1. Introduction
Malignant lymphoma is one of the most important diseases for the differential diagnosis of lymphadenopathy in the mediastinal or abdominal cavity. non-Hodgkin’s lymphoma (NHL) is the most common malignant lymphoma and is further divided into three major categories in terms of biological behavior: indolent lymphoma, aggressive lymphoma, and highly aggressive lymphoma. The treatment strategy for NHL is generally similar to the entities within each group, although the outcomes of the individual entities within each group can vary significantly [
1]. Therefore, the categorization of NHL diagnosis by using lymphadenopathy is important for determining treatment strategies. For thoracic or abdominal lymphadenopathy, the European Society of Gastrointestinal Endoscopy recommends endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) for the diagnosis and subclassification of malignant lymphoma [
2]. However, malignant lymphoma may be difficult to diagnose in lesions where EUS-FNA cannot be performed.
Studies have reported that contrast-enhanced endoscopic ultrasound (CE-EUS) could be a promising diagnostic modality for pancreatic tumors and lymph nodes [
3,
4,
5,
6,
7,
8,
9,
10]. However, according to these studies, various enhancement patterns have been recognized in malignant lymphoma, thus making it difficult to differentiate malignant lymphoma from other diseases with enhancement patterns by using qualitative evaluation, such as CE-EUS. The various enhancement patterns shown by malignant lymphoma in CE-EUS might depend on biological behavior. However, no study has evaluated the correlation between the enhancement pattern and biological behavior of malignant lymphoma in CE-EUS. We have previously reported that the diagnostic accuracy of CE-EUS with qualitative and quantitative analyses for differentiating malignant from benign lymphadenopathy was significantly better than the qualitative assessment of CE-EUS [
11]. Thus, this study was conducted to evaluate the diagnostic capability of the qualitative and quantitative evaluations of CE-EUS in differentiating biological behavior in terms of indolent NHL versus aggressive NHL.
2. Materials and Methods
2.1. Patient Selection
This retrospective cohort study was conducted at a single center—namely, Gifu University Hospital (Gifu, Japan)—between September 2016 and August 2019. The inclusion criteria were patients who underwent CE-EUS and EUS-FNA for lymphadenopathy in the mediastinum or abdominal cavity and were diagnosed with NHL on the basis of the pathological findings obtained by EUS-FNA. However, patients with a major axis of less than 10 mm on endoscopic ultrasound (EUS) were excluded because it was difficult to determine the contrast pattern. The classification of lymphoma was based on the World Health Organization classification of lymphoid neoplasms [
12], and the categories of NHL were based on the World Health Organization classification of lymphomas [
13]. The study was performed in accordance with the Declaration of Helsinki. The study protocol was approved by the institutional review board of Gifu University Hospital, and this study was registered in the UMIN Clinical Trials Registry (UMIN000047907).
2.2. EUS Procedure
All patients were moderately sedated with midazolam and pentazocine during the EUS procedure. EUS was performed with a convex-type echoendoscope (GF-UCT260; Olympus, Tokyo, Japan) connected with US systems (ProSound F75 or Pro-Sound alpha 10; FUJIFILM, Tokyo, Japan). Before CE-EUS, the size, shape (round or oval), border (sharp or fuzzy), echogenicity (hyperechoic or hypoechoic), and echo texture (heterogeneous or homogeneous) of each lymph node was evaluated. With regard to CE-EUS, the mechanical index was set at 0.2, and the transmitting frequency was set at 5 MHz, respectively. The extended pure harmonic detection mode was used for CE-EUS. Sonazoid® (Daiichi-Sankyo, Tokyo, Japan, or GE Healthcare, Milwaukee, WI, USA) was injected as the contrast agent for CE-EUS. A measure of 0.015 mL/kg body weight of the contrast agent was injected through a peripheral vein by bolus injection. After injection of the contrast enhancer, the images acquired by CE-EUS were recorded for 90 s. Digital Imaging and Communications in Medicine (DICOM) format and Audio Video Interleaved (AVI) format were used for recording all the examinations. EUS-FNA was performed for all lymph nodes, followed by CE-EUS. Patients were observed in the endoscopy unit for at least 2 h after the procedure.
