1. Study Population and conventional SUVmax value for differentiating between lymph node metastasis and reactive hyperplasia manifestations in bladder cancer
A total of 153 patients with bladder urothelial carcinoma were included in this retrospective study. A total of 82.35% (n = 126) of patients were more than 60 years old, and the majority of patients were males (84.97%; n = 130). The number of pT0-T2 and pT3-T4 staged patients were similar. High-grade urothelial carcinoma accounted for 96.73% (n = 148). A total of 250 lymph nodes showed clear pathological findings, including metastatic lymph nodes (n = 145, 58%) and lymph node reactive hyperplasia (n = 105, 42%) (Table 1). The distribution of metastasis and reactive hyperplasia were dominated by bilateral external iliac lymph nodes (Table 2). Therefore, close attention should be paid to the lymph nodes in the external iliac region during imaging to prevent missed diagnosis.
Table 1
Characteristics of patient tumors and lymph nodes
Variable | N (%) |
age | |
≥60 | 126 (82.35%) |
< 60 | 27 (17.65%) |
Gender | |
male | 130 (84.97%) |
female | 23 (15.03%) |
Primary tumor status | |
pT0-T2 | 77 (50.33%) |
pT3-T4 | 76 (49.67%) |
Differentiated | |
High grade | 148(96.73%) |
Low grade | 5(3.27%) |
Nodal status | |
hyperplasia | 145 (58%) |
metastasis | 105 (42%) |
Table 2
Location and number of lymph nodes in patients
Lymph node location | Metastasis | Hyperplasia |
Right common iliac | 12 | 3 |
left common iliac | 7 | 4 |
Right internal iliac | 1 | 4 |
left internal iliac | 6 | 2 |
Right external iliac | 49 | 50 |
left external iliac | 46 | 35 |
Right obturator | 13 | 2 |
left obturator | 7 | 0 |
Right groin | 2 | 1 |
left groin | 2 | 4 |
Both lymph node metastasis and reactive hyperplasia have shown imaging manifestations of enlarged lymph node volume and increased FDG metabolism[10], which makes it difficult to diagnose lymph node metastasis by PET/CT. Previous studies have suggested a SUVmax cut-off of 2.0[16] or 2.5[17, 18] to identify lymph node metastasis in bladder carcinoma. SUVmax value of 2.0 or 2.5 has high sensitivity in diagnosing lymph node metastasis. However, in our study, as shown in Table 3, using either cut-off of SUVmax value of 2.0 or 2.5 resulted in a low diagnostic specificity for lymph node metastasis in bladder urothelial carcinoma, and the inability to effectively differentiate metastatic lymph nodes hyperplasia.
Table 3
Conventional SUVmax values in lymph nodes
Parameter | Sensitivity | Specificity | suspicious lymph nodes detected (n) | metastasis lymph nodes detected (n) |
SUVmax | | | | |
2.0 | 95.2% | 26.2% | 225 | 138 |
2.5 | 91.7% | 43.8% | 207 | 133 |
DSUVmax | | | | |
2.0 | 98.5% | 13.3% | 238 | 144 |
2.5 | 97.8% | 26.7% | 224 | 143 |
Further, we analyzed if delayed 18F-FDG PET/CT imaging can be useful to find more suspicious lymph nodes or metastatic lymph nodes(Table 3). When the cut-off value of SUVmax was 2.0, 13 more suspicious lymph nodes were found, 6 of which were metastatic lymph nodes in delayed imaging compared with standard imaging. 4.1% more metastatic lymph nodes were found in delayed imaging than standard imaging. When SUVmax = 2.5 was taken as the cut-off value, 17 more suspicious lymph nodes were found, 10 of which were metastatic lymph nodes in delayed imaging compared with standard imaging. 6.9% more metastatic lymph nodes were found in delayed imaging than standard imaging.
2. Metabolic parameters for metastatic and reactive hyperplasia lymph nodes of bladder cancer in dual time point imaging
Since tumors and hyperplasia may differ in the rate of uptake and excretion of FDG[12, 19], delayed imaging may help differentiate lymph node reactive hyperplasia from lymph node metastases[20, 21]. There are few previous studies on the changes in metabolic parameters before and after delayed imaging of metastatic lymph nodes and reactive hyperplasia in bladder cancer.
