Idiopathic multicentric Castleman - TAFRO syndrome can mimic sepsis and myeloid neoplasia

Castleman disease, first described in the 1950s, is a heterogeneous lymphoproliferative disorder clinically categorized as unicentric or multicentric.¹ Contemporary clinically oriented frameworks emphasize that management and prognosis are guided primarily by clinical context and recognize additional categories such as oligocentric Castleman disease and asymptomatic multicentric Castleman disease (MCD), while subdividing MCD into human herpesvirus-8 (HHV-8/KSHV)-associated MCD, POEMS-associated MCD, and idiopathic MCD (iMCD), with clinical phenotypes including iMCD-TAFRO, iMCD-idiopathic plasmacytic lymph-adenopathy, and iMCD-not otherwise specified (NOS).² The main variants of Castleman disease have been incorporated into the most recent World Health Organization (WHO)-classification of lymphoid neoplasms^.3 TAFRO syndrome refers to the constellation of thrombocytopenia, anasarca, fever/marked inflammation, reticulin fibrosis/ renal dysfunction, and organomegaly and can occur in association with iMCD (iMCD-TAFRO), but importantly it may also present without lymphadenopathy, making a diagnosis of Castleman disease impossible by definition and often limiting access to diagnostic lymph node histopathology.⁴ Clinically, TAFRO frequently presents as a fulminant, sepsis-like hyperinflammatory syndrome and may overlap with hematologic malignancy-associated findings, creating substantial diagnostic pressure toward infectious or myeloid neoplastic etiologies.^5,6 In the case presented here, generalized lymphadenopathy enabled biopsy support for iMCD-TAFRO, yet the presentation mimicked fulminant sepsis with multiorgan failure and marrow findings suggestive of a myeloid neoplasm – illustrating key pitfalls and the need for a multidisciplinary, clinicopathological approach to timely immunomodulatory therapy.

Because of its hyperinflammatory presentation, iMCD-TAFRO can closely resemble sepsis, and infections may also complicate the course under treatment. Here, we highlight two underappreciated pitfalls: procalcitonin can exceed 100 ng/mL in iMCD-TAFRO, and bone marrow findings may strongly mimic a myeloid neoplasm, yet both may rapidly resolve with interleukin (IL)-6-directed therapy.

A previously healthy 20-year-old woman of Ghanaian origin was transferred to our Intensive Care Unit with high fever, abdominal pain, severe thrombocytopenia, acute kidney injury with a tubular proteinuria, and a markedly inflammatory laboratory profile (Table 1). A sepsis-like condition rapidly developed, with a Sepsis-related Organ Failure Assessment (SOFA) score of up to 16 points, requiring invasive ventilation, dialysis, and high-dose catecholamine. A comprehensive infectious work-up was initiated promptly.

Contrast-enhanced computed tomography showed polyserositis, generalized lymphadenopathy, hepatosplenomegaly, and pulmonary infiltrates (Figure 1). Blood and urine cultures remained negative. Bronchoscopy with lavage, pleural and ascites taps, and a lymph-node biopsy revealed inflammatory changes without organisms. Broad eubacterial, pan-fungal, and mycobacterial polymerase chain reaction analyses were negative. Given a recent vacation in southern Germany, leishmaniasis was assessed and excluded. Despite broad anti-infective therapy with ceftriaxone, meropenem, levofloxacin, and piperacillin/tazobactam, markers of infection did not improve. The patient required intubation and high-dose vasopressor support.

Pericardial fluid aspiration showed hemorrhagic cytology with presumed myeloid precursor cells. Microbiology was negative. This was followed by bone marrow aspiration and biopsy. Histopathology showed 100% cellularity with architectural disturbance, extensive atypical megakaryocytic proliferation, and grade 1 myelofibrosis with open sinusoids (Figure 2A-C). Mild increases in plasma cells and mature T cells were present (a CD4:CD8 of approximately 1:1). Assays for Leishmania and Epstein-Barr virus were negative. Overall, the bone marrow findings were strongly suggestive of a myeloid neoplasia with mild myelofibrosis. Targeted Illumina myeloid next-generation sequencing and fluorescence in situ hybridization for TP53 alterations did not reveal any pathogenic mutations or fusions. Whole genome sequencing from blood was unremarkable. Core biopsy of an axillary lymph node showed lymphadenitis with a marked increase of polytypic plasma cells (Figure 2D-F). Epstein-Barr virus-encoded small RNA in situ hybridization and studies for HHV8 were negative. There were no indications of IgG4 proliferation (<100 IgG4-positive plasma cells/mm²). One major histopathological criterion for the diagnosis of HHV8-negative-iMCD was fulfilled with a grade 3 plasmacytosis (sheet-like polytypic plasmacytosis) in the interfollicular space. (Figure 2E, F).⁵ In addition, regressive B-cell follicles were observed, in line with the histological criteria of iMCD-TAFRO.⁷ Although it would have been preferred, lymph node extirpation could not be performed in the acute phase due to hemodynamic and hemostatic reason. Flow cytometry failed to detect any malignant cells in either the pericardial effusion or the blood. Paroxysmal nocturnal hemoglobulinuria was ruled out by fluorescence-activated cell sorting analysis.

