Operative ultrasonography for lung cancer surgery

J

THoRAc CARDIOVASC SURG

1989;98:540-5

Operative ultrasonography for lung cancer surgery During 60 operations for lung cancer, high-resolution (7.5 MHz) operative ultrasonography was performed to evaluate direct cardiovascular invasion of tumor (24 operations), lymph node metastasis (30 operations), and liver metastasis (13 operations). Immediately after thoracotomy or sternotomy but before tissue dissection, operative scanning enabled delineation and evaluation of the vessels and heart (atrium) behind or within the tumor and detection of regional lymph nodes. The accuracy of operative ultrasound in diagnosing the presence or the extent of cardiovascular invasion was 91.7% (22 of 24 operations), which was significantly (p < 0.02) higher than preoperative studies (62.5%), including computed tomography and angiography. Of the 24 surgical procedures, 23 were consistent with operations proposed on the basis of operative ultrasound findings, whereas only 16 were consistent with preoperatively proposed operations (p < 0.01). Operative ultrasound provided the capability of depicting lymph nodes as small as 3 mm. More lymph nodes (8.0 ± 1.84 nodes per cancer) were detected with operative ultrasound than with computed tomography (4.8 ± 1.56 nodes) (p < 0.(01); however, the sensitivity and specificity of operative ultrasound in determining lymph node metastasis were 82.4 % and 67.3%, respectively. No liver metastasis was identified. The information provided by operative ultrasound regarding cardiovascular invasion and lymph node and liver metastasis was considered helpful in selecting the type of surgical procedure and in avoiding unnecessary tissue dissection.

Junji Machi, MD, PhD, Ryozo Hayashida, MD, Toshihiko Kurohiji, MD, Yutaka Nishimura, MD, Shinzo Edakuni, MD, Yuichi Yamashita, MD, Jinryo Takeda, MD, Teruo Kakegawa, MD, Kurume, Japan, and Bernard Sigel, MD, Philadelphia. Pa.

In the treatment of lung cancer, diagnosis has improved with advances in preoperative imaging methods, including chest roentgenography, bronchography, angiography, and, more recently, computed tomography (CT) and magnetic resonance imaging (MRI). However, the degree of local tumor invasion or lymph node metastasis cannot always be diagnosed correctly before operation. In such instances, the final decision in selecting the type of surgical procedure, such as lobectomy or pneumonectomy with or without lymph node dissection, is usually made on the basis of inspection and palpation during surgical exploration, and this often necessitates extensive tisFrom the First Department of Surgery, Kurume University School of Medicine, Kurume, Japan, and the Department of Surgery. Medical College of Pennsylvania, Philadelphia, Pa. Received for publication April 21, 1988. Accepted for publication Jan. 31. 1989. Address for reprints: Junji Machi. MD, PhD. Department of Surgery. Medical College of Pennsylvania. 3300 HenryAve.. Philadelphia. PA 19129. 12/]/11255

540

sue dissection during thoracotomy. To improve the diagnostic accuracy of evaluating lung cancer extension and determining cancer staging before such tissue dissection during operations, we have evaluated high-resolution operative ultrasonography, which has already been used in various surgical fields. 1-4

Patients and methods From May 1985 until December 1987, operative ultrasonography was used during 60 lung cancer operations, 34.9% of a total of 172 lung cancer operations performed during this period in our institute. Informed consent for this study was obtained from each patient. The three purposes for operative ultrasonic examination in this study were to evaluate the cardiovascular invasion of lung cancer, lymph node metastasis, and liver metastasis. Operative ultrasound was used for one of these purposes in 53 operations and for two of these purposes in sevenoperations. Cardiovascular assessment was performed in 24 operations (14.0% of the total operations) when preoperative imaging studies indicated the possibility of tumor invasion to the vessels or the heart. or when the result of imaging studies was equivocal in diagnosing such tumor invasion. All 24 patients had both CT and angiography, and in addition MRI was used in 10 pa-

