The Diagnostic Performance of Multiparametric Ultrasound in the Qualitative Assessment of Inconclusive Cervical Lymph Nodes

Simple Summary Enlarged cervical lymph nodes (CLNs) are routinely evaluated for distinction between benign infectious enlargements and malignant diseases. Ultrasound (US) is the first-line imaging modality for the assessment of CLNs. The aim was to investigate the diagnostic performance of the addition of the multiparametric applications shear wave elastography (SWE) and contrast-enhanced US (CEUS) to B-mode US and color-coded duplex sonography (CCDS). Our results show that the diagnostic performance of multiparametric US (mpUS) is significantly higher than that of B-mode US and CCDS. MpUS may help to distinguish between malignant and benign CLNs and might aid the decision of choosing between the watch-and-scan strategy and surgery in primary cases as well as between repeated surgery and short-term follow-up in oncological patients. Abstract Background: Enlarged cervical lymph nodes (CLNs) can result from infection or malignancies, and a definitive diagnosis requires histological examination. Ultrasound (US) remains the first-line imaging modality for detection, and new US techniques may improve characterization. The aim of our study was to investigate whether the qualitative assessment of multiparametric US (mpUS) can improve diagnostic performance in the differentiation of benign and malignant CLNs. Methods: 107 CLNs in 105 patients were examined by preoperative mpUS consisting of B-mode US, color-coded duplex sonography (CCDS), shear wave elastography (SWE) and contrast-enhanced US (CEUS). US images were evaluated in consensus by two experienced US operators. Histopathological examination was used as reference standard. Results: SWE and CEUS combined showed the highest overall diagnostic performance (91% sensitivity, 77% specificity, 87% positive predictive value (PPV), 83% negative predictive value (NPV), 90% accuracy, χ2 (1) = 51.485, p < 0.001) compared to B-mode US and CCDS (87% sensitivity, 44% specificity, 73% PPV, 65% NPV, 73% accuracy χ2 (1) = 12.415, p < 0.001). In terms of individual techniques, SWE had higher specificity than B-mode and CCDS (71% sensitivity, 90% specificity, 92% PPV, 64% NPV, 78% accuracy, χ2 (1) = 36.115, p < 0.001), while qualitative CEUS showed the best diagnostic performance of all investigated US techniques (93% sensitivity, 85% specificity, 91% PPV, 87% NPV, 90% accuracy, χ2 (1) = 13.219, p < 0.001). Perfusion patterns, homogeneity, presence of necrosis, and malignancy differed significantly between malignant and benign CLNs (p < 0.001). Conclusions: SWE and CEUS can facilitate the differentiation of inconclusive CLNs when performed to supplement B-mode US and CCDS. MpUS may thus aid the decision between surgery and a watch-and-scan strategy in enlarged CLNs.


Introduction
Cervical lymph nodes (CLNs) can be enlarged due to infectious conditions or in the presence of metastasis from malignancies such as head and neck squamous cell carcinoma (HNSCC) or lymphatic diseases.Differentiation requires a histological examination [1].Many HNSCCs first present with signs and symptoms of CLN metastasis [2], where CLN involvement is a prognostic factor and influences the choice of treatment [3,4].Close monitoring of CLNs is recommended in HNSCC patients [5], with ultrasound (US) being the image modality of choice [6].In B-mode US, benign CLNs have an oval shape, a well-defined border and a hilus sign.Deviations from any of these morphological features are considered indicators of potential malignant involvement [7].Even though the combination of B-mode US and CCDS shows a higher diagnostic performance than computer tomography or magnetic resonance imaging of the neck, its sensitivity of 66-91% and specificity of 61-78% are not yet satisfying [6,8].Recent studies have demonstrated the added value of utilizing preoperative quantitative multiparametric US (mpUS) in inconclusive CLNs [9][10][11].MpUS has recently been developed to include multiple modalities such as shear wave elastography (SWE) and contrast-enhanced ultrasound (CEUS) in addition to regular B-mode US and color-coded duplex sonography (CCDS) [12].SWE was found to allow differentiation of metastatic from benign CLNs based on metric evaluation of tissue stiffness, which resulted higher in malignancy [13].Moreover, SWE was found to enable the distinction between malignant and benign CLNs even when none of the morphological B-mode US criteria were present [9].Quantitative CEUS, which depicts CLN vessel structure and gives a real-time image of a tissue's microvascularization, identified no significant differences between benign and malignant nodes [9,13].This observation might seem counterintuitive given that most metastatic CLNs show a higher density of blood vessels [14] and, depending on their size, are also characterized by the presence of necrosis [15].Thus, this study aimed to evaluate whether the qualitative assessment of mpUS images, which already were assessed quantitively [9], can improve diagnostic performance in the differentiation of benign and malignant CLNs.