2.3. Image Analysis
As for image analysis of CE-EUS, the enhancement patterns were categorized as homogeneous or heterogeneous, and the vascular patterns were classified as hypervascular or hypovascular, respectively. Vascular and enhancement patterns were evaluated at 30–45 s. We previously reported that vascular patterns of hypervascular and isovascular are difficult to distinguish in a lymph node because of the connective tissue around the organ; therefore, vascular patterns characterized as hypervascular and isovascular were all considered hypervascular [
11]. Similar to the previous study [
11], hypervascular lesions were defined as being similar to higher intensity of enhancement compared to the surrounding connective tissue; on the other hand, hypovascular lesions were defined as a lower intensity of enhancement compared to the surrounding connective tissue in this study.
2.4. Time Intensity Curve (TIC) Analysis
All the examinations of CE-EUS were recorded in DICOM format, and TIC analysis was performed using DAS-RS1 (Hitachi-Aloka Medical, Tokyo, Japan). For the TIC analysis, the average echo intensity within the region of interest was used. The region of interest was set as to trace the outline of the lesion at the time of administration of the contrast medium. TIC analysis was performed for up to 60 s after the administration of the contrast medium. The baseline intensity (I
base, dB) was defined as intensity before administration of contrast medium. The intensity at 60 s was defined as I
60 (dB). Maximum intensity was I
max (dB); maximum echo intensity was from I
base to I
60, time to maximum intensity was T
max (seconds); intensity gain from time from I
max to I
0 was I
max − I
base (dB); intensity gain from I
base to I
max and velocity of reduction (VR) was [I
max − I
60]/[60 − T
max], (dB/s); and VR from I
max to I
60 was evaluated by TIC analysis (
Figure 1). TIC analysis was carried out separately for lesions with homogeneous and heterogeneous enhancement by CE-EUS in the current study, similar to that in a previous study [
11].
2.5. Statistical Analysis
In order to assess the efficacy of qualitative evaluation in CE-EUS, the sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy were calculated. For the comparison between the two groups, either Fisher’s exact test or the chi-square test was used for nominal variables. The Mann–Whitney U test was used for continuous variables. To determine sensitivity, specificity and accuracy with the optimal cut-off value, we performed Receiver operating characteristic (ROC) analysis on TIC parameters with statistical significance. A p-value < 0.05 was considered statistically significant. All statistical analyses were carried out with JMP software version 11.2 (SAS Institute, Cary, NC, USA).
4. Discussion
The diagnostic ability of the qualitative evaluation of CE-EUS for differentiating indolent NHL and aggressive NHL in mediastinal or abdominal lymphadenopathy showed a sensitivity, specificity, and accuracy of 61%, 72%, and 66%, respectively. Quantitative evaluation with TIC analysis improved the sensitivity, specificity, and accuracy to 94%, 69%, and 82%, respectively, when combined with qualitative and quantitative evaluations. The accuracy rate of CE-EUS combining qualitative and quantitative evaluations in differentiating indolent NHL and aggressive NHL was significantly higher than that of the qualitative evaluation of CE-EUS (p = 0.0389).
Several studies have reported the differentiation of NHL using sonographic features. Gu et al. [
14] evaluated the B-mode sonographic characteristics of thyroid ultrasound for primary thyroid lymphoma (PTL) in 27 patients. They reported in the discussion section that of 25 PTLs with echogenic strands, all (8/8) MALTs and 73.3% (11/15) of DLBCLs had linear echogenic strands, and two DLBCLs had destructive linear strands; therefore, it is difficult to estimate the pathological type of lymphoma on the basis of sonographic characteristics. Javier et al. [
15] evaluated the sonographic findings of transabdominal ultrasound for the liver involvement of histologically proven NHL with histologic subtype in 17 patients. They reported that there were no differences in sonographic patterns (hepatomegaly, splenomegaly, focal liver lesions, diameter, echogenicity, and lymphadenopathies) between aggressive NHL in thirteen patients and indolent NHL in four patients. In the current study, there were no significant differences in sonographic characteristics (long axis, short axis, shape, border, echogenicity, and echotexture) from B-mode EUS between aggressive NHLs and indolent NHLs. It seems to be difficult to distinguish between aggressive NHL and indolent NHL by using only sonographic characteristics based on B-mode EUS.