Volume 98 Number 4 October 1989

tients. The operative ultrasound findings were compared with these preoperative tests, based on surgical exploration and histologic results. Lymph nodes were assessed on a random basis in 30 operations (17.4% of the total operations) when preoperative CT displayed at least one enlarged lymph node at the pulmonary hilum or the mediastinum, and the numbers of lymph nodes depicted by CT and ultrasound were compared. The size of each lymph node was also ultrasonically estimated, and qualitative assessment of the presence or absence of metastasis was made by comparing the operative ultrasound findings with the histologic results of the removed lymph nodes. Operative ultrasound of the liver was performed in 13 operations (7.6(fc of the total operations), because preoperative studies by CT and percutaneous ultrasound were indeterminate in diagnosing or excluding liver metastasis. Operative ultrasonography was performed with a high-resolution, real-time, B-mode instrument (Aloka SSD-330, Aloka Co. Ltd., Tokyo, Japan), which used electronic linear-array and sector transducers. In the majority of operations, T- or l-shaped linear-array probes were used, and sector probes were used only when needed. Scanning of intrathoracic structures such as the cardiovascular system and lymph nodes was performed with 7.5 MHz transducers; 5.0 MHz transducers were added for liver scanning so that the ultrasound beam could penetrate more deeply. Operative scanning was done immediately after lateral thoracotomy or median sternotomy. Pleural adhesions when present were lysed. No other surgical tissue dissection was needed. After deflation of the lung, the probe, sterilized by an ethylene oxide gas, was placed over the thoracic tissue. For appropriate acoustic coupling, saline solution was introduced into the thoracic cavity. For evaluation of cancer invasion to the vessels or the heart, the probe was placed on the tumor or the deflated lung tissue so that the cardiovascular system behind or within these tissues could be imaged. For investigation of lymph nodes, the probe was systematically moved from the upper to the lower part of the mediastinum and toward the hilum. For examination of the liver from the thoracic cavity, the probe was positioned on the diaphragm and the liver was visualized through the diaphragm (transdiaphragmatic scanning).

Results The ages of the 60 patients ranged from 32 to 80 years with a mean age of 64.1 years. Forty-two patients were men and 18 were women. Fifty-three patients had histologically proved primary lung cancer and seven patients had metastatic lung cancer. Cardiovascular invasion. Operative ultrasound was performed to evaluate cardiovascular invasion in 24 operations; the lung cancer was central in 21 patients and peripheral in three patients. In the case of peripheral cancer, mediastinal lymph node metastasis rather than a primary tumor was suspected as the possible cause of invasion to the cardiovascular system. All 24 cancers were primary lung carcinoma; histologic diagnoses indicated squamous cell carcinoma in 15 cancers, adenocarcinoma in four, small cell carcinoma in two, large cell carcinoma in two, and carcinoid tumor in one. Twelve cancers were

Operative ultrasonography

54 I

Fig. 1. Operative sonogram of a lung cancer at the right upper lobe. The tumor (T) invaded the right pulmonary artery (PA). as shown by arrowheads. An arrow indicates one upper lobe branch of the pulmonary artery.

situated in the right lung and 12 in the left lung. The operation was performed by right thoracotomy in 10 patients, left thoracotomy in nine, and median sternotomy in five. The relationship of the tumor to the vessels, such as the pulmonary artery and vein, or to the heart could be imaged by operative ultrasonography, even though surgical exploration was impossible or ineffective at the time of scanning. Ultrasound scanning was continued to determine whether direct invasion of the tumor to the pulmonary artery and vein, vena cava, aorta, or the heart was present or not. Abutment of the tumor on the vessels or the heart per se on ultrasound images was not judged to be tumor invasion. On the other hand, when ultrasound revealed that the vessel wall or heart wall was distorted or destroyed by tumor, a diagnosis of cardiovascular invasion was made. Fig. 1 shows invasion of an upper lobe cancer to the pulmonary artery. In 24 lung cancers, preoperative imaging studies and operative ultrasound were used to evaluate possible cardiovascular invasion at the following anatomic locations: the pulmonary trunk and pulmonary artery and its

The Journal of

542

Machi et al.

