Baseline patient characteristics
A total of 93 newly-diagnosed CD patients (49 males; 44 females) were enrolled in the study, with a median age of 53 years (range, 14~78 years). The baseline characteristics of patient population are shown in Table 1. The proportions of HV, PC and Mixed are significantly different statistically different between UCD (85%, 10% and 5%) and MCD (19%, 63% and 11%), p < 0.001. The proportion of patients with clinical symptoms in MCD was significantly higher than that in UCD (p < 0.001), and the proportion of IL-6 positives in MCD was significantly higher than that in UCD (p = 0.049). In addition, among 48 patients who underwent HHV-8 testing, only 4 (8.3%) were positive for HHV-8, and all were from MCD.
Table 1
Baseline patient characteristics
Variable | UCD (n = 20) | MCD (n = 73) | p |
Age (years, median) | 46 (23 ~ 76) | 54 (14 ~ 78) | 0.133 |
Gender | | | 0.074 |
Male | 7 (35.0%) | 42 (57.5%) | |
Female | 13 (65.0%) | 31 (42.5%) | |
Histology type | | | 0.001 |
Hyaline vascular | 17 (85.0%) | 19 (26.0%) | |
Plasma cell | 2 (10.0%) | 46 (63.0%) | |
Mixed cell | 1 (5.0%) | 8 (11.0%) | |
IL−6 | | | 0.049 |
Positive | 7 (70.0%) | 59 (90.4%) | |
Negative | 3 (30.0%) | 4 (9.6%) | |
Unknown | 10 | 10 | |
Complications | | | |
Involvement | 7 (35.0%) | 52 (71.2%) | 0.003 |
No involvement | 13 (65.0%) | 21 (28.8%) | |
Clinical symptoms | | | |
Symptomatic | 8 (40.0%) | 59 (80.8%) | 0.001 |
B-symptoms | 2 (10.0%) | 28 (38.4%) | 0.012 |
Splenomegaly | 3 (15.0%) | 30 (41.1%) | 0.036 |
Oedema/effusion | 2 (10.0%) | 36 (49.3%) | 0.002 |
Active pneumonia | 4 (20.0%) | 17 (23.3%) | 1.000 |
IL-6: interleukin 6; B-symptoms: night sweats: fever, drenching, and a weight loss of more than 10% over 6 months; p < 0.05: signifcant |
18 F-FDG PET/CT findings
A total of 477 involved lymph node stations were detected by PET/CT, including 375 (78.6%) by CT and 458 (96.0%) by PET. Among the 13 lymph node distribution areas, PET detection capability was significantly better than CT in all areas except for the abdomen and bilateral inguinal regions (Fig. 2). The most common lymph node involvement area in UCD was the cervical (n = 9/20; 45%), followed by the mediastinal/hilar (n = 3; 15%), axilla (n = 3; 15%), abdominal/retroperitoneal (n = 4; 20%) and pelvis (n = 1; 5%). The most common lymph node stations in MCD were the cervical (n = 61/73; 83.6%), axilla (n = 51; 69.9%), mediastinal (n = 48; 65.8%), inguinal (n = 36; 49.3%), iliac (n = 35; 47.9%), and paraaortic (n = 32; 43.8%), while uncommon lymph node stations were located in the hilar region (n = 24; 32.9%) and abdomen (n = 22; 30.1%). The median LNS in MCD was 6 (IQR, 7; rang, 2 ~ 13).
For all patients, the maximum short diameter of involved lymph nodes ranged 0.80~10.90cm (median, 1.70; IQR, 0.92), and the highest SUVmax ranged 1.40~28.18 (median, 4.86; mean, 5.95 ± 3.80). The proportion of lymph nodes with increased metabolism was 96%. UCD had a significant statistical difference from MCD in FDG metabolic parameters (SUVmax, p = 0.004; MLV, p < 0.001; TLG, p < 0.001; LLR, p < 0.001; LMR, p = 0.001; SLR, p < 0.001) (Table 2). The metabolic parameters (MLV, p = 0.002; TLG, p = 0.002; LLR, p = 0.024; SLR, p < 0.001) were significantly different among histological subtypes (HV, PC and Mixed). In adition, the metabolism of PC was higher than Mixed, and Mixed was higher than HV (Table 3).