Patient Selection
In this single-center study, we prospectively screened 105 eligible patients referred by the otorhinolaryngological department's outpatient clinic between August 2019 and September 2021 using the following inclusion criteria: (a) age 18 or older, (b) enlarged CLN(s) for more than two weeks, (c) suspicious CLN(s) during oncological aftercare and (d) additional histological confirmation.Histological confirmation was obtained by analyzing samples of biopsy, lymph node dissection, neck dissection or US-guided core needle biopsy.US examinations were performed as part of presurgical workup using a standardized examination protocol.The study was approved by the institutional ethics committee of Charité-Universitätsmedizin Berlin (protocol code EA1/087/19, date of approval: 2 May 2019), and written informed consent was obtained from all participants according to the Declaration of Helsinki.

Imaging Protocol
All US examinations were performed by a trained otolaryngologist or an experienced radiologist using a single high-end US system with a 4-10 MHz multifrequency lineararray transducer and a center frequency of 7 Mhz (Acuson Sequoia, Siemens Healthineers, Erlangen, Germany).B-mode US of CLNs was performed to assess size, shape, echogenicity and homogeneity.CCDS was used to determine the vascularization of the nodes.SWE was performed using the 2D SWE approach.
Shear waves are generated by the ultrasound probe by using an acoustic radiation force pulse sequence.These waves spread vertically to the ultrasound beam, which causes tissue displacement.The velocity distribution of each shear wave at each pixel of the image is measured by the computer and relates to the shear modulus, which is known Cancers 2023, 15, 5035 3 of 11 as the absolute measure of the elastic properties of tissue.The measured velocities are coregistered with the B-mode image and are depicted in a color-coded SWE-map [16].Five consecutive images were obtained, and for the purpose of the quantitative assessment which was presented in a previous study [9], a circular region of interest (ROI) adapted to the CLN size was placed in dual-image mode for assessment of the whole CLN (B-mode image left, color-coded SWE map on the right, ROI for measurement visualized in both images).The color-coded SWE map represents CLN stiffness from soft to hard in blue, green, yellow and red as illustrated in Figure 1.To optimize visualization the standardized penetration depth was adapted to each patient while gain was kept steady in order not to influence SWE measurement.
tered with the B-mode image and are depicted in a color-coded SWE-map [16].Five consecutive images were obtained, and for the purpose of the quantitative assessment which was presented in a previous study [9], a circular region of interest (ROI) adapted to the CLN size was placed in dual-image mode for assessment of the whole CLN (B-mode image left, color-coded SWE map on the right, ROI for measurement visualized in both images).The color-coded SWE map represents CLN stiffness from soft to hard in blue, green, yellow and red as illustrated in Figure 1.To optimize visualization the standardized penetration depth was adapted to each patient while gain was kept steady in order not to influence SWE measurement.
The same high-end system was used for CEUS examinations.We used a state-of-theart CEUS-specific protocol and a low mechanical index (<0.1) to avoid early microbubble destruction.A 90 s cine loop was recorded while a bolus of 1.2 mL of a second-generation ultrasound contrast agent (SonoVue ® , Bracco Imaging, Milan, Italy) was injected intravenously followed by a 10 mL flush of 0.9% saline solution.The contrast agent consists of microbubbles filled with sulfur hexafluoride and is eliminated through the lungs within 10-15 min.It causes almost no adverse events (0.088%), with headaches or nausea representing the most recent mild adverse events (0.047%) [17].The same high-end system was used for CEUS examinations.We used a state-of-theart CEUS-specific protocol and a low mechanical index (<0.1) to avoid early microbubble destruction.A 90 s cine loop was recorded while a bolus of 1.2 mL of a second-generation ultrasound contrast agent (SonoVue ® , Bracco Imaging, Milan, Italy) was injected intravenously followed by a 10 mL flush of 0.9% saline solution.The contrast agent consists of microbubbles filled with sulfur hexafluoride and is eliminated through the lungs within 10-15 min.It causes almost no adverse events (0.088%), with headaches or nausea representing the most recent mild adverse events (0.047%) [17].

