Stevens-Johnson syndrome precipitating days after treatment for dostarlimab-induced autoimmune encephalitis: a case report

Abstract

Background

Immune checkpoint inhibitors (ICIs) are increasingly used in the first-line and maintenance treatment of advanced and recurrent endometrial cancer. Their expanding use has been accompanied by a rising incidence of immune-related adverse events (irAEs), which vary widely in presentation and severity. Severe neurologic and cutaneous irAEs are rare but can be life-threatening.

Case presentation

A 69-year-old woman with FIGO stage IVB, mismatch repair-deficient endometrioid endometrial cancer receiving maintenance dostarlimab presented with altered mental status. When her encephalopathy persisted despite treatment of an identified infection and metabolic derangements and improved only with corticosteroids, she was diagnosed with presumed autoimmune encephalitis. Days later, approximately one to two weeks after receiving beta-lactam antibiotics and about five weeks after her last dostarlimab dose, she developed biopsy-confirmed Stevens-Johnson syndrome and toxic epidermal necrolysis (SJS/TEN), ultimately involving roughly 30% of her body surface area and requiring burn-center care. She recovered fully with no residual deficits.

Conclusion

We propose that checkpoint inhibition, evidenced by the preceding neurologic irAE, primed the patient for a severe hypersensitivity reaction to a subsequently administered, ordinarily tolerated medication. This case underscores the importance of heightened vigilance for severe reactions to common drug exposures in patients with a recent irAE, even after immunotherapy has ended.

Highlights

• •Dostarlimab-linked autoimmune encephalitis preceded life-threatening SJS/TEN in one patient.
• •Checkpoint blockade may prime severe reactions to ordinarily tolerated drugs.
• •SJS/TEN followed a beta-lactam given 5 weeks after the last dostarlimab dose.
• •Vigilance for hypersensitivity is warranted even after ICI therapy has ended.

1. Introduction

Recent advances in the treatment of primary advanced and recurrent endometrial cancer have established a role for immune checkpoint inhibitors (ICI) in first-line therapy. Dostarlimab, in particular, is now used in both the first-line and maintenance settings (Mirza et al., 2023). With ICI, however, clinicians face the challenge of a broad immune response that frequently manifests as immune-related adverse events (irAEs). An estimated 30 to 60% of patients experience an irAE, most often involving the skin, gastrointestinal tract, endocrine glands, and liver (Gao et al., 2022).

Dermatologic events are among the most common ICI toxicities, affecting 34% to 42% of patients on anti-PD-1 monotherapy (Quach et al., 2021). Most are mild and self-limiting, such as morbilliform or lichenoid eruptions, while severe cutaneous adverse reactions including Stevens-Johnson syndrome and toxic epidermal necrolysis (SJS/TEN) are rare but life-threatening (Quach et al., 2021). Milder eruptions are usually managed with topical steroids without interrupting immunotherapy, whereas severe reactions require systemic corticosteroids, supportive care, and withdrawal of the checkpoint inhibitor along with any other possible culprit drug (Quach et al., 2021).

Neurologic irAEs are far less common but can be severe. Across clinical trials, neurologic events of any severity occurred in 3.8% to 12.0% of patients depending on the regimen, and most were mild to moderate and nonspecific, with headache the most frequently reported, followed by encephalopathy and meningitis (Cuzzubbo et al., 2017). Serious events are considerably rarer, with reported estimates of 0.2% to 0.8%, and include encephalitis, meningitis, meningoencephalitis, myasthenia, myositis, and vasculitis (Buckley et al., 2025). The median time from initiation of therapy to immune-related encephalitis is approximately 2 months, although onset more than a year after the first dose has been reported, and neural antibodies are found in roughly half of those tested (Buckley et al., 2025). Management generally involves discontinuing immunotherapy, a neurologic work-up, and immunosuppression such as corticosteroids, intravenous immunoglobulin, or rituximab (Buckley et al., 2025).

We present a rare case of autoimmune encephalitis preceding Stevens-Johnson syndrome (SJS) and toxic epidermal necrolysis (TEN) in a patient receiving checkpoint inhibitor therapy. We propose that the encephalitis reflected a primed, overactive immune state that predisposed her to a severe skin reaction after she received a beta-lactam antibiotic she would ordinarily have tolerated.

