Conventional ultrasound combined with contrast-enhanced ultrasound quantitative analysis for diagnosing lung cancer metastases in neck lymph nodes

1. Introduction

Lung cancer remains one of the most common malignant tumors worldwide. It is ranked first in cancer incidence and mortality in men and is also at the forefront in cancer incidence and mortality in women [1, 2]. However, due to its deep location and indolent course, some young patients might have already developed lymph node metastasis by the time of diagnosis [3]. The first symptom of lung cancer is supraclavicular lymph node metastasis due to cervical lymph node drainage. Thus, identifying cervical lymph node metastases, especially in supraclavicular lymph nodes, is important to accurately stage lung cancer, evaluate patients’ prognosis and select appropriate treatments [4].

There are many methods for detecting cervical metastatic lymph nodes in clinics, including palpation, 2D ultrasound, Color Doppler Flow Imaging (CDFI), trans intravenous contrast ultrasound imaging, Computer Tomography (CT), Magnetic Resonance Imaging (MRI), etc. Among all imaging techniques, conventional ultrasound is more convenient as it can show the morphological characteristics obtained that are not easily obtained by CT and MRI, but it is still difficult to accurately distinguish benign and malignant lymph nodes [5]. Conventional ultrasound can be used to observe the size, shape, boundary, and hilar structure of lymph nodes. Color Doppler flow imaging can be used to preliminarily observe the blood flow signals inside and around lymph nodes. As a new qualitative ultrasound diagnostic technique, contrast-enhanced ultrasound has been widely used in clinical practice. There are relatively mature research results in the qualitative analysis of abdominal organs, superficial organs, lymph nodes, and so on. For the differential diagnosis of benign and malignant cervical lymph nodes, certain reference information can be provided for clinical differential diagnosis by observing the contrast agent perfusion method of the target lymph node, whether the perfusion is uniform, compared with surrounding tissues, the relative intensity of the target lymph node perfusion, the speed of perfusion and clearance, and whether there is a perfusion defect area [6]. The trans-intravenous injection of ultrasound contrast agent microbubbles is commonly used in clinics. This contrast agent enters systemic circulation through the pulmonary capillaries to reach the lesion via systemic blood flow and produces a large amount of backscattering in the ultrasonic acoustic field to enhance the contrast between blood flow within microvessels and tissue and observe tissue microcirculation perfusion in real time. Contrast-enhanced ultrasound (CEUS) is more sensitive than color Doppler flow imaging for detecting low-speed microvascular blood flow [7]. The quantitative analysis of ultrasound contrast uses software to analyze and process the region of interest (ROI) during ultrasound contrast to form the time-intensity curve (TIC), obtain specific contrast quantitative imaging parameters and provide a valuable objective reference for clinical differential diagnosis.

2. Methods

2.4 Quantitative analysis of contrast ultrasound

Digital Imaging and Communications in Medicine (DICOM) data were exported using the Vuebox CEUS quantitative analysis software (Bracco Suisse SA-Software Applications, 31 route de la Galaise, CH-1228 Plan-les-Ouates, Switzerland). The software preset automatically optimized the ultrasound instrument model and superficial high-frequency probe, and lymph nodes were delineated as ROI. An attempt was made to have the ROI area cover the boundary of the target lymph node and compensate for motions. A TIC was prepared, and the analysis report was exported before selecting the peak enhancements (PE), rise time (RT) and fall time (FT) for statistical analysis (Fig. 1).

Fig. 1.Vuebox analysis interface for metastatic lymph nodes in the neck of lung cancer.

3. Results

3.1 Pathological results

An ultrasound-guided crude needle puncture histological biopsy was performed for final pathological analysis, which identified 26 cases of lung cancer cervical lymph node metastasis (Fig. 2) and 13 cases of benign reactive hyperplasia of cervical lymph nodes (Fig. 3, Table 2). And divided into metastatic group and benign group according to pathological results.

Fig. 2.Pathological image of cervical lymph node metastasis of lung cancer. HE: haematoxylin and eosin; PCK: pan cytokeratin; TTF-1: thyroid transcription factor-l; CK7: cytokeratin 7; EGFR: epidermal growth factor receptor; P63: recombinant tumor protein P63.

