Discussion
In our study, elevated serum IgG4 levels and infiltration of IgG4 + cells in tissues were observed in RDD patients with either nodal or extranodal involvement, mimicking IgG4-RDs clinically, serologically and pathologically. This underlines the necessity to distinguish RDD from IgG4-RDs in our clinical work. Our study analyzed a group of patients with RDD mimic IgG4-RD and characterized the clinical, laboratory and histopathological features of RDD and IgG4-RD. Extranodal involvement was found in both diseases. However, RDD mimic IgG4-RD was likely to involve the nervous system but rarely to involve LSGs, which are commonly affected in IgG4-RD. RDD mimic IgG4-RD patients had an elevated serum IgG4 level, but the level was still lower than that in IgG4-RD patients in general. Pathologically, there were features that remained exclusively to either RDD or IgG4-RD, which could be regarded as the gold standard for distinguishing one disease from the other. Moreover, although we discovered enrichment of IgG4 + plasma cells in affected tissues in RDD, a small gap remained between RDD and IgG4-RDs with regard to the proportion of IgG4 + plasma cells.
Rosai–Dorfman disease, or sinus histiocytosis with massive lymphadenopathy, is a rare, benign histiocytic proliferative disease. It occurs predominantly in children and youth, manifesting as chronic inflammation and painless lymph node masses, and extranodal involvement is discovered in 25% to 40% of patients [
10]. Characteristic pathological changes associated with this disease are phagocytosis of lymphocytes (emperipolesis) and the proliferation of S-100-positive histiocytes.
IgG4-RD is an immune-mediated condition that is likely to form tumorous lesions in multiple organs and is characterized by elevated serum IgG4 level and IgG4-positive plasma cell infiltration in affected tissues. However, these two features, both serological and pathological, are observed in many other diseases, suggesting that these diseases may mimic IgG4-RD.
Consistent with previous studies, our study showed that RDD could mimic IgG4-RD. We thus analyzed three groups of patients clinically, serologically and pathologically.
Clinically, for RDD with extranodal involvement, masses and polyploid protrusions are the most common presentation. It is difficult to distinguish these clinical manifestations from those of IgG4-RD, which usually presents as organ enlargement and sclerosis. The most commonly affected extranodal site in RDD is the skin. Other organs that may be affected include the central nervous system, soft tissues, the upper respiratory tract, the nasopharynx, breasts, testes, kidneys, thyroid glands and the appendix [
11‐
14]. For IgG4-RD, the organs most often involved are LSGs, the pancreas, the bile duct, the retroperitoneum, thyroid glands, kidneys and lungs. We obtained the same results as previous studies showing that extranodal involvement often occurs in the nervous system in RDD, while LSGs are the sites mostly involved in IgG4-RD. However, contrary to previous studies that indicate that most RDD patients have lymph node involvement, our study showed no statistically significant difference in lymph node involvement between RDD and IgG4-RD. This may be due to the way the patients in the total RDD group were diagnosed. In addition to the 7 patients with RDD mimic IgG4-RD, all of the patients had extranodal involvement, and patients in the RDD group were mostly recruited from surgery departments and diagnosed based on histopathology postsurgery; thus, these patients were likely to have extranodal involvement, which resulted in a lower percentage of RDD patients with lymph node involvement. Overall, we may roughly distinguish these two diseases based on their distinct clinical features. However, the RDD patients in our study also displayed lacrimal gland swelling, which may misguide the diagnosis of IgG4-RD. Therefore, clinical manifestations can only provide indications but not solid evidence for diagnosis.
With regards to laboratory parameters, elevated serum IgG4 concentration is recognized as the dominant serological characteristic of IgG4-RD. However, reliance upon this feature for establishing the diagnosis of IgG4-RD is usually highly problematic because of the poor specificity (60%) when the cut-off value of serum IgG4 concentration is set at > 1350 mg/L. Many studies have reported that serum IgG4 levels could be increased in various diseases, including connective tissue disease, chronic infection, lymphoma, and Castleman disease, making this parameter unreliable as the only criterion for the diagnosis of IgG4-RD. Our results support this notion, showing that 6 out of 7 RDD mimic IgG4-RD patients had elevated serum IgG4 levels. However, we also found that patients with RDD mimic IgG4-RD had a lower level of serum IgG4 levels than IgG4-RD patients. Furthermore, we tried to set a new cut-off value of serum IgG4 levels to distinguish these two diseases. Research indicates that the cut-off value should be adjusted in some situations to improve the diagnosis of IgG4-RDs [
15]. Since the present cut-off value may perform even more poorly when used to distinguish IgG4-RD from diseases mimicking it, in our study, we discovered the cut-off value of serum IgG4 level > 6531 mg/L using a ROC curve, performed for data in Fig.