Thoracic and Cardiovascular Surgery

Table I. Results and accuracy ofpreoperative studies and operative ultrasonography in diagnosing cardiovascular invasion of lung cancer Preoperative studies True diagnosis False diagnosis

15

22

9

2

Accuracy

62.57, (p

Fig. 2. Operative sonogram ofsubaortic lymph nodes scanned through the aorta. In addition toa 13 by 4 mmlymph node (arrowhead), twosmall lymph nodes (arrows). 3 mminsize, which were not recognized preoperatively, were detected byoperative ultrasound. A, Aortic arch; PA. left pulmonary artery.

branches in 20 instances; the pulmonary vein in eight instances; the vena cava in three instances; the aorta in four instances; and the atrium in three instances. The accuracy of operative ultrasonography and preoperative studies in diagnosing the presence or the extent of cardiovascular invasion is shown in Table I. Often, operative ultrasound was used to examine the heart or vessels at more than one site during each operation. In Table I, the judgment of a true diagnosis for preoperative studies or operative ultrasonography was made only when the diagnosis concerning cardiovascular invasion was correct (true positive or true negative) for all sites examined by these methods in each operation. Of 24 operations, the diagnoses based on preoperative studies were true in 15 operations, with an accuracy rate of 62.5%, whereas the diagnoses based on operative ultrasound were true in 22 operations, with an accuracy of 91.7% (p < 0.02 by x2 test).

Operative ultrasonography

< 0.02)

91.7'k

To determine the diagnostic impact of both of these imaging methods in making decisions regarding surgical procedures, we compared operations proposed by preoperative studies and by operative ultrasound with operations actually performed (Table 11). Operations performed included eight pneumonectomies, nine lobectomies, three combined atrial or aortic resections, and four thoracotomies without tumor resection. Of these 24 operations, surgical procedures proposed by preoperative studies were consistent with those actually performed in 16 operations (66.7%). In the remaining eight operations, procedures were changed because of misdiagnosisof cardiovascular invasionby preoperative studies. On the other hand, in 23 of 24 operations (95.8%), surgical procedures proposed by operative ultrasound were exactly the same as those performed. The number of surgical procedures accurately proposed by operative ultrasound was significantly different from the number proposed by preoperative studies (p < 0.01 by X 2 test). Lymph node metastasis. In 30 operations in which lymph node metastasis was evaluated by operative ultrasound, 26 cancers were primary and four cancers were metastatic. Microscopic examination of primary lung cancers indicated squamous cell carcinoma in eight cancers, adenocarcinoma in 14, adenosquamous carcinoma in two, small cell carcinoma in one, and carcinoid tumor in one. Eighteen were right lung cancers and 12 were left lung cancers. Operative ultrasound was able to image both ipsilateral and contralateral mediastinal and pulmonary hilar lymph nodes before surgical tissue dissection. Fig. 2 shows subaortic lymph nodes detected immediately after thoracotomy. Although surgical exploration of these lymph nodes required massive tissue dissection, operative ultrasound examination was accomplished quickly and noninvasively. Operative ultrasound had the capability of depicting lymph nodes as small as 3 mm, as demonstrated in Fig. 2. In addition, because ultrasound scanning of lymph nodescould be done from various positionsand directions in real time, it was possible to obtain three-dimensional

Volume 98 Number 4

Operative ultrasonography

October 1989

543

Table II. Surgical procedures performed and comparison of diagnostic impact of imaging studies on choice of procedures

Operation performed

Operations same as those proposed by preoperative studies

No. of operations performed

Pneumonectomy Pneumonectomy and left atrial resection Lobectomy Lobectomy and aortic resection Thoracotomy only (open biopsy only) Total operations Total operations same as proposed

Operations same as those proposed by operative ultrasound

8

5

8

2

1

2

9 I 4 24

7 2

8 I 4

16

23

(66.7'k) 8 (33.3%)

(95.8%) I (4.2%)

I

Total operations not same as proposed

(p<0.01)

