Pediatric SJS-TEN: Where are we now?

Introduction

Skin (cutaneous) reactions are among the most common types of adverse reactions to medications in children, accounting for 36% of any adverse drug reaction (ADR)¹. Many of these reactions are mild and self-resolving, but the rare severe ADRs can be associated with significant morbidity and even mortality. The diagnosis and treatment of cutaneous ADRs (cADRs) in children are challenging for several reasons. First, children are more commonly infected with viruses as compared with adults; each year, children in their first years of life average six to 10 respiratory viral infections and older children and adolescents average three to five such illnesses². Many of these viruses, such as Epstein–Barr virus, adenovirus, and enteroviruses, are frequently associated with cutaneous reactions that can be misinterpreted as cADRs. Second, triggers of severe cADRs, such as Stevens–Johnson syndrome (SJS), are different in adults and children; medications more often are implicated in adults and infections more commonly cause SJS in children³. Third, causality tools currently available—for example, Naranjo, algorithm of drug causality for epidermal necrolysis (ALDEN), and the US Food and Drug Administration’s division of drug experience—have been developed and studied primarily in adults and not applied to children. Finally, owing to the overall rarity of severe cADRs in children, there is little evidence to guide treatment.

ADRs can be either type A (predictable, caused by on-target drug action) or type B (heterogeneous, immunologically and metabolically mediated) (Table 1). cADRs can be classified clinically by morphology and severity on the basis of the presence (complex) or absence (simple) of fever and other systemic symptoms (Table 2).

Given their rarity, complex or severe cADRs are difficult to study systematically and pediatric-specific data are limited. These challenges have led to the adoption of adult paradigms for diagnosis and management in pediatric practice, where better evidence exists.

In SJS and toxic epidermal necrolysis (TEN), severe blistering of the skin and mucous membranes related to either medications or infections occurs. The differential diagnosis in early cases where diffuse erythema is common may include viral exanthems, Kawasaki disease, and acute generalized exanthematous pustulosis. As widespread blistering begins to develop, thermal burns, toxic erythema of chemotherapy, pemphigus of all types, staphylococcal scalded skin syndrome (which spares the mucous membranes), acute graft-versus-host disease, acute syndrome of apoptotic pan-epidermolysis associated with systemic lupus erythematosus, and generalized bullous fixed drug eruption become the primary considerations.

In recent years, significant developments have occurred in the category of severe blistering drug reactions—SJS and TEN—the focus of this review. The watershed evolutions in this area will be reviewed and ultimately support the need for evolution of the diagnostic categories for severe blistering cADRs in children.

Incidence

Until 2017, estimates of the incidence of SJS and TEN were based on retrospective surveys and databases from the 1990s and a European registry^6–9. More recently, larger datasets have been used in attempts to better describe these rare events in pediatrics.

A recent US pediatric database cohort study suggests that SJS-TEN spectrum disorders occur in 7.5 per 100,000 hospitalized children¹⁰; incidences of 6.3 and 0.5 per 100,000 hospitalized children per year for SJS and TEN, respectively, were reported. A second US cross-sectional study that sourced data from the same 2009–2012 time frame found much lower rates, of 5.3 (SJS) and 0.4 (TEN) per million children per year, in the overall population¹¹. Similar findings in both studies suggest that pediatric SJS-TEN results in a substantial health-care burden, although mortality is less when compared with adult data. There are many limitations to using large datasets to examine rare events such as SJS-TEN, including poor validity of International Classification of Diseases 9/10 (ICD-9/10) codes, inability to determine cause, and the lack of a standardized diagnostic approach across institutions. Given that the clinical features of severe cADRs can be challenging to interpret for diagnosis, standardization in case identification and validation is needed.

Clinical features and refining diagnosis: drug versus bug

Etiologies

Recent pediatric retrospective studies identify a drug as the cause of SJS and TEN in 72 to 90% of cases^20–23. Idiopathic cases, where no cause (medication or infection) can be identified, make up 5 to 17% of cases in retrospective reviews^20,21,24. A limited number of medications, including antibiotics, anticonvulsants, and non-steroidal anti-inflammatory drugs, are implicated in the majority of pediatric cases. As stated above, MP is a common trigger of severe mucocutaneous reactions that have been considered SJS-TEN in the past; however, the clinical presentation and outcomes suggest a unique disease process.