2. Case presentation

The patient was a 69-year-old woman with FIGO stage IVB, mismatch repair-deficient (microsatellite instability-high [MSI-H]) endometrioid endometrial cancer. Her cancer was initially treated with neoadjuvant carboplatin and paclitaxel, with cisplatin substituted for carboplatin at the second cycle owing to the national carboplatin shortage. She continued this regimen until she developed a mild paclitaxel hypersensitivity reaction and did not tolerate rechallenge, prompting a switch to nab-paclitaxel for the fourth cycle. After 4 neoadjuvant cycles she underwent total abdominal hysterectomy, bilateral salpingo-oophorectomy, omentectomy, ileocecectomy, and tumor debulking to no gross residual disease. She then received 3 cycles of adjuvant carboplatin and nab-paclitaxel with dostarlimab added through cycle 7, followed by maintenance dostarlimab every 6 weeks; she had completed 3 maintenance cycles with no evidence of disease, her most recent dose approximately 5 weeks before presentation. She otherwise tolerated her antineoplastic therapy without adverse reaction, and her only other documented drug allergy was a mild rash attributed to enoxaparin.

She presented with acute altered mental status, including hallucinations. Initial work-up showed significant electrolyte abnormalities, including hypomagnesemia and hypercalcemia, which were subsequently corrected, and a Klebsiella urinary tract infection treated with ceftriaxone. Her mentation initially improved until hospital day (HD6), when she developed a new fever and worsening agitation. A chest radiograph showing new mild pulmonary edema, right hilar fullness, and possible developing consolidation raised concern for aspiration pneumonia, and coverage was broadened from ceftriaxone to cefepime. Cefepime was discontinued the following day after neurology raised concern that it could be contributing to her encephalopathy, and therapy was changed to ciprofloxacin and metronidazole. After a witnessed aspiration event, piperacillin-tazobactam was started, likely for broader empiric coverage (Fig. 1).

Fig. 1.

Schematic timeline of the inpatient course showing medication exposures relative to the two sequential immune-related adverse events, autoimmune encephalitis and SJS/TEN, separated by rash onset.

Despite treatment of the infectious and metabolic contributors, her mentation did not improve and her hallucinations persisted. Neurology was consulted, and empiric intravenous methylprednisolone (approximately 1 mg/kg, 62.5 mg daily) was started on HD7 for presumed autoimmune encephalitis while antimicrobial therapy continued. The diagnostic work-up included a non-contrast brain MRI (HD10) showing only nonspecific chronic microvascular and involutional changes, without metastatic disease or an acute intracranial process, and an electroencephalogram (HD11) showing severe diffuse encephalopathy without epileptiform discharges or seizures. Lumbar puncture was deferred by neurology given the waxing and waning nature of her symptoms. A serum anti-GAD65 antibody later returned low-positive at 0.15 nmol/L, a level the reference laboratory noted carries low positive predictive value for neurological autoimmunity (in contrast to values ≥20.0 nmol/L). No other neural autoantibodies were tested. The treated infection and corrected metabolic derangements did not account for her persistent symptoms, which improved only after corticosteroids were started. Taken together, these findings supported a working diagnosis of possible autoimmune encephalitis (CTCAE version 5.0 grade 3) (National Cancer Institute, 2017). Her agitation and hallucinations continued to improve over the next several days, and on HD13 she was transitioned to oral prednisone 80 mg daily (1 mg/kg) with a planned outpatient taper of 1 mg/kg per week.

She developed a pruritic morbilliform rash on the chest and arms around HD16. Dermatology assessed it as a morbilliform drug reaction and favored the recently administered beta-lactam antibiotics as the most likely cause. Over the next few days, she developed oral mucositis followed by progressive skin involvement, with flaccid bullae, skin denudation, and a positive Nikolsky sign over the back, flanks, buttocks, inner thighs, and perineum. She also had hemorrhagic crusting of the lips and denuded areas of the tongue and oral mucosa, while ocular involvement was limited to the eyelid margins and spared the conjunctivae (Fig. 2). Because of the worsening mucocutaneous involvement and concern for a severe cutaneous adverse reaction, a shave biopsy of the right hip was performed on HD23. It showed acute epidermal necrosis consistent with SJS/TEN (Fig. 3), and direct immunofluorescence was negative, which argued against an autoimmune blistering disorder. On the day before transfer, the dermatology team estimated body surface area involvement at approximately 20% and calculated a SCORTEN of 3, corresponding to an estimated mortality of about 35% (Bastuji-Garin et al., 2000). By the time she was transferred to a specialized burn center the next day (HD24), involvement had progressed to roughly 30% of the body surface area, consistent with TEN.

