S c a n d J Thor Cardiovasc Su r g 24: 207-21 1 , 1990

COMPUTED TOMOGRAPHY AND T H E TNM CLASSIFICATION O F LUNG CANCER Jgrgen S p ar u p , M e r e t e Friis, Jorn Brenee, Hans Vejlsted, Bjarne Villumsen, K n u d P. Olesen, Sven Borgeskov and Svend Bertelsen From the Department of Thorcicic and Viiscirltrr Surgery and the Depciriment of Rrrdiology, Bispebjerg Hospital, Copenhagen. Denmurh

Scand Cardiovasc J Downloaded from informahealthcare.com by University of Sydney on 01/02/15 For personal use only.

(Accepted for publication July 3 I , 1989)

A h s t m c f . Computed tomography (CT) of the thorax and upper abdomen was prospectively evaluated in 84 patients with potentially operable lung cancer. Invasion into the thoracic wall and the mediastinal structures was not accurately demonstrated by CT. For metastatic mediastinal lymph nodes, the sensitivity and specificity of CT were, respectively, 86% and 61 % and the positive and negative predictive indices 49% and 91%. For T I , T2 and T3 tumours the negative indices were 100%, 9 6 4 and 71 %. Positive predictive index did not differ between squamous cell carcinoma and adenocarcinoma. Adrenal metastases were CT-suspected in 17 cases and liver metastases in eight. but were verified by ultrasonography in only one and four cases. CT should be used in preoperative investigation of lung cancer. irrespective of stage. Demonstration of thoracic-wall or mediastinal invasion need not exclude tumour resection. Preoperative mediastinoscopy is indicated if CT shows nodal metastases or if there are signs of tumour invasion, but not in CT-negative TI or T2 tumour. .4bdominal metastases indicated by CT should be investigated with CT-guided needle biopsy.

KO words: lung cancer, computed tomography. preoperative evaluation, tumour invasion, metastases.

time 5 sec) with 8 mm continuous intervals from the apex of the lung to the lower hepatic edge. Subsequently, after a bolus injection of urographin 76% or omnipaque 350. mediastinal scanning was done from the pulmonary apex to 2 cm below the carina. After conclusion of the study. the computed tomograms were re-evaluated 'blind' by two radiologists. The tumour was then classified according to size and invasion into mediastinum and thoracic wall and comparison made with the findings at thoracotomy. Mediastinal nodes larger than 1 cm at CT were classified as metastases and were recorded in accordance with the lymph-node mapping scheme of the American Thoracic Society (4). Corresponding records were made of all metastatic or benign lymph nodes larger than I cm seen at mediastinoscopy and thoracotomy, when all suspected nodes were biopsied. Abdominal ultrasonography (US) was performed on all patients in order to verify metastases shown by CT and needle biopsy was made of any pathologic findings. Statistical significance was evaluated with Fisher's exact test. The accuracy of CT was expressed by the following definitions: Sensitivity

Computed tomography (CT) is increasingly used in preoperative investigation of lung c a n c e r , mainly in the search f o r nodal metastases in the mediastinum, where the n u m b e r of false positive diagnoses is considerable, b u t false negatives are fewer (2, 12, 17-19). We have investigated t h e value of CT for preoperative TNM classification regarding t u m o u r size and invasion into the mediastinum and the thoracic wall, nodal metastases in the mediastinum and abdominal metastases.

correct positive

correct positive Specificity

index (PPI) Negative predictive index (NPI)

100

+ false negative

correct negative x 100

correct negative Positive predictive

X

+ false positive

correct positive x 100

correct positive

+ false positive

correcf negcifive x 100

correct negative

+ false negative

M A T E R I A L AND METHODS

RESULTS

In 1987. 84 consecutive patients with histologically proven bronchogenic carcinoma underwent CT of the thorax and upper abdomen before planned mediastinoscopy. The median age of the 53 men and 31 women was 67 (range 38-83) years. CT was done on a Siemens Somatom 2 (scan

The histologic t u m o u r t y p es were sq u am o u s cell carcinoma (29 cases) , adenocarcinoma (26), large cell carcinoma (20), small cell carcinoma (3, adenosquamous (2) and carcinosarcoma (2). Mediastin-

Scand Cardiovasc J Downloaded from informahealthcare.com by University of Sydney on 01/02/15 For personal use only.

