Gottschalk plots strategy to preserve V/Q diagnosis

Ventilation/perfusion imaging for detection of pulmonary embolism fills in when CT angiography/venography is contraindicated

Disappointed by some results from the Prospective Investigation of Pulmonary Embolism Diagnosis trial, Dr. Alexander Gottschalk is advocating changes in the way CT, nuclear medicine, and clinical evaluations should work together to diagnose pulmonary embolism.

Gottschalk, a professor of diagnostic radiology at Michigan State University, fears a possible risk of breast cancer for patients exposed to ionizing radiation from multislice CT for PE diagnosis. As an alternative to CT, he recommends ventilation-perfusion (V/Q) imaging to diagnose PE for women in their child-bearing years. V/Q should be performed whenever CT is contraindicated and when contradictions between clinical evaluations and CT for PE arise, he said.

The multicenter PIOPED II trial concluded that a reference diagnosis and CT angiography together fall short of rendering definitive diagnoses. CTA's sensitivity for identifying the presence of PE was just 83% for 824 patients, though combining a clinical evaluation with CTA and CT venography raised the sensitivity to 90%. Both approaches produced specificities of about 95%. The PIOPED II results were published in the June 1 issue of The New England Journal of Medicine (NEJM 2006;354[22]:2317-2327).

Publication corresponded with 2006 Society of Nuclear Medicine meeting, where Gottschalk, chair of the nuclear medicine working group of PIOPED II, was awarded the SNM's prestigious Benedict Cassen Prize. He used the Cassen Oration to place PIOPED II in context, citing the rising clinical popularity of multislice CT for diagnosing PE and drawing on more than 25 years of experience with nuclear medicine ventilation/perfusion (V/Q) studies for PE detection.

Growth of MSCT for PE diagnosis is undeniable, Gottschalk said. A Temple University Hospital study presented at the conference the previous day showed that its physicians ordered CT three times more often than V/Q studies to diagnose PE in 2005. Compared with the previous year, CTA utilization for diagnosing PE rose 20%, while V/Q utilization for PE fell 20%. The speed, convenience, and round-the-clock accessibility of CT are frequently cited as sources of its popularity.

"It can be a one-stop shop for venous thromboembolic disease, which is really what we're after," he said.

But CT is not the best test for all patients with suspected PE. The PIOPED II group, led by principal investigator Dr. Paul D. Stein, director of research education at St. Joseph Mercy Oakland Hospital in Pontiac, MI, found that contrast-enhanced CTA/CTV is contraindicated for 20% of patients with suspected PE. V/Q is prescribed as an alternative for pregnant patients and those with contrast allergies or high creatinine enzyme levels.

The Temple study found that 51% of PE-related V/Q studies were performed on patients undergoing renal failure. Physician preference was cited as the reason for 23% of V/Q scans. Nondiagnostic CT led to 12% of the V/Q tests, and pregnancy was cited as the rationale for 6%.

RADIATION CONCERNS

The biggest problem with pulmonary CTA stems from radiation exposure, Gottschalk said. Research performed by Dr. Eric N.C. Milne at the University of California, Irvine, and Dr. Mark S. Parker at the Medical College of Virginia Hospitals persuaded him that CT pulmonary angiography should not be performed on women during their child-bearing years. Direct measurements, including thermolucent dosimetry at various points on the breast performed on 64 patients by Milne, indicated that pulmonary CTA exposes the female breast to an average of 7 rads (70 mGy).

Parker's study of 1325 CT pulmonary angiograms found the effective minimum dose to the breast for the average women was 2 rads (20 mGy) per breast. In a letter to the American Journal of Roentgenology (AJR 2006;186:E24), Milne equated the radiation dose of the breast during CTA to 30 to 100 mammograms or 300 to 1000 chest radiographs.

High radiation exposure translates into a higher risk of cancer-related death. For evidence, Gottschalk cited a Canadian program that used serial fluoroscopy with therapeutic pneumothorax to treat tuberculosis patients from 1930 to 1952. Of 30,000 women in the program, those exposed to more than 10 rads had a 26% higher risk of breast cancer-related death than women exposed to less radiation. The risk was especially high among younger women.

"Dose to the breast from CTA is not a trivial problem," Gottschalk said.

V/Q testing can address the problem, he said. It can involve breast radiation 70 times lower than that delivered by pulmonary CTA.

Gottschalk believes that the data cited indicate that if patients are stratified to those with normal or near-normal chest x-rays and a very low probability V/Q scan, reliability can be comparable to CTA, and radiation exposure to many female patients in their reproductive years can be avoided.

While CTA may be the first choice for diagnosing PE, PIOPED II results indicate that it should not be the only test used to determine if the patient should receive PE therapy. CT alone cannot be trusted, Gottschalk said.

