| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, University of Washington Medical Center, 1959 NE
Pacific St., RR215, Box 357115, Seattle, WA 98195-7115.
2 Department of Radiology, Harborview Medical Center, Seattle, WA.
3 Harborview Injury Prevention Research Center, Seattle, WA.
Received September 26, 2006;
accepted after revision May 13, 2007.
Address correspondence to Y. Anzai.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. We evaluated the diagnostic accuracy of MDCT in 336 consecutive patients undergoing evaluation for nontraumatic SAH with both CTA and 3D DSA within 48 hours. The diagnostic performance of CTA was assessed by radiology reports using DSA as the gold standard. Analyses were performed per aneurysm and per patient, the results were stratified by aneurysm size and location, and the MDCT data—16-MDCT data versus 4-, 8-, and 16-MDCT combined data—were compared.
RESULTS. The overall sensitivity and specificity of CTA per aneurysm
was 83% (CI, 0.78–0.87) and 93% (0.88–0.97), respectively. CTA
failed to detect 49 of the 284 aneurysms. Thirty-nine (80%) of these 49 missed
aneurysms were 
3 mm, nine (18%) were 4–6 mm, and one (2%) was
7–10 mm. The sensitivity and specificity of CTA per patient was 95%
(0.91–0.97) and 97% (0.92–0.99), respectively. Of 211 patients, a
primary aneurysm was missed on CTA in 11 patients.
CONCLUSION. CTA showed excellent diagnostic performance for aneurysm
detection. The high negative predictive value (91.2%) for the per-patient
analysis indicates that CTA has merit as a screening tool. Most aneurysms
missed were 3 mm and in patients in whom a primary aneurysm had already
been detected.
Keywords: aneurysm • cerebral vasculature • CT angiography • digital subtraction angiography • MDCT angiography • neurovascular injury • rotational angiography • subarachnoid hemorrhage
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The results of numerous studies evaluating CTA in the detection of intracranial aneurysms have been published [5–17]. More recent studies have suggested that with MDCT, the diagnostic accuracy of CTA approaches that of DSA [18–23]. MDCT uses multiple detector elements aligned in the z-axis (usually 4, 8, 16, 32, or 64 rows) and uses continuous scanning while the patient moves through the gantry. The volumetric data set is used to produce maximum-intensity-projection, volume-rendered, 2D, or 3D representations of the data with readily available postprocessing software. Some authors have asserted that improvements in resolution and the ability of multiplanar reformations allow MDCT to detect aneurysms as small as 3 mm [20–22].
In many of the previously published studies, CTA was reviewed by two or more expert reviewers or by neurointerventional radiologists and neurosurgeons in consensus. This may have resulted in a positive bias toward the diagnostic accuracy of CTA. To accurately assess how well CTA performs in a general clinical setting, we analyzed radiology reports of 336 consecutive patients who presented with nontraumatic SAH.
Our institution is a high-volume level 1 trauma center where approximately 22,000 head CT and 800 head CTA examinations are performed each year. The standard of care at our institution for patients undergoing evaluation for nontraumatic SAH is to undergo CTA first. Subsequently, most of these patients undergo 3D rotational DSA to better assess the relationship of the aneurysm to the parent vessel and adjacent vessels and to find potentially CTA-occult aneurysms.
The purpose of this study was to evaluate the effectiveness of CTA performed with MDCT in patients undergoing evaluation for nontraumatic SAH. The diagnostic performance of CTA was assessed in 336 consecutive patients for an 18-month period using retrospective review of radiology reports, with DSA as the gold standard. We also compared the diagnostic accuracy of CTA performed with 4-, 8-, and 16-MDCT versus CTA performed with 16-MDCT alone.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Data Extraction
This study was approved by our institutional review board for a
retrospective chart review and data extraction. For the 336 patients, the
following information was recorded by review of CTA and DSA reports and
electronic patient records: basic demographics, presence of SAH on unenhanced
head CT or lumbar puncture, and size and location of aneurysm or aneurysms. If
no aneurysm was detected, the alternative diagnosis to explain the presence of
SAH or clinical presentation was recorded.
