Endobronchial Ultrasound Combined with Clinical Data for Predicting Malignant Peripheral Pulmonary Lesions

Figures & data

Table 1 Clinical and Histopathologic Data of Patients in the modelling group (Mean±Standard Deviation)

Variables	Number
Sex
Male	84
Female	51
Mean age, yr	55.5±16.0
Tumor size, cm	36.5±19.9
Total EBUS duration, min	4.3±1.9
Neoplasm	77
Adenocarcinoma	42
Squamous cell carcinoma	24
Small cell carcinoma	5
Adenosquamous carcinoma	1
Bronchioalveolar carcinoma	1
Non-small cell cancer	2
Others	2
Non-neoplasm	58
Pneumonia	16
Pulmonary tuberculosis	11
Lung abscess	7
Inflammatory pseudotumor	3
Pulmonary fibrosis and infection	4
Chondromatous hamartoma of lung	3
Fungus infection	5
Pulmonary cyst	3
Others	6

Abbreviation: EBUS, endobronchial ultrasonography.

Table 2 Clinical and Histopathologic Data of Patients in the validation group (Mean±Standard Deviation)

Variables	Number
Sex
Male	31
Female	19
Mean age, yr	61.2±18.0
Tumor size, cm	33.20±16.55
Total EBUS duration, min	4.1±1.8
Carcinoembryonic antigen	14.7±8.09
Neoplasm	34
Adenocarcinoma	20
Squamous cell carcinoma	9
Small cell carcinoma	1
Infiltrating adenocarcinoma	2
Micro-infiltrating adenocarcinoma	1
Adenocarcinoma in situ	1
Non-neoplasm	16
Pneumonia	9
Pulmonary tuberculosis	2
Lung abscess	3
Aspergilloma	2

Abbreviation: EBUS, endobronchial ultrasonography.

Figure 1 A 50-year-old man had a peripheral pulmonary lesion in the right lung. (A) A round lesion (arrow) could be seen in the right middle lobe with a size of 9 mm×9 mm. (B) Endobronchial ultrasonography revealed a heterogeneous lesion with hyper- and hypoechoic areas. (C) Tumor and normal cells existed in the same area with necrosis and fibrosis in the middle (HE ×100).

Figure 2 A 65-year-old woman had a peripheral pulmonary lesion (inflammatory pseudotumor) in the lung back segment. (A) A lesion of 30 mm×28 mm (arrow) could be seen in the back segment with an air bronchogram. (B) Endobronchial ultrasound could detect a heterogeneous lesion with diffuse short signals. (C) A few distorted and expanded bronchia existed in the inflammatory cells and normal cells (HE ×100).

Table 3 Comparisons of Clinical Factors and EBUS Image Patterns in Predicting Malignancy in Patients

Variables	Neoplasm	Non-Neoplasm
Age
<50	59	51
≥50	18	7
Current or former smoker
Yes	52	20
No	25	38
Cancer history
Yes	5	1
No	72	57
Serum CEA	59	51
<15	15	27
≥15	62	31
Lesion configuration
Circular or oval	38	26
Irregular	39	32
Continuous hyperechoic margin
Yes	21	12
No	56	46
Distinct borderline
Yes	45	18
No	32	40
Echo density
Homogenicity	19	36
Heterogenicity	58	22
Anechoic areas
Yes	21	4
No	56	54
Air bronchogram
Yes	25	38
No	30	4
Concentric circle
Yes	8	18
No	69	40
Internal vessels
Yes	41	35
No	36	23

Abbreviations: EBUS, endobronchial ultrasonography; CEA, carcinoembryonic antigen.

Table 4 Logistic Model for Predicting Peripheral Lung Cancer Using EBUS and Clinical Features

Variables	Regression Coefficient (B)	P	Dominance Ratio	95% Confidence Interval
Constant	−2.986	0.000	0.051
CEA	2.293	0.000	9.908	2.96–33.09
Smoking history	1.993	0.000	7.338	2.42–22.21
Distinct borderline	1.552	0.003	4.723	1.71–13.05
Heterogeneous echo	1.616	0.002	5.033	1.82–13.93
No air bronchogram	−2.011	0.000	7.463	2.56–20.00
Anechoic areas	1.748	0.029	5.745	1.20–27.54

Abbreviations: CEA, carcinoembryonic antigen; EBUS, endobronchial ultrasonography.

Table 5 Testing of the Logistic Equation with Pathological Results in 50 Patients

Logistic Equation	Pathological Results		Total
	Malignant	Benign
Malignant	30	12	42
Benign	4	4	8
Total	34	16	50

Figure 3 Receiver operating characteristic curve for the direction model for diagnosis of malignant peripheral pulmonary lesions (PPLs). The prediction model was developed based on data from 135 consecutive patients.