mucositis, vesicles, redness, and crusted wounds. Manifestations involving genitourinary tract included vesicles, papules, blisters, erosions with pain, bleeding, crusting or purulent discharge, ulcers on genital mucosa, urethral lesions, dysuria, balanitis, erythema of meatus, and anogenital swelling. Esophageal and gastrointestinal involvement included manifestations such as esophageal erosions, eating difficulties, ileus, diarrhea, vomiting, painful defecation and painful bowel movements, rectal bleeding, constipation, abdominal pain, upper gastrointestinal tract bleeding, abdominal compartment syndrome. Endoscopy was used to confirm esophageal erosions in one patient (1.6%).
Frequencies of reported complications are reported in Table 2.
Two patients (3.1%) had reported previous history of allergic reaction to penicillin. Testing for drug hypersensitivity was performed in 4 patients (6.3%) with variable results: one patient had positive patch test and LTT for amoxicillin and negative for mefenamic acid [23]; one patient had negative patch test, intradermal test and LTT for amoxicillin/clavulanic acid, while ampicillin-specific T-cell lines (TCLs) proliferated against penicillin and carboxyfluorescein diacetate succinimidyl ester (CFDA-SE) assay results were positive for amoxicillin in CD4þ cells exclusively [26]; one patient had negative intradermal test for amoxicillin/clavulanic acid, negative LTT for penicillin, ampicillin, amoxicillin and amoxicillin/clavulanic acid, and negative TCLs for amoxicillin and amoxicillin/clavulanic acid [50]; one patient had negative patch and intradermal test for amoxicillin/clavulanic acid, negative LTT for penicillin, ampicillin, amoxicillin and amoxicillin/clavulanic acid, while TCLs were positive for penicillin and negative for ampicillin, amoxicillin and amoxicillin/clavulanic acid [50]. None of the patients was re-challenged with the drug. Confirmation of diagnosis by skin biopsy was reported in 14 patients (21.9%).
Almost all patients were treated in a hospital setting (n ¼ 63; 98.4%). It was specified that 10 patients (15.6%) were managed in the intensive care units, while 11 patients (17.2%) were treated in specialized burn units. The specified total length of hospital stay ranged from 3 to 70 days (median: 16 days). The summary of treatment for each patient is given in Supplementary Table 1. Treatment with amoxicillin and amoxicillin/clavulanic acid was reported to be discontinued in 27 patients (42.2%) either after establishing diagnosis of the reaction or when the therapy course was completed. Use of some form of supportive and symptomatic care (e.g., local care, intravenous fluids, nutritional support, antihistamines, antiseptics, analgesics, mechanical ventilation … ) was reported in 49 patients (76.6%). Systemic corticosteroids were reported to be administered in 29 (45.3%) patients and intravenous immunoglobulin (IVIG) in 11 (17.2%) patients (5 of these patients received both systemic corticosteroids and IVIG). Systemic cyclosporin use was reported in 2 patients (3.1%), while plasmapheresis was performed in 3 patients (4.7%). Systemic and topical antibiotics were administered in 17 (26.6%) and 13 (20.3%) patients, respectively. Adverse events related to treatment were reported only in one patient (1.6%): a single dose of carbapenem antibiotics administered due to postoperative infection triggered an anaphylactic shock.
Outcomes of individual patients are reported in Supplementary Table 1. The majority of the patients survived (n ¼ 54; 84.4%), while 10 patients died (15.6%). Deceased patients were generally older (5 of 10 had between 64 and 80 years) and the majority of them suffered from TEN (8 of 10). Their clinical course was generally complicated by other serious pathologies which mostly involved sepsis/septic shock, systemic inflammatory response syndrome, multiple organ failure, acute renal failure, respiratory failure, acute respiratory distress syndrome and/or pulmonary embolism which eventually led to death.
