Professional Documents
Culture Documents
Radiology
Published online before print MATERIALS AND METHODS: A radiology department file search revealed 36
10.1148/radiol.2281020623 patients with a thickened gastric wall at CT who underwent double-contrast barium
Radiology 2003; 228:166 –171 suspension upper gastrointestinal tract examinations within 6 weeks before or after
CT. The authors reviewed the CT images without knowledge of the final radiologic,
Abbreviation:
UGI ⫽ upper gastrointestinal endoscopic, or pathologic findings to determine the degree of gastric wall thicken-
ing and the symmetry, distribution, and enhancement of the thickened wall. The
1
sensitivity and specificity of these findings for detection of malignancy were calcu-
From the Department of Radiology,
Hospital of the University of Pennsyl- lated.
vania, 3400 Spruce St, Philadelphia,
PA 19104. Received May 28, 2002; RESULTS: Two of 36 patients had two gastric abnormalities each. The final diag-
revision requested July 26; revision re- noses in the 38 cases were gastritis in 19, hiatal hernia in four, benign ulcer in three,
ceived September 19; accepted No- benign (n ⫽ 3) or malignant (n ⫽ 8) gastric neoplasm in 11, and no gastric
vember 18. Address correspondence
abnormality in one case. Mean wall thickness was 1.5 cm (range, 0.7–7.5 cm). The
to E.K.I. (e-mail: insko@rad.upenn.
edu). finding of gastric wall thickness of 1 cm or greater had a sensitivity of 100% but a
Current address: specificity of only 42% for detection of malignant or potentially malignant stomach
2
Department of Radiology, New York lesions. The finding of focal, eccentric, or enhancing wall thickening had a sensitivity
University Medical Center, New York, of 93%, 71%, or 43%, respectively, and a specificity of 8%, 75%, or 88%, respec-
NY.
tively, for detection of these lesions. Gastric wall thickening that was 1 cm or greater
and was focal, eccentric, and enhancing had a specificity of 92% but a sensitivity of
only 36% for detection of these lesions.
CONCLUSION: Gastric wall thickness of 1 cm or greater at CT had a sensitivity of
100% but a specificity of less than 50% for detection of malignant or potentially
malignant stomach lesions that necessitated further diagnostic evaluation.
© RSNA, 2003
When evaluating the stomach with computed tomography (CT), it is important to be able
to differentiate benign conditions such as gastritis from ulcers or neoplasms that necessi-
tate further assessment with barium or endoscopic examinations. However, evaluation of
the stomach at CT may be limited by a variety of factors, including gastric peristalsis,
Author contributions: incomplete distention, an inadequate volume of oral contrast material, and retained food.
Guarantor of integrity of entire study, The results of several previously performed studies (1– 4) have shown that a normal gastric
E.K.I.; study concepts and design, all wall usually has a thickness of 5 mm or less at CT, whereas a gastric wall involved by tumor
authors; literature research, E.K.I.; clin-
ical studies, B.A.B., J.E.J.; data acquisi-
usually has a thickness of 1 cm or greater. However, in the radiologic literature, apart from
tion, all authors; data analysis/inter- the data on degree of wall thickness, there is little information regarding the CT criteria for
pretation, E.K.I., M.S.L.; statistical differentiating normal stomach or benign conditions such as gastritis from potentially
analysis, E.K.I.; manuscript prepara- malignant lesions that necessitate further diagnostic evaluation.
tion, definition of intellectual content,
Our experience has been that patients who are suspected of having gastric neoplasms
and editing, E.K.I., M.S.L.; manuscript
revision/review and final version ap- because of a thickened wall at CT not infrequently are found to have gastritis or even a
proval, all authors. normal stomach at barium or endoscopic examinations. Considering the established CT
© RSNA, 2003 criteria for differentiating benign from malignant lesions of the small or large bowel (5), we
wondered whether the degree of gastric wall thickening combined with other criteria such
166
as symmetry, distribution, and enhance- col. All patients received oral contrast abnormally thickened gastric wall (1,2).
ment of the thickened gastric wall might material (600 – 800 mL of 2%–3% dia- The thickened wall was also evaluated for
enable radiologists to better differentiate trizoate meglumine and diatrizoate so- symmetry (ie, eccentric or asymmetric vs
benign from malignant conditions in- dium [Gastrografin]; Bristol-Myers Squibb, circumferential or symmetric), distribu-
volving the stomach at CT. With this Princeton, NJ, or a 2.1% wt/vol barium tion (ie, focal vs diffuse), and presence or
background, we performed a blinded ret- suspension [Readi-cat], E-Z-Em, Westbury, absence of enhancement after intrave-
Radiology
rospective study involving a large series NY) 30 – 45 minutes before the CT exam- nous contrast material administration.
