| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Department of Radiology, Massachusetts General Hospital, 55 Fruit St, Boston, MA 02114-2698. Received September 8, 2004; revision requested November 12; revision received November 26; accepted December 30. Address correspondence to S.E.J. (e-mail: sejones{at}partners.org).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MATERIALS AND METHODS: Investigational review board approval was obtained, informed consent was waived, and the study was HIPAA compliant. Retrospective review of 3612 CT pulmonary angiography reports created between July 1, 2001, and July 1, 2003, was performed with a keyword search for "indeterminate," "nondiagnostic," or "inadequate" (thereafter, all defined as "indeterminate") and yielded studies from 237 patients (mean age, 57 years; 117 men, 120 women). Randomly selected diagnostic studies were used to form a control group of 25 subjects (mean age, 64 years; eight men, 17 women). Electronic medical records were reviewed for follow-up imaging (repeat CT pulmonary angiography, conventional pulmonary angiography, ventilation-perfusion scintigraphy, or lower-extremity ultrasonography [US]), use of anticoagulation, placement of inferior vena cava (IVC) filters, clinical outcomes, and comments regarding indeterminate reading of CT angiograms. Studies (in patients and control subjects) were reviewed for PE, contrast attenuation in the main pulmonary artery (MPA), motion artifacts, image noise, and flow artifacts. Findings were compared with two-sample t tests assuming unequal variance.
RESULTS: The cause cited for indeterminism was most often motion (74%), followed by poor contrast enhancement (40%). Contrast attenuation in the MPA was 245 HU ± 80 (standard deviation) in patients and 339 HU ± 88 in control subjects (P < .001). Only 46% of indeterminate studies met institutional criteria for adequate contrast attenuation in the MPA. Rereview of studies demonstrated five missed PEs. A total of 81 patients (33%) underwent follow-up imaging within 5 days, with one positive pulmonary angiogram and four positive lower-limb US scans. Reread or follow-up images depicted thromboembolic disease in 4.2% of patients. Nineteen patients (8%) with indeterminate final result were treated for thromboembolic disease with either anticoagulation or IVC filters. Reports on 22% of indeterminate studies contained recommendations for follow-up imaging, and those recommendations nonsignificantly increased the rate for those examinations from 13% to 19%. Review of discharge summaries showed 22% of studies are clinically interpreted as negative.
CONCLUSION: The two major causes of indeterminism are motion artifacts and poor contrast enhancement.
© RSNA, 2005
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
With the high sensitivity and specificity of recently developed CT pulmonary angiography techniques (a sensitivity up to 94%–96% and specificity up to 94%–100%) (3,5,11), treatment protocols resulting from a diagnostic study are well defined: A positive study results in anticoagulation treatment or placement of vena cava filter, while a negative study necessitates an alternative diagnosis (which is often supplied as an incidental finding from chest CT). However, the clinical outcome resulting from indeterminate studies is not well described.
The potentially serious clinical effect of an indeterminate CT pulmonary angiogram is raised when noting the clinical outcomes of indeterminate ventilation-perfusion (V/Q) scintigrams. V/Q studies are indeterminate or intermediate (the terms are used interchangeably in the literature) in up to 70% of cases (12). Follow-up conventional angiograms demonstrate PE in 33% of these cases (13). Kember et al (12) reported that 35% of indeterminate V/Q scans are misinterpreted, and patients with these indeterminate scans do not undergo treatment, a finding with major implications. Clinicians often treat an indeterminate study as an end point in the investigation; thus, patient care is compromised (14). The purpose of our study, therefore, was to retrospectively review both the imaging characteristics of CT pulmonary angiograms indeterminate for PE and the patient outcome.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Hardware and Technique
Images were acquired in a caudal-cranial direction on either a four– or a 16–detector row CT scanner (GE Medical Systems, Milwaukee, Wis). Our institutional protocol is to perform CT pulmonary angiography at 380 mA, scanning helically from the lung apices to the adrenal glands. Automated dose reduction techniques were not used. To allow a 4 mL/sec flow of 135 mL iopamidol (Isovue 300; Bracco Diagnostics, Princeton, NJ), an 18- or 20-gauge antecubital catheter was used. The patient was asked to hold his or her breath after full inspiration. Imaging was started when the contrast material bolus was maximally in the pulmonary arterial system, which was typically 20–30 seconds after injection as determined according to institutional protocol and depending on body habitus and scanner type. No bolus tracking technology was used. At 210 seconds after pulmonary scanning, delayed imaging of the pelvis and lower extremities (from the inferior popliteal region to the aortic bifurcation) provided a CT venogram for evaluation of deep venous thromboembolism.
