ORIGINAL ARTICLE

ACTA RADIOLOGICA

Multislice Computed Tomography Coronary Angiography at a Local Hospital: Pitfalls and Potential K. KOLNES, H. OSE VELLE, S. HAREIDE, K. HEGBOM & R. WISETH Departments of Radiology and Internal Medicine, Volda Hospital, Volda, Norway; Department of Cardiology, Trondheim University Hospital and Department of Circulation and Medical Imaging, Norwegian University of Science and Technology, Trondheim, Norway Kolnes K, Ose Velle H, Hareide S, Hegbom K, Wiseth R. Multislice computed tomography coronary angiography at a local hospital: pitfalls and potential. Acta Radiol 2006;47:680–686. Purpose: To evaluate whether the favorable results achieved with multislice computed tomography (MSCT) of coronary arteries at larger centers could be paralleled at a local hospital. Material and Methods: Fifty consecutive patients with suspected coronary artery disease scheduled for invasive investigation with quantitative coronary angiography (QCA) at a university hospital underwent MSCT with a 16-slice scanner at a local hospital. Diagnostic accuracy of MSCT for coronary artery disease was assessed using a 16segment coronary artery model with QCA as the gold standard. Results: Segments with diameter v2 mm, as defined by QCA, segments distal to occlusions, and stented segments were excluded. Of the remaining 489 segments, 73 (15%) were not assessable with MSCT. Sensitivity, specificity, and positive and negative predictive values for significant (w50%) stenosis for the 416 assessable segments were 92%, 82%, 53%, and 98%, respectively. Conclusion: Our beginners’ experience demonstrated favorable results regarding sensitivity and negative predictive value. The positive predictive value, however, was unsatisfactory. Calcifications were identified as the most important factor for falsepositive results with MSCT. With widespread use of MSCT coronary angiography, there is a risk of recruiting patients without significant coronary artery disease to unnecessary and potentially harmful invasive procedures. Key words: Adults; arteriosclerosis; cardiac; CT angiography; heart; observer performance Kristian Kolnes, Department of Radiology, Volda Hospital, NO-6100 Volda, Norway (tel.+47 70 05 82 78, fax. +47 70 05 83 44, e-mail. [email protected]) Accepted for publication 16 March 2006

Multislice computed tomography (MSCT) of the coronary arteries has in recent years shown increasing accuracy in step with a rapid technological development and with the increasing expertise of dedicated practitioners of the method (1, 10, 16). In particular, the ability to rule out significant coronary artery disease at both the segment- and patient-based levels has steadily improved (8, 12, 17). This aspect of the method may particularly be useful at local hospitals without facilities for invasive investigations. Most studies published on MSCT coronary angiography are performed at larger centers where there is also experience in invasive coronary angiography, and it is not known whether the favorable results obtained in these studies are

reproducible in less-specialized centers. As MSCT of coronary arteries is adopted by several minor hospitals, this is considered a highly relevant question. The aim of this study was to assess diagnostic accuracy of MSCT for coronary artery disease when performed at a local hospital with only general expertise in radiology and cardiology. Material and Methods From January 2004 to March 2005, 50 consecutive patients scheduled for invasive coronary angiography due to suspected coronary artery disease were examined with MSCT at a local hospital prior to invasive investigation at a university hospital. The patient characteristics are demonstrated in Table 1. DOI 10.1080/02841850600806332

# 2006 Taylor & Francis

MSCT Coronary Angiography at a Local Hospital Table 1. Patient characteristics. Male Age, years Body mass index, kg/m2 Heart rate during scanning, bpm

35/50 61¡10 27.4¡4.1 58¡8

(70%) (35–80) (16.4–38.9) (42–80)

