is an Instructor of Radiology,
is an Associate Professor of Radiology, and
is a Professor of Radiology and Oncology, and the Director of
Diagnostic Imaging and Body CT, Johns Hopkins University School
of Medicine, Baltimore, MD. Dr. Fishman is also a member of the
editorial board of this journal.
The development of multidetector computed tomography (MDCT) has
provided improved visualization and evaluation of the stomach. Due
to their ability to provide excellent mucosal detail, endoscopy and
upper gastrointestinal imaging (UGI) series have traditionally been
considered the primary modalities for the detection of gastric
carcinoma. Previously, this exquisite detail was unmatched by
conventional single-slice computed tomography (CT). With
advancements in computer and scanner technology, as well as
development of 3-dimensional (3D) software programs, MDCT now
allows improved detection and staging of stomach cancer. In
addition, evaluation of intramural, extraluminal, and metastatic
disease is not possible with endoscopy or UGI, but is readily
performed with the cross-sectional capabilities of MDCT. This
article illustrates the usefulness of MDCT and 3D CT for the
evaluation of gastric carcinoma.
Thorough examination of the stomach requires optimal gastric
distension and properly timed intravenous (IV) contrast
administration. The stomach wall enhances significantly with
contrast administration, and this enhancement pattern can be
altered by underlying pathology. Therefore, IV contrast
administration is essential to detect disease. With the faster
scanning capabilities of MDCT, this enhancement can be optimally
timed to detect subtle mural abnormalities. In addition, the
stomach must be adequately distended with proper oral contrast.
Underdistension of the stomach can cause lesions to be missed or
can produce pseudolesions, which can lead to over- or misdiagnosis.
Gastric distension can be accomplished using positive oral contrast
agents (diluted barium sulfate or iodinated solution), neutral
substances (water), or negative contrast material (air).
Water has emerged as an optimal oral contrast agent and several
papers have shown its usefulness in delineating gastric disease.
Water is well-tolerated, inexpensive, results in good distension,
and does not interfere with 3D manipulation of data sets. The
relatively low attenuation of water (0 to 10 HU) produces excellent
contrast differentiation between the enhancing wall of the stomach
and the intraluminal water (Figure 1). Subtle enhancement
differences of mural pathology may be masked by positive contrast
agents in the adjacent lumen. The gastric wall may enhance up to
120 HU, which can be silhouetted by the positive oral contrast
In addition, inadequate mixing of positive contrast agents and
retained gastric products has led to pseudotumors.
At our institution, the patient is given 750 mL of water
approximately 15 to 30 minutes before the study. Immediately prior
to scanning, an additional 250 mL of water is given. Nonionic
contrast material (120 mL of Omnipaque 350, GE Healthcare,
Princeton, NJ) is routinely injected at a rate of 3 mL/sec.
Arterial-phase images are acquired 25 seconds after the beginning
of contrast injection. Venous-phase images are obtained 50 seconds
after the start of the injection.
For detailed imaging of the stomach and 3D reconstruction, we
utilize our 16-slice MDCT scanner (Sensation 16, Siemens Medical
Solutions, Malvern, PA) to obtain 0.75-mm slices, acquired at a
rate of 32 slices per second. These thin slices acquired at a fast
rate result in high-resolution data sets with minimal artifact.
This dual-phase protocol produces angiographic-quality delineation
of gastric arteries and veins as well as visualization of liver
Previously, data acquired could be evaluated only in the axial
plane, which is inadequate for the complex anatomy of the stomach.
Although MDCT still acquires data in the axial plane, data is
acquired as a volume set rather than sequences of adjacent slices.
For this reason, the data can be seamlessly manipulated with 3D
software, allowing it to be viewed in multiple planes with the
addition of depth perception, which allows for better understanding
and evaluation of anatomy and pathology. In addition, gastric
pathology can be easily visualized in relationship to adjacent
structures. For the stomach, coronal or coronal oblique planes are
often helpful for delineation of pathology. Referring physicians
often like this view, as it simulates the traditional familiar UGI
This 3D software allows manipulation of window width and level
to accentuate desired structures, such as soft tissue, vessels, or
even bone. CT angiography (CTA) can be performed using the same
data set and can reveal effects of gastric pathology on adjacent
vessels. This information can be valuable in determining a
potential approach for surgery. The CT data set can also be
windowed and manipulated to produce endoscopic images, giving a
virtual gastroscopy, similar to virtual colonoscopy pictures. In
this setting, the gastric mucosal folds are accentuated and
intraluminal pathology is highlighted (Figures 2 and 3).
Adenocarcinoma accounts for approximately 95% of all malignant
gastric neoplasms, typically occurring between the ages of 50 and
70 years with a male predominance. The distribution tends to be
geographic, with the highest frequency in Japan, Chile, and
Predisposing conditions include atrophic gastritis, pernicious
anemia, gastric polyps, partial gastrectomy (such as Billroth II),
and Ménétrier's disease.