Long-term consequences (sequelae) were reported in 10 of 54 patients who survived (18.5%) and more frequently in younger patients (8 of 10 had 30 years or less). A 17-year-old boy had slow healing of corneal abrasions with pseudomembranes and symblepharon, along with persistent dryness [39]. year old girl at the 1-year follow-up had maintained a best-corrected visual acuity of 20/50 in each eye, advanced symblepharon, severe lid margin complications, moderately dry eyes and an intact ocular surface with the help of soft bandage contact lenses, while 2-year-old girl at 5-month follow-up had mild lid margin complications, minimally dry eyes and the vision of 20/25 in each eye [55]. A 3-year-old girl had three long-term sequelae: dysphagia as a result of a long esophageal stricture requiring gastrostomy followed by serial retrograde esophageal dilatations, bilateral blindness secondary to severe corneal abrasions and residual skin hyperpigmentation with scattered scarring [40]. A 37-year-old woman developed progressive cholestatic jaundice resembling sclerosing cholangitis also related to amoxicillin with permanent scarring and dilatation of the intra-hepatic bile ducts, while 6 weeks after initial TEN presentation, TEN threatened to recur after ciprofloxacin use for a urinary infection [20]. An 18-year-old girl had residual skin efflorescence [49], a 30-year-old woman [31] and 54-year-old woman [32] had hyperpigmentation, while a 2-year-old boy had asthenia, phimosis in addition to marked hyperpigmentation [36]. A 26year-old woman had severe ocular problems with permanently reduced visual acuity and recurrent ingrown eyelashes and visible facial scarring accentuated by variable pigmentation in the affected areas [58].

Discussion
Our study showed that amoxicillin, either administered alone or in combination with clavulanic acid, can cause SJS/TEN, both in patients with and without previous allergic reactions to beta-lactam antibiotics. These serious adverse effects of amoxicillin occurred both in children and adults, wherein the culprit drug had been used in approved indications in all patients included in this review. These amoxicillin-induced reactions in almost all cases involved severe skin changes, with often oral, ocular, genitourinary and gastrointestinal manifestations. SJS/TEN may occur promptly after administration of amoxicillin, usually within the first few days, but it could also be delayed adverse effect of amoxicillin occurring after several weeks from the drug initiation. Amoxicillin-induced SJS/TEN is accompanied by frequent occurrence of serious infections and other complications, long-term ocular and skin sequelae and high mortality rate.
Amoxicillin was the most commonly used antibiotic in the United States in 2020, with a total prescription number of more than 31 million, while amoxicillin/clavulanic acid was in third place with 21 million prescriptions [64]. Similarly, amoxicillin and amoxicillin/clavulanate are the most consumed antibacterial agents in most European countries [65]. In addition to a good pharmacokinetic profile, amoxicillin is also believed to be a fairly safe drug, which can be used in children, pregnancy and lactation. The most common adverse effects of this drug (frequency 1/1000 to <1/100) are diarrhea, nausea, skin rashes, urticaria, vomiting and vaginitis [8,66]. SJS and TEN are listed as possible undesirable effects in the summary of product characteristics (SmPC) both of amoxicillin (as very rare, in <1/10,000 of patients) [10] and amoxicillin-clavulanic acid (not known frequency) [66]. It seems that the use of amoxicillinclavulanic acid is associated with a higher frequency of gastrointestinal, hepatic and haematological adverse effects compared to amoxicillin alone [67]. The possible association between the use of amoxicillin and the occurrence of these serious skin side effects was observed in several dozen of patients. The pathogenesis of SJS/TEN has not been fully elucidated, but it is believed to be immune-mediated [68]. SJS and TEN are based on cytotoxic cellular immune reaction directed at keratinocytes, where CD8þ T lymphocytes, activated macrophages and CD4þ T cells in the skin attack normal cells, provoked by recognition of foreign antigen on the target cells [69]. This immunological reaction leads to large increase of tumor necrosis factor concentration in the affected tissue, followed by apoptosis of keratinocytes and attraction of other inflammatory cells [69]. Besides, lytic Fas ligand (FasL, CD95L) is another possible cause of apoptosis; it is overexpressed by keratinocytes, and its soluble form is also increased. How much binding of the FasL for Fas (CD95, a cell surface death receptor) contributes to apoptosis is not clear yet, but it cannot be overlooked [70]. Drugs may cause SJS or TEN acting as haptens and mimicking foreign antigens, or by their oxidant properties that decrease anti-oxidant capacity of keratinocytes and make them prone to damage by free radicals emerging for other reasons (e.g., exposure to ultraviolet light or else) [71].
