https://orcid.org/0000-0002-7615-3949
Abstract
As a rare lymphoproliferative disorder, many patients with HHV-8/HIV-negative Castleman disease (CD) have hypoalbuminemia. However, data is limited on whether hypoalbuminemia is an independent predictor of CD. We retrospectively collected data from 230 patients diagnosed at 12 medical centers in China and the U.S. Different classifications included 147 patients with unicentric CD (UCD) and 83 with idiopathic multicentric CD (iMCD). Adjusted smooth curve fitting showed that the relationship between albumin and all-cause death of patients with CD and iMCD was linear. Cox proportional hazards regression modeling showed a negative association between the risk of death and albumin level (hazard ratio [HR]: 0.84; 95% CI, 0.76, 0.93). Using the Kaplan–Meier method, we determined that hypoproteinemia was a risk factor for poorer prognosis in patients with CD, UCD, and iMCD. Albumin was independently and negatively associated with the risk of death in CD patients, especially those with iMCD.
Introduction
Castleman disease (CD) is a group of at least four disorders that share a spectrum of characteristic histopathological features but have a wide range of etiologies, presentations, treatments, and outcomes. CD was first described in the 1950s by Benjamin Castleman as localized mediastinal lymph node enlargement characterized by increased numbers of lymphoid follicles with germinal center involution and marked capillary proliferation, including both follicular and interfollicular endothelial hyperplasia [Citation1]. By the mid-1980s, CD was divided into unicentric CD (UCD), which involves a single enlarged lymph node or region of lymph nodes, and multicentric CD (MCD), which involves multiple lymph node stations [Citation2]. Alternatively, one-third of MCD cases without Human immunodeficiency virus (HIV) and Human herpes virus-8 (HHV-8) positive is known as ‘idiopathic’ (iMCD) [Citation3,Citation4]. In the 2010s, Takai et al. identified a severe form of HHV-8− or iMCD in which patients had a homogeneous constellation of abnormal laboratory tests and clinical features that they termed thrombocytopenia, ascites, reticulin fibrosis, renal dysfunction, organomegaly (TAFRO) syndrome [Citation5,Citation6].
The clinical manifestations of CD are heterogeneous in terms of different clinical classifications. In general, UCD patients typically present with either compressive symptoms, or nodes are found incidentally [Citation4]. The clinical presentation of iMCD varies from mild constitutional symptoms to life-threatening cytokine storms, organ failure, or even death. Laboratory hallmarks of CD include leukocytosis, anemia, thrombocytosis or thrombocytopenia, elevated erythrocyte sedimentation rate (ESR), increased C-reactive protein (CRP) and fibrinogen, hypergammaglobulinemia, and hypoalbuminemia [Citation7]. Complete surgical resection is the primary method of treatment for UCD, while the treatment strategy of iMCD varies from ‘watchful waiting’ to high-dose chemotherapy based on the treatment of non-Hodgkin Lymphoma, and numerous other single-agent therapy (anti-CD20 antibodies, anti-IL-6 antibodies and anti IL-1 antibodies) have been proposed [Citation4,Citation8,Citation9]. Due to the low incident of CD, no generally accepted prognostic factors for CD have been found to help determine treatment strategies.
Several laboratory hallmarks have been found in patients with CD, such as high CRP levels and ESR, increased lactate dehydrogenase (LDH) and β2-microglobulin (β2-MG), hypoproteinemia, and inverse albumin/globulin ratios [Citation4,Citation10]. Significantly decreased albumin level is associated with overall survival (OS) in CD. In recent studies, the results indicated that hypoproteinemia remains independently associated with OS [Citation10,Citation11]. The role of albumin in the progression of CD, however, remains unknown. We hypothesized here that serum albumin levels at admission might reflect the severity of systemic inflammation and thus can serve as a predictive factor for CD outcomes. To address this question, we performed a retrospective study to compare the outcome in patients with or without hypoalbuminemia and to explore the impact of albumin in the prognosis of CD. Thus, we analyzed the clinical, laboratory, pathologic variables, and outcomes of 230 patients with CD organized from 12 major academic medical centers in China and the U.S. to better understand the association between serum albumin and death as a result of CD.
