Diagnostic Imaging Asia Pacific
December 2001
Asian Oceanian Congress of Radiology Conference Reporter
CT, MRI, CAD expand options for chest imaging
Innovations make lung screening for cancer and tuberculosis a possibility
By Dan Krotz
A steep rise in the incidence rates of pathologies such as lung cancer and AIDS-related tuberculosis has made the role of chest imaging in curbing mortality increasingly important. At the same time, technological advances such as spiral CT, three-dimensional imaging, and computer-aided detection have bolstered radiologists’ ability to detect these diseases at earlier stages.
It’s little wonder then that the organizers of the 9th Asian Oceanian Congress of Radiology devoted six sessions to chest imaging. Radiologists from several continents discussed the latest techniques and explored the touchy subject of how and when to implement screening programs.
Acute respiratory distress syndrome has a fatality rate of about 50%, according to the American Lung Association. It is characterized by rapid onset of lung malfunction, due in part to extensive inflammation and small blood vessel injury in the surrounding tissue. Complicating matters, the disease is often misdiagnosed, largely because it is commonly triggered by more generalized pathology such as trauma, systemic infection, diffuse pneumonia, or shock.
Because of these confounding factors, radiographic characterization of the disease is difficult, according to Dr. Lawrence R. Goodman, cochief of thoracic imaging at the Medical College of Wisconsin in the U.S. Until a few years ago, CT imaging had no favorable effect on these patients’ clinical outcome, mostly because the disease was too far advanced by the time they were scanned.
Goodman, however, has performed 70 high-resolution CT scans on 33 patients with early- and late-stage acute respiratory distress syndrome to learn the early radiological manifestations of the disease, which can in turn be used to ensure that patients receive early-stage therapy. Patients with early-stage secondary forms of the disease have symmetric ground-glass opacification of the lungs, while patients suffering from primary forms of the disease exhibit asymmetric alveolar opacification. In both cases, air bronchograms and pleural effusions are almost always present. These distinctions blur roughly two weeks into the disease’s progression, as the pathology matures and becomes more fibrotic.
Goodman is also starting to debunk some long-held beliefs.
“When acute respiratory distress syndrome was initially described, the lungs were thought to be stiff, meaning it’s difficult to ventilate the patient. But CT has shown that lungs are actually not stiff,” he said.
CT has also shown that acute respiratory distress syndrome is not a homogeneous disease of the lung, Goodman said. It’s not a stationary disease either, especially over the first few days. Understanding these characteristics should change the way radiologists image the disease.
Despite CT’s imaging prowess, radiography remains the preferred modality, Goodman said. CT is used only after the presence of the disease is clinically or radiographically confirmed and the need to search for any suspected complications is established.
Tuberculosis is another common chest imaging indication. The World Health Organization estimates that TB kills two million people annually. About eight million people become sick with TB each year, with nearly three million of these cases occurring in Southeast Asia, according to Prof. Jung-Gi Im, a radiologist with Seoul National University Hospital in Korea and a longtime pulmonary tuberculosis imaging specialist.
High-resolution CT is extremely helpful in tracking the spread of tuberculosis lesions, Im said. The radiological indications of the disease vary widely and are affected by the robustness of the patient’s immune system, but yellowish-white nodules and air-space consolidation, as well as a strong likelihood of cavitations, often characterize pulmonary tuberculosis. The first signs of bronchogenic spread of tuberculosis lesions may appear on high-resolution CT scans as 2 to 4-mm nodules or branching lesions with a distinct tree-in-bud appearance. In primary tuberculosis, the infection often spreads to the hilar and mediastinal lymph nodes. The subsequent enlarged lymph nodes can often be detected on contrast-enhanced CT images because of their low attenuation and peripheral rim enhancement.
Complete healing of a tuberculosis lesion is rare, and it’s not uncommon for a patient to have symptoms for 20 to 30 years, according to Im.
Although tuberculosis is an age-old disease, the combination of tuberculosis and AIDS is a modern and particularly nettlesome diagnostic dilemma. The connection between the two diseases is simple: The human immunodeficiency virus (HIV), which causes AIDS, attacks T lymphocytes, which are the body’s chief defense against TB.
The need to take a close look at the distinct radiological findings of the simultaneous presence of these two diseases was underscored by a 1989 study in The New England Journal of Medicine that found that roughly 14% of patients who tested positive for HIV developed tuberculosis within two years. Other studies have determined that up to 31% of tuberculosis patients in the U.S. and 73% of tuberculosis patients in Zambia are infected with HIV.
The radiological findings of a patient with tuberculosis and HIV infection hinge on how severely the body’s immune system is suppressed, Im said. Although CT scans and chest radiographs of such patients are often equally useful in determining the extent of the infections, CT outshines radiographs in determining the presence or absence of mediastinal lymphadenitis and the cavitation of pulmonary lesions.
Chest CT and MRI
Multislice CT and highly sensitive MRI protocols are picking up where chest radiography leaves off. Many studies have determined that CT is better than plain film in both detecting and characterizing peripheral pulmonary nodules, according to Dr. Shuji Adachi, an associate professor of radiology at Kobe University School of Medicine in Japan.
“By combining the advantages of a single breath-hold acquisition with improved vascular contrast, multidetector CT’s role in imaging many chest pathologies is dramatically improved,” Adachi said.
As with many advances in radiology, the benefit of CT lies not only in how well it images pathologies, but in its ability to replace more invasive procedures. In pulmonary emboli cases, for example, CT angiography provides more vascular information than pulmonary angiography.
