Abstract
OBJECTIVE To assess the accuracy of multi –slice spiral CT (MSCT) with imaging reconstruction in judging central pulmonary vascular involvement from central lung cancer, and to explore its ability to predict the resectability of lung cancer
METHODS MSCTs were conducted on 48 patients who were diagnosed preoperatively with central lung cancer. Images of pulmonary arteries and veins that might affect lobectomy or pneumonectomy were reconstructed by means of imaging processing techniques. Then the relationship of the tumor to the vessels was assessed prospectively on both axial CT images and axial CT images plus reconstructed images(CT-RI) in comparison to subsequent pathologic and surgical findings.
RESULTS MSCTs were obtained on all 48 patients whom 42 underwent thoracotomy, lobectomy or pneumonectomy. Compared with the axial CT images, CT –RI was more accurate in judging the relationship of the central pulmonary vessels to the tumor based on subsequent pathologic 78 vessels studied and surgical findings (186 vessels studied) (0.01 <P<0.05). The sensitivity and positive predictive value of unresectability of the vessels were all remarkably higher with CT –RI (P <0.01).
CONCLUSION MSCT with imaging reconstruction can improve the recognition of neoplastic invasion of central pulmonary vessels. It can be used to predict preoperatively the resectability of central lung cancer and to plan surgery.
keywords
CT is one of the most useful imaging techniques employed for evaluation of lung cancer and it is used both in diagnosis and in preoperative selection of patients for radical pulmonaryresection. With the recent introduction of multi-slice spiral CT (MSCT) scanners, the scans can be performed more quickly and with higher longitudinal spatial resolution than was possible with single-slice spiral CT scanners. Currently multi-slice 4-channel CT scanners gatherup to 4 channels of data simultaneously from interweaving helices.[1,2]
The purpose of this study was to test the accuracy of CT with and without imaging reconstruction in the detectionof invasion of central pulmonary vessels in patients undergoing pulmonary resectionfor central lung cancer in order to assess resectability.
MATERIALS AND METHODS
Subjects
Since May 2001, MSCT angiography was obtained from 48 patients who were diagnosed preoperatively as having central lung cancer. The tumors were located in the patients as follows: the left upper lobe in 20, the left lower lobe in 6, the right upper lobe in 15, the right middle lobe in 2 and the right lower lobe in 5. Forty-two underwent thoracotomy, lobectomy or pneumonectomy, and were confirmed to have central lung cancer. Thirty-four patients were male, and 8 were female, with a mean age of 54.6 years.
Scanning protocols and reconstruction techniques
Scanning was perform using a GE lightspeed plus scanner equipped with a windows workstation (General Electric Medical System, Milwaukee,WI). We used 2.5 mm collimation and a pitch of 6:1 to produce a 2.5 mm thick image at 1.0 mm increments. The scanning delay was from 20 to 40 s. Images of pulmonary arteries and veins that might affect lobectomy or pneumonectomy were reconstructed by means of imaging processing techniques such as shaded surface display (SSD), maximum intensity projection (MIP), curved planar reformation (CPR), multi-planar reformation (MPR), volume rendering (VR), and CT virtual endoscopy (CTVE), to produce 2-dimensional and 3-dimensional models (Fig.1).
Image review
Vessel integrity on axial CT image studies and axial CT with reconstructed image(CT-RI) studies were assessed on a 3-grade scale: grade 0, normal, with a fat plane or normal lung tissue between the tumor and the vessel; grade 1, loss of a fat plane between the tumor and the vessel, flatting or slight irregularity of one side of the vessel; grade 2, encased vesselwith the tumor extending around at least 2 sides (two thirds of the perimeter), altering itscontour and producing concentric narrowing or occlusion of the lumen (Fig.2).
Pulmonary arteries and veins that might affect lobectomy or pneumonectomy were graded on this scale. The original presumption was that vessels with grade 0 and 1 would be dissectable or resectable, whereas a grade 2 would be undissectable or unresectable.
The pathologic and surgical findings of the related vessels were also graded on this scale. By comparison with pathologic and surgical findings respectively, we worked out the correlation rate, i.e., the rate of accuracy in judging pulmonary vascular involvement from the tumor.
Statistical analysis
We used the chi-square analysis of contingency to test the difference in the rate of correlation in nominal measurement between the axial CT images and CT-RI.