Firstly, we analyzed two-time-point PET imaging of metastatic lymph nodes and reactive hyperplasia. We found that after delayed imaging, FDG uptake of reactive hyperplasia lymph nodes is easier to gradually aggregate to the central area or other local areas of the lymph node than metastatic lymph nodes, and the FDG uptake at the edge of the lymph node decreased, even showing a star or spot pattern (Fig. 2, case 1, 2). But most of the cancerous lymph nodes showed a persistent increase in FDG uptake, and the FDG uptake in the whole lymph node mainly increased (Fig. 2, case 3, 4). Compared to metastatic lymph nodes, reactive hyperplasia lymph nodes were more likely to show increased lymph node-localized FDG uptake on delayed FDG imaging, whereas metastatic lymph nodes were more likely to have whole, diffuse increased FDG uptake on delayed imaging.
We further compared the changes of SUVmax, SUVmean, MTV before and after delayed imaging of metastatic lymph nodes and reactive hyperplasia. Among them, SUVmax and SUVmean in both types of lymph nodes were significantly increased after delayed imaging (Fig. 3A-C). 87.6% of metastatic lymph nodes displayed increased SUVmax, and 81% of reactive hyperplasia lymph nodes showed increased SUVmax after delayed imaging. Similarly, SUVmean values increased in 77.9% of metastatic lymph nodes, 63.8% of reactive hyperplasia lymph nodes showed increased SUVmean after delayed imaging (Fig. 3A-C).
Interestingly, following delayed imaging, 85.5% of metastatic lymph nodes had increased MTV values, while more hyperplasia lymph nodes (61.9%) had decreased MTV values. The MTV parameter of metastatic lymph nodes significantly increased, whereas the MTV of reactive lymph nodes decreased significantly (Fig. 3A, D).
3. Comparison of PET/CT-related parameters between metastatic lymph and reactive hyperplasia nodes before and after delayed imaging
We compared the PET/CT-related parameters and analyzed the differences in lymph node length, short axis, and metabolic parameters in dual time point imaging between metastatic lymph nodes and reactive hyperplasia. We used the Mann-Whitney U test to compare the differences between the two groups. All indicators showed significant differences except DISUVmax (Table 4). We performed multivariable logistic regression analysis to identify predictive factors for metastatic lymph nodes. The short axis lymph node diameter, SUVmean, and DIMTV were shown to be independent factors for the diagnosis of lymph node metastasis, and the delayed imaging DIMTV index was the best parameter for predicting lymph node metastasis (Table 5).
Table 4
Univariate analysis for diagnosing lymph node metastases
Parameter | Lymph node metastases | Z | P |
Hyperplasia | Metastasis |
Long-axis diameters | 10.835 ± 2.779 | 14.71 ± 5.632 | -6.117 | < 0.001 |
Short-axis diameters | 6.811 ± 1.896 | 9.905 ± 3.993 | -7.154 | < 0.001 |
SUVmax | 3.009 ± 2.043 | 7.285 ± 4.628 | -9.893 | < 0.001 |
SUVmean | 2.044 ± 0.794 | 4.259 ± 1.725 | -10.336 | < 0.001 |
MTV | 0.558 ± 0.530 | 1.177 ± 1.670 | -3.891 | < 0.001 |
DSUVmax | 3.932 ± 2.189 | 10.304 ± 8.085 | -9.541 | < 0.001 |
DSUVmean | 2.597 ± 1.105 | 4.873 ± 2.171 | -9.326 | < 0.001 |
DMTV | 0.488 ± 0.575 | 1.645 ± 2.152 | -8.661 | < 0.001 |
DISUVmax | 0.426 ± 0.548 | 0.414 ± 0.416 | -0.215 | 0.830 |
DISUVmean | 0.313 ± 0.445 | 0.183 ± 0.333 | -2.687 | 0.007 |
DIMTV | -0.060 ± 0.526 | 0.6088 ± 0.840 | -8.950 | < 0.001 |
Table 5
