Diagnosis of Gastric Cancer with MDCT Using the Water-Filling Method and Multiplanar Reconstruction: CT Histologic Correlation

Transcription

1 MDCT of Gastric Cancer Gastrointestinal Imaging Original Research Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved Kensaku Shimizu 1 Katsunori Ito Naofumi Matsunaga yame Shimizu Yasuhiko Kawakami Shimizu K, Ito K, Matsunaga N, Shimizu, Kawakami Y DOI: /JR Received pril 26, 2004; accepted after revision December 6, ll authors: Department of Radiology, Yamaguchi University School of Medicine, Minamikogushi, Ube, Yamaguchi , Japan. ddress correspondence to K. Shimizu. JR 2005; 185: X/05/ merican Roentgen Ray Society Diagnosis of Gastric Cancer with MDCT Using the Water-Filling Method and Multiplanar Reconstruction: CT Histologic Correlation OJECTIVE. The purpose of this study was to assess the utility of MDCT with a thinsliced multiplanar reconstruction (MPR) technique and water-filling method for the diagnosis of gastric cancers. MTERILS ND METHODS. Sixty-five patients with gastric cancers were preoperatively examined with MDCT using the water-filling method. The abdomen was dynamically scanned at 30 and 80 sec after the start of contrast medium administration. MPR images were reconstructed with a slice width of 1.25 mm and a slice interval of 1 mm. The detection rate and accuracy of T staging for gastric cancer were evaluated on MPR images and compared with 5-mm-slice axial images. In addition, MDCT images were correlated with pathologic findings. RESULTS. The detection rate of all gastric cancers using the MPR technique was 65%. The detection rate of advanced gastric cancers was 96.2% (25 of 26), whereas that of early gastric cancers was 41.2% (14 of 34). There was a statistically significant difference (p < 0.05) in the detection rate of early gastric cancers between MPR and 5-mm-slice axial images. The MDCT appearances of gastric cancers were well correlated with pathologic findings such as mucinous component or differences in infiltration of cancer cells. The overall accuracy of CT staging was 85%. MPR images were superior to axial images for the evaluation of the z-axis extent of tumor. CONCLUSION. MDCT with the water-filling method has advantages in acceptable evaluation of depth invasion of gastric carcinomas and in visualization of histologic changes in the tumors. MPR images may be a useful guide for the evaluation of the z-axis extent of tumor. onventional CT has been considered unsuitable for the detection C and staging of gastric cancer and has been limited to the evaluation of lymph nodes or distant metastasis. MDCT, which has recently been introduced into clinical CT practice, allows a high speed of scanning, resulting in increased z-axis coverage, and routine use of thin slice collimation during a breath-hold acquisition. s a result, the quality of multiplanar reconstruction (MPR) images obtained using thin slice collimation has dramatically improved. High-quality MPR images allow visualization of the fine anatomic details of the gastric wall in any plane. Studies have indicated that the water-filling method with gastric CT allows clear depiction of the gastric wall and gastric tumor without overshooting artifacts by air in the lumen [1 5]. The combination of high-quality MPR with the water-filling method has the potential to improve diagnostic performance for detecting lesions and determining the depth of invasion of gastric carcinomas. In this study, CT findings of gastric carcinomas shown by MDCT performed using MPR techniques and the waterfilling method were correlated with pathologic findings to assess the utility and limitations of MDCT for the diagnosis of gastric carcinoma. Materials and Methods Patients Sixty-five patients with pathologically proven gastric cancers were examined with MDCT using the water-filling method between May 2000 and November In five patients, gastric cancer was not resected and they were excluded from this study. Surgical or endoscopic resections were performed in 60 patients at our institute within 1 month after CT examination. The study group was composed of 44 men and 16 women who ranged in age range from 29 to 87 years (mean, 62.7 years). Pathologically, the tumors were classified as pt1 (invasion to the mucosa or the submucosa) in 34 patients, pt2 (invasion to the proper muscle) in 12 patients, pt3 (invasion to the serosa) in nine patients, and pt4 (invasion to adjacent organs or structures) in five patients JR:185, November 2005