RESULTS
MSCT images were obtained from all 48 patients. Foury two patients underwent thoracotomy,lobectomy or pneumonectomy and central lung cancer was a confirmed diagnosis. The encasementgrade assigned to the central pulmonary vessels of these patients was equal to both CT techniques (axial CT images and CT-RI). A total of 78 vessels were studied in comparison to subsequent pathologic findings, along with 186 vessels examined at surgery.
Compared with the axial CT images, CT-RI was more accurate in judging the relationship of the central pulmonary vessels to the tumor (0.01<P<0.05) Table 1.
Considering grades 0 and 1 as predictors of resectability, and grade 2 as predictor of unresectability, CT-RI was more valuable. The sensitivity, specificity, positive and negative predictive values of undissectability of the vessels were all higher, particularly the positive predictive value and sensitivity were remarkably higher (P < 0.01) Table 2.
DISCUSSION
CT angiography has become an alternate angiographic technique in judging central pulmonary vascular involvement with central lung cancer. Advanced knowledge of the encasement of central pulmonary vessels is very important because it renders the prospect of resection difficult, requiring appropriate planning and mobilization of available resources. However, vascular encasement cannot be detected thoracoscopically or bronchoscopically. Dynamic CT scanning is accurate in predicting the unresectability of lung cancer. With recent advancements in CT scanning, detection and staging of lung cancer are improving.[3-6]
In our study we explored the thesis that MSCT angiography with imaging reconstruction canimprove the prediction of tumor resectability by improving the detection of vascular invasion. Indeed, in our selected patients with lung cancer in the hilar area without distant metastasis, we found that CT-RI was more accurate than axial CT images alone in judging the relationship of central pulmonary vessels with the tumor both in the pathological group (78 vessels studied) and in the surgical group (186 vessels studied). The sensitivity and positive predictive value of unresectability of the vessels were all higher. Our results indicate that MSCT with imaging reconstruction can improve the accuracy in the detection of invasion of the central pulmonary vessels in patients undergoing pulmonary resection for central lung cancer and to predict the resectability.
Considering that the operative modality is affected by many factors, such as surgeons having different operative skills, whereas patients vary in general conditions. So in our series, we just evaluated the encasement and resectability of pulmonary vessels, but did not assume a particular operative modality. Nevertheless, after knowing the involvement of the related vessels, an experienced surgeon can easily establish an operative modality.
The inability of CT to accurately distinguish vascular encasement may be due to insufficient vascular en hancement or the relatively decreased resolution of cross-sectional imaging along the z-axis. Thus, subtle changes in the caliber of the central pulmonary vessels may be missed. Some of these problems may be reme died with the used of spiral CT and CT angiography, which both increases vascular opacification and, in ad dition to cross-sectional CT imaging, produces angio graphic images. However, spiral CT angiography does not improve resolution, and artifacts from image-pro cessing techniques and involuntary motion may further degrade the z-axis resolution.
MSCT has the capability of rapidly scanning a large longitudinal (z) volume with high z-axis resolution. A recent study indicated that 4-slice spiral CT can provide equivalent imagequality at 2 to 3 times the volume coverage speed of single slice spiral CT.[7] At the same time, artifacts from image-processing techniques and involuntary motion are decreased.
In distinguishing vascular encasement, axial images alone often were insufficient becausemost vessels do not lie on the axial plane. MSCT with imaging reconstruction (2-dimensional, 3-dimensional models and CT virtual endoscopy) provided much more useful information (Figs. 1,2). Related vessels can be assessed by viewinguall aspects, especially multi-planar reformation (MPR) which overcame the visual limitation and thus improved the accuracy.
We conclude that MSCT with imaging reconstruction provides useful information about neoplastic invasion of the central pulmonary vessels. Compared with axial CT imaging alone, it improves staging for resectability and can be used to plan surgery and predict cases in which resection may prove to be difficult.
With the development of CT technology, more rapid scanning can be employed. MSCT will play more and more important roles in both diagnosis and in preoperative selection of patients for radical pulmonary resection.
- Received June 21, 2005.
- Accepted July 11, 2005.
- Copyright © 2005 by Tianjin Medical University Cancer Institute & Hospital and Springer