Subjects, materials, and methods
Patient information
This study was executed and approved by the ethics committee. We identified 428 patients with CD who had been diagnosed and treated at 12 large medical centers in China and the U.S. between January 2005 and February 2017. We collected detailed clinical, laboratory, and pathological data based on diagnosing and analysis of these patients. We excluded patients with concomitant malignancies, HIV positivity, and POEMS (polyneuropathy, organomegaly, endocrinopathy, monoclonal plasma-proliferative disorder, and skin changes) syndrome, and patients without albumin or prognosis information. This study was approved by our institutional review board and conducted following the Declaration of Helsinki.
A total of 230 patients had detailed clinical data on albumin, which were collected during diagnosis, and had follow-up records available for analysis in this discovery cohort. Clinical, laboratory, and diagnostic materials from lymph nodes, tissues, or organs were evaluated in accordance with generally accepted guidelines to confirm a diagnosis of CD. shows the inclusion process for the study population. The data on patients, including general information, associated clinical complaints, laboratory test results, radiological imaging reports, histological phenotypes of biopsy specimens, and treatment and follow-up data, were recorded and analyzed. Their clinical classifications were based on radiological examination and surgical findings. MCD was defined by the involvement of more than two lymph nodes in at least two separate regions, whereas cases with localized involvement were classified as UCD. Pathological diagnosis was based on the histopathological characteristic of biopsy specimens with review by at least two experienced pathologists. Hepatosplenomegaly, lymphadenopathy, and luminal effusion (including hydrothorax, ascites, pelvic effusion, and pericardial effusion) was confirmed by computerized tomography (CT) examination of involved regions and B-type ultrasonography (USG-B) examination of superficial lymph nodes (neck, axillary, groin, abdominal, and pelvic cavities) and organs (cardiac, liver, pancreas, spleen, and kidney). B-symptoms were identified based on fevers with a temperature more than 38 °C, night sweats, or weight loss of more than 10% over the previous 6 months. The diagnoses of anemia and hypoproteinemia were based on the results of laboratory tests, whose hemoglobin was lower than 120 g/L for males, 110 g/L for females, and albumin was converted from a continuous variable to a categorical variable (< 30 g/L, 30 − 40 g/L, ≥ 40 g/L) based on clinical experience.
Figure 1. Flowchart review of the literature. Abbreviation. OS: overall survival.
The regimen of treatments administered varied across different subtypes (Supplementary Figure S1 and Table S1), which was based on the severity of clinical symptoms and drug accessibility, including waiting and watching, surgery, radiotherapy, siltuximab (anti-IL-6 antibodies), tocilizumab (anti-IL-6 receptor antibodies), rituximab (anti-CD20 antibodies), thalidomide, prednisone, cyclophosphamide (CTX), and other chemotherapies. Dosing for single-agent therapies was as follows: 11 mg/kg siltuximab intravenously every 3 weeks per protocol or every 6 weeks at the investigator’s discretion; 375 mg/m2 rituximab intravenously weekly for 4 weeks; 8 mg/kg tocilizumab intravenously every 4 weeks per protocol; 0.5–1 mg/kg prednisone; 0.2 − 0.4 g/day CTX; 100 − 150 mg/day thalidomide; or 15 − 5 Gy radiotherapy. Chemotherapies included CHOP (Cyclophosphamide, hydroxyl doxorubicin, hydrochloride, vincristine, and prednisone), CHOP-like chemotherapy regimens, or CHOP regimens plus rituximab (R-CHOP). Follow-up information was generated from a review of each visit record until the time of last follow-up or death. Castleman Disease Collaborative Network (CDCN) response criteria, based on the evaluation of biochemical, lymph node, and symptom response, were used to assess treatment response [Citation12]. These criteria were also used by independent radiologists who reviewed our results. We also evaluated OS, where OS was defined as the duration from the date of diagnosis to death or last follow-up. Follow-up was assessed through February 28, 2018.
Statistical analysis
Patient characteristics and survival outcomes were summarized using descriptive statistics. The different albumin levels groups were compared using Chi-square and T-tests. The Kaplan-Meier method was used for survival analysis, and a log-rank test was used to analyze the survival rate between the three groups. The independent association of albumin with the risk of death was evaluated using Cox proportional hazards regression modeling (hazard ratio [HR] and 95% CI), with an adjustment for major covariables in three models. Model 1: crude model; Model 2: adjusted for age, gender, ethnicity; Model 3: additionally adjusted for pathology, UCD/iMCD, hemoglobin, B symptom, luminal effusion, and hepatosplenomegaly. In the regression analysis model, the previous variables were selected because of their clinical importance or statistical significance in our univariable analysis. A negative association of albumin levels with all-cause death was assessed using a generalized additive model (GAM) and a fitted smoothing curve (penalized spline method). Moreover, possible modifications on the association between albumin and the risk of death were also evaluated by stratified analyses and interaction testing. A p value < 0.05 was considered statistically significant. All statistical analyses were performed using R (http://www.R-project.org, The R Foundation) and Empower (R) (www.empowerstats.com; X&Y Solutions, Inc., Boston, MA).