208

J . Sparup et al.

Fig. 1 . CT scan showing a central cancer in the right lung with false positive invasion in the posterior rnediastinum and oesophagus. Atelectasis is also seen. At pneumonectomy no turnour invasion was found.

oscopy was performed on only 73 patients, as I 1 were found to be inoperable before the planned examination. The 49 patients with normal findings underwent thoracotomy, with lobectomy in 17 cases, pneumonectomy in 27. and exploration only (because of mediastinal tumour invasion) in five. Tumour invasion in medir~stinirrir and thoracic wall ITS). The C T evaluation of invasion was compared with the surgical findings for each of the mediastinal structures and the thoracic wall (Table I). Invasion was found in 14 cases at thoracotomy, but was correctly demonstrated by C T in only six. Three of the six had aortic invasion and were inoperable, but resection was possible in the other three. In all of the false positive cases, but only six of the eight false negative cases the tumour could be resected. Nodal metastases in mediastinurn ( N 2 ) . In 7 3 patients the nodal findings at C T were compared with the findings a t mediastinoscopyithoracotomy. Table I1 shows the nodal findings and the accuracy of CT in relation to the T classification. There were 22 patients with nodal metastases (30%), and CT demonstrated 19 of them correctly. In the TI group both cases were identified by CT, in the T2 group nine of ten, and in the T3 group eight of ten. Sensitivity, specificity, PPI and NPI were 100% for the TI tumour group. Although these values fell with Sc unJ J Tiior C'urdiouusc 24

increasing T stage. the difference5 were n o t significant (p>0.05). Between squamous cell carcinoma, adenocarcinoma and large cell carcinoma there was no significant difference @>0.05) in regard to sensitivity, specificity, PPI and NPI (Table 111). Metastases t o the abdomen ( M I ) . C T indicated abdominal metastases in 25 of the 84 patients with lung cancer (Table IV). Three patients had CT signs of metastases in the liver and both adrenal glands, and two of them also in the retroperitoneal lymph nodes. Adrenal hyperplasia with suspected metastasis was demonstrated by CT in 17 cases, bilateral in five. Only one case was verified by US and needle biopsy, which showed bilateral metastasis of adenocarcinoma. In a unilateral case a tumour demonstrated by US was found at needle biopsy to be a pheochromocytoma, and in another case US showed bilateral hyperplasia. Autopsy was performed on only one patient with normal US and revealed normal adrenals, Only four of eight cases of CT-suspect liver metastases were confirmed by US, one of them also with needle biopsy. All four were inoperable. DISCUSSION Tumour invasion in the mediastinurn and thoracic, wall. Because of the many false positive and false negative C T findings, the sensitivity and specificity

Table I . CT findings of tumour invasion in the mediustinum and thoracic wall cornpared with the thoracotomy findings Figures in brackets

=

Mediastinal structures

vein Pulmonary artery ~~l~~~~~~ artery atrium Aorta Vena Pericardium

no. of lung resections CT positive

CT negative

True

False

False

2 (2) 3 (3) l ( 1 ) 2 (2)

True

1(1)

+

1

3

Oenonhaeus - ..-r--0--

Thoracic wall

Total

6

9

8

26

Scand Cardiovasc J Downloaded from informahealthcare.com by University of Sydney on 01/02/15 For personal use only.