of patients who underwent abdominal ination and an additional 400 –500-mL
CT, as well as barium suspension exami- dose of the oral contrast material fol-
Final Diagnosis
nation, endoscopy, and/or surgery, to de- lowed by 3 g of an oral effervescent agent
termine the sensitivity and specificity of dissolved in 8 ounces of water (Baros; The reports from the double-contrast
CT criteria in the differentiation of be- Lafayette Pharmaceuticals, Lafayette, Ind) UGI examinations, as well as those from
nign and malignant disease involving immediately before the examination. All the endoscopic, surgical, and/or patho-
the stomach in patients with a thickened patients also received a 150-mL injection logic examinations, were reviewed by
gastric wall at CT. of 60% iodinated contrast material (dia- one author (E.K.I.). If the patient under-
trizoate meglumine [Hypaque] or iohexol went double-contrast UGI examination
MATERIALS AND METHODS [Omnipaque 300]; Nycomed Amersham, but not endoscopy, the UGI findings
Princeton, NJ) through an antecubital were accepted as the final diagnosis or
Patient Selection vein at a rate of 2–3 mL/sec. The CT im- the reference standard. If the patient un-
A computerized search of our hospi- ages were routinely obtained with the pa- derwent endoscopy or surgery, as well as
tal’s radiology files by one author (E.K.I.) tient in a supine position during full in- double-contrast UGI examination, the
during a 3-year period from 1997 to 1999 spiration. In one case, prone-position CT endoscopic or histopathologic findings
revealed 152 patients with a thickened images of the stomach also were ac- in the endoscopic biopsy specimens or
gastric wall at abdominal CT (keyword quired. CT images of the upper abdomen resected surgical specimens were ac-
searches: thickened, bulky, prominent, were obtained by using either 5-mm col- cepted as the reference standard. Al-
abnormal) who had also undergone dou- limation and a pitch of 1.5:1.0 or 7-mm though it is conceivable that one or more
ble-contrast barium suspension examina- collimation and a pitch of 1.3:1.0 (200 – malignant tumors could have been
tions of the upper gastrointestinal (UGI) 220 mAs); transverse images were recon- missed on the barium images, we believe
tract. One hundred sixteen patients were structed with a soft-tissue algorithm. that the findings on these images were a
excluded from analysis for any of the fol- reasonable reference standard in the ab-
lowing reasons: (a) The CT study was per- sence of endoscopic or pathologic corre-
formed without intravenously adminis- UGI Examination Technique lation, because double-contrast UGI stud-
tered contrast material, (b) the CT images ies have been shown to have a sensitivity
All patients underwent double-con-
could not be retrieved from our image of greater than 95% in the detection of
trast examinations of the UGI tract that
archives, (c) a blinded review of the CT gastric carcinomas (7). The final diagno-
were performed by using digital fluoro-
images revealed no definite abnormali- sis was based on the UGI findings in 16
scopic equipment (Diagnost 76 Plus;
ties of the gastric wall and no inadequate patients, on the histopathologic findings
Philips, Eindhoven, the Netherlands).
distention of the stomach, (d) the inter- at endoscopy alone in 13 patients, on the
These studies, which were performed as
val between CT and barium suspension endoscopic and surgical findings in six
biphasic examinations by using an effer-
examination was longer than 6 weeks, patients, and on the surgical findings
vescent agent (Baros) and a 250% wt/vol
and (e) the patients had undergone gas- alone in one patient. Two patients each
barium suspension (E-Z-HD; E-Z-Em) fol-
tric surgery previously. were found to have two separate abnor-
lowed by a 50% wt/vol barium suspen-
The remaining 36 patients comprised malities in the stomach. The diagnoses of
sion (Entrobar; Lafayette Pharmaceuti-
our study group. These 36 patients had a these lesions were based on the his-
cals), have been described previously (6).
total of 38 suspected gastric abnormali- topathologic findings at endoscopy alone
All UGI examinations were performed by
ties at CT. The mean interval between CT in one patient and on the findings at
a supervised resident (E.K.I.) or an at-
and double-contrast UGI barium suspen- endoscopy and surgery in the other pa-
tending UGI radiologist (M.S.L.).