Table 1.Reference ranges of various clinical chemistry parameters and the patient’s baseline and post-treatment values.

Figure 1.Contrast-enhanced computed tomography scan taken during the patient’s stay in the intensive care unit, showing coronal and axial views with findings typical of TAFRO syndrome. (A) Marked hepatomegaly (25x17x23 cm), extensive ascites (black asterisk) and mesenteric, periportal and axillary lymphadenopathy with lymph nodes having a short axis diameter of up to 16 mm (white arrows). (B) Pleural effusions on the left side (white asterisk) as well as further lymphadenopathy in the right hilum (white arrow). (C) Axillary lymphadenopathy (white arrows) as well as the right-sided pleural effusion (white asterisk). The right-sided pleural effusion was drained before computed tomography. TAFRO: thrombocytopenia, anasarca, fever, reticulin fibrosis/renal insufficiency, organomegaly.

Figure 2.Histopathology of the trephine biopsy and core biopsy of the lymph node, showing specific changes associated with idiopathic myulticentric Castleman disease - TAFRO at the time of diagnosis. (A) Hematoxylin & eosin (H&E) stain showing hypercellular bone marrow in the trephine biopsy; 40x magnification. (B) H&E stain showing hyperplastic hematopoiesis; 400x magnification. (C) CD61 staining pointing at groups and clusters of atypical megakaryocytes; 400x magnification. (D) H&E staining of the lymph node core biopsy¸showing interfollicular expansion of polytypic plasma cells; 40x magnification. (E) In situ hybridization for k light chains; 100x magnification. (F) In situ hybridization for λ light chains; 100x magnification. TAFRO: thrombocytopenia, anasarca, fever, reticulin fibrosis/renal insufficiency, organomegaly.

Multidisciplinary rounds including infectious diseases, hematology, oncology, rheumatology, nephrology, and intensive care continued the exclusion work-up. Duodenal, colonic, and skin biopsies showed no malignancy or infection, including Whipple disease. Additional blood polymerase chain reaction testing for histoplasmosis and parvovirus were negative. Immunofixation and broad autoimmunity testing were unremarkable. There was no evidence of thrombotic microangiopathy: normal ADAMTS-13 activity, Shiga toxin negative, no schistocytes.

Based on the presence of the TAFRO clinical constellation together with Castleman disease-compatible lymph node histopathology – and negative testing for relevant mimickers (including HHV-8) – we made the biopsy-supported diagnosis of iMCD-TAFRO. Major alternative diagnoses such as bacterial sepsis, hemophagocytic lymphohistiocytosis (HLH)/macrophage activation syndrome (MAS), autoimmune disease, lymphoma, and other hematologic malignancies were systematically excluded.²

In view of the persistently very high procalcitonin levels, we decided against the use of steroids and opted for intensified IL-6 blockade. The patient (body weight 70 kg) received siltuximab 11 mg/kg (770 mg) once, followed by tocilizumab 8 mg/kg (560 mg) on 3 consecutive days as a short rescue strategy. This resulted in rapid clinical improvement: the fever resolved, the thrombocytopenia was corrected, and procalcitonin level normalized. The woman was weaned from the ventilator and dialysis and was ambulatory on the ward within 2 weeks. The clinical course, together with the absence of myeloid driver alterations, argued against an underlying myeloid neoplasm; in clinicopathological correlation, the megakaryocytic proliferation and reticulin fibrosis were deemed reactive in the context of TAFRO.⁸ The patient was discharged on treatment with siltuximab every 3 weeks. At last follow-up she remains in sustained remission with normal platelet counts, normal levels of inflammatory markers and normal renal function. There is no longer any evidence of serositis.