Fig. 2.

De-identified clinical photographs demonstrating the mucocutaneous involvement (denuding bullae and mucosal erosions) consistent with SJS/TEN. (A) Hemorrhagic crusting of the lips with denuded oral mucosa. (B) Denudation and hemorrhagic crusting with erosions over the trunk.

Fig. 3.

Skin biopsy of the right hip (hematoxylin and eosin) demonstrating acute epidermal necrosis consistent with SJS/TEN; direct immunofluorescence was negative. (A) Detached epidermis with blister formation and frequent dyskeratotic keratinocytes (H&E, original magnification ×100). (B) Higher-power view demonstrating epidermal necrosis and epidermal detachment (H&E, original magnification ×400).

At the burn center she was treated with anti-infectives, atovaquone, and silver dressings, with support from nutrition, physical therapy, and occupational therapy. After 45 days she was discharged to a skilled nursing facility. Her steroid taper was completed, her rash resolved fully, and she had no lasting neurologic or cutaneous deficits. Dostarlimab was permanently discontinued and not resumed. Imaging in January 2025 showed no evidence of disease; she subsequently developed recurrent metastatic endometrial cancer and died in May 2025 from complications of disease progression, unrelated to the immune-related toxicities described here.

3. Discussion

The notable feature of this case is the occurrence of two separate, severe immune-mediated toxicities in one patient, a neurologic irAE that required systemic corticosteroids, followed days later by a life-threatening skin reaction. We propose that these events were linked, and that the broad immune activation from dostarlimab that produced the encephalitis left her in a primed state in which an ordinarily tolerated antibiotic provoked a far more severe reaction than expected.

The neurologic event was most consistent with autoimmune encephalitis, which varies in presentation but often includes seizures, subacute memory loss, mood or behavioral changes, and movement disorders. Supporting findings can include elevated cerebrospinal fluid protein or pleocytosis, an abnormal brain MRI, or neural autoantibodies. These may target cell surface or synaptic proteins such as NMDAR, LGI1, CASPR2, AMPAR, and the GABA receptors, or intracellular antigens such as GAD65 (Graus et al., 2016). Intracellular targets are more common in immune-related encephalitis, though low titers such as our patient's may be nonspecific (Buckley et al., 2025). The consensus criteria for possible autoimmune encephalitis proposed by Graus and colleagues permit a working diagnosis based on subacute onset of altered mental status, at least one supportive ancillary finding, and reasonable exclusion of alternative causes, even when autoantibody testing is inconclusive (Graus et al., 2016). Her presentation, the persistence of symptoms despite treatment of competing causes, and her response to immunosuppression fit that framework, although she did not meet the ancillary criterion, since imaging was nonspecific, cerebrospinal fluid was not obtained, and the electroencephalogram showed diffuse slowing without epileptiform activity. Outcomes are generally favorable, with partial or complete recovery in about 80% and relapse in roughly 10%, occurring during corticosteroid taper or on re-exposure to ICI (Buckley et al., 2025).

The skin reaction, by contrast, pointed away from dostarlimab. Her most recent dose had been given about 5 weeks before admission, whereas the rash appeared around HD16, one to two weeks after she began the beta-lactams ceftriaxone (HD2 to HD6), cefepime (HD7 to HD8), and piperacillin-tazobactam (HD9 to HD12). Several other drugs were administered concurrently, including ciprofloxacin (HD8 to HD9), olanzapine (HD2 to HD16), and pantoprazole (HD1 to HD18; Fig. 1), and attribution can be difficult when a patient on ICI is receiving several medications (Quach et al., 2021). Dermatology considered the one-to-two-week interval typical for a beta-lactam reaction and favored those agents as the most likely cause.