Indication for invasive investigation was based on established criteria for coronary angiography. The MSCT examination was performed at a mean of 3.7 (2–6) days before the invasive study. Patients over the age of 18 with sinus rhythm were offered inclusion. Pregnancy and contraindication to iodinated contrast medium were criteria for exclusion. Three patients were excluded after the MSCT examination due to technical mishaps during image recording or reconstruction due to user errors (n52) or lack of motivation for invasive coronary angiography (n51). MSCT The examination was carried out with a 16-slice scanner (Lightspeed 16, GE Medical Systems, Milwaukee, Wisc., USA) (collimation 1660.625 mm, 500 ms rotation time, 400 mA, 120 kV, pitch 0.3) under ongoing intravenous infusion of the contrast agent iodixanol (Visipaque, GE Healthcare, Little Chalfont, UK; 320 mg/ml I, 140 ml, 4 ml/s) and with retrospective ECG synchronization. Bolus tracking technique was employed. Patients with heart rate w65 bpm were given beta-blockers (metoprolol 5– 10 mg i.v.). Mean heart rate during scanning was 58 (42–80) bpm. For each patient, 10 sets of images were generated for various parts of the heart cycle. These were transferred to a dedicated workstation (Advantage Windows 4.1, GE Medical Systems, Milwaukee, WI, USA) for further processing. Image data were evaluated multiplanarly with thin- (1–1.5 mm) and thick-slab maximum-intensity projections. Other reconstruction methods (volume rendering, vessel analysis program) were used when appropriate. Nine patients had previous coronary artery bypass operation; in these, only native coronary arteries were assessed. The coronary artery tree was divided into 16 segments in accordance with the American Heart Association classification (3). Each segment was analyzed semiquantitatively and classified into one of four categories; normal, (50% stenosis, w50% stenosis or occluded, non-assessable with MSCT. The degree of stenosis was defined by consensus between a radiologist with general MSCT competence and a cardiologist. Both were blinded to the results of the invasive investigation. For each segment, an estimate was made of calcium content on a scale from 0–3.

681

Our training background before starting the study included visits to two centers abroad that master the method; 1 week in total for a cardiologist, a radiologist, and a CT technician. Before including the study patients, a pilot phase with examination of 20 patients was performed. Invasive coronary angiography Invasive coronary angiography was performed using standard techniques. The angiograms were assessed by two experienced invasive cardiologists, both blinded to the result of MSCT. Segmental disease was analyzed in each segment using the same segment model employed for MSCT analysis. Stenosis severity was determined using a quantitative coronary angiography (QCA) program (Philips Medical Systems, Eindhoven, The Netherlands), and a maximal lumen diameter stenosis w50% in any projection was defined as significant. Segments with proximal diameter v2 mm as determined by QCA were excluded from the analysis. The study complied with the Declaration of Helsinki and was approved by the regional ethics committee. All patients gave their written informed consent. Statistical methods For analysis of sensitivity, specificity, and positive and negative predictive values, lesion severity, as defined by QCA, was used as the gold standard. Results No MSCT examination was excluded due to poor image quality. Examples of MSCT images and corresponding invasive angiograms are given in Figs 1–3. Segment-based evaluation (Table 2) A total of 188 segments with diameter v2 mm, as defined by QCA, 61 segments distal to an occlusion, and 14 stented segments were excluded from the analysis. Of the remaining 489 segments, 73 (15%) were unassessable by MSCT. When omitting these 73 segments from the analysis, the sensitivity for detecting significant coronary stenosis was 92% (95% CI 86–98) and the specificity 82% (95% CI 78–86). Positive and negative predictive values were 53% (95% CI 44–62) and 98% (95% CI 96–100), respectively. There were six false-negative segments in a total of five patients, distributed in segments 1, 6, 8, and 11 (Table 2). Four of these segments had borderline lesions (50–55%) by QCA. Acta Radiol 2006 (7)

682

K. Kolnes et al.

Patient-based evaluation Among the 34 patients with at least one significant stenosis in segments w2 mm, 33 (97%) were correctly identified with MSCT as having significant coronary artery disease. The one patient that was not identified had a significant stenosis in segment 8, assessed to 50–55% by QCA. This lesion was not subject to intervention. In the other patients with at least one non-recognized significant stenosis on MSCT, significant lesions were detected in other segments, thus defining the patients as having significant coronary artery disease. Of 16 patients who did not have any significant stenosis in segments w2 mm, eight were correctly identified with MSCT, leaving an equal number with a falsepositive result.

Fig. 1. Example of MSCT volume-rendered image of proximal parts of the coronary tree in a patient with no significant stenosis. Ao: aorta; LA: left atrium; LAD: left anterior descending artery; LM: left main artery; PT: pulmonary trunk; RCA: right coronary artery.

Among the 73 segments that could not be assessed with MSCT, QCA showed significant stenosis in seven segments. When including these segments in the analysis, the sensitivity and negative predictive value were 84% and 95%, respectively.