Fig. 3.Pathological image of cervical benign reactive hyperplasia lymph nodes. HE: haematoxylin and eosin; PCK: pan cytokeratin; TTF-1: thyroid transcription factor-l; CK19: cytokeratin 19.

3.2 Conventional ultrasound results

Of the 26 cervical lymph nodes with lung cancer metastasis, 22 had a full morphology, while 4 were irregularly shaped. 8 of the 13 cases of benign reactive hyperplasia of the cervical lymph nodes showed morphological fullness (Fig. 4), while the remaining 5 showed a normal morphology. The three morphologies (full, normal, irregular) significantly differed between the two groups (p < 0.001) (Table 3).

Moreover, the lymphatic hilum structure was indistinct in 24 of the 26 cervical lymph nodes containing lung cancer metastasis. The blood flow was mixed or peripheral type (Fig. 5), and only two cases of hilum of lymph gland exhibited portal-type blood flow. Only three of the 13 cases with benign reactive hyperplasia of the cervical lymph nodes showed unclear hilar structures, presenting with nonhilar blood flow. Comparatively, the remaining 10 cases showed visible hilar blood flow signals. The rate of unclear lymphatic hilum significantly differed between the two groups (p < 0.001), which was found to be higher in the metastasis group than in the benign group (92.30% vs. 23.08%) (Table 3).

Fig. 4.Benign reactive hyperplastic lymph nodes. FR: frame rat; CHI: contrast harmonic imaging; Frq: frequency; Gn: Gain; D: depth; AO: acoustic output; PRF: pulse repetition frequency; L/A: line density/frame average; WF: wall filter; S/P: spatial filter/ packet size; CF: color flow.

Fig. 5.Metastatic lymph nodes of lung cancer. FR: frame rat; CHI: contrast harmonic imaging; Frq: frequency; Gn: Gain; D: depth; AO: acoustic output; PRF: pulse repetition frequency; L/A: line density/frame average; WF: wall filter; S/P: spatial filter/packet size; CF: color flow.

Table 3.Wilcoxon rank sum test for difference analysis.

Index	Metastatic lymph nodes of lung cancer (n = 26)	Benign reactive hyperplastic lymph nodes (n = 13)	Z/c2	p
PE
	Median (Min, Max)	4852.29 (2547.47, 7733.31)	11,464.70 (5470.65, 21,077.96)	4.349	0.037
RT
	Median (Min, Max)	4.93 (4.41, 6.68)	4.54 (4.03, 5.365)	1.350	0.245
FT
	Median (Min, Max)	9.11 (8.15, 14.40)	8.46 (7.095, 9.425)	1.878	0.171
Lymphatic hilum
	unclear	24 (92.30%)	3 (23.08%)	19.500	<0.001
	clear	2 (7.70%)	10 (76.92%)
Form
	full	21 (80.77%)	8 (61.54%)	15.486	<0.001
	normal	0 (0.00%)	5 (38.46%)
	irregular	5 (19.23%)	0 (0.00%)
Note: PE was significantly lower in the metastasis group than in the benign group (4852.29 vs. 11,464.70). The rate of unclear lymphatic hilum was higher in the metastasis group than in the benign group (92.30% vs. 23.08%). PE: peak enhancements; RT: rise time; FT: fall time.

3.4 Diagnostic efficacy of quantitative parameters

PE cutoff value ≤1330.855 units showed good diagnostic results for diagnosing malignant lymph nodes (p = 0.037), while the diagnostic effectiveness of RT and FT were low (p > 0.05). Quantitative CEUS analysis showed that the PE sensitivity, specificity, accuracy and AUC were 84.60%, 7.70%, 91.66% and 0.293, respectively. PE combined with the presence or absence of lymphatic hilus at a cutoff value of 0.902 units showed good diagnostic ability for detecting metastatic lymph nodes (p = 0.004). The sensitivity, specificity, accuracy and AUC of the CEUS quantitative analysis for PE combined with the presence or absence of lymphatic hilus were 73.10%, 76.90%, 48.73% and 0.790, respectively (Fig. 6 and Table 4).

Fig. 6.ROC curve figure. PE: peak enhancements; RT: rise time; FT: fall time; ROC: receiver operating characteristic.

Table 4.ROC curve table.