Results
Association of clinical presentation and albumin
All 230 patients with HHV-8 and HIV negative were diagnosed in 12 large medical centers in China and the U.S. between January 2005 and February 2017. Of all included patients, 147 (63.9%) were clinically classified as UCD, and 83 (36.1%) were classified as having iMCD. Among the patients with CD, 161 (70%) were hyaline-vascular (HV) types, with 62 (27%) being plasma cell (PC), and only 7 (3%) having mixed cellular variants (Mix). There were 123 patients ≤ 40 years of age and 107 patients > 40 years of age at the time of diagnosis. We found that there was a higher incidence of hypoalbuminemia in patients > 40 years of age than that in patients < 40 years of age (p < 0.05). There were no significant differences in the sex distribution and ethnicity between patients with different albumin levels (). We observed a strong relationship between centricity (UCD vs iMCD), hepatosplenomegaly, histopathological subtypes (HV subtype vs PC and Mix subtype), and different albumin level (p < 0.001). Patients with hypoalbuminemia had higher rates of luminal effusion (p = 0.01) and B-symptoms (p < 0.001) than those with normal albumin. According to the radiological imaging data from these 193 patients, there was no significant difference compared with mass volume ().
Table 1. Clinical characteristics of the study population according to albumin (230 cases).
With detailed results from laboratory test, we calculated the distribution rates of various items between different albumin level groups (). According to the results of quantitative analyses of various laboratory tests, patients with normal albumin (≥ 40 g/L) had a higher level of hemoglobin (p < 0.001) and platelets (p = 0.006) than those with hypoalbuminemia. Levels of inflammatory markers such as β2-MG (p < 0.001), ESR (p < 0.013), CRP (p < 0.001) were elevated in patients with albumin less than 40 g/L, and levels of LDH were no significant differences ().
Association between albumin and all-cause death risk
We next conducted a univariate analysis of OS for patients with CD. There were nine significant risk factors for CD patients identified by the results of a log-rank test, including age, histopathological subtype, B symptom, luminal effusion, hepatosplenomegaly, hemoglobin levels, platelet levels, and elevated ESR and CRP. There were six significant risk factors for iMCD patients identified by the results of a log-rank test, including age, histopathological subtype, luminal effusion, hepatosplenomegaly, hemoglobin levels, and elevated ESR. Univariate analysis did not identify statistically significant differences related to any investigated factors in UCD patients, which are listed in .
Table 2. Univariate analyses of overall survival of 230 patients with CD.
Adjusted smooth curve fitting showed that the relationship between albumin and all-cause death of patients with CD and iMCD was linear (). The results suggested that the risk of death decreased when albumin levels increased. In this study, we constructed three models for analyzing the independent association between albumin and the risk of death in CD patients (). The models were adjusted with covariables that were selected based on their clinical importance and statistical significance in univariable analysis (). In a fully adjusted model (model 3), the risk of death increased 9% for a 1 unit decrease in albumin (HR: 0.91; 95% CI, 0.86, 0.97). When the normal albumin (≥ 40 g/L) was used as the reference group, the risk of death of the mild hypoalbuminemia (30–40 g/L) and hypoalbuminemia (< 30 g/L) groups increased 3.48 times and 7.25 times, respectively (30–40 g/L, HR: 3.48; 95% CI, 1.11, 10.89; < 30 g/L, HR: 7.25; 95% CI, 1.89, 27.84). As shown in , we fit the association between albumin and all-cause death of iMCD. Additionally, the risk of death inecreased 9% for a 1 unit decrease in albumin (HR: 0.91; 95% CI, 0.85, 0.98). The risk of death of the mild hypoalbuminemia (30–40 g/L) and hypoalbuminemia (< 30 g/L) groups showed 3.4-fold and 6.15-fold increased, respectively (30–40 g/L, HR: 3.4; 95% CI, 0.84, 13.84; < 30 g/L, HR: 6.15; 95% CI, 1.29, 29.29), implying a significant negative relationship between the risk of death by iMCD and albumin level. The HRs (95% CIs) were 2.71 (0.08, 93.87) and 38.56 (2.43, 612.26). A similar result was also found in UCD (Table S2 in the Supplementary Appendix). The results suggested that the risk of death due to CD and iMCD decreased with elevated albumin level.