Software
The imaging protocol used at our institution specifies contiguous thin sections (1.25 mm) obtained by using a standard reconstruction algorithm. Images are reviewed on a picture archiving and communication system monitor (IMPAX version 4.1; AGFA, Teterboro, NJ) by using PE settings (window level, 100 HU; window width, 700 HU). After evaluation for PE, the images are reviewed again at different window settings for lung parenchyma (window level, –600 HU; window width, 1500 HU) and soft tissue (window level, 40 HU; window width, 350 HU).
Patient Selection
Patient population selection was initiated by searching the electronic medical record for all chest CT pulmonary angiography examinations conducted during a 24-month period between July 1, 2001, and July 1, 2003, with a total of 3612 studies. Of these, indeterminate studies were identified with an electronic word search of the full radiology reports by using the three keywords "indeterminate," "nondiagnostic," and "inadequate," including variations in spelling and hyphenation. (Unless otherwise indicated, for the remainder of this article the term "indeterminate" will represent all three keywords.) A total of 258 studies were identified, and thereafter 21 studies were excluded for one of six reasons: (a) patient enrolled in the PIOPED II study and therefore a complete work-up, including V/Q scanning and/or angiography, automatically ensued; (b) patient with known PE was already undergoing treatment; (c) study that was initially read as positive was subsequently revised to indeterminate, with treatment already under way; (d) study was obtained following previous CT pulmonary angiography conducted within 5 days; (e) contrast material had failed to be administered for technical reasons; and (f) patient had deep venous thrombosis diagnosed on concurrent lower extremity and pelvic CT venograms. No attempt was made to separate studies on the basis of a resident radiologist assisting the staff radiologist versus the unassisted staff radiologist. Similarly, no separation was made on the basis of the setting of presentation—for example, an emergency department CT pulmonary angiogram versus an outpatient CT pulmonary angiogram.
After exclusions, the patient set for the remaining analysis included 237 patients with 237 indeterminate studies, for a rate of 6.6% (Table 1). The mean patient age was 57 years, and there were 117 men and 120 women. A control group of 25 patients was selected at random from the initial list of 3612 studies, excluding any studies that were read as indeterminate or positive for PE. The mean age in the control group was 64 years, and there were eight men and 17 women. The differences in age distributions (P = .09) and sex distributions (P = .14) were not statistically significant, and the statistical analysis was not adjusted for age or sex.
|
Rereview of Images
All 237 studies were rereviewed for PE and image quality by the senior author (C.W., 5 years of experience and a central reader for PIOPED II), who was blinded to the radiology report. PE was evaluated in the standard fashion by either observation of occlusive pulmonary arterial filling defects on contiguous images or direct observation of an endoluminal clot with peripheral flow (5,11). Consensus with the first author (S.E.J.) was obtained for any PE discovered at rereview.
Image quality was assessed for the categories of motion, contrast agent bolus enhancement, and noise. Motion was identified by means of blurred parenchymal detail and motion artifacts as the "seagull wing" appearance of pulmonary vessels. The contrast attenuation was quantified by means of measurement of the mean Hounsfield units and standard deviation on the largest image of the main pulmonary artery within a region of interest with a diameter equal to half the diameter of the artery. The criterion used at our institution to define optimal opacification at this location is a Hounsfield unit value greater than 250 HU. Noise was identified as excessive image heterogeneity within large regions of contrast enhancement, such as the heart chambers and great vessels. Each category was separately evaluated at the lobar, segmental, and subsegmental levels and each in the upper, middle or lingula, and lower lobes, for a total of 27 subcategories. Scores for the left and right lobes were combined.
