Severe Cutaneous Drug Reactions: SJS/TEN and DRESS

A spectrum, not a single disease

Drug eruptions run from trivial to catastrophic, and telling them apart is the whole skill. At the mild end sits the maculopapular (morbilliform) exanthem — the ordinary "drug rash": symmetrical pink-to-red macules and papules over the trunk, itchy, appearing days into a course, and settling once the drug stops. It is a nuisance, not a threat. But certain features are alarms that a benign rash has become one of the severe cutaneous adverse reactions (SCARs): pain rather than itch, dusky or blistering skin, peeling, involvement of the lips, eyes or genitals, facial swelling, fever, and enlarged lymph nodes. Three named emergencies live at that severe end — Stevens–Johnson syndrome / toxic epidermal necrolysis (SJS/TEN), DRESS, and acute generalised exanthematous pustulosis (AGEP). Recognising the shift from ordinary rash to SCAR is the moment that saves a life, and it links directly to the Principles of Pharmacology chapter on adverse drug reactions, where these figure as the archetypal type B (unpredictable, immune-mediated) hypersensitivity reactions.

SJS/TEN: a burn from within

Stevens–Johnson syndrome and toxic epidermal necrolysis are two ends of one disease, separated only by how much skin detaches: under 10% of the body surface is SJS, over 30% is TEN, and the band between is the overlap. Mechanistically it is a T-cell–mediated attack in which cytotoxic (CD8) T cells and natural killer cells, activated by the drug, order the keratinocytes of the epidermis to die. They do it through the death-signal molecules granulysin, perforin and granzyme, and the Fas/Fas-ligand pathway — the same apoptotic machinery covered in the Immunology section. The result is confluent keratinocyte apoptosis: the epidermis dies as a sheet and lifts off the dermis beneath. Clinically the skin is painful and dusky, blisters coalesce, and a light sideways pressure shears the top layer away — a positive Nikolsky sign. Because the mucous membranes are epithelium too, they detach as well: the mouth, the eyes (risking blindness), the airway and the urogenital tract. The patient effectively suffers a second-degree burn generated from within, and dies — when they die — of the same things a burns patient does: fluid loss, infection, and failure of the skin's barrier.

The classic culprits

A short list of drugs accounts for most SJS/TEN — knowing them is half the diagnosis. The offenders cluster into recognisable groups. Allopurinol is now one of the commonest triggers worldwide. The aromatic anticonvulsants — carbamazepine, lamotrigine, phenytoin, and phenobarbital — are a classic cluster, which ties this topic to the CNS/epilepsy chapter where their monitoring is taught. Sulfonamide antibiotics (co-trimoxazole) and other sulfa drugs are long-standing culprits, linking to the Antimicrobials section. Nevirapine, an older antiretroviral, is notorious. NSAIDs, especially the oxicam class (piroxicam, meloxicam), round out the list. The reaction characteristically appears within the first several weeks of a new drug — usually one to three weeks in — because the immune system needs time to sensitise to it. That latency is a fingerprint: a patient reacting on day 2 of a drug they have taken for years is unlikely to have SJS from it, whereas the newly started anticonvulsant three weeks ago is a prime suspect.

Pharmacogenetics: screening before you prescribe

One of the most elegant advances in all of pharmacology grew directly out of SJS/TEN: the ability to predict who will react, from their genes, before a single dose. The immune attack begins when the drug (or its metabolite) is presented to T cells on a particular HLA molecule — the body's antigen-display platform — and certain HLA types present certain drugs in a way that provokes a violent response. Two associations are strong enough to be used clinically. HLA-B*15:02 carries a markedly raised risk of carbamazepine-induced SJS/TEN, concentrated in Han Chinese and other Southeast Asian populations; guidelines now recommend genotyping before starting carbamazepine in those groups, and choosing another drug if positive. HLA-B*58:01 predicts allopurinol-induced SCARs and is tested for in populations where it is common (Han Chinese, Thai, Korean). This is the flagship real-world example of the pharmacogenetics and HLA screening introduced in the Principles of Pharmacology chapter — a case where reading the patient's DNA turns an unpredictable idiosyncratic reaction into a preventable one.