CT arid the TNM cIussiJic,utio)iqf’lirng cancer

Tumour

No. of

classification

pat-

TI T2 T3 Total

5 43 25

ients

CT positive True

False

2 9 8

73

CT negative

0

10 10 20

19

True

False

(%)

(%I

index (%)

Negative predictive index ( % )

3 23 5

0 1 2

100 90 80

I00

100 47

I00 96

.. 33

44

71

x6

61

49

91

31

Pohitivr

3

Sensitivity

Specificity

predictive

(43% and 74%) were low in this study, as in earlier reports (2, 13, 18, 21). Only a few cases of aortic invasion were accurately demonstrated at CT. Atelectasis is a common cause of misjudging mediastinal invasion at CT, and when the central vessels are involved it is especially difficult to show invasion in the pulmonary artery (18). Two-thirds of the CT evaluations of pulmonary artery and vein in the present study proved to be wrong, while all pericardial and oesophageal diagnoses were wrong, with overdiagnosis of all the oesophageal findings (Fig. I ) . In CT it is difficult to distinguish between tumour adhesion to the pleura and invasion in the deeper layers of the thoracic wall. This leads to great variations in sensitivity and specificity ( I . 12, 16, 17). In three of our cases there was underdiagnosis, as CT indicated pleural adhesions, but the clinicohistologic picture was invasion. All three tumours were resectable, however. Nodal mrtustusrs in the nirdiusritiirm. Some authors did not recommend C T for investigation of TI turnours, because of low incidence (0-10%) of nodal metastases and failure to detect them (14, I S , 18). We found such metastases in two of the five TI tumours, and therefore advocate C T because of its high accuracy. Of the normal nodes in the mediasti-

NO. of

Tumour

histology

patients

CT positive

70

num, 95-99rC are smaller than 1 cm. whereas 97% of nodal metastases are larger (8. 9, 21). The commonly used borderline between benign and pathologically enlarged nodes therefore is I cm. and repcrted corresponding values for sensitivity are 74-9552, for specificity 68-94%. and for positive and negative predictive index S&88% and 81-97% respectively. In our study specificity and PPI were lower because of the many false positive findings. The commonest cause of false positive diagnosis of mediastinal node metastasis is benign hypertrophic nodes caused by postobstructive pneumonia ( 2 , 3 , 5 , 13, 17-19]. Accordingly. 14 of our 20 patients with false positive mediastinal CT had nodal hypertrophy, seven in the T2 and seven in the T3 group. These figures agree with the circumstance that most of the tumours were central, where postobstructive pneumonia can often be expected. In the T2 group one false positive assessment could be explained by scar tissue from an earlier mediastinoscopy, but the two remaining cases could not be explained. In the T3 group the cause of three false positive findings was misinterpretation of central tumour invasion. As expected, false positives were most frequent in the T3 cases, all of which showed central tumour invasion. and where specificity and PPI were lowest.

CT negative

True

False

True

False

(%)

I%)

Pobitive Negative predictive predictive index ( Q ) index ( % )

6 6

9 3

I! I1

I I

86 86

.. ii

40

79

67

92 92

5

4

9

1

83

69

56

YO

Sensitivity

Specificil)

Squarnous cell

carcinoma 27 Adenocarcinoma 21 Large cell 19 carcinoma

209

Scand Cardiovasc J Downloaded from informahealthcare.com by University of Sydney on 01/02/15 For personal use only.

210

J . Sparup et al.

Table IV. CT-suspected abdominal mi~tcr.sttr.\c*.v- which cannot be hit with the biopsy needle, and where later C T does not show signs of growth (9. with US c~~nprtrisori 14, 19). In 17 (20%) of the present case series, C T = metastasis demonstrated at US. ( ) * = benign tu( showed metastasis-suspect adrenal hyperplasia, but niour demonstrated at U s the lesion was also demonstrated by US in only Cases with CT-suspected three cases, and needle biopsy showed metastasis metastasis in one of them ( I .2 9%). If CT-guided needle biopsy had been used in our study, more metastases might S i k of suspected Operable Inoperable ( n = 131 (n=12) metahtasis have been demonstrated. That C T was more sensitive than U S could be explained by the greater suitability of US when retroperitoneal fat is sparse Adrenal (unilateral) 9 (I)* 3 Adrenal (bilateral) 1 (1)' 4 (1) (20). As in an earlier study (13). we found hepatic Liver 2 6 (4) metastasis at C T in 9 % of the cases. U s was less Retroperitoneal nodes 3 sensitive than CT. confirming only half of these 1 P\oas muscle metastases. Subcutaneous 1 Total site5 of metastasis