Causality and risk reduction

It is important to avoid implicating medications used to treat the prodromal symptoms of SJS and TEN as causative agents¹⁹. Survivors of SJS and TEN are often told to avoid all medications taken just prior to the reaction, limiting treatments that can be used for future illnesses²⁵ and causing patients and families concern about the use of medications in general. Evaluation of possible implicated medications requires utilization of an effective causality tool. Many available ADR causality tools, including the Naranjo and Liverpool tools, are non-specific to the ADR phenotype. The ALDEN is specific for severe cADRs. These tools are helpful when considering the timing of drug exposure to reaction onset, the probability of drug presence in the body, prior exposure to the same drug regardless of reaction at that time, the presence of drug beyond the progression phase, drug notoriety as an ADR cause, and the presence or absence of other etiologies²⁵. Although these tools can be helpful, their usefulness is limited by variability in results between users and testing methods. Further advancements in diagnostics and assessment tools have the potential to enhance causality assessment in the future²⁶.

Testing to identify the causative agent in SJS-TEN remains controversial and is not widely used. In vivo (varied methods of re-exposure of the patient to potential trigger medications) testing includes patch testing or delayed intradermal testing. These tests can be performed once the acute reaction has resolved but within a year. Potentially cross-reacting medications can also be identified with patch testing²⁷. Unfortunately, the results are not reliable for all medications, limiting their clinical utility. Oral re-challenge, recommended for other types of drug reactions, is not recommended for SJS and TEN given the serious risk of a second potentially fatal episode²⁸. In vitro testing with the lymphocyte transformation test (LTT) during the recovery phase, 4 to 8 weeks after the reaction, is controversial as LTT historically has had low sensitivity in SJS-TEN and many false-positive and -negative results²⁹. Recent reports support better sensitivity (86%) and specificity (74%) in SJS-TEN, even for low-risk drugs³⁰. The drug-specific interferon-gamma–releasing cell assay is highly specific (95% specificity and 79% sensitivity for allopurinol) and can be carried out in the acute phase³¹. A recent publication highlighted its advantages over LTT: higher rates of causative drug identification (73.9% versus 52.2% for LTT) and use during the acute phase (versus recovery phase for LTT)³².

Prevention of SJS and TEN is becoming a reality with the identification of risk factors that can be screened prior to drug initiation. Human leukocyte antigen (HLA) screening for ethnicity-specific risk alleles prior to administration of aromatic anticonvulsants (including carbamazepine), allopurinol, and abacavir reduces the risk of SJS-TEN (Table 4)³³. Polymorphisms that reduce drug clearance can also increase the risk of severe cADRs³⁴ When a risk allele is identified in a patient, the medication should be avoided³⁵. As a proof of concept, in Taiwan, where pre-carbamazepine screening for HLA-B*1502 was adopted in 2010, the incidence of SJS-TEN has drastically decreased³⁶. In Thailand, where HLA-B*1502 pre-treatment screening is also routine, HLA test results are printed on a wallet card that patients can carry with them as a “pharmacogenomic ID card” for future health-care interactions³⁵.

To treat or not to treat, that is the question… but we still don’t know the answer

SCORe of Toxic Epidermal Necrolysis (SCORTEN), a TEN severity-of-illness score³⁷, was developed for adults but has been used to predict both morbidity and mortality in pediatric patients^38,39. The SCORTEN has multiple factors, but three of them—age of at least 40 years, malignancy, and heart rate of at least 120 beats per minute—are less important in children because (1) they are young, (2) they have low rates of malignancy, and (3) their normal heart rate can be over 120 beats per minute if they are younger than four. The SCORTEN, or a pediatric-adapted scoring system, could be used prospectively in studies to predict treatment efficacy by comparing actual with predicted mortality and morbidity³⁸. Despite the availability of a validated severity score that helps to predict which patients will have more severe outcomes, the absence of a gold-standard effective therapy for pediatric SJS-TEN means that the decision between supportive care and directed (immunosuppressive) therapy remains challenging.