Image Interpretation
All B-mode US images, CCDS and CEUS loops were retrospectively evaluated in consensus by an experienced radiologist and an experienced otolaryngologist (DEGUM (Deutsche Gesellschaft für Ultraschall in der Medizin, German Society of Ultrasound in Medicine and Biology) Level II in head and neck US).B-mode US features were homogeneity, CLN shape and borders [18].Inhomogeneous and irregular-sized and -shaped CLNs were rated as malignant.CCDS findings were classified as central, peripheral or no hypervascularization; CLNs showing a hilus sign were rated as benign [19].SWE was rated by interpreting the color-coded-SWE-map, where blue and light green were rated as benign and dark green, yellow and red were rated as malignant CLN.Please note that the ROI that was placed in every CLN was only used to assess the quantitative SWE-measures which are presented in [9]; these results are not part of this study.CEUS features included the contrast enhancement pattern and distribution homogeneity (centrifugal or centripetal pattern, general hypo-or hyperenhancement) and presence of avascular areas, which were rated as necrosis, based on the malignancy criteria described by Xiang et al. [20].CLNs were rated by two readers and indicated malignancy when showing an irregular centripetal distribution pattern or necrosis [21,22].MpUS (combination of all four parameters) was rated as malignant if one application showed a morphologic pattern of malignancy.

Statistical Analysis
Ultrasound findings and histopathological reports of each patient were analyzed retrospectively for diagnostic performance by testing sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) as well as accuracy using cross tables.A two-sided significance level of α = 0.05 was defined appropriate to indicate statistical significance.Relationships between categorical variables were tested using Pearson's chi-square test.Group differences (malignant and benign histopathological result) in perfusion patterns or presence of necrosis were calculated using the Mann-Whitney U-test.
A univariate logistic regression of individual multiparametric ultrasound parameters was performed to assess the parameter's association with predicting malignancy in CLN.All four parameters served as covariates in the multivariate logistic regression.All statistical analyses were performed using the SPSS software (IBM Corp. Released 2016.IBM SPSS Statistics for Windows, Version 27.0.Armonk, NY, USA: IBM Corp.).
The combination of all four parameters (B-mode US, CCDS, SWE and CEUS combined) resulted in a sensitivity of 97%, specificity of 36%, PPV of 73%, NPV of 88% (χ 2 (1) = 18.072, p < 0.001) and an accuracy of 83%.Details of the results are listed in Table 3. CEUS was the only mpUS parameter showing a significant association with the prediction of CLN in the multivariate logistic regression (p < 0.001).For further details, see Table 4.

Discussion
The results of our study show that complementing US examinations of CLNs by means of mpUS improves their diagnostic performance in qualitatively evaluating for the presence of metastatic involvement.The combination of SWE and CEUS as added modalities shows higher performance for all calculated parameters compared to the commonly used B-mode US and CCDS.In particular, CEUS-derived information on vascularization can help to differentiate between benign and malignant nodes.An example is presented in Figure 2.