2, which had high specificity (85.71%) but low sensitivity (50%). This value could be important for diagnosing patients with diseases mimicking IgG4-RD. It is found in clinical practice that the elevation in serum IgG4 concentration is more significant when other IgG isotypes remain normal, but this feature may not be so meaningful if the concentrations of all IgG isotypes are elevated. Along those lines, elevation in IgG4 alone was found in our study in IgG4-RD patients. Another interesting result was that a higher serum IgG2/IgG level was found in RDD mimic IgG4-RD patients than in IgG4-RD patients. Serum IgG2 is usually produced in response to bacterial capsular polysaccharide antigens [
16‐
18]. Elevated serum IgG2 levels were observed in some IgG4-RD cases [
19], but it has never been studied in RDD. It has been reported that the class switching of the IgG subclass proceeds through an ordered pattern from IgG3 to IgG1 to IgG2 to IgG4 [
20]. Therefore, one possibility for the elevated IgG2/IgG proportion is that IgG2 in produced en route to IgG4, and the speed of class switching may be different between RDD and IgG4-RD. Laboratory test results in our study suggested that as serum IgG4 level becomes specifically higher, there is a higher possibility to diagnose IgG4-RD, but this serum feature is not sufficient to rule out RDD.
Recently, the presence of IgG4 + plasma cells has been pathologically discovered in a considerable number of RDD cases. Research performed by Zhang et al. [
5] demonstrated that of 26 RDD cases, 73.1% exhibited more than 10 IgG4 + plasma cells/hpf, and 46.2% displayed more than 30 IgG4 + cells/hpf. Meanwhile, in tissues, the IgG4/IgG proportion in some RDD cases could be as high as in IgG4-RDs [
6,
21]. Because of the similar histopathological features, these 2 diseases are speculated to fall within the same [
7] or overlapping [
22] disease spectrum. However, others consider that RDD is not a kind of IgG4-RD because in affected tissues, IgG4 + plasma cells, IgG4/IgG proportion and foxp3 + Tregs are still lower than those in classic IgG4-RD but similar to those in reactive hyperplastic lymph nodes [
10]. The results in our study display apparent enrichment of IgG4 + plasma cells and elevated IgG4/IgG ratio in RDD patients, which mimic the pathological characteristics of IgG4-RD. Although we found that the infiltration of IgG4 + plasma cells in RDD was not as pronounced as that in IgG4-RD, no convincing standard for diagnosis could highlight this feature.
Nonetheless, IgG4-RDs and RDD display diverse and specific pathological features. Pathological features, including storiform fibrosis, obliterative phlebitis and eosinophilic infiltration, were found specifically in IgG4-RD patients but absent in RDD patients in our analysis, consistent with findings in previous studies [
10]. These characteristics are of great importance because, as emphasized in the consensus statement on the pathology of IgG4-RDs [
4], these are key for an unequivocal diagnosis of IgG4-RDs, even if IgG4 + plasma cell infiltration is confirmed. On the other hand, pathological characteristics exclusive to RDD were also confirmed in our study, including S-100( +), CD38( +), CD68( +) and emperipolesis in histopathological analysis in all samples from the RDD mimic IgG4-RD group but not in the IgG4-RD group. In recent years, studies of RDD using next-generation sequencing found neoplastic conditions, point mutations occur in 33–50% of RDD cases, affecting KRAS, NRAS, SMAD4, ARAF, or MAP2K1, which is not found in IgG4-RD [
23‐
25]. This gives a possible way for differentiating RDD mimic IgG4-RD using next-generation sequencing in some cases. These findings all support the notion that these 2 diseases fall along distinct spectra, and a clear separation of these diseases should be made during diagnosis.
Treatment is different for RDD and IgG4-RD. As most IgG4-RDs show a good response to corticosteroids, the treatment for RDD varies according to the severity of the disease. Some of the RDD cases are self-limited and no treatment is required, and some can be controlled using corticosteroids, while others are lymphoma-like and need chemotherapy. As for RDD mimic IgG4-RD, in these cases, response to corticosteroids is not as good as it is in IgG4-RD, and most of them need treatment of high intensity, such as combining with immunosuppressive drug or using chemotherapy. Therefore, diagnosing RDD from IgG4-RD correctly, especially for those mimicking IgG4-RD clinically and pathologically, can lead to standard treatment with better prognosis.
There are some limitations in our study regarding time of follow-up and number of patients. Firstly, the number of patients with RDD mimic IgG4-RD is limited in our study. Because both RDD and IgG4-RD are rare diseases, very few numbers of patients could fulfill our requirements and be enrolled in our research, leaving the problem of not enough samples to be analyzed. A study with more patients is warmly welcomed to give a better cut-off value of serum IgG4 to distinguish these two diseases. Secondly, follow-up time for prognosis is not long enough. In our study, the follow-up time period for each patient post diagnosis is no longer than 1 year, which is adequate to notice the improvement after treatment but not long enough to discover a relapse of the disease. Further study with a longer period of follow-up should be conducted to estimate long-term prognosis.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.