Figure 2. Albumin and the HRs for all-cause death. (a) In the whole CD patients. (b) In the iMCD patients. Abbreviations. HR: hazard ratio; CD: Castleman disease; iMCD: human immunodeficiency virus- and human herpesvirus 8-negative multicentric Castleman disease, idiopathic MCD.
Table 3. Relationship between albumin and death in patients with CD.
Table 4. Relationship between albumin and death in patients with iMCD.
Subgroup analysis
A predefined subgroup analysis of the effect of albumin level on the risk of death of CD was done using the following variables: age, gender, UCD/iMCD, B symptom, luminal effusion, hepatosplenomegaly, Mass ≥ 5 cm, WBC, hemoglobin, and PLT (). Among patients with CD, subgroup analysis suggested that albumin level had a negative effect on the risk of death (P for interaction > 0.05) except age.
Figure 3. Subgroup analyses of the effect of albumin on the risk of death. Each subgroup analysis adjusted, if not stratified, for gender, age, UCD/iMCD, B symptom, luminal effusion, hepatosplenomegaly, mass, WBC, hemoglobin, and PLT. Abbreviations. UCD: unicentric castleman disease; iMCD: human immunodeficiency virus- and human herpesvirus 8-negative multicentric Castleman disease, idiopathic MCD; Luminal effusion: ascites and/or pleural effusion and/or pericardial effusion; WBC: white blood cell count; PLT: platelets. CI confidence interval.
Survival analysis of patients with CD
Among the 230 patients with detailed clinical data, the median follow-up duration was 48 months (range, 5–275 months). A total of 33 patients died, including 7 cases with UCD and 26 with iMCD. All of these 33 patients were accompanied by clinical complaints. Using the Kaplan–Meier method, we calculated the survival rates at 3 and 6 years. The 3- and 6-year survival curves from patients with different albumin levels with CD, UCD, or iMCD are shown in . Among patients with CD, the 3-year survival rates, respectively, were ≥ 40 g/L: 97.9% vs 30–40 g/L: 86.4% vs < 30 g/L: 54.2% (p < 0.001 by the log-rank test), and the 6-year survival rates were ≥ 40 g/L: 96.5% vs 30–40 g/L: 76.3% vs < 30 g/L: 45.8% (p < 0.001) (). Among patients with iMCD, the 3- and 6-year survival rates were 93.9 and 90.9% in the ≥ 40 g/L group, 76.4 and 61.8% in the 30–40 g/L group, and 50.0 and 37.5% in the < 30 g/L group (p < 0.001 by the log-rank test) (). Considering the excellent prognosis of patients with UCD receiving primary lesion resections, we only analyzed the 6-year survival rates, which were 98.2% in the ≥40 g/L group, 96% in the 30–40 g/L group, and 75% in the <30 g/L group (p < 0.001 by the log-rank test) ().
Figure 4. Survival curves of albumin in patients with CD. (a) In all CD patients. (b) In iMCD patients. (c) In UCD patients. Abbreviations. HR: hazard ratio; CD: castleman disease; UCD: unicentric castleman disease; iMCD: human immunodeficiency virus- and human herpesvirus 8-negative multicentric Castleman disease, idiopathic MCD.
Discussion
To better understand this disease, using clinical, laboratory, and pathologic data, as well as a review of treatment outcomes with a follow-up duration of 10 years from a multicenter study, we have performed a comprehensive evaluation of CD to date. Due to its rarity, our understanding of CD is mainly from retrospective studies and case reports. More notably, our study demonstrated an association between albumin level and the risk of death based on comprehensive information, and this analysis should advance our understanding of CD and its treatment options.
Serum albumin is a multi-functional protein that is able to bind and transport numerous endogenous and exogenous compounds [Citation13] and participate in modulating the inflammatory reaction [Citation14]. We speculate that albumin is a surrogate marker of host reactions to systemic inflammation in CD. Our study found that patients with iMCD were more likely to develop hypoalbuminemia than those with UCD, which might be associated with a series of systemic inflammation responses in iMCD patients. Autoimmune and autoinflammatory mechanisms were previously proposed as etiological drivers of iMCD [Citation15], and increased IL-6 signaling is an established driver of iMCD symptomatology and pathogenesis [Citation16]. Capillary leak syndrome and lymph node hypervascularity have been observed in iMCD [Citation17], which leads to the spilling of albumin. The significant higher of the levels of β2-MG, CRP, and ESR in the albumin <40 g/L group was also shown in our study. We speculate that this reduction in albumin can partly reflect the degree of inflammation of each patient.