There have been two reports on the usefulness of contrast-enhanced ultrasound (CE-US) in differentiating between indolent and aggressive NHL. Jing et al. [
16] reported 76 lymph nodes of NHL, including indolent NHL in 14 lymph nodes and aggressive NHL in 62 lymph nodes, with respect to the value of CE-US. They reported that 1 out of 14 cases of indolent NHL and 10 out of 62 cases of aggressive NHL showed heterogeneous enhancement. There was no significant difference between indolent NHL and aggressive NHL in terms of enhancement pattern (
p = 0.3540); however, aggressive NHL tended to show heterogeneous enhancement compared with indolent NHL. Xuelei et al. [
17] reported 140 lymphomatous lymph nodes, including aggressive NHL in 136 lymph nodes and indolent NHL in four lymph nodes, concerning the application of CE-US. They reported that 81.6% of aggressive NHLs and 100% of indolent NHLs showed rapid well-distributed hyperenhancement and that 18.4% of aggressive NHLs showed rapid heterogeneous hyperenhancement. In this study, rapid well-distributed hyperenhancement indicated hypervascularity with homogeneous enhancement, and rapid heterogeneous hyperenhancement indicated hypervascularity with heterogeneous enhancement. The two reports indicate that many NHLs show homogeneous enhancement in CE-US, and it is difficult to distinguish between indolent NHL and aggressive NHL in CE-US with qualitative assessment alone. In the present study, 63% of aggressive NHL cases and 37% of indolent NHL cases showed heterogeneous enhancement, with a statistically significant difference. Our study results showed that aggressive NHL might show more heterogeneous enhancement, although it is difficult to distinguish between aggressive NHL and indolent NHL by using only the qualitative evaluation of CE-EUS for mediastinal or abdominal lymphadenopathy.
In this study, TIC analysis was performed on lesions showing homogeneous and heterogeneous enhancements in the qualitative evaluation. No significant difference was found in heterogeneous enhancement. However, aggressive NHL resulted in significantly faster VR than indolent NHL in lesions with homogeneous enhancement (
p < 0.0001). Jiang et al. [
16] reported on the usefulness of quantitative evaluation with TIC analysis of CE-US for the differentiation of indolent and aggressive NHL. They performed CE-US on lymph nodes for 90 s, set the region of interest around the margin of lesion in the lymph node, and calculated the maximum intensity, time to maximum intensity, and VR, similar to the current study. There were no statistical differences in the TIC parameters, including maximum intensity, time to maximum intensity, and VR, in the identification of indolent NHL and aggressive NHL; however, the VR showed a faster tendency in aggressive NHL than in indolent NHL (
p = 0.088). They speculated that lesions with a higher degree of aggressiveness might have richer tumor vascularity, which might increase blood velocity. This might be the reason why aggressive NHL resulted in significantly faster VR than indolent NHL in this study.
Several studies [
18,
19,
20] reported that aggressive NHL showed a significantly higher SUV max than that of indolent NHL with PET-CT. In this study, PET-CT was performed in 40 patients and showed that aggressive NHLs had a significantly higher SUV max than indolent NHLs (aggressive NHL vs indolent NHL: 15.12 vs 8.76;
p = 0.0263). Based on the results of ROC analysis (AUC of 0.7133 with the cut-off level of 4.4), the diagnostic accuracy of aggressive NHL by PET-CT was 72%. CE-EUS qualitative and quantitative analysis for these 40 cases showed an accuracy rate of 90%, which was significantly higher in comparison with the results of PET-CT (
p = 0.0416). Although it was a small number of cases, CE-EUS may be more useful than PET-CT in differentiating between indolent and aggressive NHL.
By using a combination of qualitative and quantitative evaluations in CE-EUS for mediastinal or abdominal lymphadenopathy, the current study showed that the accuracy rate for distinguishing between indolent NHL and aggressive NHL was 82%. In 13 out of 34 patients who showed as hypervascular with homogeneous enhancement via the qualitative evaluation of CE-EUS, it was difficult to distinguish between indolent NHL and aggressive NHL by using qualitative evaluation alone. However, when combining qualitative and quantitative evaluations, CE-EUS was able to distinguish between indolent NHL and aggressive NHL for 31 out of 34 patients. (VR: aggressive NHL, 0.182 dB/s; indolent NHL, 0.127 dB/s;
p < 0.0001). Although there were some lesions in which the differentiation of aggressive NHL from indolent NHL was difficult even with the application of a combination of qualitative and quantitative evaluations, Ribeiro et al. [
21] showed that indolent NHL might be difficult to diagnose even with EUS-FNA. Performing CE-EUS before EUS-FNA for mediastinal or abdominal lymphadenopathy may still assist in diagnosis in cases wherein it is difficult to determine the biological grade of malignant lymphoma.
This study has several limitations. A retrospective study design with a small cohort may have caused bias in patient selection. There was no comparison arm to evaluate CE-EUS in distinguishing indolent NHL from aggressive NHL. The most notable strength of this study is that this is the first study that distinguished indolent NHL and aggressive NHL by using CE-EUS.