Table III. Geometric criteria for assessing metastasis of lymph nodes in lung cancer by operative ultrasonography Length * (longest plus shortest axis of nodes) Squamous cell carcinoma

Location of lymph nodes

II mm

Highest mediastinal, paratracheal, pretracheal, and retrotrachcal nodes Tracheobronchial nodes Subaortic, paraaortic nodes Subcarinal nodes Hilar, lobar nodes

24mm 21 mm 25 mm (18 mm)t 18 mm (12 mm)f

Adenocarcinoma 7mm 16 mm (13 mm)t 9mm 13 mm 12mm

"Lymph nodes of this length or greater were interpreted as positive for metastasis. tThe length in parentheses was used for retrospective modification of the criteria.

precise measurement oflymph nodes. A number oflymph nodes that were not recognized by preoperative studies were detected by ultrasound scanning. The total number of lymph nodes found by CT in 30 lung cancers was 144; the average number per cancer was 4.8 ± 1.56 (mean ± standard deviation). On the other hand, a total of 240 lymph nodes was detected by operative ultrasound in 30 operations; the average was 8.0 ± 1.84. The difference in the number of lymph nodes detected by each method was significant (p < 0.001 by t test). For the qualitative assessment of whether or not lymph node metastasis was present, the initial criteria from our previous study of lymph nodes were used. In this previous investigation, three-dimensional measurements were obtained for 718 lymph nodes removed during 54 lung cancer operations, and then the lymph nodes were examined histologically to determine the presence or absence of metastasis.i In establishing criteria for measurement, we used the length of the longest axis plus the length of the shortest axis of a lymph node (which provided the best

accuracy in diagnosing lymph node metastasis in the previous study) as geometric measures for each lymph node location and histologic type of lung cancer (Table III). For example, in squamous cell carcinoma, tracheobronchiallymph nodes of 24 mm or more (i.e., 24 mm representing the sum of the longest plus the shortest axis on ultrasound images) were ultrasonically interpreted as positive for metastasis. According to this initial criterion, 127 lymph nodes in 30 lung cancers, for which node-by-node analysis by operative ultrasound and histologic type was possible, were evaluated. The results revealed that the sensitivity of operative ultrasound in diagnosing lymph node metastasis was 82.4% and the specificity was 67.3% (Table IV). No difference in the sensitivity (p = 0.66) and specificity (p = 0.14) was noted between squamous cell carcinoma and adenocarcinoma. Of 127 lymph nodes, 17 lymph nodes were histologically confirmed as metastatic. All of these 17 lymph nodes were depicted by operative ultrasound; however, depending on the criteria, three of 17 lymph nodes were judged

The Journal of

544

Thoracic and Cardiovascular

Machi et al.

Surgery

Table IV. Results and accuracy of operative ultrasonography in assessing lymph node metastasis of lung cancer

True positive True negative False positive False negative Total Sensitivity Specificity

Squamous cell carcinoma

Adenocarcinoma

Total

11 42 15 2 70 84.6% 73.7%

3 32 21 1 57 75.0% 60.4%

14 74 36 3 127 82.4% 67.3%

to be negative for metastasis (false negative) by ultrasound. These false negative results were caused by micrometastases to relatively small lymph nodes. To eliminate the false negative interpretation and thereby make surgical removal of all metastasized lymph nodes possible and avoid leaving any metastasized lymph nodes behind, we retrospectively modified the initial criteria by decreasing the length of lymph nodes at three sites, as shown in Table III. This modification increased the sensitivity of operative ultrasound for lymph node metastasis diagnosis to 1()()%, although the specificity decreased to