sion examination was ⫾12.1 days (range, tient. There were no discrepancies be-
1–32 days). The mean age of the patients tween the UGI examination findings and
was 52 years (age range, 35– 80 years). the endoscopic or histopathologic find-
Image Review
Seventeen patients were men and 19 ings in the 19 patients who underwent
women. Our institutional review board The images acquired at CT scanning in endoscopy or surgery and barium (ie,
approved all aspects of this retrospective the 36 patients were reviewed by consen- UGI) examinations.
study and did not require informed con- sus between two CT radiologists (B.A.B.,
sent from the patients whose records 13 years experience; J.E.J., 10 years expe-
Criteria for Evaluating Accuracy
were included in our study. rience) who had no knowledge of the
of CT
final radiologic, endoscopic, or patho-
logic findings. The greatest visible gastric For the purposes of this study, all pa-
CT Technique
wall thickness was measured with elec- tients with a final diagnosis of a gastric
All patients included in the study un- tronic calipers at a picture archiving and ulcer or neoplasm were considered to
derwent helical CT of the abdomen (with communications, or PACS, workstation. have potentially malignant lesions that
a CT HiSpeed Advantage or HiSpeed CT/I All 38 cases involved a gastric wall thick- warranted further diagnostic evaluation
unit; GE Medical Systems, Milwaukee, ness of greater than 5 mm, which is a with a double-contrast UGI or endo-
Wis) according to an established proto- previously established CT criterion for an scopic examination. Ulcers were in-
Volume 228 䡠 Number 1 CT Differentiation of Benign and Malignant Stomach Lesions 䡠 167
RESULTS
Volume 228 䡠 Number 1 CT Differentiation of Benign and Malignant Stomach Lesions 䡠 169
further evaluation with barium or endo-
scopic examinations for a definitive diag-
nosis. Nevertheless, such an appearance
of the thickened gastric wall was present
at CT in only 36% of patients with ma-
Radiology
was focal, eccentric, and enhancing almost 3. Moss AA. Computed tomography in the phy. J Comput Assist Tomogr 1984;
always indicated malignant tumor, but staging of gastrointestinal carcinoma. Ra- 4:699 –700.
diol Clin North Am 1982; 20:761–780. 10. Halvorsen RA. Thompson WA. Com-
this constellation of findings was present 4. Scatarige JC, DiSantis DJ. CT of the stom- puted tomography of the gastrointestinal
in only a small percentage of cases. Radiol- ach and duodenum. Radiol Clin North junction. Crit Rev Diagn Imaging 1984;
ogists should be aware of the usefulness of Am 1989; 27:687–706. 21:83–228.
these specific criteria in the evaluation of 5. Macari M, Balthazar EJ. CT of bowel wall 11. Kaye MD, Yound SW, Hayward R, Cas-
thickening: significance and pitfalls of in- tellino RA. Gastric pseudotumor on CT
gastric wall thickening at CT to better dif-
terpretation. AJR Am J Roentgenol 2001; scanning. AJR Am J Roentgenol 1980; 135:
ferentiate gastritis or a normal stomach 176:1105–1116. 190 –193.
from malignant or potentially malignant 6. Levine MS, Rubesin SE, Herlinger H, Laufer 12. Komaki S. Normal or benign gastric wall
lesions that warrant further diagnostic I. Double contrast upper gastrointestinal thickening demonstrated by computed
evaluation with barium or endoscopic ex- examination: technique and interpreta- tomography. J Comput Assist Tomogr
tion. Radiology 1988; 168:593–602. 1982; 6:1103–1107.
aminations. 7. Low VHS, Levine MS, Rubesin SE, Laufer 13. Rockey DC, Halvorsen RA, Higgins JL,
I, Herlinger H. Diagnosis of gastric carci- Cello JP. Prospective evaluation of pa-
References noma: sensitivity of double-contrast bar- tients with bowel wall thickening. Am J
1. Lee KR, Levine E, Moffat RE, Bigongiari LR, ium studies. AJR Am J Roentgenol 1994; Gastroenterol 1995; 90:99 –103.
Hermreck AS. Computed tomographic 162:329 –334.
Volume 228 䡠 Number 1 CT Differentiation of Benign and Malignant Stomach Lesions 䡠 171