Limited sinonasal Rosai-Dorfman disease presenting as chronic sinusitis

Abstract

Aims: The sinonasal tract is a common extranodal site for Rosai-Dorfman disease (RDD). Recently, histiocytes with features of RDD were identified in the clinical setting of chronic sinusitis. This study evaluates whether this phenomenon should be considered part of the RDD spectrum or classified separately as RDD-like histiocytes.

Methods and results: We prospectively collected 13 cases showing histological features of RDD in chronic sinusitis patients and identified 14 with similar findings (3.5%) via retrospective review of 403 sinus contents over 2 years. All 27 cases displayed nodular aggregates of eosinophilic histiocytes with intermixed lymphoplasmacytic inflammation, prominent eosinophils and emperipolesis. The histiocytes were positive for S100 protein and cyclin D1 and negative for CD1a and CD207. All patients presented with severe chronic sinusitis without tumour formation or systemic symptoms. Twelve patients with follow-up (55%) required repeat sinus surgery compared with just 43 other sinusitis patients evaluated (11%); features of RDD were present in their additional specimens. Two cases that underwent targeted next-generation sequencing (20%) had oncogenic mutations in NF1 and KEAP1.

Conclusions: Overall, these findings confirm diagnostic histological and immunohistochemical features of RDD in a subset of chronic sinusitis specimens. While patients uniformly lack systemic involvement or tumefactive growth, they have a high risk of recurrent sinus disease. Although the relatively subtle nature of the findings raises consideration of separate classification, the presence of occasional oncogenic mutations and evidence of consistent MAPK/ERK pathway activation via cyclin D1 positivity suggests that this phenomenon represents a unique limited manifestation of RDD.

Keywords: Rosai-Dorfman disease; chronic sinusitis; immunohistochemistry; molecular diagnostics; sinus histiocytosis.

Figure 1

Cases displayed largely intact sinonasal mucosa with markedly oedematous underlying stroma and variable degrees of polyp formation (A). Prominent aggregates of histiocytes and lymphocytes were scattered throughout the stroma, conferring a checkerboard‐like appearance at low power (B). These histiocyte aggregates were mostly nodular, compact and even granuloma‐like with discrete borders (C), although a subset were more expansile and confluent (D).

Figure 2

Although most histiocytes were scattered throughout oedematous stroma (A), a subset had a prominent perivascular distribution without true vasculitis (B). Other groups of histiocytes surrounded seromucinous glands (C) or abutted the epithelial surface (D).

Figure 3

The histiocytes were large and epithelioid (A), spindled (B) and occasionally foamy (C) with abundant eosinophilic cytoplasm. The nuclei were round to oval and lacked grooves or indentations (D).

Figure 4

These histiocytes were closely intermixed with lymphocytes and plasma cells (A), with emperipolesis evident on haematoxylin and eosin in a subset of cases as indicated by arrowheads (B). Almost all cases also showed a dense infiltrate of eosinophils (C) with occasional eosinophilic emperipolesis as indicated by arrowheads (D).

Figure 5

All cases were positive for S100 protein, with diffuse expression in most cases (A) and patchy positivity in a few (B). The S100 protein staining also highlighted the emperipolesis in cases where it was not obvious on haematoxylin and eosin (C). All cases tested also were positive for cyclin D1 within the histiocyte aggregates (D).