2 MDCT of Gastric Cancer Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved Fig year-old man with protruding type of early gastric cancer., Coronal multiplanar reconstruction shows gastric wall as threelayer structure on CT. High-attenuating tumor (arrows) is adjacent to third layer. * = inner layer with high attenuation, ** = middle layer with low attenuation, *** = outer layer with isoattenuation., Photomicrograph shows inner layer corresponds to mucosa (*), middle layer to submucosa (**), and outer layer to proper muscle with serosa (***). Tumor is adjacent to proper muscle (arrows), as shown on multiplanar reconstruction (). Scanning Technique ll examinations were performed with an MDCT scanner (Volume Zoom, Siemens-sahi Medical). ll patients ingested ml of tap water to distend the stomach just before CT. CT scans were obtained with the patient in the prone position in 57 patients, and they were acquired with the patient in the supine position in the remaining three patients with the tumor located in gastric fundus to effectively opacify the lesions by water. fter undergoing unenhanced CT, each patient received 100 ml of nonionic contrast medium IV by means of a mechanical power injector (utoenhance -50, Nemotokyourindo) for the contrast-enhanced study. Contrast material was injected at a rate of ml/sec with a monophasic rate of injection. Two-phase scanning was begun 30 and 80 sec after the start of the injection. MDCT was performed in a breath-hold acquisition during each phase with parameters of 4 1 mm collimation and 8 mm/sec table speed. Images were reconstructed at 5-mm intervals for the screening of the abdomen. In addition, thin-slice reconstruction was performed with a slice width of 1.25 mm and a slice interval of 1 mm as source images for MPR of the stomach. Then, MPR images were routinely obtained in the coronal and sagittal directions with a slice width of 1.25 mm and a slice interval of 1 mm. MPR images sliced in two directions (coronal and sagittal views) and thin-slice axial source images were displayed on a workstation connected to the CT unit. Image nalysis Images were interpreted by two independent radiologists experienced in abdominal CT without prior knowledge of endoscopic, surgical, or histologic findings. Disagreements in their interpretations were resolved by consensus after discussion under the attendance of a third radiologist as a tiebreaker. The review procedure was performed during two separate sessions. First, 5-mm-slice axial images were reviewed for the presence of tumor. During the second session, the reviewers reviewed thin-slice MPR images sliced in two directions (coronal and sagittal views) with thin-slice axial images to evaluate the value of MPR images for the detection of tumor. During the third session, thin-slice MPR images with thin-slice axial images were reviewed to determine the depth of invasion of the tumor. When the presence of gastric tumor was suspected, oblique MPR images perpendicular to the gastric wall at the suspected region were, in addition, reconstructed for the evaluation of depth invasion. The presence of tumor and depth of invasion were evaluated on the basis of changes in the layer structure of the gastric wall. Typical CT findings of the normal layer structures of the gastric wall are shown in Figure 1. CT revealed a three-layered structure consisting of an inner layer with high attenuation, a middle layer with low attenuation, and an outer layer with intermediate attenuation between an inner and a middle layer. The presence of tumor was suspected if thickening or irregularity of an inner layer of the gastric wall was evident or wall thickening with abnormal contrast enhancement was evident. Staging of depth invasion and corresponding CT findings were as follows: T1 (invasion to the mucosa or the submucosa), when the tumor was not visible on CT or the tumor was confined to the inner or middle layer of the gastric wall; T2 (invasion to the proper muscle), when the tumor invasion extended to the outer layer or outer surface of the gastric wall but the outer surface of the gastric wall was smooth; T3 (invasion to the serosa), when a nodular or irregular outer surface of the gastric wall or perigastric infiltration was evident; and