13

17

The cause of false negative mediastinal assessment may be malignant involvement of nodes smaller than 1 cm and hence usually defined as benign at C T ( I . 3 , 9 . 10, 12). Such nodes have been reported to comprise about 3 % of metastases (14). Thus we found only a single false negative case (4.5%) in the T2 group, a 5 mm metastatic node located in region IOR. We found no false negative C.1 in the aortopulmonary window, region 5 , where C T diagnosis is difficult ( 2 , 3, 5 , 11). Invasion of central turnours into the mediastinum and atelectasis may mask enlarged nodes and give false negative findings with low negative predictive index (2. 4, 6). In the T3 group two metastatic nodes were overlooked for this reason-one measuring 2 cm in region 9 (inferior pulmonary ligament) and a 1 cm node in region 7 . NPI was lowest for the T3 tumours. Metastatic mediastinal nodes were most common in adenocarcinoma. Some authors found more false positive C T evaluations and lower PPI in squamous cell carcinoma than in adenocarcinoma, because of more central turnours with postobstructive pneumonia (2, 5. 7 ) . Our figures seemed to confirm this trend. but the difference was not significant (p>O.OS). Metiisruses to the ubdorneii. CT has been stated to show metastasis-suspect enlarged adrenal glands in about 10% of patients with lung cancer, whereas CT-guided needle biopsy showed metastases in only 3-57? (8-10. 14. 19). Most of such lesions probably are benign hyperplasia or small adenoma

Conclusions

CT should be used in all investigation of lung cancer, including peripheral TI tumours. If C T demonstrates nodal metastases in the mediastinum, and in all T3 tumours, mediastinoscopy should be performed before thoracotomy. If nodal metastases are not found in cases of TI or T2 tumour, mediastinoscopy may be omitted. CT-demonstrated invasion of the mediastinum or the thoracic wall does not contraindicate thoracotomy with the aim of pulmonary resection. For demonstrating abdominal metastasis, C T is more sensitive than US. If US also shows the metastasis, the finding is confirmed, but histologic diagnosis by needle biopsy should be required before the patient is declared inoperable. As negative US probably does not always preclude malignancy. CT-guided needle biopsy or repeated C T should be used more frequently. ACKNOWLEDGEMENT The study was supported by grants from the Danish research administration, Copenhagen, Denmark.

REFERENCES Backer CL, Shields TW. Lockhart CG. Vogelzang R. LoCicero J . Selective preoperative evaluation for possible Nzdisease in carcinoma of the lung. J Thorac Cardiovasc Surg 1987: 93: 337-343. Baron RL. Levitt RG, Sagel SS, White MJ. Roper CL, Marbarger JP. Computed tomography in the preoperative evaluation of bronchogenic carcinoma. Radiology 1982; 145: 727-732. Brion JP, Depauw L. Kuhn G , de Francquen Ph. Friberg J. Rocrnans P. Struyven J. Role of computed

Scand Cardiovasc J Downloaded from informahealthcare.com by University of Sydney on 01/02/15 For personal use only.