The risk of SJS/TEN is the greatest immediately after the initiation of drug, decreases over time and almost completely disappears after 8 weeks of drug administration [68]. The results of this review show that time elapsed from the first dose of amoxicillin to the onset of the first symptoms of SJS/TEN ranged from 1 to 38 days, with a median of 3 days. A 3 days median time latency between the beginning of drug administration to the appearance of SJS/TEN was also observed with macrolide antibiotics (range 1-14 days) [72] and acetaminophen (range 1-21 days) [73]. However, SJS/TEN occur significantly later after the administration of nevirapine (10-240 days, median 12 days) [74], carbamazepine (12-20 days, median 15 days), phenytoin (16-33 days, median 24 days), phenobarbital (9-40, median 17 days) and allopurinol (14-32 days, median 20 days) [75]. Since SJS/TEN occur very soon after the administration of acetaminophen, macrolides and amoxicillin, and these drugs are frequently prescribed for the treatment of infections and flu-like symptoms such as fever, there is a risk of false causality between these drugs and SJS/TEN. This protopathic bias is very difficult to eliminate when assessing the causality between SJS/ TEN and the use of antibiotics and acetaminophen, because there is a possibility that these drugs have been used to treat prodromal symptoms of SJS/TEN caused by another trigger [76]. On the other hand, amoxicillin, macrolides and acetaminophen are much more commonly used in the general population than antiepileptics and nevirapine, so it is possible that many people had become sensitized to these drugs [77]. This may also be the explanation for the significantly shorter time to onset of SJS/TEN after administration of antibiotics and acetaminophen [78].
Cases with SJS/TEN have been reported in all age categories [79], which is in accordance with the result of our review since the youngest patient was 1.5 years old boy [19] while the oldest patient had 80 years [56]. The population of patients with SJS/TEN associated with the use of amoxicillin was significantly younger compared to those in a multinational, casecontrol study conducted in Europe between 1997 and 2001 (EuroSCAR-Study) which evaluated the risk of medications to induce severe cutaneous adverse reactions (median 24.5 years vs. 50 years). Among the patients included in this study, approximately 42% were pediatric patients, which is also significantly higher compared to the proportion of children with macrolide-induced SJS/TEN (11%) [72]. This can be explained by the fact that amoxicillin is the drug of choice for the treatment of upper respiratory tract infections in children [80], which are particularly common in this age [81]. Regarding gender distribution, SJS and TEN affect men and women almost equally [79], which was also observed among the cases included in this review.
Almost all of the cases included in this review had typical symptoms and signs of SJS/TEN, with the exception of the two cases in whom the use of amoxicillin led to the development of incomplete SJS without skin involvement [31,53]. There were no cases without skin involvement among patients with macrolides and acetaminophen-induced SJS/TEN [72,73]. In addition to severe skin damage, epidemiological data indicate that the involvement of the oral, genital and/ or ocular mucosa may occur in more than 90% of the patients with SJS/TEN [68]. Oral mucosa was affected in about 63% of patients with amoxicillin-induced SJS/TEN, which is significantly less compared to oral mucosa involvement among patients in whom SJS/ TEN was associated with macrolide antibiotics (92,6%) [72] and acetaminophen (86,1%) [73]. When it comes to ocular involvement, approximately 55% of patients with amoxicillin-induced SJS/TEN had ocular manifestations, which is a proportion similar to those in the population of patients with acetaminopheninduced SJS/TEN (52.8%) [73]. However, it could be said that ocular manifestations are also less common in amoxicillin/induced SJS/TEN compared to SJS/ TEN caused by macrolide antibiotics (70,4%) [72]. Genitourinary region was affected in approximately 40% of all three populations of SJS/TEN patients [72,73]. Although SJS-TEN induced by amoxicillin is characterized by less involvement of mucous membranes, it is accompanied by a significant incidence of complications, long-term sequelae and a high mortality rate. Long-term ocular, skin and other functional sequelae were reported in 18.5% of patients with amoxicillin-induced SJS/TEN who survived, which is similar to the proportion of patients with long-term consequences among population of SJS/TEN induced by macrolides (14.8%) [72] and acetaminophen (13.9%) [73]. The mortality rate in patients with amoxicillin-associated SJS/TEN was 15.6% which is significantly higher in relation to SJS/TEN induced by acetaminophen (0%) [73], valproate (4.8%) [82] and macrolide antibiotics (11.1%) [72]. Mortality was significantly higher in patients in whom amoxicillin provoked the occurrence of TEN (8 patients, 26.67%) compared to patients with amoxicillin-associated SJS (2 patients, 7.1%). This is in accordance with epidemiological data indicating that TEN has a worse prognosis and a much higher mortality rate of 25-30% compared to SJS, where the mortality rate is 1-5% [2].
