Background: Idiopathic multicentric Castleman disease (iMCD) is a rare and complex disorder involving multicentric lymphadenopathy, systemic inflammation, and multiple organ system dysfunction due to a poorly understood cytokine storm-driven pathology. Diagnosis is hindered by reliance on subjective interpretation of a broad range of non-specific lymph-node (LN) histopathological features. These features include the size and morphology of secondary follicles, specifically their corresponding germinal centers (GCs) and mantle zones (MZs). There is also significant heterogeneity between the histopathological characteristics observed in different clinical subtypes of iMCD. These include iMCD-TAFRO (thrombocytopenia, anasarca, fever, renal dysfunction, organomegaly), iMCD-not-otherwise-specified (iMCD-NOS), and across histopathological subtypes (hypervascular, plasmacytic, mixed). Development of quantitative morphometric benchmarks could shorten diagnostic delay by improving inter-observer agreement and reveal biological insights across clinical and histopathological subtypes of iMCD.

Methods: H&E-stained sections from diagnostic LNs of 20 iMCD patients (iMCD-NOS = 12, iMCD-TAFRO = 8) and eight metastasis-free sentinel LNs, functioning as controls, were digitally annotated in Aperio ImageScope by three blinded reviewers. Secondary follicles were identified and outlined manually. GC area, secondary-follicle area, and MZ width and area were measured. GC density (GCs per mm² of tissue) was also calculated using LN size measurements. iMCD vs control comparisons used two-sample t tests; subtype analyses used one-way ANOVA with Tukey post-hoc tests.

Results: We quantified a total of 1927 follicles from iMCD patient LNs and 43 follicles from control LNs. iMCD LNs had on average nine-fold more GCs per LN area than control LNs (63.3GCs/cm^2 vs 7.81 GCs/cm^2), but there was no significant difference between iMCD-TAFRO and iMCD-NOS. When both iMCD-TAFRO and iMCD-NOS were included together, the size of GCs, follicles, and MZ size and width did not differ significantly compared to controls. However, when iMCD-TAFRO and iMCD-NOS were evaluated separately, marked differences emerged compared to controls. GCs in iMCD-TAFRO LNs were significantly smaller than control LNs. Conversely, MZ size and width were markedly increased in iMCD-TAFRO LNs compared to control LNs. Follicle size in iMCD-TAFRO LNs did not differ from controls, as the increase in MZ area balanced the decrease in GC size.

We also compared the same metrics across the various histopathological subtypes represented in the above cohort: mixed (N=7), hypervascular (N=7), and plasmacytic (N=5) histopathological subtypes. Plasmacytic follicles had the largest areas, whereas mixed follicles were the smallest. Similarly, plasmacytic GCs had the largest areas, while mixed GCs were the smallest. All three iMCD histopathological subtypes had similar MZ sizes and widths. Plasmacytic LNs had the highest density of GCs per area. These results indicate that both follicle size and density can be used to distinguish the three histopathological subtypes.

Conclusions: Our quantitative morphometric analysis demonstrates that iMCD LNs, when grouped together regardless of subtype, are distinguished from controls primarily by a dramatic increase in follicle density. However, when clinical subtypes are separated, there are significant differences in follicular architecture within these subtypes compared to controls. Similar findings were revealed with histopathological subtypes. Together, these objective measures of follicle density and architecture provide a reproducible framework for differentiating iMCD from typical LNs and for discriminating among iMCD clinical and histopathologic subtypes.

This content is only available as a PDF.
2025
Sign in via your Institution