Independent parameters by stepwise multivariate logistic regression analysis in early and delayed images for diagnosing lymph node metastases
Parameters | B | S.E. | OR | 95% C.I | P |
Short-axis diameters | 0.172 | 0.085 | 1.188 | 1.005–1.403 | 0.027 |
SUVmean | 1.121 | 0.188 | 3.069 | 2.125–4.432 | < 0.001 |
DIMTV | 1.445 | 0.390 | 4.242 | 1.976–9.106 | < 0.001 |
5. Roc Curve And Nomogram Predicted Lymph Node Metastasis
Short axis lymph node diameter, SUVmean, and DIMTV predicted lymph node metastasis. We further used ROC curve analysis to evaluate Short axis lymph node diameter, SUVmean, DIMTV, and the combination of the above three indicators for differentiating between lymph node metastasis and reactive hyperplasia. The results showed that Short axis lymph node diameter, SUVmean, and DIMTV diagnosed metastatic lymph nodes, and when these indicators were combined, the AUC was 0.938, and sensitivity and specificity for predicting lymph node metastasis were 82.8% and 89.5%, respectively. These results showed that when the three indicators had combined the prediction of lymph node metastasis was higher than using a single indicator or joint index of short axis diameter and SUVmean (Table 6, Fig. 4A). To assist clinicians more intuitively in diagnosing lymph node metastasis, we drew a nomogram using the independent factors, Short axis lymph node diameter, SUVmean, and DIMTV to predict lymph node metastasis. DIMTV and SUVmean were more important indicators for predicting lymph node metastasis compared with the short axis diameter (Fig. 4B).
Table 6
Comparison of receiver operating characteristic curves of prediction models for lymph node metastasis
Parameter | AUC | P value | 95% C.I | Threshold value | Sensitivity | Specificity |
Short-axis diameters | 0.764 | < 0.001 | 0.706–0.822 | 7.530 | 67.6% | 73.3% |
SUVmean | 0.883 | < 0.001 | 0.841–0.925 | 2.590 | 83.8% | 81.9% |
DIMTV | 0.832 | < 0.001 | 0.778–0.885 | 0.032 | 85.5% | 78.1% |
Short-axis diameters + DIMTV | 0.887 | < 0.001 | 0.846–0.928 | 0.493 | 82.1% | 83.8% |
short axis diameters + SUVmean + DIMTV | 0.938 | < 0.001 | 0.891–0.958 | 0.530 | 82.8% | 89.5% |
6. Differentiate lymph node metastasis from lymph node inflammatory hyperplasia by visual analysis or a combination of three parameters
For the convenience of the clinical application better, we divided all of the lymph nodes into four groups (Group A) according to the thresholds of short axis diameter, SUVmean, and DIMTV: a high risk group (short axis diameter ≥ 7.53, SUVmean ≥ 2.59, and DIMTV ≥ 0.03), a moderate-highrisk group (Of the three parameters, any two of them are above the cutoff value ), a moderate-low risk group (Of the three parameters, only one of them is above the cutoff value), a low risk group (All three parameters are below the cutoff value). The probabilities of being lymph node metastasis in four groups were 97.6%, 75.9%, 36.8% and 3.6%, respectively, and probabilities of hyperplasia were 2.4%, 24.1%, 63.2%, 96.4%, respectively. There was a significant difference in the prediction rate of metastasis lymph node among the four risk groups (P < 0.001) (Table 7).
Table 7
Prediction rate of metastasis and hyperplasia lymph nodes in high, moderate-high, moderate-low, and low risk group by the combination of these three predictors
Group (A) | No. of lymph nodes | Probality (%) | P |
Metastasis | Hyperplasia |
High risk | 83 | 97.6% | 2.4% | < 0.001 |
Moderate-High risk | 54 | 75.9% | 24.1% | < 0.001 |
Moderate-Low risk | 57 | 36.8% | 63.2% | < 0.001 |
Low risk | 56 | 3.6% | 96.4% | < 0.001 |