Managing SJS/TEN: stop, support, modulate

The single most effective intervention costs nothing and comes first: stop the drug. Management has three layers. First and above all, identify and immediately withdraw the culprit drug — every earlier day of withdrawal improves survival, and continuing it is the single worst thing you can do. Where several drugs are possible suspects, timing and the known-culprit list guide which to stop; when in doubt, stop them all. Second, meticulous supportive care, ideally in a burns unit or intensive care: fluid and electrolyte replacement for the losses through denuded skin, temperature control, scrupulous wound and skin care to prevent infection, nutritional support, urgent ophthalmology input to save the eyes, and attention to the mouth and urogenital mucosa. Third, immunomodulation to switch off the immune assault — though this layer is the most debated. Ciclosporin, a calcineurin inhibitor that suppresses T-cell activation, has the best contemporary evidence and is widely favoured. Intravenous immunoglobulin (IVIG) is used, sometimes with ciclosporin. Systemic corticosteroids and the TNF-inhibitor etanercept are used in some centres but remain contested. Prognosis is estimated with SCORTEN, a validated score built from variables such as age, heart rate, malignancy, body-surface detachment, urea, glucose and bicarbonate, which predicts mortality and guides the intensity of care.

DRESS: the slow, multi-organ reaction

DRESS — drug reaction with eosinophilia and systemic symptoms — is the other great SCAR, and it behaves completely differently from SJS/TEN. It is slow: the latency is long, typically two to eight weeks after starting the drug, far later than an ordinary rash, which is a frequent trap because clinicians forget a drug begun two months ago. And it is systemic: the skin rash (often a widespread exanthem with striking facial oedema) is only the surface of a multi-organ inflammatory illness. The defining tetrad is rash, fever, eosinophilia (a surge of eosinophils in the blood), and internal organ involvement — most characteristically hepatitis (the liver is the commonest and most dangerous organ hit), but also interstitial nephritis, pneumonitis, myocarditis, and lymphadenopathy with a reactive, sometimes atypical-lymphocyte, blood picture. A curious and important feature is its link to herpesvirus reactivation, especially HHV-6: the drug-driven immune storm reawakens latent virus, which in turn drives the prolonged, relapsing course that makes DRESS so treacherous. The culprit list overlaps with SJS/TEN but has its own flavour: allopurinol, the aromatic anticonvulsants, sulfonamides, vancomycin, and minocycline are the classics.

DRESS is managed like an autoimmune flare, not a burn. The principles echo SJS/TEN but the emphasis shifts. Stop the culprit drug immediately — the same first commandment. Then, because the disease is an internal inflammatory storm rather than a skin barrier failure, the mainstay is systemic corticosteroids, tapered slowly over weeks to months, alongside close organ monitoring: serial liver and kidney function, blood counts to track the eosinophilia, and an eye on the heart and thyroid, since a late autoimmune thyroiditis can appear weeks after apparent recovery. The slow taper matters because DRESS is notorious for relapsing if steroids are withdrawn too quickly — the underlying immune activation, amplified by viral reactivation, outlasts the initial treatment. This is the same corticosteroid and immunosuppressant logic taught in the Inflammation section, applied here to a reaction that can smoulder for months.

The universal rules of drug allergy

Whichever reaction you are facing, four rules apply to all of them, and they are what the exam — and the coroner — care about most. Identify and stop the culprit: this is the treatment, not merely a precaution. Document the allergy clearly and prominently in the record, naming the drug and the reaction, so no one prescribes it again by accident. Avoid cross-reactors: sensitivity to one aromatic anticonvulsant (carbamazepine) predicts risk with the others (phenytoin, phenobarbital); a sulfonamide-antibiotic SCAR warns against related sulfa drugs; and where one HLA-linked drug caused the reaction, its HLA-sharing relatives are suspect. And above all, never deliberately re-challenge a patient who has had a severe cutaneous reaction: a memory T-cell response is primed, so the second exposure is faster and more ferocious than the first — a re-challenge can be lethal. There is no allergy test that safely licenses giving the drug back; the reaction itself is the verdict.

A modern cousin lives in the Oncology chapter. It is worth contrasting these classic drug eruptions with a newer class of immune skin toxicity: the cutaneous immune-related adverse events (irAEs) of the immune-checkpoint inhibitors — pembrolizumab, nivolumab, ipilimumab and their kin. There, the skin (rash, itch, vitiligo, and occasionally a genuine SJS/TEN-like eruption) is attacked not because the drug became an antigen, but because the drug deliberately released the brakes on the immune system, letting it turn on normal tissue. The clinical parallel is striking: both are T-cell–driven, and both severe cases are managed by holding the drug and giving systemic corticosteroids to call off the immune assault — exactly the logic detailed in the Oncology checkpoint-inhibitor chapter. The mechanisms differ, but the therapeutic reflex — stop the trigger, suppress the immune attack — is the same.