Index	Cutoff	Sensitivity	Specificity	Accuracy	AUC (95% CI)	p
PE	1330.855	0.846	0.077	0.9166	0.293 (0.116, 0.470)	0.037
RT	4.385	0.808	0.462	0.6362	0.615 (0.425, 0.805)	0.245
FT	8.995	0.538	0.769	0.4116	0.636 (0.450, 0.822)	0.171
PE + lymphatic hilum	0.902	0.731	0.769	0.4873	0.790 (0.648, 0.931)	0.004
PE + lymphatic hilum + form	0.941	0.692	0.462	0.5997	0.467 (0.254, 0.681)	0.743
Note: the area under the ROC curve was greater than 0.5, suggesting that the diagnostic test had certain diagnostic significance. PE cutoff value ≤1330.855 units showed good diagnostic accuracy for diagnosing malignant lymph nodes (p = 0.037). PE combined with the presence or absence of lymphatic hilus at a cutoff value of 0.902 units showed good diagnostic accuracy for detecting metastatic lymph nodes (p = 0.004). ROC: receiver operating characteristic; AUC: area under the curve; CI: Confidence Interval; PE: peak enhancements; RT: rise time; FT: fall time.

4. Discussion

Given the rapid economic development and increase in population aging in an increasing number of countries and regions, controlling the incidence of lung cancer remains challenging as its risks and incidence are known to increase with age. Developments in medical technology have helped the early diagnosis of lung cancer. Specifically, assessing the status of cervical lymph node metastasis is an important step for accurately staging lung cancer because metastasis in this region is closely related to the prognosis of lung cancer and the choice of treatments.

Solely relying on the size of cervical lymph nodes may not accurately differentiate between benign and malignant lymph nodes as they usually depend on the partition of the neck and the patient’s age [9]. Moreover, metastatic lymph nodes usually appear fuller and tend to be round, whereas lymph nodes with benign reactive hyperplasia are usually larger but flatter and rounder [10]. The presence or absence of hilum of lymph gland can also be used as identification criteria for benign and malignant cervical lymph nodes [11]. Although the above differential diagnostic characteristics of two-dimensional ultrasound are not absolute, they have considerable diagnostic value. Using multicenter imaging omics, some scholars developed an ultrasonic imaging omics model based on cervical lymph nodes using two-dimensional ultrasound images to improve the differential diagnostic efficiency of cervical lymph node tuberculosis, cervical lymphoma, cervical lymph node reactive hyperplasia and metastatic lymph nodes [12]. Benign reactive lymph nodes have a normal lymph node morphology and vascular anatomy with a normal hilar structure, while metastatic lymph nodes have neovascularization that penetrates the capsule. Conventional color Doppler ultrasound can analyze the flow and perfusion pattern of lymph nodes, and the injection of an ultrasound contrast agent can significantly improve the sensitivity and specificity of malignant lymph node neovascularization [13]. It is reported that its resolution may even surpass that of CT and MRI [14], indicating ultrasound imaging to be of great value for the follow-up of malignant tumors after radiotherapy and chemotherapy and to evaluate the presence of recurrence or lymph node metastases [15].

The qualitative diagnosis obtained visually by observing the ultrasound contrast process might be too subjective. TIC obtained by processing and analyzing DICOM data with quantitative analysis software, such as Vuebox, can provide an objective basis. At present, quantitative analyses in clinics using PE and AUC can reflect the contrast concentration of contrast perfusion, while RT, arrival time (AMT) and FT can reflect the efficiency of contrast perfusion and regression to provide valuable differential diagnosis information for different types of pathological lesions [16].

Ying et al. [10] used arrival time parametric imaging technology to analyze the arrival time of contrast agents around the lymph nodes, which can provide more information for the differential diagnosis of superficial lymph nodes [17]. Nie et al. [18] quantitatively analyzed the TIC, PE, RT and AUC for three parameters and combined this information with observations of lymph node ultrasound images in contrast perfusion mode and enhancement mode to improve the differential diagnostic accuracy of head and neck lymph node status [18].

This study collected the conventional two-dimensional ultrasound imaging features of the cervical lymph nodes of lung cancer patients mainly including the enlargement of lymph node morphology and the absence of lymph node hilum. The corresponding node ultrasound contrast DICOM data were used to generate TIC with time as the abscissa and the contrast perfusion intensity as the ordinate. This curve extracted accurate quantitative data, including PE, RT and FT. RT and FT can objectively reflect the node contrast perfusion and regression rates such as fast forward and fast backward, slow forward and slow backward, fast forward and slow backward, etc. Sonja Schwarz et al. [19] used contrast ultrasound TIC analysis of the liver to show that the RT in the TIC time index could help distinguish between benign and malignant liver lesions. Further, Lu Qing et al. [20] found that differences in the degrees of differentiation of liver cells and metastatic liver cancer differed in the contrast ultrasound TIC of multiple time indices.