Other CT applications include establishing the presence of arteriovenous malformations and the sequestration of the pulmonary vessels, as well as assessing vascular invasion of hilar lung cancer. Three-dimensional vascular CT cannot depict anatomy outside the vessels, however, meaning conventional imaging procedures will remain in use.
As for MRI, the modality is particularly useful as an adjunct to x-ray imaging in deciphering complicated cases, such as distinguishing lung cancer from tuberculoma and evaluating responses to tumor therapy, Adachi said.
“High-resolution images of the entire thorax can be obtained with a single 25 to 30-second breath-hold sequence,” he said. “MRI also shows particular promise if we can develop disease-specific contrast agents.”
The next generation of contrast agents will target the genetic and molecular underpinnings of disease, instead of morphological anomalies. Researchers at Washington University in St. Louis, Missouri, are developing a contrast that homes in on fibrin, which accumulates on arterial plaque wherever it has suffered tiny fractures-a sure sign of instability that can lead to heart attack and stroke. Such contrast agents could revolutionize MRI’s role in the early detection of disease.
MRI offers high spatial resolution for morphological imaging, and high temporal resolution for functional imaging, according to Prof. Dr. Maximilian Reiser of the University of Munich. Contrast-enhanced MRI can image pulmonary and coronary arterial blood flow.
“MRI allows for reliable, precise, and reproducible imaging of the global and regional ventricular function,” Reiser said. “And it is easy to combine spatial and temporal images.”
Specifically, by combining respiratory gating and electrocardiogram-triggering, MRI can obtain a continuous 3-D data set of the heart, Reiser said. This technique circumvents the problems associated with 3-D breath-hold coronary angiography, which has an acquisition time that is too long to image a beating heart or a breathing lung with high resolution.
When compared with MRI, multidetector CT offers superior spatial resolution, inferior temporal resolution, and faster scanning, he said. CT is also less user-dependent. But CT requires radiation, and Reiser concludes that MRI is better than CT in many chest applications, especially when imaging coronary arteries, when MRI’s functional capabilities can be utilized.
Lung Cancer Screening
Although Japan hasn’t suffered the high breast cancer incidence rates that plague many Western nations, the same can’t be said about lung cancer. Last year, there were about 48,000 new cases of the disease, and 40,000 deaths, according to the World Health Organization.
“Lung cancer is the leading cause of cancer death in men and women in Japan,” said Prof. Michio Kono, director of the Hyogo Medical Center for Adults in Kobe, Japan.
Still, no major advisory organization, including the American Cancer Society, recommends routine screening for lung cancer, Kono said. Screening-detected lung cancers are usually too far advanced to affect patient mortality. Radiography, the conventional screening tool, fails to detect many well-differentiated adenocarcinomas smaller than 2 cm. Although the modality provides excellent spatial resolution of the lungs, surrounding tissue such as the mediastinal, cardiac, and hemidiaphragmatic areas are often underexposed, meaning many abnormalities go undetected.
Even so, the Japan Ministry of Welfare supports using chest x-rays and sputum cytology to screen very high risk patients such as symptomatic chronic smokers. Hoping to net more early-stage, treatable cancers, a growing number of imaging specialists are starting to suggest using CT as a screening tool, particularly when directed toward high-risk asymptomatic patients who may have low-density nodules less than 2 cm in size. These patients are especially receptive to therapy yet are often given a clean bill of health under conventional screening methods.
As with any screening program, particularly one that employs a state-of-the-art modality, cost-effectiveness is a key consideration. According to a study conducted at the Kanagawa Cancer Center in Yokohama, Japan, and published in the journal Cancer (2000;89:2489-2493), CT screening is more expensive but also much more effective than x-ray screening. The study authors compared the present method of lung cancer screening-x-ray and sputum cytology-with a CT-based screening program. Based on their findings, they concluded that CT is the most appropriate screening method. Kono offers a similar rationale.
“With spiral CT, the cost is US$4000 per year of saved life, which is an acceptable cost range,” Kono said. “If small nodular shadows are found on chest radiography, but pathological diagnosis cannot be detected, careful follow-up should be done by high-resolution CT every three months, six months, and one year.”
The Case for CAD
One of the long-standing problems in any image-based screening program is not the technology, but the overwhelming number of images that must be interpreted. If low-dose CT is ever used to form the backbone of a lung cancer screening program, the modality’s highly touted anatomical resolution will likely add to the mountains of data that must be interpreted.
“The more images we are asked to look at, the more errors, observationally and perceptually, we will make,” said Dr. Ronald A. Castellino, medical director of R2 Technology in Los Altos, California, and a professor of radiology at Stanford University.
Conversely, the amount of data that must be analyzed also exacerbates the very real problem of labeling otherwise benign lesions as suspicious.
“Overdiagnosis will also be a problem,” Castellino said. “We may find nodules that are so small the patient is more likely to die of a heart attack than lung cancer. We could run the risk of turning healthy people into patients in terms of finding small nodules that will never be symptomatic.”
That’s where computer-aided detection comes in. CAD in lung cancer imaging doesn’t take a radiologist’s place, according to Castellino. Rather, the sophisticated, algorithm-based software is essentially a spell check of medical images.
So far, experimental forays into CAD have posted promising results. The U.S. Food and Drug Administration approved the following statement: For every 100,000 cancers detected by screening mammography, the use of CAD could result in the early detection of an additional 20,500 breast cancers.
Future CAD protocols will boast increased sensitivity and a decrease in false diagnosis, Castellino said. He conceded that it isn’t yet known whether lung screening reduces mortality.
“We would be doing so much more, and we don’t have the resources now,” Castellino said. “In the future, to reduce the burden, CAD could be used as a first screening, because we are drowning in data overload now.”