T4 (invasion to adjacent organs and structures), when direct extension or invasion of the tumor into a contiguous organ or structure was evident. CT findings of tumors were correlated with histologic findings. Statistical analysis was performed using a McNemar test. Results Detection Rate of Gastric Cancer Pathologically, 34 patients showed early gastric cancers (pt1) and 26 patients were found to have advanced gastric cancers (pt2 pt4). The detection rate of all gastric cancers on thinsliced MPR images was 65% (39 of the 60 patients), whereas it was 50% (30 of the 60 patients) on 5-mm-slice axial images (Table 1), TLE 1: Tumor Detection Rates Using Multiplanar Reconstruction (MPR) Images Versus xial Images Detection Rate (%) Cancer Detected MPR Images 5-mm xial Images ll gastric cancers dvanced gastric cancers Early gastric cancers a Mucosal cancer Submucosal cancer a p < 0.05 for comparison of detection rates (McNemar test). JR:185, November

3 Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved Fig. 2 Early gastric cancers., 69-year-old man with protruding type of early gastric cancer. Oblique multiplanar reconstruction perpendicular to gastric wall shows protruding mass (arrow) of gastric cardia., 77-year-old man with superficial and depressed type of early gastric cancer. Coronal multiplanar reconstruction shows irregularly enhanced lesion (arrows) at lesser curvature of lower gastric body. Fig year-old woman with protruding type of early gastric cancer., xial CT scan shows polypoid mass with thick vessels (arrows) at basement of tumor., Photomicrograph also shows thick vessels (arrows) at basement of tumor JR:185, November 2005

4 MDCT of Gastric Cancer indicating no statistically significant difference (p > 0.05) between the two. However, there was a statistically significant difference (p < 0.05) in the detection rate of early gastric cancers between thin-sliced MPR images (14 of 34 patients, 41.2%) and 5-mm-slice axial images (7 of 34 patients, 20.5%). The locations of early gastric cancers that were detected on thinsliced MPR images but were not visible on 5- mm-slice axial images were as follows: greater curvature of the gastric angle and antrum (n = 4), lesser curvature of the gastric angle and Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved Fig year-old man with protruding type of advanced gastric cancer., Sagittal multiplanar reconstruction shows low-attenuating area (*) in tumor., Photomicrograph reveals that low-attenuating area histologically corresponds to mucinous component (*) in tumor. Fig year-old man with diffusely infiltrative type of advanced gastric cancer (pt3)., Coronal multiplanar reconstruction shows thickened gastric wall with abnormal three-layer structure: high-attenuating first layer (*), low-attenuating second layer (**), and high-attenuating third layer (arrow)., Photomicrograph shows that high-attenuating first layer pathologically corresponds to where cancer cells are more closely linked, mainly at mucosal and submucosal layers (*), and that low-attenuating second layer (**) corresponds to where tumor cells are diffusely scattered, mainly at proper muscle layer. In addition, high-attenuating third layer of outer line of gastric wall corresponds to closely linked proliferation of cancer cells from subserosal layer to serosa (arrow). JR:185, November

5 Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved antrum (n = 2), and large curvature of the gastric body (n =1). In 34 early gastric cancers, thin-sliced MPR images depicted 68.8% (11/16) of submucosal cancers, in which cancer cells invaded to the submucosa, and they showed 16.7% (3/18) of less invasive mucosal cancers, in which cancer cells invaded within the mucosa. In contrast, thin-sliced MPR images depicted 96.2% (25 of 26) of advanced gastric cancers. One advanced gastric cancer not visible on thin-sliced MPR images was a minute ulcerative tumor, a so-called stage IIc like advanced cancer. CT ppearance of Gastric Cancer (CT Histologic Correlation) Grossly, gastric cancers were categorized as one of the following types: protruding, superficial or depressed, ulcerative and localized, ulcerative and infiltrative, diffusely infiltrative, or nonclassifiable. ll protruding early gastric cancers (n = 6) were clearly depicted as protruding masses on CT (Fig. 2). In one patient, two dilated vessels were