To quantify the degree of indeterminism, one point was given for each subcategory evaluated as indeterminate for evaluation of PE. Thus, there was a maximum total of nine points for each category of motion, bolus enhancement, and noise. Note was also made of any pulmonary disease, specifically recording any degree of emphysema, consolidation or pneumonia, atelectasis, tumor larger than 3 cm in diameter, ventilation, lobectomy or pneumonectomy, pleural effusion greater than 2 cm, pulmonary edema, and chronic interstitial lung disease (Table 2). Technical factors regarding contrast enhancement were recorded, specifically contrast agent bolus failure, poor timing, or transient interruption of contrast material, which is an inspiration-associated artifact causing transient nonopacification of the cardiopulmonary system as contrast material is diverted retrograde into the inferior vena cava (15). Last, quantified factors from rereview of the images were compared with the corresponding factors originally mentioned in the radiology reports; roughly, this represents a comparison of objective image measures with subjective image measures.
|
Control Group
Analysis of the 25 determinate CT pulmonary angiography studies, which represented the control set, proceeded exactly in parallel to analysis of the 237 indeterminate CT pulmonary angiography studies, as detailed in the previous sections. This included rereview of both the images (C.W.) and the reports (S.E.J.).
Statistical Analysis
Statistical samples were compared by using a two-sample t test assuming unequal variances. P values were computed by using a one-tailed distribution, unless otherwise indicated. P values from contingency tables were computed by using a two-tailed Fisher exact test. A P value of less than .05 was considered to indicate a significant difference. Data were collected and analyzed by using Excel (Microsoft, Seattle, Wash).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
Recommendations for radiologic follow-up were provided for 21.9% of indeterminate studies, and recommendations were mostly for angiography (47 studies, 19.8%), V/Q scanning (two studies, 0.8%), or both (three studies, 1.3%).
Imaging Characteristics
A total of five PEs were identified that had been missed at the initial reading, for a rate of 2.1% (Table 4). Of these, an artifact causing indeterminism was a likely factor on two studies (Figs 1), (Figs 2 ). A total of 10 studies demonstrated transient interruption of contrast material. Significantly delayed bolus timing, as indicated by the lack of contrast enhancement in the superior vena cava, was identified on five studies. Conversely, on three studies there was poor superior vena cava contrast enhancement caused by superior vena cava stenosis.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
The possible sequela of missed PE was examined by reviewing the electronic medical records for readmission. A total of 53 (22.4%) patients with indeterminate CT pulmonary angiograms were readmitted to our hospital or to our sister institutions within 3 months of the study. Of these 53 patients, none were readmitted for PE or the sequela of PE. All admissions were for other causes, which were often related to comorbidities that existed during the initial admission at which CT pulmonary angiography was performed. In patients who were not readmitted within 3 months and who had continued entries available in the medical record, there was no indication of any patient being readmitted to an outside institution during that period.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The two major causes of indeterminism are excessive motion and poor bolus enhancement. In our study, the correlation of subjective with objective assessments for excessive motion and poor bolus enhancement revealed that excessive motion is often inaccurately ascribed as a cause for indeterminism, while poor bolus enhancement is more accurately assessed but is undercalled. The assessment of poor bolus enhancement is largely unaffected by the presence of motion, whereas poor bolus enhancement slightly reduces the number of studies ascribed as indeterminate because of motion. The cause for poor bolus enhancement was most likely inadequate timing of bolus administration. The infrequently found cause of body habitus was likely due to increased x-ray absorption, which caused increased image noise confounding evaluation for small filling defects within the minor pulmonary arteries.
Since the main cause of motion during CT pulmonary angiography is the inability of dyspneic patients to hold their breath during the scanning time, the increased use of 16– and 64–detector row CT with substantially faster scanning times should lower the number of indeterminate CT pulmonary angiograms in the future. An estimate of this number could be drawn from our subjective data that indicate that a total of 176 reports (74%) mentioned motion as a contributing cause to indeterminism. It is likely that, in the future, the rate of CT pulmonary angiograms rendered indeterminate because of motion will be substantially decreased. A similar reduction in the second leading cause, poor bolus enhancement, could be accomplished with increased attention to detail during contrast agent administration. Optimal timing for contrast agent administration could be determined by using prior intravenous injection of a small test bolus and then continually scanning the pulmonary artery, noting the time of maximal enhancement.