Discussion
The results of our study show that complementing US examinations of CLNs by means of mpUS improves their diagnostic performance in qualitatively evaluating for the presence of metastatic involvement.The combination of SWE and CEUS as added modalities shows higher performance for all calculated parameters compared to the commonly used B-mode US and CCDS.In particular, CEUS-derived information on vascularization can help to differentiate between benign and malignant nodes.An example is presented in Figure 2.  C,D).This is a centripetal perfusion pattern with subsequent short hyperenhancement in E and diffuse washout (F).These criteria suggest malignancy.
CCDS and CEUS, which visualize vascular patterns, showed the highest sensitivity at 87% and 93%, respectively, and high PPV (82% and 91%) and accuracy (79% and 90%).This supports our hypothesis that altered vascularization resulting from changes in vessel density in malignant CLNs [14] can be visualized after administration of an ultrasound contrast agent, thus supporting the detection of malignant CLNs.Vascularization patterns differed significantly between malignant and benign CLNs.Benign CLNs were identified by showing a centrifugal enhancement pattern, which is characterized by a hilus sign, from where the contrast agent spread throughout the node, ultimately resulting in homogeneous enhancement.Smaller malignant CLNs showed a centripetal pattern in conjunction with hyperenhancement.This is in line with histopathological findings showing higher blood vessel density in metastatic CLNs within the scope of tumor neoangiogenesis than in reactively enlarged CLNs [14,23].Tumor neoangiogenesis is characterized by an irregular vessel distribution throughout the node.In larger metastatic CLNs, we additionally observed non-enhancing areas, suggesting necrosis surrounded by an enhancing rim.During neoangiogenesis, vessels grow irregularly, which leads to an avascular center with rim enhancement [24].This can be explained by the fact that tumor necrosis is not  C,D).This is a centripetal perfusion pattern with subsequent short hyperenhancement in (E) and diffuse washout (F).These criteria suggest malignancy.
CCDS and CEUS, which visualize vascular patterns, showed the highest sensitivity at 87% and 93%, respectively, and high PPV (82% and 91%) and accuracy (79% and 90%).This supports our hypothesis that altered vascularization resulting from changes in vessel density in malignant CLNs [14] can be visualized after administration of an ultrasound contrast agent, thus supporting the detection of malignant CLNs.Vascularization patterns differed significantly between malignant and benign CLNs.Benign CLNs were identified by showing a centrifugal enhancement pattern, which is characterized by a hilus sign, from where the contrast agent spread throughout the node, ultimately resulting in homogeneous enhancement.Smaller malignant CLNs showed a centripetal pattern in conjunction with hyperenhancement.This is in line with histopathological findings showing higher blood vessel density in metastatic CLNs within the scope of tumor neoangiogenesis than in reactively enlarged CLNs [14,23].Tumor neoangiogenesis is characterized by an irregular vessel distribution throughout the node.In larger metastatic CLNs, we additionally observed non-enhancing areas, suggesting necrosis surrounded by an enhancing rim.During neoangiogenesis, vessels grow irregularly, which leads to an avascular center with rim enhancement [24].This can be explained by the fact that tumor necrosis is not only associated with angiogenesis but also with hypoxia and inflammation [25].One characteristic sign of inflammatory processes is rim enhancement, which is defined as a hypovascular area surrounded by a hypervascular border [26], for example in abscesses [27].The differentiation of infectious from malignant processes as the source of this imaging phenomenon has to be made clinically (i.e., clinical presentation, blood tests) [24].Rim enhancement in nodal metastasis is often seen in magnetic resonance imaging or computed tomography [28], but it is also visible in CEUS.A recent study has shown that rim enhancement in CEUS of axillary lymph nodes is associated with breast cancer metastasis [29].These irregular vascular patterns highlight the value of additional qualitative assessment by the investigator, since quantitative time-intensity curve analyses performed by the perfusion software are not suitable in CLNs with larger areas of central necrosis (showing no dynamic perfusion in necrotic areas).This may explain why quantitative CEUS assessment showed no statistically significant difference between benign and malignant CLN in previous studies [9] and the discrepancy in our current study addressing qualitative assessment.In general, necrotic areas may be easier to detect through dynamic real-time CEUS examination without the need for quantitative assessment.
The transformation from benign to HNSCC-metastatic lymph nodes is not only associated with an increase in vascularization, but also leads to higher stiffness due to tumor cell infiltration [30,31].This is corroborated by both our earlier quantitative SWE measurement, which showed higher shear wave velocity, and therefore higher stiffness in malignant CLNs compared to benign ones [9], and the qualitative evaluation of SWE in the current study.SWE achieved the highest diagnostic performance in terms of specificity (90% versus 46% for conventional B-mode US).Furthermore, SWE is the US technique with the highest PPV, followed by CEUS.During the US examination, the system displays SWE results in real time in a color-coded map of the distribution of tissue stiffness values in the ROI of the target area.Such maps allow easy and fast visual evaluation, making US-based SWE a feasible tool in the evaluation of inconclusive CLNs.One limitation of the examination is that there is no consensus whether to place one large ROI covering as much of the CLN as possible or if it is of advantage to use multiple smaller ROIs within the target lymph node.Further studies and a standardized training program are needed to establish a commonly used evaluation system which not only makes the comparison of study results easier but also ensures higher quality in US diagnostics [32].
Finally, it is important to highlight, that US examinations consist of the combined use of different modalities [33].Our results do not only show the diagnostic value of SWE and CEUS as single but also as combined modalities.The addition of SWE to standard US (B-mode and CCDS) increased the accuracy from 73 to 76%, while adding CEUS to standard US increased the accuracy to 77%.Adding both new parameters to standard US led to an increase in diagnostic accuracy by 10%.
Overall, our results suggest a benefit of mpUS in the assessment of inconclusive CLNs, which can not only be used in initial diagnostic workup but also within followup proceedings after treatment.The high specificity of SWE suggests its usefulness in confirming benign CLNs, while the high sensitivity of CEUS makes it a useful parameter to further evaluate CLNs suggestive of malignancy.While the quantitative analysis of mpUS is time-consuming and, for CEUS, requires dedicated postprocessing software, a qualitative interpretation of findings is made by the operator during the US examination.SWE is a noninvasive and fast-to-perform (approximately 5 s per image) diagnostic tool requiring merely a specific ultrasound probe, while CEUS involves intravenous contrast agent administration, which necessitates prior preparation such as the placement of an intravenous line and an assistant who administers the agent and a saline flush.We therefore suggest the implementation of SWE as part of routine head-and-neck US and to complement the examination by CEUS in case of inconclusive SWE findings.