clearly displayed at the basement of the tumor on CT (Fig. 3). Superficial or depressed early gastric cancers (n = 28) were revealed as lesions with irregular enhancement of the inner layer of the gastric wall (Fig. 2), but the detection rate of this type of gastric cancer was low (28.6%). Protruding advanced gastric cancers (n = 3) were revealed as a protruding mass in all patients. Two lesions were histologically diagnosed as mucinous adenocarcinomas and had a lowattenuating area in the tumor (Fig. 4). The low-attenuating area histologically corresponded to large mucin pools, mainly in the submucosal and proper muscle layers (Fig. 4). Ulcerative and localized advanced gastric cancers (n = 4) or ulcerative and infiltrative advanced gastric cancers (n = 10) were also revealed as a protruding mass with a central depression in all patients. Diffusely infiltrative advanced gastric cancers (n = 6) were revealed as a markedly thickened gastric wall with abnormal two-layer (n = 3) or three-layer (n = 3) structures (Fig. 5). high-attenuating first (inner) layer histologically corresponded to where cancer cells were more closely proliferated, mainly in the mucosal and submucosal layers. low-attenuating second layer histologically corresponded to where cancer cells were diffusely scattered, mainly in the proper muscle layers with an inflammatory or desmoplastic reaction. high-attenuating third layer, seen in 50% of this type, was revealed on the serosal surface of the gastric wall and histologically corresponded to the infiltration of cancer cells from the subserosal layer to the serosa, resulting in serosal infiltration by the tumor (Fig. 5). Depth Invasion of Gastric Cancer The accuracy results for CT staging of gastric cancers in comparison with that for pathologic staging are summarized in Table 2. The accuracy of CT staging using MPR images was 85% (51 of 60). The MPR technique was helpful for evaluation of the z-axis extent of the tumor in three cases (one pancreatic invasion and two serosal invasions of the greater curvature of the gastric angle or antrum) (Fig. 6). On the thin-sliced MPR images, four of the five patients with invasion to other organs or structures were correctly staged at CT. The accuracies of CT staging using MPR for pt1, pt2, and pt3 gastric cancers were 94.1%, 66.6%, and 77.8%, respectively. Two of the 34 patients with pt1 tumors and three of the 12 patients with pt2 tumors had tumors that were overstaged as T2 and T3, respectively (Fig. 7), whereas one of the 12 patients with pt2 tumors and two of the nine patients with pt3 tumors had tumors that were understaged as T1 and T2, respectively. TLE 2: CT and Pathologic Findings of Depth of Tumor Invasion Pathologic CT Staging Staging T1 T2 T3 T4 pt pt pt3 2 7 pt4 1 4 Discussion dequate distention of the stomach filled with water as a negative contrast agent has led to improved visualization of the gastric wall and the gastric tumor itself without overshooting artifacts derived from intraluminal air [1 5]. y combining the helical scanning technique with this water-filling method, the accuracy of the depiction of gastric cancer and its depth of invasion by CT has remarkably improved [1 5]. In this study, we combined high-quality thin-sliced MPR using MDCT with the water-filling method to assess the detection of gastric carcinomas and determine the depth of invasion of gastric carcinomas, and the diagnostic ability of these high-resolution MPR images was compared with that of 5-mm-slice axial images. In our study, the detection rate of advanced gastric cancers with thin-sliced MPR images was excellent (96.2%), indicating that thinsliced MPR using MDCT is a reliable tool for the detection of advanced gastric cancers. In contrast, there were some difficulties in the detection of early gastric cancers even when used with thin-sliced MPR, although the detection rate of early gastric cancers in MPR (41.2%) was significantly higher than that in 5-mmslice axial images (20.5%). The detection rate for mucosal cancer was very low (16.7%) in comparison with that for submucosal cancer (68.8%). In addition, the detection rate for grossly nonprotruding early gastric cancers, such as stage IIc, was comparatively low (28.6%). These results indicate that CT is not suited for detection of mucosal cancer or grossly nonprotruding early gastric cancer even when used with