iMCD is a cytokine-driven lymphoproliferative disorder characterized by systemic inflammation with multiorgan dysfunction. According to the fifth edition of the WHO classification of hematolymphoid tumors (WHO-HAEM5) clinical subtypes of iMCD include iMCD-TAFRO and iMCD-NOS.³ In WHO-HAEM5, iMCD-TAFRO is defined by five required clinical features – thrombocytopenia, anasarca, fever/marked inflammation, renal dysfunction or bone marrow reticulin fibrosis, and organomegaly – alongside iMCD lymph-node histology and appropriate exclusions.^3,5,6 This framework differentiates iMCD-TAFRO from iMCD-NOS, which more often shows thrombocytosis and hypergammaglobulinemia.^3,7 Clinically, our patient had T: severe thrombocytopenia (4×10⁹/L), A: polyserositis/anasarca, F: hyperinflammation (fever; C-reactive protein 526 mg/L; IL-6 504 pg/mL), R: dialysis-requiring renal failure, and O: hepatosplenomegaly with generalized lymphadenopathy, fulfilling all five TAFRO features. Lymph node histology and broad exclusions met essential criteria for iMCD.^3,5,6 Applying the 2019 updated TAFRO severity score, the patient reached 12/12 (grade 5, “very severe”). Note that the typical histopathology described for TAFRO lymph nodes is a combination of atrophic follicles with prominent interfollicular vascular proliferation and fewer interfollicular plasma cells compared to those in iMCD-NOS – an important nuance given that our patient’s node demonstrated atrophic B-follicles and an increase in interfollicular plasma cells, still compatible with iMCD by consensus criteria.^3,7

A sepsis phenotype (SOFA domains: thrombocytopenia, renal failure, vasopressor need) is common in iMCD-TAFRO flares, driving empiric use of broad-spectrum antimicrobials.⁹ A k ey pitfall is the procalcitonin level. Although widely used as a biomarker of bacterial infection, procalcitonin can be markedly elevated in cytokine-storm biology. Elevated procalcitonin has been reported in iMCD-TAFRO.¹⁰ and a prior report of a case of HHV8-associated MCD described procalcitonin >100 ng/mL.¹¹ Together with our patient’s procalcitonin >100 ng/mL despite repeatedly negative microbiology, these observations emphasize that extreme procalcitonin elevations do not exclude Castleman-spectrum hyperinflammation and highlight the need for improved biomarker-based differentiation across cytokine-storm syndromes (e.g., TAFRO/ HLH vs. sepsis). Thus, procalcitonin-guided ‘bacterial sepsis’ algorithms may be misleading in cytokine-driven hyperinflammation, and persistently negative microbiology should prompt consideration of iMCD-TAFRO even in the presence of extreme procalcitonin levels.

Another key mimic in fulminant cytokine-storm presentations is secondary HLH/MAS. Recent data suggest that readily available inflammatory markers can support early differentiation: sCD25 and ferritin are typically substantially higher in HLH, whereas C-reactive protein tends to be higher in iMCD-TAFRO, providing a practical biomarker pattern to guide diagnostic prioritization when microbiology is negative and clinical overlap is pronounced.¹²

Bone marrow in iMCD-TAFRO often shows hypercellularity, megakaryocytic hyperplasia, and reticulin fibrosis, which can mimic myeloproliferative neoplasms/myelodysplastic syndrome.^3,7,8 In our case, 100% cellularity with atypical megakaryocytes and grade 1 myelofibrosis prompted suspicion of a myeloid neoplasm. However, next-generation sequencing and fluorescence in situ hybridization were negative, lymph-node and clinical features supported iMCD, and rapid response to IL-6 blockade argued for reactive changes rather than clonal myeloid disease. Awareness of this mimicry is crucial to prevent misdiagnosis and inappropriate cytotoxic therapy.^3,8

International guidelines recommend anti-IL-6 therapy – siltuximab first-line, tocilizumab when siltuximab is unavailable – with steroids as needed in severe cases. Our patient’s intensified IL-6 blockade led to reversal of organ failure, consistent with the IL-6-driven biology recognized by WHO and consensus frameworks.^13,14

In fulminant sepsis-like presentations with persistently negative microbiology, iMCD-TAFRO should be considered – even when procalcitonin is extremely high – and IL-6 blockade instituted promptly when diagnostic criteria are met.

In summary, it must be emphasized that clinicians need to be very aware of the entity of TAFRO syndrome. This is due to its acute, often life-threatening course with multi-organ failure and to the extremely relevant differential diagnoses. Particularly in the setting of intensive care for critically ill patients, the almost perfect imitations of myeloproliferative neoplasia (cytopenias, fibrosis in bone marrow, splenomegaly) and sepsis (high procalcitonin values, inflammation, cytokine storm) can mean diametrically different therapeutic approaches. This can sometimes significantly delay the actual diagnosis and targeted therapy. The patient gave informed consent to the publication of this case report, which respects the ethical rules of our country.

Footnotes

• Received December 10, 2025
• Accepted April 23, 2026

Correspondence

*OH and JV contributed equally as first authors

O. HUDOWENZ - ole.hudowenz@uksh.de

Disclosures

No conflicts of interest to disclose.

Contributions