Table 2
Comparison of metabolic parameters between UCD and MCD
Variable | UCD (n = 20) | MCD (n = 73) | Total (n = 93) | p |
SUVmax | 3.9 (2.2) | 5.3 (4.3) | 4.9 (4.0) | 0.004 |
MLV | 9 (26) | 29 (48) | 24 (41) | < 0.001 |
TLG | 23 (68) | 74 (135) | 60 (129) | < 0.001 |
LLR | 1.26 (0.77) | 2.00 (1.13) | 1.86 (1.06) | < 0.001 |
LMR | 2.26 (1.36) | 3.15 (2.08) | 2.93 (1.99) | 0.001 |
SLR | 0.87 ± 0.15 | 1.12 ± 0.30 | 1.07 ± 0.29 | < 0.001 |
UCD: unicentric castleman disease; MCD: multicentric castleman disease; SUVmax: maximum standardized uptake value; MLV, metabolic lesion volume; TLG: total lesion glycolysis; LLR: lymph node to liver ratio of SUVmax; LMR: lymph node to mediastinum ratio of SUVmax; SLR: spleen to liver ratio of SUVmax; p < 0.05: signifcant |
Table 3
Comparison of metabolic parameters between histologic subtypes
Variable | HV (n = 36) | PC (n = 48) | Mixed (n = 9) | p |
SUVmax | 4.1 (2.6) | 5.3 (4.4) | 5.0 (4.1) | 0.120 |
MLV | 12 (21) | 46(46) | 24 (34) | 0.002 |
TLG | 31(67) | 124 (144) | 69(123) | 0.002 |
LLR | 1.69 (0.88) | 2.13 (1.44) | 1.92 (1.10) | 0.024 |
LMR | 2.49 (1.61) | 3.26 (2.35) | 2.65 (2.54) | 0.059 |
SLR | 0.90 ± 0.17 | 1.19 ± 0.29 | 1.07 ± 0.36 | < 0.001 |
HV: hyaline vascular type; PC: plasma cell type; Mixed: mixed cell type; SUVmax: maximum standardized uptake value; MLV: metabolic lesion volume; TLG: total lesion glycolysis; LLR: lymph node to liver ratio of SUVmax; LMR: lymph node to mediastinum ratio of SUVmax; SLR: spleen to liver ratio of SUVmax; p < 0.05: signifcant |
Metabolic parameters compared to laboratory findings
A total of 6 laboratory tests (HGB, PLT, CRP, ESR, IL-6, ALB) were included in this study to compare the differences in PET/CT metabolic parameters between the normal and abnormal groups (Suppl Table 1). LLR was significantly higher in abnormal groups of inflammatory indicators CRP and ESR than in normal groups (p = 0.031 & p = 0.021); TLG was significantly higher in abnormal groups of PLT and ESR than in normal groups (p = 0.010 & p = 0.004); The LNS was significantly higher in abnormal groups of all laboratory tests than in the normal groups (p < 0.05); while SLR was significantly higher in abnormal groups of HGB, PLT, CRP, ESR and ALB than in normal groups (p < 0.01).
Comparison of PET/CT findings and clinical symptoms
In 93 patients, 67 (72.0%) had clinical symptoms and 26 (28.0%) were asymptomatic (Table 1). In symptomatic patients, 30 (32.3%) were present with B symptoms, 33 (35.5%) with splenomegaly, 38 (40.9%) with serous cavity effusion and 65 (69.9%) with CT manifestations of lung cysts, nodules, thickening of the bronchovascular bundle or interlobular septal thickening, but only 21 (22.6%) patients had increased FDG uptake, with a SUVmax ranged 1.1 ~ 10.0 (median, 2.1). A small number of patients presented with pain, rash or digestive system symptoms. Figure 3A showed that the metabolic parameters of symptomatic group were significantly higher than those of asymptomatic group (SUVmax, p = 0.001; MLV, p = 0.002; TLG, p < 0.001; LLR, p < 0.001; LMR, p = 0.002; SLR, p < 0.001; LNS, p < 0.001; MLV, p = 0.002). In addition, 15 (16.1%) patients had combined bdominal pelvic omental, mesenteric thickening or perirenal fascia with fuzzy density, of whom 8 (8.6%) had mild increase of FDG uptake; 9 (9.7%) had bilateral parotid hyper-metabolism; 5 (5.4%) had unilateral or bilateral adrenal hyper-metabolism foci; 3 (3.2%) involved the liver; 2 (2.2%) involved the pancreas; 2 (2.2%) involved the uterus or appendages and 1 (1.1%) involved the stomach wall.