The mechanism linking these events is the priming of the immune system by checkpoint inhibition. Checkpoint inhibitors such as dostarlimab are monoclonal antibodies that block the co-inhibitory signals which normally restrain T cell activation and preserve tolerance to healthy tissue (Gao et al., 2022). Releasing this brake increases antitumor activity but also primes the immune system to react against a wider range of antigens and can induce hypersensitivity reactions (Hammond et al., 2022). Host and tumor factors may add to this risk. Mismatch repair-deficient, MSI-high tumors like our patient's carry a high mutational and neoantigen burden (Howitt et al., 2015), and this has been linked to a greater risk of irAEs, offering a plausible reason for her heightened susceptibility (Kerepesi et al., 2024).

This framework helps explain the severity of the reaction. SJS/TEN during checkpoint inhibitor therapy is rare, and a systematic review of 47 published reports identified only 50 patients, with a median latency of 23 days from the first ICI dose to symptom onset (Zhou et al., 2024). Our patient had tolerated 7 months of immunotherapy, which makes direct causation by dostarlimab unlikely. Even so, PD-1 receptor occupancy on circulating T cells remains above 70% for at least 2 months after a dose, so she likely remained in a primed, checkpoint-inhibited state when the antibiotics were given (Brahmer et al., 2010). The same sequence has been demonstrated directly. A patient who had tolerated iodinated contrast on several prior occasions developed Stevens-Johnson syndrome once atezolizumab was introduced, with causality confirmed by positive rechallenge and T-cell testing (Hammond et al., 2021). Taken together, these observations suggest that a prior severe irAE can raise the risk of a more severe reaction to another agent.

Sequential irAEs affecting different organ systems have been reported before. In one case, a patient with metastatic melanoma developed atopic dermatitis, vitiligo, autoimmune nephritis, autoimmune hepatitis, and autoimmune encephalitis after pembrolizumab, with the encephalitis appearing after therapy had been stopped (Shah et al., 2019). Our case adds two temporally distinct, severe toxicities and highlights the need for ongoing vigilance even after checkpoint inhibitor therapy has ended.

This case has limitations. Lumbar puncture was deferred given the intermittent nature of her symptoms, and the low-positive anti-GAD65 titer, the absence of a full neural autoantibody panel, and the nonspecific imaging further limit diagnostic certainty. Attribution of the skin reaction to a beta-lactam rests on timing and clinical assessment alone, without rechallenge or in vitro testing, so contributions from another agent or from her prior checkpoint inhibitor exposure cannot be excluded.

4. Conclusion

This case illustrates two sequential, severe immune-related toxicities in a single patient, in which a preceding neurologic irAE may have primed the immune system for a life-threatening cutaneous reaction to an ordinarily tolerated antibiotic. Clinicians should maintain a high index of suspicion for autoimmune encephalitis when mental status does not improve despite treatment of common reversible causes, and patients who have experienced one immune-related adverse event should be monitored for further reactions, including hypersensitivity to exposures unrelated to the checkpoint inhibitor itself.

Preprint

A preprint of this work has previously been posted (doi:https://doi.org/10.2139/ssrn.5228126). The manuscript has not been published elsewhere and is not under consideration by another journal.

CRediT authorship contribution statement

Morgan Kuchar: Writing – review & editing, Visualization, Project administration, Investigation. Akash Shah: Writing – review & editing, Investigation, Conceptualization. Rayyan Khan: Writing – original draft, Investigation. Taylor Pigg: Writing – review & editing. Ashlyn Lipnicky: Writing – review & editing. Brandol Wolfenbarger: Resources. Matthew Powell: Resources. Marian Symmes Johnson: Writing – review & editing, Supervision, Conceptualization.

Consent for publication

Written informed consent for publication of the clinical details and accompanying de-identified images was obtained from the patient. A copy of the written consent is retained by the authors and is available for review by the Editor-in-Chief on request.

Ethics approval and consent to participate

This report describes the care of a single patient and does not constitute human-subjects research; formal institutional review board approval was therefore not required.

Ethics declaration

Written informed consent to take part in the study and to publish the article has been obtained from all participants or their legal representatives. The privacy rights of participants have been observed.

This study was performed in compliance with relevant laws, regulatory frameworks and guidelines where the research took place. Ethics committee approval was not required under relevant laws and institutional guidelines. This report describes the care of a single patient and does not constitute human-subjects research; formal institutional review board approval was therefore not required. Written informed consent for publication of the clinical details and de-identified images was obtained from the patient.

This research follows the CARE guidelines and the CARE checklist.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

Supplementary material

Data availability

No datasets were generated or analyzed for this case report. All information supporting the findings is contained within the article.