The location of the aneurysm was categorized as middle, anterior, or
posterior cerebral artery (MCA, ACA, or PCA, respectively); internal carotid
artery (ICA); anterior communicating artery; posterior communicating artery;
basilar artery; or other (e.g., posterior inferior cerebellar artery,
vertebral artery, superior cerebellar artery). Detected aneurysms were divided
into four size categories: 3, 4–6, 7–10, and > 10 mm, as
measured on DSA. The treatment option for all 336 patients was recorded (i.e.,
surgical clipping, endovascular coiling, or observation). For those who
underwent surgery, concordance between the surgical findings and the CTA and
DSA findings was evaluated. An attempt was made to identify the primary
aneurysms by size, morphology, and location with respect to the distribution
of SAH or parenchymal hemorrhage. The clinical outcomes of patients with
aneurysms were categorized as death, some neurologic deficit, or no neurologic
deficit. Cases with discordant CTA and DSA findings were reviewed in more
detail.
CT Scanner Protocols
The 336 patients were evaluated using 4-, 8-, or 16-MDCT. Each CTA
examination included unenhanced and contrast-enhanced head imaging. The CTA
images were sent to PACS and postprocessing workstations (Vitrea 2, version
3.2, Vital Images). The protocol for the CTA portion of the examination was as
follows: 110 mL of iodixanol (Visipaque, Nycomed) for 4- and 8-MDCT or 80 mL
of iohexol (Omnipaque, Nycomed) for 16-MDCT followed by 30 mL of saline
infused at 3.0 mL/s for 4- and 16-MDCT and at 4.0 mL/s for 8-MDCT. Slice
thickness was 1.25 mm for 4- and 8-MDCT and 0.625 mm for 16-MDCT, and the
table interval was 0.8 mm for 4- and 8-MDCT and 0.625 mm for 16-MDCT. Table
speed was 7.5 mm per rotation for 4-MDCT, 13.5 mm per rotation for 8-MDCT, and
13.75 mm per rotation for 16-MDCT. A setting of 140 kV was used, and tube
current was 300 mA for 4-MDCT, 380 mA for 8-MDCT, and 350–380 mA for
16-MDCT. The display field of view was 16 cm.
The CTA source images were reviewed by 10 diagnostic neuroradiologists at the PACS station. Six standard views of 3D reformatted images were created by CT technologists before the radiologists' review in gray-scale. On dedicated workstations separate from the PACS station, the neuroradiologist used postprocessing software to create additional 3D and color reformation images as desired. This additional postprocessing is not standard and was performed only occasionally. At our institution, beginning in January 2004, 16-MDCT replaced 4- and 8-MDCT for CTA examinations. Overall, 191 of the CTA examinations were 16-MDCT, 137 were 4-MDCT, and eight were 8-MDCT.
DSA Protocol
DSA imaging was performed using 3D rotational angiography (Integris V3000,
Philips Medical Systems). Images were acquired in the standard projections
(anteroposterior [AP], lateral, and AP and lateral obliques).
Three-dimensional rotational angiography was routinely performed when an
aneurysm was found to better characterize the aneurysm and its relationship to
parent vessels to determine the appropriate treatment option.
Three-dimensional rotational angiography uses a mode over an angle of 180° at a frame rate of 12.5 frames per second and a rotation speed of up to 30° per second. During the run, iodinated contrast agent was injected (e.g., 300 mg/mL at a rate of 4 mL/s for 6 seconds) to provide continuous filling of the vasculature. Acquisition took place in a single 180° rotational angiography scan. Three-dimensional volumes were reconstructed on a workstation (Integris 3D-RA, Philips Medical Systems).
CTA and DSA Reporting
In this study, 10 attending neuroradiologists with various degrees of
experience and expertise in cerebral aneurysms generated CTA and DSA reports.
In most cases (334/336), the CTA examination preceded the DSA examination, and
the CTA images were available on the PACS station before the DSA images.
Data Analysis
Using DSA as the gold standard, we calculated sensitivity, specificity,
positive predictive value (PPV), and negative predictive value (NPV) on
per-aneurysm and per-patient bases, and the results were stratified by
aneurysm size and location. CIs of 95% and chi-square test values were also
calculated. The diagnostic accuracy of 16-MDCT was compared with that of the
combined scanner data.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
|
|
|
3 mm, 94 (33%) were 4–6 mm,
75 (26%) were 7–10 mm, and 44 (15%) were > 10 mm. The distribution of
aneurysm locations was as follows: 64 MCA (23%), 57 ICA (20%), 48 anterior
communicating artery (17%), 41 posterior communicating artery (14%), 27 ACA
(10%), 24 basilar artery (8%), 11 PCA (4%), and 12 other (4%).