"I often hear this in the emergency room: 'This guy's got chest pain; let's get a CT.' But that's a superficial view, and I spell superficial here as s-u-c-k-s," Gottschalk said in the lecture.

COMBINED STUDIES

Performing both CTA and CTV is essential to optimizing sensitivity. Combining CT findings with a clinical assessment using Wells, Geneva, or revised Geneva criteria boosts diagnostic confidence even higher.

Gottschalk recommended an enzyme-linked immunosorbent assay (ELISA) D-dimer clinical evaluation in some cases before performing CTA/CTV. No imaging is required when the D-dimer is less than 500 mg and the clinical assessment finds a low or intermediate risk of PE, he said.

A D-dimer test is not indicated following a high-probability clinical assessment. In this circumstance, the negative predictive values of CTA and CTA/CTV are 60% and 82%, respectively. A high-probability CTA identifying a large clot in the lobar or main pulmonary artery is a definitive diagnosis for PE even without a high-probability clinical assessment. A V/Q scan is recommended when discordance arises between the clinical assessment and the CTA/CTV imaging exam, Gottschalk said.

Certainty is achieved, however, with a moderate probability clinical assessment and a negative rapid ELISA D-Dimer. The post-test probability of PE in this instance is just 5%. Although leg ultrasound is optional in this circumstance, CT and V/Q are not necessary, and the need for treatment is unlikely.

CTA/CTV is recommended when the D-dimer test fails to exclude PE after a moderate clinical assessment. Under these conditions, its NPV and PPV are 89% and 92%, respectively. A positive finding should lead to PE treatment, he said.

PIOPED II indicated that a normal or very low probability clinical assessment, coupled with negative findings from a leg ultrasound exam and a low- or very low probability V/Q scan, can be used to exclude the presence of PE. A simple chest x-ray can also confirm a patient as normal or near-normal after a very low probability clinical assessment and, with a low-probability V/Q scan, can produce a definitive diagnosis for many of these cases, Gottschalk said.

The following criteria were used in PIOPED II for V/Q studies to rule out PE in very low probability cases. Investigators looked for nonsegmental lesions and perfusion defects that were smaller than a radiographic lesion. Solitary triple-matched defects in the mid- or upper lung zone confined to a single segment confirmed that pulmonary infarcts rarely occur outside the lung basis. A stripe sign around a perfusion defect tells the clinician that the abnormality is not PE. A pleural effusion covering more than one-third of the pleural cavity without other perfusion defects in either lobe also indicates that PE is unlikely.

Gottschalk performed his own retrospective analysis using 108 studies from a European Society of Thoracic Imaging Prospective Evaluation of Pulmonary Embolism (ESTIPEP) trial to establish the efficacy of testing for very low probability cases. A definitive V/Q reading covering PE present, PE absent, or PE uncertain was possible in 89% of the cases. The sensitivity, specificity, and accuracy were 90%, 97%, and 95%, respectively.

"The data show that the very low probability category is useful and effective," he said.

Such experience led Gottschalk to conclude that the standard dictation lexicon for V/Q studies should be changed to reflect the language used to describe CTA/CTV diagnosis of PE. Gottschalk urged colleagues to use "PE present" instead of "high probability of PE." Normal and very low probability findings could be combined in a new category of "pulmonary embolism absent." All other options could be described as either uncertain or nondiagnostic, he said.

"If that is a little too radical for you, how about 'pulmonary embolism present with a greater than 90% probability,' 'pulmonary embolism absent with less than 10% probability,' and 'nondiagnostic?'" he said.

Despite the growing popularity of CTA/CTV, V/Q scanning will survive as a diagnostic test for PE, Gottschalk said. It will be performed when there is discordance between the clinical evaluation and CTA/CTV results. It reliably diagnoses PE when CT is not feasible because of renal failure, or high creatinine levels. It should be the first choice for PE diagnosis for women during their reproductive years and for serial studies evaluating clot burden.

Mr. Brice is senior editor of Diagnostic Imaging.

---

Three Bs buzz to the rescue of V/Q

Challenge presented by CTA/CTV use requires preemptive steps

Dr. Alexander Gottschalk urges radiologists to follow the three Bs to assure the survival of nuclear ventilation/perfusion imaging for diagnosing pulmonary embolism:

- Be available. Nuclear medicine will lose to CT, if gamma cameras capable of V/Q imaging are not also staffed around the clock. X-ray technologists can be cross-trained to fill in during off-hours.

- Be good. Use a technically sound protocol to assure high-quality scanning.

- Be right. With a normal chest x-ray, being right is relative simple. -JB