Conclusions
Measures of the diagnostic performance of the qualitative US assessment in CLNs showed that mpUS exceeds standard US techniques, driven especially by the high accuracy of mpUS.Moreover, mpUS is fast and easy to perform, underlining its potential for routine use on top of B-mode US and CCDS in the differentiation between benign and malignant CLNs, and might therefore aid the decision to choose between the watch-and-scan strategy and surgery in primary cases as well as between repeated surgery and short-term follow-up in oncological patients.

Figure 1 .
Figure 1.Representative example of ultrasound image evaluation.(A): B-mode US, CLNs were evaluated in two planes for their morphological appearance (size, shape, and borders; 1-3: diameter in each plane) (B): SWE, left side with B-mode US image and color bar (blue indicating soft and red indicating hard tissue), right side showing result of SWE assessment within the ROI (yellow circle marked with "1").(C): Contrast inflow of a CLN at 15 s of a 90 s CEUS cine loop.The video was evaluated for perfusion pattern, distribution homogeneity and presence of necrosis.

Figure 1 .
Figure 1.Representative example of ultrasound image evaluation.(A): B-mode US, CLNs were evaluated in two planes for their morphological appearance (size, shape, and borders; 1-3: diameter in each plane) (B): SWE, left side with B-mode US image and color bar (blue indicating soft and red indicating hard tissue), right side showing result of SWE assessment within the ROI (yellow circle marked with "1").(C): Contrast inflow of a CLN at 15 s of a 90 s CEUS cine loop.The video was evaluated for perfusion pattern, distribution homogeneity and presence of necrosis.

Figure 2 .
Figure 2. Metastatic CLN from HNSCC, only detected by CEUS and not by B-mode US or SWE.(A): B-mode US: oval shape, regular borders, and no criteria for malignancy.(B): SWE: B-mode US on the left; color-coded map of SWE on the right.Color bar on the left side with blue representing low and red representing high stiffness.Tissue in the ROI on the right side is blue and indicates soft tissue; therefore, there is no sign of malignancy.(C-F): CEUS 18, 20, 27 and 34 s after contrast agent administration.Contrast agent does not enter the node via a hilus but from the periphery (C,D).This is a centripetal perfusion pattern with subsequent short hyperenhancement in E and diffuse washout (F).These criteria suggest malignancy.

Figure 2 .
Figure 2. Metastatic CLN from HNSCC, only detected by CEUS and not by B-mode US or SWE.(A): B-mode US: oval shape, regular borders, and no criteria for malignancy.(B): SWE: B-mode US on the left; color-coded map of SWE on the right.Color bar on the left side with blue representing low and red representing high stiffness.Tissue in the ROI on the right side is blue and indicates soft tissue; therefore, there is no sign of malignancy.(C-F): CEUS 18, 20, 27 and 34 s after contrast agent administration.Contrast agent does not enter the node via a hilus but from the periphery (C,D).This is a centripetal perfusion pattern with subsequent short hyperenhancement in (E) and diffuse washout (F).These criteria suggest malignancy.

Table 1 .
Study Population.Demographic data, CLN status, and primary cancer entities.Continuous variables are given as mean (SD), categorical variables are given as absolute/total numbers (n/N) and percentages in brackets.Abbreviations: HNSCC = head and neck squamous cell carcinoma.

Table 3 .
Diagnostic performance of the qualitative assessment of different US techniques and parameters.Sensitivity, specificity, PPV, NPV and accuracy are given in percentages with corresponding p-values and χ 2 (degrees of freedom stated in brackets).