The diagnosis of SJS/TEN is based on a combination of anamnestic data, clinical symptoms and signs and histopathological analysis obtained by skin biopsy [1]. The distinction among SJS, SJS/TEN overlap, and TEN based on histopathological findings is not possible [83]. In addition, due to the characteristic clinical picture that accompanies SJS/TEN, histopathological analyzes are often not necessary to diagnose these severe cutaneous reactions, especially in patients with a clear temporal association between drug administration and symptom onset [84]. Positive Nikolsky's sign may be a helpful clinical indicator of epidermal necrolysis, although it is not specific to SJS/TEN and can also be positive in pemphigus [1]. For determination of the causality between the use of the drug and the occurrence of SJS/TEN, the most important are the anamnestic data relating to the time that elapses from the use of the drug and the appearance of the first symptoms of SJS/TEN [79]. Causality can be assessed retrospectively using ALDEN, an instrument that cannot be used to assess causality in the acute phase of SJS/TEN [1,79]. Due to the severity of symptoms and the high mortality rate, intradermal testing and rechallenge should never be used for confirmation of the causality between the culprit drug and SJS/TEN [1,68,83]. Assessment of causality in patients with SJS/TEN may be particularly problematic in patients suspected of having these reactions as a result of antibiotic use due to protopathic bias which is often difficult to eliminate [76]. Identification of the culprit drug is also difficult in patients who have been exposed to multiple drugs [1]. In this case, it is necessary to take into account all the medications that the patient took 2 months before the onset of SJS/TEN [1]. For each drug taken by the patient clinicians should estimate the probability that the drug was present in the body at the onset of SJS/ TEN by analyzing its pharmacokinetic parameters, any renal or hepatic dysfunction and possible drug interactions, as well as notoriety of the drug to cause SJS/TEN, bearing in mind that implication of the particular drug is more likely if the patient has a previous history of a drug hypersensitivity reaction with the same or similar drug [1].
Withdrawal of the culprit drug and high-quality, multidisciplinary supportive care are crucial in the treatment of patients with SJS/TEN [1]. Rapid identification of the causative agent and cessation of its further use are directly related to a better prognosis and lower mortality rate in SJS/TEN [68,69]. A multidisciplinary approach should be provided in the treatment of patients with SJS/TEN, including the mandatory participation of specialists in dermatology, plastic surgery, ophthalmology and skincare nursing [1]. If other organs are affected, specialists from other clinical disciplines should be included in the management team [1]. Careful local care of damaged skin and the use of bland emollients are implied [1]. A major aim of supportive care in SJS/TEN is fluid resuscitation and prevention of the development of organ hypoperfusion and shock [1]. Patients with >10% BSA epidermal loss should be treated in intensive care or burn units [1]. The treatment of acute cutaneous pain is obligatory. Opioid analgesics should be used in SJS/TEN patients with moderate to severe pain [1]. Although the results of some studies indicate that systemic corticosteroids [85], IVIG [86], cyclosporine [87], combined use of corticosteroids with IVIG [88] significantly reduce mortality in patients with SJS/TEN, according to the UK guideline for the management of SJS/TEN [1] no active therapeutic regimen with unequivocal benefit exists for SJS/TEN. In addition, another immunosuppressant drug, etanercept, has been considered lately as a potential treatment of SJS/TEN. The results of the recent metaanalysis showed that the use of etanercept in SJS/TEN patients was followed by the lowest mortality rate compared to cyclosporine, corticosteroids and IVIG, so etanercept in combination with adequate supportive care may provide the best results in patients with TEN or SJS/TEN [89]. Of the 10 patients with amoxicillin-associated SJS/TEN who died, 4 received systemic corticosteroids, 1 IVIG, 1 combination of steroids and IVIG, while 4 patients received only supportive therapy.
This systematic review has the following limitations: (1) a relatively small number of reported cases with amoxicillin-induced SJS/TEN was analyzed; (2) completeness of the included cases varied, and some important information was missing or incompletely presented (such as dosage of administered amoxicillin, duration of treatment before the onset of symptoms, information about causality assessment by the authors, etc.); (3) in the vast majority of the analyzed cases assessment of the potential contribution of concomitant therapies is missing; (4) it should be noted that even if cases contain all the essential information, it is often not possible to establish definitive conclusion regarding causality. However, we believe that this review article can help clinicians recognize and adequately treat amoxicillin-induced SJS/TEN.
In conclusion, clinicians should be aware that amoxicillin alone or combined with clavulanic acid can cause SJS/TEN in patients of all ages. Given the high mortality rate that accompanies amoxicillin-associated SJS/TEN, prompt recognition and withdrawal of the culprit drug along with supportive care is of crucial importance. However, this serious side effect of amoxicillin is very rare and this antibiotic should still be considered safe for use in adequate indications even in vulnerable populations. On the other hand, the awareness that amoxicillin can cause such a severe adverse effect could positively affect its rational use and prevent unnecessary prescribing of this antibiotic.

Disclosure statement
No potential conflict of interest was reported by the authors.

Funding
The study was partially funded by grant No 175007 given by Ministry of Education, Science and Technological Development of the Republic of Serbia.