thin-sliced MPR because distinction between well-enhanced mucosal or nonprotruding early gastric cancer and well-enhanced normal mucosal layers may be difficult on CT. Takao et al. [5] reported that the accuracy of detection of early gastric cancers was only 23% using axial images obtained on a singledetector helical CT scanner with the waterfilling method, which is similar to our result using 5-mm-slice axial images. In our seven early gastric cancers detected at MPR but not visible on 5-mm-slice axial images, six were located at the large or lesser curvature of the gastric angle or antrum. Detection of early gastric cancers with 5-mm-slice axial images may be difficult because of partial volume artifacts where the gastric wall is tangent to the axial scan, such as at the greater or lesser curvature of the gastric antrum or angle. In this study, the MDCT findings of the gastric cancers were well correlated with pathologic findings. In the protruding type of early gastric cancers, thick vessels were shown at the basement of the tumor both on MDCT and at pathology (Fig. 3). ecause thick vessels in the tumor sometimes become the cause of massive bleeding in endoscopic polypectomy, this information may be useful on the occasion of polypectomy. In the diffusely infiltrative type of advanced gastric cancers, abnormal two- or three-layer structures were depicted in tumors. Histologically, a high-attenuating abnormal inner layer corresponded to where cancer cells were closely proliferated, mainly in the mucosal and submucosal layers, whereas a low-attenuating abnormal second layer corre JR:185, November 2005

6 MDCT of Gastric Cancer Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved Fig year-old man with diffusely infiltrative type of advanced gastric cancer (pt4)., Coronal multiplanar reconstruction shows inhomogeneous mass (arrows) invading pancreas., Pancreatic invasion is not clearly evident on axial CT scan. sponded to where cancer cells were diffusely scattered, mainly in the proper muscle layer with desmoplastic change and inflammatory reaction. In addition, the presence of a highattenuating abnormal third layer on the serosal surface of the gastric wall suggested the serosal infiltration of the tumor because this high-attenuating layer histologically corresponded to a closely linked proliferation of cancer cells from the subserosal layer to the serosa, indicating direct serosal infiltration by the tumor. These CT findings associated with the infiltration of cancer cells were helpful in determining the depth invasion of the tumor. characteristic CT appearance was also noted in mucinous gastric carcinomas. The two mucinous carcinomas in our study also had a large and very low-attenuating area in the tumor that histologically corresponded to a large mucin pool (Figs. 4 and 7). The presence of this large low-attenuating area in the tumor may assist in distinguishing mucinous from nonmucinous gastric carcinomas. Park et al. [6] also reported the helical CT findings of 21 patients with mucinous gastric carcinoma and concluded that a diffusely thickened, low-attenuating middle or outer layer and preservation of the thin high-attenuating inner layer could be helpful findings that suggest mucinous gastric carcinoma. Mucinous Fig year-old man with protruding type of advanced gastric cancer (pt2)., xial CT scan shows large protruding mass with nodular deformation (arrow) of outer surface of gastric wall., Photomicrograph shows extraprotrusion of serosa by mucinous lake (arrow), but serosal infiltration by cancer cells is not histologically revealed. JR:185, November

7 Downloaded from by on 01/21/18 from IP address Copyright RRS. For personal use only; all rights reserved carcinoma is rare but is considered to have a poor prognosis [7, 8]. Therefore, it is clinically important to distinguish mucinous from nonmucinous gastric carcinomas on CT. It is well known that the depth of invasion of a tumor is one of the significant prognostic factors for patients with gastric cancer [9 11]. In addition, minimally invasive surgery or laparoscopic gastrectomy for gastric cancer has been recently developed to obtain better postoperative quality of life [12, 13]. To select a proper indication for these minimally invasive resections, the accuracy of the preoperative diagnosis to decide the depth of tumor invasion is important [12, 