Discordance by Aneurysm
The overall sensitivity and specificity of CTA per aneurysm was 83%
(0.78–0.87) and 93% (0.88–0.97), respectively. The PPV was 96%
(0.93–0.98), and the NPV was 72% (0.64–0.78). Sensitivity was then
calculated for each size category. For aneurysms 3 mm, sensitivity was
45%. For aneurysms 4–6, 7–10, and > 10 mm, sensitivity was 90%,
99%, and 100%, respectively (p < 0.001). The sensitivity,
specificity, PPV, and NPV of CTA were not significantly altered when 16-MDCT
data were separated from the combined MDCT data for subgroup analyses
(Fig. 1).
CTA failed to detect 49 (17%) of the 284 aneurysms identified on DSA.
Thirty-nine (80%) of these 49 aneurysms were 3 mm; nine (18%) were in the
4- to 6-mm range, and one (2%) was in the 7- to 10-mm range. None of the
aneurysms that was > 10 mm were missed on CTA (Fig.
2A,
2B). The locations of
CTA-missed aneurysms are illustrated in
Figure 3 (listed above the
total number of CTA-detected aneurysms in that location). Of the 49 CTA
misses, 16 (33%) were aneurysms of the MCA; 12 (24%), ICA; six (12%),
posterior communicating artery; five (10%), PCA; five (10%), ACA; four (8%),
other; and one (2%), anterior communicating artery. There were a total of nine
false-positives on CTA: two each of the posterior communicating artery,
basilar artery, and ICA; and one each of the MCA, ACA, and PCA (Fig.
4A,
4B,
4C,
4D).
|
|
|
|
Discordance by Patient
Among 336 patients, 211 patients were found to have at least one aneurysm
and 125 patients did not have an aneurysm. When calculated on a per-patient
basis (i.e., whether a patient was found to have an aneurysm on CTA), the
sensitivity was 95% (0.91–0.97), specificity was 97% (0.92–0.99),
PPV was 98.0% (0.95–0.99), and NPV was 91.2% (0.86–0.96). The
sensitivity, specificity, PPV, and NPV of CTA were not significantly altered
in subgroup analyses when 16-MDCT data were separated from the combined MDCT
data.
There were 49 false-negatives and nine false-positives on CTA, resulting in 58 discordant aneurysms in 46 patients. Thirty-one (67%) of the 46 patients had multiple aneurysms. In 15 of the 46 patients, there was discordance between CTA and DSA findings with regard to detection of a primary aneurysm (Table 1). Of these 15 patients, CTA findings were false-negative in 11 patients and false-positive in four. Therefore, 11 (22%) of the 49 aneurysms missed on CTA were primary aneurysms thought to be responsible for SAH, and 38 (78%) of 49 aneurysms were secondary or multiple aneurysms. Thus, CTA missed 11 (5.2%) of 211 primary aneurysms.
|
Of the 11 patients with false-negative CTA examinations for primary aneurysms, five underwent surgery and one underwent endovascular coiling, confirming the presence an aneurysm. The remaining five patients were observed. The clinical outcome of nine of these 11 patients resulted in no neurologic deficit. The other two patients, one who had a 2-mm ICA aneurysm and another who had a 3-mm posterior communicating artery aneurysm missed on CTA, died due to massive brain swelling that occurred immediately postoperatively and to myocardial infarction thought to be due to a preexisting heart condition, respectively (Fig. 5A, 5B, 5C). Thus, two (0.5%) of 336 patients evaluated had primary aneurysms that were missed on CTA and negative clinical outcomes.
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our study showed a sensitivity of 83% and specificity of 93% per aneurysm, which was slightly lower in sensitivity but higher in specificity than reported in more recent studies using MDCT. In a meta-analysis of 21 published studies of more than 1,250 patients, Chappel et al. [5] found CTA sensitivity to range from 75% to 100%, with a cumulative sensitivity of 93%, and specificity to range from 50% to 100%, with a cumulative specificity of 88%.
