Lenalidomide‐Dexamethasone as an Effective and Well‐Tolerated Treatment Option for Refractory Rosai‐Dorfman‐Destombes Disease: A Clinical and Pharmacophenomic Study

Le résumé fourni par la source

A combined clinical and computational pharmacophenomic study from the Coastal Rare Inflammatory Diseases Program and Every Cure. Rosai-Dorfman-Destombes (RDD) disease is a rare histiocytic neoplasm that commonly presents with skin and subcutaneous masses, lymphadenopathy, and bone lesions [1]. Management is guided by disease extent, symptom burden, presence of targetable mutations, and treatment response [1, 2]. Although recent studies have highlighted the role of the mitogen-activated protein kinases-extracellular signal-regulated kinases (MAPK–ERK) pathway mutations in histiocytic disorders [3, 4], approximately 60% of patients with RDD lack targetable mutations on next generation sequencing (NGS), underscoring the need for alternative therapeutic strategies. Immunomodulatory drugs (IMiDs), including thalidomide and lenalidomide, exert anti-inflammatory and antineoplastic effects through modulation of cytokine signaling and cereblon-mediated pathways [5, 6]. Given that RDD pathophysiology involves cytokine-driven inflammation—including tumor necrosis factor and interleukin-6—lenalidomide represents a biologically rational therapeutic candidate. The National Cancer Care Network (NCCN) guidelines list lenalidomide as an “other recommended regimen,” and expert consensus supports its use in refractory disease [1, 7]. Clinical activity of IMiDs in RDD has been described in numerous case reports [8-10], and a recent phase II study demonstrated an overall response rate of 87% in 23 patients treated with lenalidomide and dexamethasone (len/dex) [11]. Computational pharmacophenomics is an emerging approach for systemic drug repurposing that evaluates the therapeutic potential of approved agents across diseases and disease subtypes [12]. Given the vast amount of data available: with approximately 4000 FDA-approved drugs and 18 500 recognized human diseases, manual identification alone of the 75 000 000 possible combinations is impossible. Biomedical knowledge graphs enable this approach by integrating diverse biomedical data sources, including drug–disease associations and mechanisms of action, to identify therapies with shared biological or pathophysiologic features [13]. This approach is particularly valuable in rare diseases such as RDD, where conventional large-scale trials are often infeasible. In this study, we conducted a combined clinical and pharmacophenomic analysis of len/dex in RDD. First, we conducted a retrospective review of patients enrolled in the Coastal Rare Inflammatory Diseases (CoRID) program, a Canadian rare disease program directed by one of the senior authors (LC). Ethics approval was obtained from the University of British Columbia Clinical Research Ethics Board. Eleven patients with histopathologically confirmed RDD treated with len/dex were included. Demographic, clinical, radiographic, and outcome data were abstracted from medical records, with a data cutoff of October 31, 2025. Treatment response was assessed by clinical evaluation and/or imaging (PET-CT, CT, or MRI) and categorized as complete response, partial response, stable disease, or progressive disease based on adapted consensus response criteria for histiocytoses [14]. Second, a computational pharmacophenomics approach (manuscript in preparation) was used to evaluate drug repurposing opportunities for RDD. An ensemble machine learning model trained on known drug–disease relationships was applied to a biomedical network integrating the RTX-KG2 and ROBOKOP knowledge graphs [15, 16]. Network nodes were embedded using Node2Vec, and drug and disease embeddings were used to generate prediction scores for drug–disease pairs. Model outputs were normalized to produce a ‘treat score’ (0–1), reflecting the predicted biological plausibility of a drug for RDD, with 0.00 being the lowest and 0.99 being the highest. Results were then re-ranked according to various strategic criteria, including consideration of unmet medical need, existing evidence (both preclinical and clinical), and relevant drug safety information. Eleven patients were included (six males, five females), with a median age at diagnosis of 55 years (range 17–80 years), six of whom were included in previous publications [2, 17-19]. Extranodal disease was present in ten patients (Table I). Tissue NGS was performed in seven cases, identifying one KRAS p.K117N mutation three MAP2K1 mutations across two patients. Five patients received corticosteroids as first-line therapy. Additional systemic agents—including rituximab, sirolimus, cladribine, and methotrexate—and local interventions such as surgery or radiation did not provide clinical response. One patient with a KRAS mutation achieved a partial response to trametinib but discontinued therapy after 6 months due to cutaneous toxicity. All eleven patients received len/dex across lines of therapy ranging from first to fifth (Table 1). Lenalidomide was administered at doses ranging from 10 to 25 mg daily for 21 of every 28 days, in combination with dexamethasone at doses of 8 to 40 mg weekly based on standard doses at our cancer center [https://www.bccancer.bc.ca/chemotherapy-protocols-site/Documents/Lymphoma-Myeloma/MYLDREL_Protocol.pdf], along with low dose aspirin or anticoagulation. The median time on treatment was 18.1 months (range 4.0–59.2 months). Ten patients demonstrated clinical and/or radiographic benefit, and only one experienced progressive disease while on treatment (Figure 1). The median follow-up time from the date of diagnosis to the date of data cut-off was 33.1 months (range 11.0–78.7 months). KRAS p.K117N FUBP1 p.I443V ERBB4 p.V949I Computational pharmacophenomics analysis evaluated 2975 FDA-approved drugs for predicted activity in RDD. Among the highest-ranked candidates were cytotoxic agents and corticosteroids currently used in clinical practice (Figure 2A,C). Dexamethasone ranked 4th overall (treat score 0.9986), and lenalidomide ranked 23rd (0.9967). Dexamethasone, which often accompanies lenalidomide in clinical practice in this disease space, is the highest-rated corticosteroid. When excluding corticosteroids and traditional chemotherapies, lenalidomide emerged as the third-highest ranked agent, exceeding several NCCN “preferred” cytotoxic therapies, including cladribine and cytarabine (Figure 2B). In this series of patients with relapsed or refractory RDD, lenalidomide combined with dexamethasone was associated with consistent and clinically meaningful benefit across molecular subtypes and treatment lines. These findings address a key therapeutic gap in RDD, particularly for patients without actionable MAPK–ERK pathway mutations or those who have exhausted conventional systemic options. While targeted inhibitors have transformed management in mutation-positive disease, their utility is limited by mutation prevalence, access to molecular testing, drug availability, and toxicity. In contrast, lenalidomide and corticosteroids are broadly accessible even in lower-income countries, and do not require genomic stratification. Some cases are illustrative of the responses that can be seen with this regimen. An 80 year old man (case 3) presented with large peri-orbital masses affecting his vision and had no actionable mutations on NGS. He had minimal clinical or radiological response to rituximab, sirolimus, cladribine, and local radiation. Lenalidomide and dexamethasone given in fifth-line resulted in substantial clinical and radiographic improvement (Figure S1C,D). Case 8 had an 8.6 cm right thigh soft tissue lesion causing significant discomfort from femoroacetabular impingement. Although she had a MAP2K1 mutation, medication coverage issues precluded the use of targeted therapy in first line. After 7 months of lenalidomide and dexamethasone, the mass was no longer palpable and symptoms resolved (Figure S1A,B). Lenalidomide and dexamethasone were also generally well tolerated, with only low-grade adverse events observed and toxicities were consistent

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.