A

Factors causing false-positive results Overestimation of lesion severity by MSCT occurred to a large extent in segments with calcifications. A total of 49 out of 61 (80%) falsepositive segments were calcified. In 47 of 61 (77%) segments judged false positive with MSCT, there were extensive vessel-wall irregularities on invasive angiography, several of these representing borderline lesions. In three patients, a hypoplastic right coronary artery was misinterpreted as occlusion in the proximal segments of the artery. Overestimation of lesion severity occurred most frequently in segments 1, 6, and 11 (Table 2), that is, in the

B

Fig. 2. MSCT thin-slab (2 mm) maximum-intensity projection (MIP) of LAD segment 7, demonstrating non-calcified plaque. The resultant stenosis was assessed to be w50% by MSCT (A), and the lesion severity was confirmed by QCA (B). Acta Radiol 2006 (7)

MSCT Coronary Angiography at a Local Hospital

A

683

B

Fig. 3. MSCT thick-slab (15 mm) MIP from a patient that is ineligible for MSCT evaluation, demonstrating several false-positive segments. There is extensive calcification in the right coronary artery (RCA), circumflex artery (Cx), and left anterior descending artery (LAD), as well as in the left main artery. Moreover, there are slight rhythm irregularities during recording, and in general a poor signal-to-noise ratio (A). QCA of left side demonstrates only a solitary significant stenosis at the origin of the first diagonal of the LAD (B). There was no significant stenosis in the RCA.

proximal segments of the main branches of the coronary tree.

which are segments in the periphery of the coronary tree. Of the seven non-assessable segments with significant stenosis, the main reasons for nonassessability were small diameter, poor SNR, and motion artifacts.

Non-assessable segments A variety of factors, often in combination, caused non-assessability with MSCT. Several segments were judged to have a diameter v2 mm with MSCT, while QCA demonstrated a diameter w2 mm. This was frequently combined with poor signal-to-noise ratio (SNR), calcifications with resulting blooming artifacts, respiratory artifacts, extrasystoles, or increased heart rate. Non-assessability with MSCT occurred most frequently in segments 8, 12, and 14 (Table 2),

Discussion In this study we have demonstrated beginners’ experience when introducing MSCT coronary angiography in the diagnostic armamentarium for coronary artery disease at a local hospital. Like others, we demonstrate a high sensitivity for

Table 2. Diagnostic accuracy of MSCT in the three main vessels and in different segments of the coronary tree. RCA Segment no

LAD

Cx

1

2

3

4

5

6

7

8

9

10

11

12

13

14

True positive 9 False positive 10 True negative 26 False negative 1 MSCT non-assessable, QCA negative 2 MSCT non-assessable, QCA positive 0 Segment excluded, v2 mm diam. (QCA) or not existing 0 (peripheral segments) Segment excluded, stent 2 Segment excluded, distal to occlusion (QCA) 0 Sum 50

6 4 25 0 2 2 4

3 2 24 0 2 0 7

0 1 11 0 6 1 19

4 8 38 0 0 0 0

10 12 22 2 3 0 0

16 7 19 0 3 0 1

2 0 20 1 9 2 8

1 5 10 0 0 1 30

2 0 6 0 5 0 31

5 9 27 2 1 0 4

5 2 17 0 11 0 12

4 1 19 0 6 1 17

1 0 7 0 8 0 30

2 5 50

1 11 50

0 12 50

0 0 50

1 0 50

2 2 50

0 8 50

1 2 50

0 6 50

2 0 50

1 2 50

0 2 50

2 2 50

Results from segments 15 and 16 are not reported in this table as, in these segments, only 21 segments with a diameter w2 mm were identified by QCA. Cx: circumflex artery; LAD: left anterior descending artery (including left main artery); RCA: right coronary artery. Acta Radiol 2006 (7)

684

K. Kolnes et al.

detecting significant coronary artery stenosis and a high negative predictive value using standard invasive coronary angiography as the gold standard. However, compared to other studies, the specificity and in particular the positive predictive value were lower (11, 18). Overestimation of calcified and borderline significant lesions as well as misinterpretation of hypoplastic right coronary arteries were identified as important factors causing discrepancies between results obtained by MSCT and invasive coronary angiography with QCA. MSCT is a promising method for noninvasive evaluation of patients with suspected or definite coronary artery disease, although the exact role of this method in clinical practice has yet to be defined (19). For a number of reasons, the method has the potential to partly decentralize coronary artery imaging, which is a cornerstone in the evaluation of ischemic heart disease. Of importance in this context is that, in the course of a few years, the necessary equipment will be available at many minor hospitals. Furthermore, the method does not require invasive expertise. Compared to other CT examinations, MSCT coronary angiography is more demanding regarding patient preparation, image recording, and image interpretation (14). It is therefore a challenge to raise the level of expertise in this field at smaller hospitals, as increasing demand for this examination both from clinicians and patients is to be expected in the near future. In recently published clinical competence statements on cardiac imaging with computed tomography, considerations are given on competence recommendations for coronary CT angiography (5, 20). The innate gravity of coronary artery disease requires a high sensitivity and negative predictive value when MSCT is used in patients suspected of having this disease. We have demonstrated that this can be achieved, even with beginners’ experience. Our results, however, demonstrate a positive predictive value inferior to that published by others. The figures improved somewhat when lifted to patient-based evaluation due to higher prevalence at a patient compared to segment level. Of 41 patients that would have been referred to invasive coronary angiography based on the MSCT results, eight would have been referred without having significant coronary artery disease as defined by the invasive investigation. One out of 34 patients with significant coronary stenosis would have been missed due to false-negative conclusions on MSCT evaluation, and this particular patient was not a candidate for revascularization. Accordingly, if results from MSCT had been used in a triage for the group of Acta Radiol 2006 (7)