The effect of a radiologist's recommendation for further imaging in the report was surprisingly low, with no statistically significant effect (P = .27). This means that if the radiologist is certain that follow-up imaging is required, stating so in the report is insufficient. Direct communication with the medical staff is required.
The radiologist should be aware that in a nonnegligible fraction of examinations, an indeterminate conclusion will be interpreted as a negative conclusion, as indicated in 22% of discharge summaries and clinical medical records. This could result in many patients with PE left untreated. Misinterpretation of the CT pulmonary angiography results could be minimized with improved communication with the medical staff about the meaning of "indeterminate." For example, an indeterminate CT pulmonary angiogram has a different meaning than an indeterminate V/Q scan, which is typically equated with an intermediate probability V/Q scan. If clinicians do not clearly understand the importance of such findings, a period of education may be necessary.
A study limitation was that during a sporadic review of some of the 3612 CT pulmonary angiography reports with determinate results, it became apparent that some reports actually described an indeterminate study without ever using one of the three specified keywords. Reports of this nature were not included in the study because they are not amenable to an electronic keyword search, and their inclusion would therefore entail reading 3612 reports. Other less frequently used terms, such as "limited study," were not included because it was believed that they did not sufficiently convey the sense of an indeterminate study. The total number of such reports can be estimated by searching for all nonindeterminate reports for suggestive but noncommittal terms such as "limited," "degraded," "poor," or "suboptimal," which yields a percentage of 41%. This represents an upper limit, since it is problematic to assume that such a blind word search means 41% of all CT pulmonary angiography studies are actually indeterminate; most of those chosen terms are likely to be used in other contexts. However, it raises concern that there are possibly substantially more images that the radiologist believes to be indeterminate but includes no direct language in the report to convey that impression.
A technical problem regarding CT pulmonary angiography is that the patient population is typically dyspneic and anxious. Although the scanning time for 16–detector row CT through the thorax is about 10 seconds, patients undergoing CT pulmonary angiography are often in such respiratory distress that they cannot hold their breath for that period of time. Thus, there is a high rate of motion artifacts leading to indeterminate conclusions. A second problem is the timing of the contrast agent bolus: At an optimal CT pulmonary angiography examination, the contrast agent bolus is centered on the pulmonary arterial system. If the bolus is too "early" or too "late," an indeterminate conclusion is likely reached. Reasons for such errors include technician error, anatomic variations, or, most likely, different perfusion rates depending on patient's cardiac status. Some of these limitations may be overcome by use of a test bolus to directly measure optimal timing.
The increased use of low-molecular-weight heparin, such as dalteparin (Fragmin; Pharmacia and Upjohn, Kalamazoo, Mich), for treatment of PE is a limitation regarding our conclusions on treatment effects resulting from the indeterminate CT pulmonary angiography studies. The medical record search for anticoagulant use related to CT pulmonary angiography was focused solely on unfractionated heparin, the use of which was easily indicated by a high partial thromboplastin time. No readily accessible assay for low-molecular-weight heparin is available at our institution, and indication for its use would necessitate a more unreliable and difficult search of medication orders and administration.
The possibility that patients could have been lost to follow-up for readmissions within 3 months reduces the strength of our finding that no readmissions occurred for PE. Since our institution and our sister institutions compose a large fraction of the metropolitan medical care, it is unlikely that a sizeable percentage of patients would have sought medical care elsewhere without leaving some reference in the subsequent medical record.