The slightly lower sensitivity of our study (83%) compared with others reported in the literature might be explained by multiple factors; CTA was interpreted as a part of a clinical examination by one of our 10 neuroradiologists with variable degrees of experience and expertise in the diagnosis of cerebral aneurysms on CTA. Dedicated viewing on a 3D workstation might have improved the diagnostic accuracy of CTA. However, this was not routinely performed in our institution (beyond the six standard technologist-generated reconstructions). The degree of scrutiny of the CTA might have been compromised as a result of the reviewers' knowledge that most patients at our institution with SAH undergo DSA after CTA. This knowledge could have affected missing small incidental aneurysms in patients in whom a primary aneurysm had already been detected on CTA. An alternative explanation for the lower sensitivity of CTA in our study may be due to advances in the accuracy of the gold standard itself. Most of our patients underwent 3D rotational DSA performed by dedicated neurointerventional radiologists. The use of 3D DSA may lead to the detection of smaller aneurysms than standard DSA [24, 25]. Our study showed a higher specificity than most of the published data, suggesting that false-positive cases in a "real" clinical setting are fewer than that by the expert reviewers.
Our results are concordant with the published data showing increased
sensitivity of CTA with increasing aneurysm size, with the threshold for
detection being > 3 mm. The sensitivity of CTA for the detection of
aneurysms 3 mm was 45% and that for aneurysms > 3 mm was 95.3%. Eighty
percent of the aneurysms missed on CTA in our study were 3 mm. Only one
aneurysm > 6 mm was missed on CTA, and that aneurysm was a cavernous ICA
aneurysm in a patient with six other aneurysms, including multiple bilateral
ICA aneurysms.
The diagnostic accuracy of CTA was much higher when calculated on a per-patient basis—approaching 95% sensitivity and 97% specificity—than on a per-aneurysm basis. More important, the NPV of CTA in this study was 91.2%, which is critical for the use of CTA in the context of a screening test. Thirty-one (67%) of the 46 patients with aneurysms missed on CTA were patients with multiple aneurysms. One should not underestimate, however, the importance of detecting multiple aneurysms. Secondary aneurysms can be treated at the same time as the primary aneurysm, either by a surgical or endovascular approach. Given the morbidity and costs associated with cerebral aneurysm treatment, accurate detection of all aneurysms before making a treatment decision is essential. Similarly, false-positive CTA results could have resulted in unnecessary surgical exploration associated with potentially substantial morbidity and mortality if CTA were to completely replace DSA for the diagnostic workup for patients with SAH.
The subgroup analyses comparing 16-MDCT data with the combined data set showed no significant difference in accuracy. Technologic improvement, particularly in 3D volume-rendering or reconstruction techniques, may improve the accuracy of CTA in the detection of aneurysms < 3 mm.
Our study has several limitations. It is a retrospective review using radiology reports rather than prospective blinded reviewers. The interpretation of CTA examinations was not uniform among our 10 neuroradiologists. Many of our neuroradiologists view a set of source images with only a few default 3D reformations. Also, we enrolled only patients who underwent both CTA and DSA for aneurysm detection, targeting a population with a high prevalence of aneurysm. For this reason, our study likely underestimates the number of true-negative CTA results by excluding patients who may have undergone evaluation with CTA alone. Despite these limitations, to our knowledge our study includes the largest number of consecutive patients in the literature from a single institution where all patients underwent both CTA with MDCT and 3D rotational DSA within 48 hours. Our results provide a fair representation of how well CTA performs in a general clinical setting and reflect the effectiveness of CTA for the detection of aneurysms in patients with nontraumatic SAH.
In conclusion, the results of our study showed that CTA was accurate for
the detection of intracranial aneurysms in a routine clinical setting at a
high-volume level 1 trauma center. The majority of aneurysms that were missed
on CTA were 3 mm (80%) and were found in patients with multiple aneurysms
(67%). The high NPV (91.2%) yielded from our per-patient analysis suggests
that CTA has merit in the context of screening. The diagnosis of aneurysms
3 mm on CTA to assess preoperative mapping of all aneurysms remains
challenging. Future studies are needed to determine whether additional 3D
viewing or 64-MDCT improves the diagnostic accuracy of CTA with MDCT and
provides adequate characterization of the morphology of aneurysms for
preoperative imaging workup.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