This aggressive disease has an overall 5-year survival rate of
However, detection of early carcinoma offers the possibility of a
Some surgeons opt for laparotomy in every case of gastric
carcinoma, with the outcome either being curative resection or
palliative therapy. Other physicians rely on imaging modalities for
staging and therapy planning. CT is the most widely used modality
for staging of gastric adenocarcinoma. Reported accuracy for CT
staging of gastric cancer varies due to differences in equipment
and scanning protocols. In a study by Hori et al
using single-detector CT with IV contrast and water as oral
contrast, examinations detected 95% of advanced gastric cancers,
93% of elevated carcinomas, and 18% of early depressed carcinomas.
With MDCT allowing more narrow collimation and thinner, more
closely spaced scanning, it is probable that this accuracy will
In addition, though controversial, dual-phase imaging may be
beneficial in detecting and staging stomach cancer. In studies by
Hundt et al
and Mani et al,
dual-phase imaging was performed with water as an oral contrast
agent. In a study of 40 patients, Hundt and colleagues
were able to detect 97.5% of cancers correctly with CT
staging-79.4% accuracy compared with pathologic staging. Mani and
coworkers studied 20 patients and concluded that earlier-phase CT
was highly accurate in determining the depth of tumor invasion
through the wall, correctly determining the depth in 17 of 20
patients. Later-phase images are useful in detecting metastatic
spread, especially liver metastasis. Optimal imaging protocols for
detection and staging are still under investigation.
Adenocarcinoma of the stomach has varying presentations. The
purpose of CT is to detect gastric tumors, determine the stage, and
analyze for metastasis.
With optimal gastric distension, the wall of the stomach is very
thin and its folds become essentially effaced. Normal gastric wall
thickness, on average, measures 5 to 7 mm. Walls thicker than 8 to
10 mm are abnormal. On CT, gastric tumors may appear as a discrete
mass with or without ulceration or may appear as an infiltrative
lesion. These appearances were described in a paper by Ba-Ssalamah
that attempts to correlate tumor pathologic type and depth of wall
invasion using MDCT (Figure 4). Infiltrating tumors can be focal,
segmental, or diffuse. Diffusely infiltrating tumors can cause
gastric rigidity, resulting in linitis plastica shown by limited
distension after administration of oral contrast. The average wall
thickness for this malignant neoplasm is 2 cm, with a range of 6 mm
to 4 cm.
Though there have been significant technologic CT advances, the
limitation of spatial resolution hampers distinction of gastric
layers and, therefore, distinction of depth of tumor invasion.
Moss et al
developed a widely used staging system for gastric adenocarcinoma:
Stage I disease-intraluminal mass invading gastric mucosa without
deeper tumor extension; stage II disease-gastric wall thickness
>1 cm, which is compatible with tumor spread into the muscularis
propria; stage III disease-tumor invasion through the muscularis
propria and serosa with or without local adenopathy; and stage IV
disease-tumor invasion of adjacent organs or distant
Stage I and stage II disease are considered limited and are
treated with partial gastrectomy (Figure 5). Stages III and IV are
considered locally advanced and are treated with chemotherapy. In
advanced disease, surgery is reserved for palliative measures.
Stage III disease indicates local disease composed of
extragastric extension. This tumor spread can be subtle, manifested
as slightly increased attenuation of the normally homogenous
perigastric fat to complete obliteration of fat planes between
organs with replacement by soft tissue due to tumor. This
distinction can be difficult and can be mimicked by concurrent
inflammatory conditions, such as pancreatitis. The probability of
transmural extension increases in proportion to mural thickness of
the stomach wall. Disease commonly spreads along ligaments and
peritoneal reflections to extend to nearby organs.
Common pathways of extension include the liver via the
gastrohepatic ligament, the pancreas via the lesser sac, and the
transverse colon via the gastrocolic ligament. Carcinoma of the
cardia may extend to directly involve the distal esophagus, while
tumor of the antrum may spread to the duodenum. Three-dimensional
manipulation of the data sets may help with the detection of local
spread (Figure 6).
Lymph node involvement occurs in 74% to 88% of people with
gastric malignancy because of the extensive lymphatic vessels of
Lymphatic involvement increases relative to the size and depth of
the primary lesion. Nodal malignancy decreases the median survival
rate by 65%.
The best CT indicator of tumor extension to lymph nodes is
enlargement. This spread commonly occurs in the region of the
gastrohepatic ligament. Perigastric nodes are suspicious for nodal
spread of disease when they measure >8 mm in diameter.
Normal-size nodes may contain tumor, although enlarged ones do not.
Multiple enlarged nodes are considered suspicious. The sensitivity
for detection of nodal spread by CT ranges from 47% to 97%,
primarily limited by inability to detect microscopic spread
Gastric cancer frequently produces hematogenous metastases to
the liver via gastric drainage through the portal vein. Other less
frequent target organs that develop metastases include the lungs,
adrenal glands, kidneys, bones, and brain. Peritoneal seeding may
occur in advanced cases. Drop metastases to the ovaries are known
as Krukenberg tumors. Multidetector CT can detect this spread of
disease and allow for appropriate planning and treatment
Gastric adenocarcinoma is a deadly disease. Presentation varies
from subtle mural abnormalities to widespread disease. Technologic
advances, including MDCT and 3D capabilities, allow for improved
detection and staging of stomach disease in conjunction with
conventional imaging methods. With these improved imaging
capabilities, MDCT may play an increasing role in evaluation and