Immune checkpoint inhibitors (ICIs) are monoclonal antibodies used to treat many solid tumors. In targeting immune regulatory pathways, ICIs can inadvertently trigger a spectrum of immune-related adverse events (irAEs). Rheumatological irAEs are reported in 5%–22% of patients, and common arthralgias and arthritis are generally manageable (1). However, rare rheumatologic irAEs can pose significant diagnostic and therapeutic challenges. One rare entity is eosinophilic fasciitis (EF), an autoimmune fibrotic scleroderma mimic. EF is characterized by erythema and edema of the extremities and trunk, followed by diffuse collagenous thickening of the subcutaneous fat and fascia (2, 3). Physical exam findings of EF include peau d’orange skin typically sparing the hands and feet with a “woody induration” and a “groove sign”, while laboratory tests may reveal hypergammaglobulinemia and peripheral eosinophilia (2, 4). There are no widely accepted criteria for diagnosis, but proposed classification criteria include the Pinal-Fernandez and Jinnin criteria (4, 5). Management generally involves holding the inciting ICI and utilizing immunosuppressive agents such as corticosteroids, immunosuppressives, biologics, or a combination (6).

Several case reports have documented the association of ICI-induced EF with the use of pembrolizumab, nivolumab, atezolizumab, and avelumab weeks to several months after initiation of therapy (7, 8). In all these previous reports, patients were actively undergoing treatment when they developed EF. Herein, we present two cases of delayed EF after discontinuation of ICI therapy: one case at 1 month following cessation of avelumab for treatment of metastatic bladder cancer and the other case at 9 months after ceasing nivolumab for treatment of locally advanced melanoma.

Although several case reports have previously documented the occurrence of ICI-induced EF, we report two unique presentations of delayed EF, defined as onset of EF after ICI cessation. Case 1 occurred 1 month following discontinuation of avelumab after 32 months of well-tolerated administration, and Case 2 occurred 9 months after completing 1 year of adjuvant nivolumab. On review, no other inciting events were identified with rheumatology for both patients prior to their initial dermatologic presentation including inciting medications (no PPIs prior to prednisone or statins), environmental exposures (chemical solvents and pesticides), or recent infections (viral, bacterial, or parasitic). Of note, the patient in case 2 did have a paternal uncle with a history of systemic lupus erythematosus, but there was no other personal or family history of rheumatologic disorders or cancers in both patients. These presentations highlight the need to remain vigilant for irAEs even following discontinuation of ICI therapy and are distinct from most ICI-induced EF described during active ICI therapy (7–10). Our two cases met the most widely accepted diagnostic framework for EF, the Pinal-Fernandez criteria, which require either two major criteria or one major plus two minor criteria, while excluding systemic sclerosis (4). Case 1 fulfilled both major and two minor criteria (groove sign and muscle weakness), while Case 2 fulfilled both major and three minor criteria [peripheral eosinophilia (1.1 × 10⁹/L), groove sign/peau d’orange, and MRI evidence of fascial involvement].

Pharmacovigilance data have shown a link between ICIs and EF. One French analysis identified ICIs as the most commonly implicated drug class in EF with an overall reporting odds ratio (ROR) of 75.3 [95% confidence interval (CI): 46.9–120.8], with nivolumab demonstrating the highest association (ROR: 104.3; 95% CI: 61.0–175.0), followed by pembrolizumab and ipilimumab (39.8; 95% CI: 14.6–108.2 and 35.1; 95% CI: 11.1–110.8, respectively) (8). To our knowledge, only one prior case has documented EF post-immunotherapy discontinuation occurring 1 month after pembrolizumab cessation for metastatic melanoma (11). Both of our presented cases of EF had ICI therapy for ≥12 months, and neither experienced irAEs during active treatment. Prior reports show a mean EF onset of 376 days (12.4 months) after ICI initiation (6, 7, 9), although the only case reported with avelumab occurred 1.5 months into treatment (10).