59.1%. Liver metastasis. Of 13 lung cancers for which operative ultrasound was performed to evaluate liver metastasis, nine were primary lung cancers and four were metastatic lung cancers. Eight cancers were located at the right lung and five at the left lung. Eight right thoracotomies and five left thoracotomies were performed. In all operations, the entire liver could be scanned from the thoracic cavity through the diaphragm regardless of the thoracotomy side. The images of the liver obtained during transdiaphragmatic operative scanning with highfrequency instruments were just as clear and detailed as those obtained during operative ultrasound in the abdominal cavity. The adrenal gland and the kidney could also be examined by this ultrasound technique. In 13 operations, no liver metastasis was detected by transdiaphragmatic operative ultrasound; however, in one operation a 3 by 4 mm preoperatively unrecognized tumor suspected of being a hepatic hemangioma was detected, and in another operation preoperatively suspected liver metastasis was excluded by operative ultrasound. Discussion Accurate staging of lung cancer is a prerequisite for determining the indication for operation and the type of surgical procedure.' Although preoperative imaging methods, in particular CT, have contributed remarkably

to lung cancer staging, the diagnostic accuracy of these methods is subject to limitations.r!' The major limitations of these methods include difficulties in correctly diagnosing tumor invasion to the mediastinum and in making qualitative assessments of lymph node metastasis. During various operations for cancer of the alimentary tract, operative ultrasonography has been considered useful in evaluating the extent of tumor spread.s 12-17 For example, the vascular invasion of primary liver tumors and pancreatic cancers or liver metastases from colorectal cancers can be accurately examined by operative ultrasound immediately after laparotomy; therefore, we have, for the first time, applied this ultrasound modality to lung cancer. The areas where operative ultrasound was attempted included cardiovascular invasion, lymph node metastasis, and liver metastasis. Diagnosis of cardiovascular invasion of lung cancer is particularly important for decisions regarding the type of surgical procedure to be followed. When the results of preoperative imaging methods are indeterminate in this diagnosis, exploratory inspection and palpation should be performed during the operation. Surgical exploration of the cardiovascular area behind or within the lung tumor, however, often requires extensive tissue dissection, which is usually time-consuming and may cause operative complications such as massive bleeding and nerve injury. The present study indicated that operative ultrasound was significantly more accurate in diagnosing cardiovascular invasion oflung cancer than the preoperative studies with CT and angiography. Because of its superior accuracy in cardiovascular assessment, opera tive ultrasound appeared to have a favorable impact on the surgical management of lung cancer. One third of the preoperatively proposed operations had to be changed to other types of operations because of inaccurate preoperative diagnosis, whereas all except one of 24 surgical procedures performed were correctly predicted by operative ultrasound examination. This suggests that decisions regarding the selection of the type of operation may be more reliably based on operative ultrasound diagnosis; therefore, unnecessary exploratory tissue dissection and possible complications may be eliminated or diminished. Also, operating time may be reduced and inappropriate operations proposed by preoperative studies may be avoided. For evaluating lymph node metastasis in lung cancer, this study indicated that operative ultrasound provided better detection of lymph nodes than CT and more accurate three-dimensional measurement of detected lymph nodes. In the qualitative assessment of the presence or absence of cancer metastasis to lymph nodes, the ultrasound values for sensitivity and specificity were not better than those previously reported for the accuracy of