In conclusion, our results show that 6.6% of CT pulmonary angiograms are indeterminate, mostly because of respiratory motion, followed by poor bolus enhancement. The radiologist has a lower threshold for the presence of motion effects than for poor bolus enhancement. Although the percentage of indeterminate studies positive for PE at further analysis is relatively low (4.2%), this represents 0.28% of all CT pulmonary angiography examinations performed at our institution. If 450 000 such examinations are conducted annually in the United States, 1245 patients who undergo CT pulmonary angiography with indeterminate results have PE. If we assume an untreated mortality of 30%, about 370 patients die needlessly each year. This number may be improved with (a) the increased use of higher-order multi–detector row CT, (b) attention to contrast agent bolus details, and (c) improved communication with medical staff about the meaning of indeterminism and the necessity for follow-up recommendations. Clear language in the radiology report about the reason for, level of, meaning of, and recommendations based on an indeterminate conclusion at CT pulmonary angiography is essential.
|
FOOTNOTES |
|---|
Abbreviations: PE = pulmonary embolism • PIOPED II = Prospective Investigation of Pulmonary Embolism Disease II • V/Q = ventilation-perfusion
Authors stated no financial relationship to disclose.
Author contributions: Guarantors of integrity of entire study, S.E.J., C.W.; study concepts/study design or data acquisition or data analysis/interpretation, S.E.J., C.W.; manuscript drafting or manuscript revision for important intellectual content, S.E.J., C.W.; approval of final version of submitted manuscript, S.E.J., C.W.; literature research, S.E.J., C.W.; clinical studies, S.E.J., C.W.; statistical analysis, S.E.J.; and manuscript editing, S.E.J., C.W.
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. Roggenland, S. A. Peters, S. P. Lemburg, T. Holland-Letz, V. Nicolas, and C. M. Heyer CT Angiography in Suspected Pulmonary Embolism: Impact of Patient Characteristics and Different Venous Lines on Vessel Enhancement and Image Quality Am. J. Roentgenol., June 1, 2008; 190(6): W351 - W359. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. D. Sostman, P. D. Stein, A. Gottschalk, F. Matta, R. Hull, and L. Goodman Acute Pulmonary Embolism: Sensitivity and Specificity of Ventilation-Perfusion Scintigraphy in PIOPED II Study Radiology, March 1, 2008; 246(3): 941 - 946. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. D. MacKenzie, J. Nazario-Larrieu, T. Cai, M. S. Ledbetter, M. A. Duran-Mendicuti, P. F. Judy, and F. J. Rybicki Reduced-Dose CT: Effect on Reader Evaluation in Detection of Pulmonary Embolism Am. J. Roentgenol., December 1, 2007; 189(6): 1371 - 1379. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. M. Heyer, P. S. Mohr, S. P. Lemburg, S. A. Peters, and V. Nicolas Image Quality and Radiation Exposure at Pulmonary CT Angiography with 100- or 120-kVp Protocol: Prospective Randomized Study Radiology, November 1, 2007; 245(2): 577 - 583. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. Scarsbrook, K. Bradley, F. Gleeson, A. M. Groves, S. J. Yates, T. Win, I. Kayani, J. Bomanji, and P. J. Ell Perfusion Scintigraphy Still has Important Role in Evaluation of Majority of Pregnant Patients with Suspicion of Pulmonary Embolism Radiology, August 1, 2007; 244(2): 623 - 625. [Full Text] [PDF]
|
|
|
|
|
|
H. Arakawa, T. Kohno, T. Hiki, and Y. Kaji CT Pulmonary Angiography and CT Venography: Factors Associated with Vessel Enhancement Am. J. Roentgenol., July 1, 2007; 189(1): 156 - 161. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. Wittram How I Do It: CT Pulmonary Angiography Am. J. Roentgenol., May 1, 2007; 188(5): 1255 - 1261. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. G. Vrachliotis, K. G. Bis, A. Haidary, R. Kosuri, M. Balasubramaniam, M. Gallagher, G. Raff, M. Ross, B. O'Neil, and W. O'Neill Atypical Chest Pain: Coronary, Aortic, and Pulmonary Vasculature Enhancement at Biphasic Single-Injection 64-Section CT Angiography Radiology, May 1, 2007; 243(2): 368 - 376. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M R Jones and J H Reid Emergency chest radiology: thoracic aortic disease and pulmonary embolism Imaging, September 1, 2006; 18(3): 122 - 138. [Abstract] [Full Text] [PDF]
|
|
Read all eLetters
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