Our EF cases demonstrate an important emerging concept of delayed immune-related events (DIREs), irAEs that occur ≥90 days after ICI cessation. Couey et al. described 23 DIRE cases with a median latency of 6 months post-ICI discontinuation (12). Such events may be underrecognized, as adverse event reporting windows in clinical trials frequently terminate within 90 days of final treatment. A retrospective study found that 15% of patients continuing ICI beyond 12 months developed delayed irAEs, and 54% of these were off therapy at the time of event onset (13). The underlying mechanism of DIREs remains incompletely elucidated. One postulated etiology is loss of immune tolerance. ICIs impair regulatory T-cell function and other immunologic checkpoints, potentially leading to dysregulated immune activation upon a secondary insult such as infection, trauma, or metabolic stress (14). In Case 1, the development of EF shortly after a subdural hematoma suggests that central nervous system trauma may have served as a precipitating event in a patient with a previously ICI-primed immune system. Another proposed mechanism is epitope spreading, a process whereby immune-mediated tissue injury releases novel self-antigens, leading to the activation of autoreactive T or B cells against previously tolerated epitopes (15). This mechanism may help explain the temporal dissociation between ICI exposure and EF onset, as even subclinical inflammation during therapy may subsequently escalate due to persistent autoreactivity, even in the absence of the drug.

In evaluating patients with suspected ICI-induced EF, differential diagnoses include systemic sclerosis, morphea (both linear and generalized), and scleromyxedema. Neither patient exhibited Raynaud’s phenomenon, digital ulcers, or systemic involvement, and serologic markers, including ENA, were negative or not contributory. Overall, these findings reduced the likelihood of systemic sclerosis. EF can also occur as a paraneoplastic syndrome in solid or hematologic malignancy, and a retrospective review conducted at the Mayo Clinic reported a 13.4% risk of associated malignancy among patients with EF, of which a proportion were classified as paraneoplastic EF (16). In both our case reports, there were no clinical or radiographic evidence of cancer recurrence at EF onset, making paraneoplastic EF unlikely.

Management of EF is complex due to its rarity and lack of consensus guidelines. First-line therapy involves glucocorticoids, often requiring high initial doses and a prolonged taper. Second-line agents include MTX, MMF, cyclosporine, IVIG, and biologic agents such as rituximab or infliximab (6, 9, 17, 18). A cohort study of 63 patients with EF reported a 64% complete response rate with MTX and prednisone combination therapy, and a mean steroid duration of 18 months (17). A review of 15 ICI-induced EF cases revealed that 90% were treated with systemic immunosuppression following ICI discontinuation, most commonly prednisone (40–80 mg/day), MMF (3 g/day), methylprednisolone (ranging from 24 mg/day to 1 g/day), MTX (15–20 mg/week), or infliximab (3 mg/kg) (9). A separate study of 32 patients reported a mean prednisone dose of 52 mg/day and better outcomes in those receiving pulse methylprednisolone (500–1,000 mg/day for 3 days) (18).

In our cases, both patients required high-dose prednisone (50–80 mg/day) and prolonged immunosuppression. Both experienced flares following early tapers, requiring escalation. Case 1 was transitioned from MTX to MMF due to thrombocytopenia and required monthly IVIG (2 g/kg over 2 days) and rituximab for progressive disease. Case 2 was treated with MTX 25 mg/week and a short course of IVIG (administered over 5 days), ultimately achieving complete remission over 33 months. Both patients required pulse steroids due to symptom progression. Steroid duration totaled 10 months (Case 1) and 33 months (Case 2) (Figures 2, 4), well above the previously reported mean duration of therapy of 18 months for complete remission in other cases of ICI-induced EF (17). These cases illustrate the importance of slow tapering, close monitoring, early use of steroid-sparing agents, and early subspecialist involvement. Importantly, the initial steroid tapers were likely too rapid in both cases, contributing to symptom recurrence and a more prolonged disease course.

Limitations of these reports include the absence of full serologic evaluation in Case 1, which, while not required for diagnosis, may have further substantiated the autoimmune etiology. Nonetheless, both patients were promptly evaluated, biopsied, and treated, and the diagnoses were confirmed using current criteria. Another limitation was the speed at which the first steroid tappers were conducted, which ultimately led to symptom recurrence, leading to a prolonged treatment course for both patients. This suggests that a slower steroid taper, early subspecialist involvement, and initiation of steroid sparing agents may be beneficial in improving clinical outcomes.

Recognizing and screening for delayed onset, post-treatment irAEs such as EF is paramount as we continue to expand the scope of ICI treatments for cancer. EF should be considered in patients with current or prior ICI exposure who present with myalgias, arthralgias, extremity edema, and cutaneous thickening, particularly if associated with peau d’orange, groove sign, and sparing of the hands and feet. Diagnostic workup should include eosinophil count, serologies (including ANA and ENA), MRI of involved areas, full-thickness skin biopsy, and consideration of the Pinal-Fernandez criteria. Early recognition of EF, subspecialist involvement, and aggressive treatment with immunosuppression including steroid-sparing agents with slow subsequent taper are important to optimizing outcomes.