CT and the T N M classijkation qf lung c m c e r tomography and mediastinoscopy in preoperative staging of lung carcinoma. J . Comput Assist Tomogr 1985: 9: 480-484. 4. Clinical staging of primary lung cancer. American thoracic Society node mapping scheme. Am Rev Respir Dis 1983; 127: 659-669. 5. Daly BDT, Faling LJ. Pugatch RD, Jung-Legg Y . Gale ME, Bite G , Snider GL. Computed tomography. J Thorac Cardiovasc Surg 1984; 88: 48-94, 6. Daly BDT. Faling LJ. Bite G. Gale ME. Bankoff MS. Jung-Legg Y. Cooper AG. Snider G L . Mediastinal lymph node evaluation by computed tomography in lung cancer. J Thorac Cardiovasc Surg 1987: 94: 664-672. 7. Ferguson MK. MacMahon H , Little AG. Colomb HM, Hoffman PC, Skinner DB. Regional accuracy of computed tomography of the mediastinum in staging of lung cancer. J Thorac Cardiovasc Surg 1986: 91: 498-504. 8. Genereux GP. Howie J L . Normal mediastinal lymph node size and number: C T and anatomic study. AJR 1984; 142: 1095-1 100. 9. Glazer G M , Gross B H . Quint L E , Francis IR. Bookstein FL. Orringer MB. Normal mediastinal lymph nodes: Number and size according to American Thoracic Society mapping. AJR 1985; 144: 261-265. 10. Glazer H S , Duncan-Meyer J , Aronberg DJ. Moran JF. Levitt RG, Sagel SS. Pleural and chest wall invasion in bronchogenic carcinoma: C T evaluation. Radiology 1985; 157: 191-194. I I . Heavey LR. Glazer GM. Gross BH, Francis 1R. Orringer MB. The role of CT in staging radiographic TINOM,lung cancer. AJR 1986: 146: 285-290. 12. Lewis JW, Madrazo BL. Gross SC. Eyler WR. Magilligan DJ, Kvale PA, Rosen RA. The value of radiographic and computed tomography in the staging of lung carcinoma. Ann Thoric Surg 1982; 34: 553-558.

21 1

13. Modini C , Passariello R. lascone C. Cicconetti F.

Simonetti G. Zerilli M. Tirindelli-Danesi D. Stipa S . TNM staging in lung cancer: Role of computed tomography. J Thorac Cardiovasc Surg 1982: 84: 569-574. 14. Oliver T W . Bernardino ME. Miller JI. Mansour K , Greene D. Davis WA. Isolated adrenal masses in nonsmall-cell bronchogenic carcinoma. Radiology 1984; 153: 217-218. 15. Pearlberg J L , Sandler MA. Beute G H . Madrazo BL. TINOM,bronchogenic carcinoma: Assessment by CT. Radiology 1985; 157: 187-190. 16. Pennes DR, Glazer GM, Wimbish KJ, Gross BH. Long RW, Orringer MB. Chest wall invasion by lung cancer: Limitations of CT evaluation. AJR 1985: 144: 507-5 1 1. 17. Rea H H . Shevland J E . House AJS. Accuracy of computed tomographic scanning in assessment of the mediastinum in bronchial carcinoma. J Thorac Cardiovasc Surg 1981; 81: 825-829. 18. Rendina EA, Bognola DA, Mineo TC. Gualdi GF. Caterino M. Di Biasi C. Facciolo F. Ricci C. Computed tomography for the evaluation of intrathoracic invasion by lung cancer. J Thorac Cardiovasc Surg 1987; 94: 57-63. 19. Richey HM. Matthews J1. Helsel RA. Cable H. Thoracic C T scanning in the staging of bronchogenic carcinoma. Chest 1984; 85: 218-221. 20. Sandler MA. Pearlberg J L , Madrazo BL. Gitschlag KF. Gross SC. Computed tomographic evaluation of the adrenal gland in the preoperative assessment of bronchogenic carcinoma. Radiology 1982: 145: 733-736. 21. Whittlesey D. Prospective computed tomographic scanning in the staging of bronchogenic cancer. J Thorac Cardiovasc Surg 1988; 95: 876-882.

S