Hypoalbuminemia is a minor criterion for the diagnosis for CD [Citation18] and an adverse prognostic factor in patients with CD. However, to date, published studies have reported that multicentricity, pathology type, B symptoms, gender, and age were prognostic factors for patients with []CD [Citation7]; and PC, hepatomegaly and/or splenomegaly, and hemoglobin and pleural effusion were identified in iMCD international prognostic index (iMCD-IPI) [Citation7,Citation20]. There are few studies focused on the prognosis of albumin in CD patients. In retrospective studies, hypoproteinemia was considered a poor prognosis factor in CD [Citation10] and MCD [Citation11]. But there was no significant difference in 1-year and 3-year survival of different albumin levels [Citation19]. This is related to the small number of cases they included, and the multivariate analysis of albumin was not performed. In these studies, the albumin was divided into two groups: < 35 g/L and ≥ 35 g/L. The association between albumin level and disease prognosis of patients with CD has not yet been fully elucidated. We converted albumin from a continuous variable to a categorical variable in this study at < 30 g/L, 30 − 40 g/L, and ≥ 40 g/L. Whether UCD or MCD, patients with albumin < 30 g/L had the poorest survival compared to other groups. Moreover, those in the 30 − 40 g/L group were more likely to achieve poor survival compared to in ≥ 40 g/L group. The risk of death of patients with CD was decreased with elevated albumin level.
The treatment of CD is often tailored based on centricity, clinical experiences, and drug accessibility. In UCD, complete surgical resection is the primary option and can receive a good prognosis followed by a watch and wait strategy [Citation4]. In our study, 93.4% of patients with UCD received surgical treatment and achieved long-term survival (Supplementary Table S1). However, our study suggests that albumin is an independent prognostic factor of CD and could provide indicators of clinical intervention for CD patients.
Unfortunately, standard regimens for the treatment of iMCD have not been established yet. In recent years, siltuximab has become a safe and effective therapy that targets a key aspect of CD []pathogenesis [Citation8], which significantly reduced the disease burden and symptoms in a large portion of patients [Citation20]. Due to our study being a retrospective analysis, treatment with IL-6 has not been generally identified as a first-line treatment of iMCD, and only 8.5% patients with iMCD received anti-IL-6 treatment (Supplementary Table S1). Regardless of the treatment, hypoalbuminemia was shown to be an independent risk factor for iMCD prognosis. Long-term albumin administration has also been shown to improve survival, which was related to a significant increase in albumin levels [Citation21]. With the application of siltuximab/tocilizumab in iMCD, albumin also may assist with predicting treatment effectiveness in future [Citation15]. Thus, we hypothesize that proper treatment can raise albumin levels and may improve the prognosis of CD patients, especially iMCD patients.
Our study had several limitations. First, it is a retrospective study, and some data were missing (especially ESR and CRP). Second, because of the heterogeneous treatments in our study (Supplementary Table S1), we could not further compare the effect of albumin in different treatments. In addition, only a small number of iMCD patients received anti-IL-6 therapy or monotherapy, meaning we could not compare the association between monotherapy efficacy and albumin level.
Conclusion
In general, our study identified clinical features and prognosis outcomes between different albumin levels in 230 patients with CD. We found the albumin was a significant risk factor in UCD and iMCD, but these were not exactly the same. For all patients with CD, hypoalbuminemia was considered to be an independent poor prognosis factor, and appropriate supportive treatment may reduce the risk of death in these cases. Prospective study about the effect of amelioration of low blood albumin in CD patients should be conducted to follow this up. We hope our study can clarify our understanding of UCD and MCD, and the unfavorable risk factors of albumin in this study can provide a reference to a therapeutic decision and prognosis assessment. It is essential to continue exploring the etiology, classification, treatment, and prognosis in future investigations.
GLAL-2022-0228-File008.docx
Download MS Word (16.7 KB)GLAL-2022-0228-File007.tif
Download TIFF Image (14.1 MB)Acknowledgements
We thank LetPub (www.letpub.com) for its linguistic assistance during the preparation of this manuscript.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
- Castleman B, Iverson L, Menendez VP. Localized mediastinal lymphnode hyperplasia resembling thymoma. Cancer-Am Cancer Soc. 1956;9:822–830.