In the last two years, there has been an explosion of retrospective and database publications on pediatric SJS-TEN which have come to similar conclusions on the relative lack of efficacy of corticosteroids and intravenous immunoglobulin (IVIG) on various outcomes^{20,22,40–43}. For treatment of SJS-TEN, there is a consensus on the importance of rapid identification and withdrawal of the causative medication. As adjuvant therapy, corticosteroids are most frequently used, followed by IVIG both as monotherapy and in combination with corticosteroids. None of these mono- or combination therapies appears to affect time to healing or length of hospital stay. The best evidence on treatment before 2017–2018 was from a large (n = 128 cases) systematic review published in 2011 that suggested that patients who received IVIG and prednisone had better outcomes than those who received supportive care only⁴⁴. Overall mortality rates for SJS-TEN in newer publications range from 0 to 6.6%, and rates of up to 25% were reported for TEN cases^{20,22,40–43}.

Recent case reports and case series document the rapid efficacy of cyclosporine 3 mg/kg per day divided twice a day and infliximab 5 mg/kg intravenously for one dose in both drug- and infection-related cases^45,46. There is emerging evidence from case reports for anti-tumor necrosis factor (anti-TNF) therapy in the acute phase^47–50. Furthermore, a recent Taiwanese randomized clinical trial of etanercept in TEN included children older than 4 years of age and showed that etanercept (25 mg twice weekly <65 kg and 50 mg if >65 kg) decreased the predicted mortality rate and reduced skin healing time compared with corticosteroids (1 to 1.5 mg/kg per day intravenously) in the group as a whole⁵¹. Both groups received treatment until their skin lesions healed⁵¹. This study included five children who were 6 to 13 years old⁵¹. Further large-scale studies are needed to confirm these promising results in the pediatric population.

Ocular complications and sequelae are significant for pediatric patients with SJS-TEN. Aggressive initial management, including adjuvant amniotic membrane transplantation, can reduce complications^52,53. Even in the absence of severe ocular involvement in acute SJS-TEN, children may develop progressive disease of the ocular surface and conjunctival inflammation over time⁵⁴. Delayed lid margin keratinization⁵⁵, vision deterioration, and corneal damage can occur after SJS-TEN⁵⁶, emphasizing the importance of close continual follow-up by ophthalmology^55,57.

The early introduction of psychological and social support is critical for pediatric patients with SJS-TEN to avoid long-term anxiety and depression. Ideally, the introduction should occur as soon as the patient is stable enough for a psychologist or child-life worker to visit. For children, the experience of complete loss of control of their body can be devastating, and early explanation and attention to the patient’s emotional needs are critical. Other chronic sequelae are summarized in Table 5.

How redefining diagnosis of pediatric Stevens–Johnson syndrome – toxic epidermal necrolysis can improve treatment

The literature on the treatment of pediatric SJS and TEN is disappointing as larger studies suggest no impact of any intervention other than drug withdrawal. In contrast, case series and reports document the effectiveness of various anti-inflammatory immunosuppressive treatments. Could it be that the way that past cases are classified results in groups that are too heterogeneous (that is, triggered by infection or drugs or idiopathic) to respond similarly to a given therapy? Considering a shift in diagnostic paradigm as discussed above may be a way to interpret the existing literature with a lens focused from cause/trigger to treatment. Adding to this a precision medicine approach that compares subpopulations of responders with non-responders to identify characteristics that permit early recognition and takes advantage of biomarkers like the granulysin rapid test may lead to a future in which we can select the right treatment for the right patient every time.

Conclusions

The future is bright for pediatric SJS-TEN. Initiatives are under way to improve our understanding of this spectrum of disorders, specifically in children. A British guideline on management was recently published⁶¹. With increasing accessibility and validation of risk factor screening (HLA and metabolism variants) and biomarkers, we may soon be able to prevent SJS-TEN in predisposed individuals and diagnose and treat it early when it occurs by chance.

Abbreviations

ADR, adverse drug reaction; ALDEN, algorithm of drug causality for epidermal necrolysis; BSA, body surface area; cADR, cutaneous adverse drug reaction; DEN, drug-induced epidermal necrolysis; HLA, human leukocyte antigen; ICD, International Classification of Diseases; IVIG, intravenous immunoglobulin; LTT, lymphocyte transformation test; MIRM, Mycoplasma pneumoniae–induced rash and mucositis; MP, Mycoplasma pneumoniae; RIME, reactive infectious mucocutaneous eruption; SCORTEN, SCORe of Toxic Epidermal Necrolysis; SJS, Stevens–Johnson syndrome; TEN, toxic epidermal necrolysis