Cervical metastatic lymph nodes of lung cancer show neovascularization of the lymph node capsule, which changes the microcirculation perfusion in the lymph nodes similar to contrast enhanced ultrasound differences between liver space occupying lesions and normal liver tissues, maybe in the time index of TIC, appeared some differences from reactive proliferative lymph nodes. Thus, the RT and FT time index were selected in this study, but our final results did not show any significant difference in these two between the two groups of lymph nodes. A possible explanation for the difference in PE between the two groups could be that enlarged lymph nodes due to lung cancer metastases may exhibit partial necrosis, partial contrast, low contrast or even a lack of contrast perfusion. Consequently, when the software analyzed the contrast agent, the overall perfusion peak intensity was shown as low and benign reactive hyperplastic lymph nodes may not show necrosis. Software analysis of the ROI of the contrast agent showed that the overall perfusion PE value was higher in the hyperplasia group than in the metastasis group [21]. Further, the ROC analysis also indicated that PE had good diagnostic effects (sensitivity 84.6%, accuracy 91.66%, p = 0.037), while its specificity remained very low (7.70%). The absence of the lymphatic hilus is one of the most common malignant criteria for conventional type B ultrasound [11]. Color Doppler blood flow imaging and ultrasound contrast imaging can be more accurate and intuitive methods to diagnose lymphatic gate structures and the presence of portal-type blood flow. In this present study, the morphology of cervical lymph nodes was recorded by visual observation. However, the specific division of cervical lymph nodes and patients’ age varied. In addition, some of the enlarged reactive hyperactive lymph nodes also exhibited relatively full image characteristics, making the image features more subjective compared with other collected parameters and imaging features. Based on the ROC assessment, combining the peak intensity of CEUS quantitative analysis with the presence or absence of the lymphatic gate structure demonstrated a good diagnostic effect for detecting metastatic lymph nodes in the neck of lung cancer patients (sensitivity 73.10%, specificity 76.90%, accuracy 48.73%, p = 0.004). Although adding the characteristics of lymph node morphology slightly increased the diagnostic accuracy (59.97%), the specificity remained low (46.20%).

Due to the regional and time limit of this study, the number of samples collected was small, leading to insufficient cervical metastatic lymph node cases with different pathological types, such as lung adenocarcinoma, squamous cell carcinoma and small cell lung cancer, for statistical analysis and comparison between groups and might have affected the final statistical results. In addition, this study did not have any limitations on the size of target lymph nodes. During the actual inspection process, it was challenging to use conventional ultrasound to identify the presence of lymphatic hilum in smaller lymph nodes, which might be resolved by increasing the sample size and using multicenter settings to analyze conventional ultrasound images via artificial intelligence imageomics technology to avoid the impact of human subjective factors.

5. Conclusions

In conclusion, conventional ultrasound observations of lymph node structures combined with quantitative analysis of the peak intensity of contrast-enhanced ultrasound have important values in diagnosing metastatic lymph nodes in lung cancer. Specifically, this strategy can help evaluate the stage, prognosis and choice of treatment options for lung cancer. However, this study remained subject to considerable limitations due to the number of samples, regional limitations and human factors, urging the need for further research to confirm these findings.

Availability of Data and Materials

All data generated or analyzed during this study are included in this published article.

Author contributions

ARW and CL—Conception and design; CL—Administrative support; MZ and SFX—Provision of study materials or patients; ARW—Collection and assembly of data, Data analysis and interpretation; All authors —Manuscript writing and Final approval of manuscript.

Ethics approval and consent to participate

The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013). The study was approved by the Ethics Committee of Jingzhou First People’s Hospital (NO.: KY202226), and individual consent for this retrospective analysis was waived.

Acknowledgment

We thank Bracco Imaging Medical Technologies (Shanghai) Co., Ltd. for the help of Vuebox Software.

Funding

This research received no external funding.

Conflict of interest

The authors declare no conflict of interest.