- Weisenburger DD, Nathwani BN, Winberg CD, et al. Multicentric angiofollicular lymph node hyperplasia: a clinicopathologic study of 16 cases. Hum Pathol. 1985;16(2):162–172.
- Liu AY, Nabel CS, Finkelman BS, et al. Idiopathic multicentric Castleman’s disease: a systematic literature review. Lancet Haematol. 2016;3(4):e163–e175.
- Yu L, Tu M, Cortes J, et al. Clinical and pathological characteristics of HIV- and HHV-8-negative Castleman disease. Blood. 2017;129(12):1658–1668.
- Takai K, Nikkuni K, Shibuya H, et al. Thrombocytopenia with mild bone marrow fibrosis accompanied by fever, pleural effusion, ascites and hepatosplenomegaly. Rinsho Ketsueki 2010;51:320–5.
- Iwaki N, Fajgenbaum DC, Nabel CS, et al. Clinicopathologic analysis of TAFRO syndrome demonstrates a distinct subtype of HHV-8-negative multicentric Castleman disease. Am J Hematol. 2016;91(2):220–226.
- Talat N, Schulte KM. Castleman’s disease: systematic analysis of 416 patients from the literature. Oncologist. 2011;16(9):1316–1324.
- Fajgenbaum DC, Kurzrock R. Siltuximab: a targeted therapy for idiopathic multicentric Castleman disease. Immunotherapy-Uk. 2016;8(1):17–26.
- Gerard L, Michot JM, Burcheri S, et al. Rituximab decreases the risk of lymphoma in patients with HIV-associated multicentric Castleman disease. Blood. 2012;119(10):2228–2233.
- Lan X, Li Z, Zhang M. Analysis of clinical characteristics and prognosis factors of 71 cases with HIV-negative Castleman’s disease: hypoproteinemia is an unfavorable prognostic factor which should be treated appropriately. J Cancer Res Clin Oncol. 2018;144(7):1265–1277.
- Zhang X, Rao H, Xu X, et al. Clinical characteristics and outcomes of Castleman disease: a multicenter study of 185 Chinese patients. Cancer Sci. 2018;109(1):199–206.
- van Rhee F, Voorhees P, Dispenzieri A, et al. International, evidence-based consensus treatment guidelines for idiopathic multicentric Castleman disease. Blood. 2018;132(20):2115–2124.
- Merlot AM, Kalinowski DS, Richardson DR. Unraveling the mysteries of serum albumin-more than just a serum protein. Front Physiol. 2014;5:299.
- Arroyo V, Garcia-Martinez R, Salvatella X. Human serum albumin, systemic inflammation, and cirrhosis. J Hepatol. 2014;61(2):396–407.
- Fajgenbaum DC. Novel insights and therapeutic approaches in idiopathic multicentric Castleman disease. Blood. 2018;132(22):2323–2330.
- Casper C, Chaturvedi S, Munshi N, et al. Analysis of inflammatory and anemia-related biomarkers in a randomized, double-blind, placebo-controlled study of siltuximab (anti-IL6 monoclonal antibody) in patients with multicentric Castleman disease. Clin Cancer Res. 2015;21(19):4294–4304.
- Fajgenbaum DC, Uldrick TS, Bagg A, et al. International, evidence-based consensus diagnostic criteria for HHV-8-negative/idiopathic multicentric Castleman disease. Blood. 2017;129(12):1646–1657.
- Dispenzieri A, Fajgenbaum DC. Overview of Castleman disease. Blood. 2020;135(16):1353–1364.
- Dong Y, Wang M, Nong L, et al. Clinical and laboratory characterization of 114 cases of Castleman disease patients from a single centre: paraneoplastic pemphigus is an unfavourable prognostic factor. Br J Haematol. 2015;169(6):834–842.
- van Rhee F, Wong RS, Munshi N, et al. Siltuximab for multicentric Castleman’s disease: a randomised, double-blind, placebo-controlled trial. Lancet Oncol. 2014;15(9):966–974.
- Caraceni P, Tufoni M, Zaccherini G, et al. On-treatment serum albumin level can guide long-term treatment in patients with cirrhosis and uncomplicated ascites. J Hepatol. 2021;74(2):340–349.
- Yu L, Shi M, Cai Q, et al. A novel predictive model for idiopathic multicentric castleman disease: the international Castleman disease consortium study. Oncologist. 2020;25(11):963–973.