Volume 98 Number 4 October 1989

CT. 7- JO It was, however, possible to raise the sensitivity of operative ultrasound for lymph node metastasis to 100% by retrospectively modifying the diagnostic criteria, if lowering of the specificity was allowed. These results indicate that operative ultrasound may be a valuable tool for detecting and facilitating the removal of all metastatic lymph nodes and for identifying areas where tissue dissection for lymph node removal is unnecessary. At present, lymph nodes that are delineated by operative ultrasound should be surgically removed for histologic examination. However, where operative ultrasound does not exhibit any lymph nodes, lymph node dissection may not be needed. Operative ultrasonography performed in the abdominal cavity is known to have a significantly higher rate of accuracy than preoperative CT and ultrasound in detecting primary and metastatic liver tumors? 16 Transdiaphragmatic operative ultrasound from the thoracic cavity during lung cancer operations was technically different but provided high-quality images of the liver because high-frequency instruments were used. Operative ultrasound, therefore, may have the potential to identify metastatic tumors in the liver during operations for lung cancer. There are limitations in the use of operative ultrasound. Tracheobronchial invasion cannot usually be assessed because of the presence of air even after lung deflation. Other than for the liver and the adrenal gland, metastases to distant organs cannot be evaluated. However, operative ultrasound with high-frequency equipment provides clear delineation of the pulmonary hilum, mediastinum, liver and even lung tissue itself. High-resolution operative ultrasonography with this capability is valuable in evaluating cardiovascular invasion and lymph node and liver metastases. Therefore, this new imaging modality may be used selectively during operations for lung cancer. REFERENCES I. Sigel B. Coelho JCU. Machi J. et al. The application of real-time ultrasound imaging during surgical procedures. Surg Gynecol Obstct 1983;157:33-7. 2. Makuuchi M. Hasegawa H, Yamazaki S. Takayasu K, Moriyama N. The use of operative ultrasound as an aid to liver resection in patients with hepatocellular carcinoma. World J Surg 1987;11:615-21. 3. Flanigan DP, Douglas OJ, Machi J. Sigel B. Schuler JJ, Buchbinder D. Intraoperative ultrasonic imaging of the carotid artery during carotid endarterectomy. Surgery 1986;100:893-9.

Operative ultrasonography 5 4 5

4. Chandler WF, Rubin JM. The application of ultrasound during brain surgery. World J Surg 1987;11 :558-69. 5. Hayashida R. Kurohiji T, Machi J, Yamashita Y. Kakegawa T. Values of operative ultrasonography in lung cancer surgery. Proceedings of the 51st meeting of the Japan Society of Ultrasonics in Medicine. 1987:411-2. [In Japanese]. 6. Moores DWO. MillerSJ Jr, McKneally MF. Lung cancer: a surgeon's approach. Curr Probl Surg 1987;24:708-22. 7. Martini N, Heelan R. Westcott J. et al. Comparative merits of conventional, computed tomographic. and magnetic resonance imaging in assessing mediastinal involvement in surgically confirmed lung carcinoma. J THORAC CARDIavASC SURG 1985;90:639-48. 8. Baron RL, Levitt RG, Sagel SS, White MJ, Roper CL, Marbarger JP. Computed tomography in the preoperative evaluation of bronchogenic carcinoma. Radiology 1982; 145:727-32. 9. Daly BOT Jr. Faling LJ, Pugatch RD, et al. Computed tomography: an effectivetechnique for mediastinal staging in lung cancer. J THORAC CARDIOVASC SURG 1984; 88:48694. 10. Glazer GM. Orringer MB. Gross BH. Quint LE. The mediastinum in non-small cell lung cancer: CT-surgical correlation. AJR 1984;142:1101-5. II. Richardson JV. Zenk BA. Rossi NP. Preoperative noninvasive mediastinal staging in bronchogenic carcinoma. Surgery 1980;88:382-5. 12. Bismuth H, Castaing 0, Garden OJ. The use of operative ultrasound in surgery of primary livertumors. World J Surg 1987;11:610-4. 13. Sigel B, Machi J, Ramos JR, Duarte B, Donahue PE. The role of imaging ultrasound during pancreatic surgery. Ann Surg 1984;200:486-93. 14. Plainfosse MC, Bouillot JL, Rivaton F. Vaucamps P, Hernigou A, Alexandre JH. The use of operative sonography in carcinoma of the pancreas. World J Surg 1987;11:654-8. 15. Machi J. Takeda J, Kakegawa T, et al. The detection of gastric and esophageal tumor extension by high-resolution ultrasound during surgery. World J Surg 1987;11:664-71. 16. Machi J. Isomoto H, Yamashita Y, Kurohiji T, Shirouzu K, Kakegawa T. Intraoperative ultrasonography in screening for liver metastases from colorectal cancer: comparative accuracy with traditional procedures. Surgery 1987; 101:678-84. 17. Boldrini G, Gaetano AM, Giovannini I, Castagneto M. Colagrande C. Castiglioni G. The systematic use of operative ultrasound for detection of liver metastases during colorectal surgery. World J Surg 1987;11:622-7.