The value of PET/CT in differentiating the severity of CD
To explore the influencing factors of disease severity in CD patients, this study included 7 metabolic parameters and 7 clinical risk factors. The univariate analysis revealed that there were significant statistical differences between the severe group (n = 31) and the mild group (n = 62) in terms of SLR (p < 0.0001), LNS (p < 0.001), hypohemoglobinemia (p < 0.0001), hypoalbuminemia (p < 0.0001), splenomegaly (p = 0.002), serous cavity ascites (p = 0.002) and age (p = 0.018) (Table 4). In the binary logistic regression analysis, stepwise regression method (forward LR) was used to screen variables with statistical differences, and finally SLR (p = 0.011, OR value = 14.8) and hypohemoglobinemia (p = 0.004, OR value = 0.044) were included as independent influencing factors with disease severity (Table 5). The subsequent receiver operating characteristic (ROC) curve revealed that SLR can relatively effectively judge the severity of CD, with a sensitivity of 77.4%, specificity of 69.4%, cut-off value of 1.04 and area under the curve (AUC) of 0.761 (Fig. 3B). In addition, to further study the relationship between SLR and high-risk iMCD, according to the stratified criteria of iMCD risk level by CDCN, a total of 33 patients (19 severe and 14 non-severe) with iMCD were included in this study, and the independent sample T test showed that the SLR value of patients with severe iMCD was significantly higher than that of non-severe (P = 0.016) (Fig. 4A, 4B).
Table 4
Comparison of risk factors between mild group and severe group
Factor | Mild (n = 62) | Severe (n = 31) | p |
SUVmax | 4.7 (3.7) | 5.1 (4.6) | 0.903 |
MLV | 20(43) | 39 (47) | 0.172 |
TLG | 54 (111) | 125(158) | 0.244 |
LLR | 1.84 (1.06) | 2.01 (1.40) | 0.728 |
LMR | 2.91 (1.69) | 3.21 (2.60) | 0.669 |
SLR | 0.98 ± 0.22 | 1.25 ± 0.33 | < 0.001 |
LNS | 3 (6) | 7 (6) | < 0.001 |
Age (year) | 48 (24) | 56 (25) | 0.018 |
Sex (m/f) | 32/28 | 16/15 | 0.877 |
Spleen size (cm) | 11.0 ± 2.8 | 12.9 ± 2.8 | 0.002 |
HGB decrease | 23(37.1%) | 30(96.8%) | < 0.001 |
ALB decrease | 19(30.6%) | 27(87.1%) | < 0.001 |
serous effusion | 18(29.0%) | 20(64.5%) | 0.002 |
Pneumonitis | 49(64.5%) | 25(80.6%) | 0.110 |
SUVmax: maximum standardized uptake value; MLV: metabolic lesion volume; TLG: total lesion glycolysis; LLR: lymph node to liver ratio of SUVmax; LMR: lymph node to mediastinum blood pool ratio of SUVmax; SLR: spleen to liver ratio of SUVmax; LNS: No. of involved lymph node stations; HGB: hemoglobin; ALB: albumin; y: years; cm: centimeter; m: male; f: famale; p < 0.05: signifcant |
Table 5
Binary logistic regression analysis of influencing factors of CD severity
Factor | p | OR value | 95% CI |
SLR | 0.011 | 14.806 | 1.833 ~ 119.597 |
HGB decrease | 0.004 | 0.044 | 0.005 ~ 0.366 |
SLR: spleen to liver ratio of SUVmax; HGB: hemoglobin; OR: odds ratio; CI: confidence interval; p < 0.05: signifcant |