13]. Our accuracy using thin-sliced MPR for T staging of gastric cancer was 85%, indicating that CT is useful in determining the depth of invasion of gastric tumor. In addition, the MPR technique was helpful for the evaluation of the z-axis extent of tumor in three cases (one pancreatic invasion and two serosal invasions of the greater curvature of the gastric angle or antrum). In these three cases, the depth of tumor invasion could hardly be determined using axial images because of partial volume artifact where lesions were tangent to the axial scan. Invasion to other organs is an important prognostic factor and, therefore, it may be important to visualize the precise relationship between the gastric tumor and other organs without partial volume phenomenon using high-quality MPR. On the other hand, accurate distinction between T2 and T3 tumors was sometimes difficult even using the MPR technique. Dux et al. [4] reviewed helical hydro-ct (using axial images) in 112 patients with gastric carcinoma and concluded that CT was inadequate for accurately diagnosing serosal infiltration of gastric carcinoma. In our study, three patients with deformities of the outer surface of the gastric wall were staged as T3 on CT, but serosal infiltration was not histologically revealed. Deformity of the outer surface of the gastric wall can occur as a result of factors other than serosal invasion, such as compression of the serosa by tumor (Fig. 7). Conversely, in two patients with pt3 tumor, serosal infiltration was not diagnosed on CT because the outer surface of the gastric wall was closely attached to the abdominal wall or the liver, resulting in understaging as T2 on CT. Thus, MDCT had some limitations in the evaluation of serosal infiltration. The limitation of our study is that normal cases, in which gastric cancer was absent, were included in our study. Therefore, it may be difficult to weigh the true value of sensitivity and specificity of CT for tumor detection. In conclusion, high-quality thin-sliced MPR using MDCT with the water-filling method has advantages in reliable detection of advanced gastric cancers, acceptable evaluation of the depth of invasion of gastric carcinomas, and visualization of histologic changes such as mucinous component or differences in infiltration of cancer cells, although there are some difficulties in the depiction of flat mucosal cancer and in accurate distinction between T2 and T3 tumors in some cases. References 1. Lee DH. Two-dimensional and three-dimensional imaging of gastric tumors using spiral CT. bdom Imaging 2000; 25: Kadowaki K, Murakami T, Yoshioka H, et al. Helical CT imaging of gastric cancer: normal wall appearance and the potential for staging. Radiat Med 2000; 18: Sohn KM, Lee JM, Lee SY, hn Y, Park SM, Kim KM. Comparing MR imaging and CT in the staging of gastric carcinoma. JR 2000; 174: [Erratum in JR 2000; 175:556] 4. Dux M, Richter GM, Hansmann J, Kuntz C, Kauffmann GW. Helical hydro-ct for diagnosis and staging of gastric carcinoma. J Comput ssist Tomogr 1999; 23: Takao M, Fukuda T, Iwanaga S, et al. Gastric cancer: evaluation of triphasic spiral CT and radiologic pathologic correlation. J Comput ssist Tomogr 1998; 22: Park MS, Yu JS, Kim MJ, et al. Mucinous versus nonmucinous gastric carcinoma: differentiation with helical CT. Radiology 2002; 223: Wu CY, Yeh HZ, Shih RTP, Chen GH. clinicopathologic study of mucinous gastric carcinoma including multivariate analysis. Cancer 1998; 83: Hyung WJ, Noh SH, Shin DW, et al. Clinicopathologic characteristics of mucinous gastric carcinoma. Yonsei Med J 1999; 40: Habu H, Takeshita K, Sunagawa M, Endo M. Prognostic factors of early gastric cancer: results of longterm follow-up and analysis of recurrent cases. Jpn J Surg 1987; 17: Isozaki H, Fujii K, Nomura E, et al. Prognostic value of tumor cell proliferation and intratumor microvessel density in advanced gastric cancer treated with curative surgery. Int J Oncol 1998; 13: Yamamura Y, Nakajima T, Ohta K, et al. Determining prognostic factors for gastric cancer using the regression tree method. Gastric Cancer 2002; 5: Zhang DS, Shimayama S, Kaminishi M. Feasibility of pylorus-preserving gastrectomy with a wider scope of lymphadenectomy. rch Surg 1998; 133: Kitano S, Shiraishi N. Current status of laparoscopic gastrectomy for cancer in Japan. Surg Endosc 2004; 18: JR:185, November 2005