patients included in our study, no patient would have lost the opportunity for revascularization. Our patient material, however, is too small for a proper patient-based evaluation. We identified several factors contributing to the overestimation of stenosis severity in this study. Important factors for suboptimal image quality were calcifications, rhythm disturbances, increased heart rate, and technically suboptimal scanning. In a recent paper by Raff et al. (17), the effect of calcium and increased heart rate on MSCT accuracy was evaluated in a broad spectrum of patients examined with 64-slice MSCT. Even with this generation of scanner, accuracy was affected by these disturbing factors. Dedicated coronary calcium measurements with Agatston score (2, 4) have not been included in the MSCT examination in our study. Calcium is known to cause blooming artifacts with CT, and thereby prevent an exact assessment of stenoses (15). Calcification of the coronary arteries was the most important factor causing false-positive interpretations in our study in which a semiquantitative calcium scoring based on the MSCT angiograms was used. An unenhanced low-dose MSCT examination as a prescan could be of use in defining patients most suitable for MSCT coronary angiography. In case of calcifications giving rise to blooming artifacts or with an Agatston score over a certain value, omitting MSCT coronary angiography should be considered, as recently proposed by others (7). Our results indicate that we systematically overestimated borderline stenoses, particularly in the proximal segments of the coronary tree. The spatial resolution of the 16-slice MSCT images does not allow for a precise stenosis assessment on the basis of standard diameter measurements as is obtained during quantitative analysis of an invasive angiogram. The assessment of stenosis degree with MSCT is semiquantitative when using 16-slice CT systems that achieve a near-isotropic spatial resolution of approximately 0.5–0.8 mm, while invasive coronary angiography may provide a resolution of 0.1 mm. To differentiate a 10–20% coronary artery stenosis, an isotropic resolution of 0.3 mm is required (6). With 64-slice MSCT, a spatial resolution of 0.4 mm is provided (17). The patient population that may benefit the most from MSCT evaluation at a local hospital probably represents a group with lower prevalence of coronary artery disease compared to the population examined in our study. This could still yield a high predictive value of a negative examination, while the positive predictive value could decrease.

MSCT Coronary Angiography at a Local Hospital

Furthermore, the method could supplement clinical investigations in individuals with intermediate cardiovascular risk who, based on relevant ischemia testing, are borderline regarding indication for invasive investigation. Patients with acute chest pain of unknown etiology could also be considered candidates for MSCT both at local and larger hospitals. In this challenging clinical situation, MSCT could rule out aortic dissection, pulmonary embolism, and coronary artery disease. Our results argue for a strict and disciplined approach to the indication for MSCT examination in the evaluation of coronary artery disease, and for ‘‘beginners’’ a close dialogue with invasive cardiologists is recommended. The examination should be ordered by a cardiologist after having performed a clinical assessment including standard ischemia testing. With definite symptoms and signs of myocardial ischemia indicating the need for coronary artery revascularization, the role of MSCT is less convincing as these patients are candidates for invasive investigation irrespective of MSCT findings. In general, when defining the indication for MSCT coronary angiography, the significant radiation dose should be kept in mind (9, 13). MSCT has a low degree of invasiveness, and this could lower the threshold for MSCT coronary angiography. There is a challenge, therefore, to avoid that this technique results in an increasing number of referrals to invasive procedures due to the method generating false-positive results. There are several limitations of this study. The number of patients included is small and does not allow for statistical analysis of the patient-based evaluation. Our intention, however, was to examine whether, on a segmental basis, we could assess stenosis severity with MSCT. The results from this study should be interpreted with the patient selection in mind. Our study population was generally more affected by coronary artery disease than the population on which the method is likely to be adopted in a clinical scenario. In another patient population, results for accuracy could have been different. In conclusion, we have demonstrated beginners’ experience when introducing MSCT of coronary arteries at a local hospital. A high sensitivity and a high negative predictive value were obtained. However, specificity and in particular the positive predictive value were lower compared to those published by others. Calcifications were identified as an important factor for false-positive results. MSCT did not prove useful in differentiating significant versus non-significant lesions in patients

685

with known coronary artery disease. With widespread use of MSCT coronary angiography, there is a risk of recruiting patients without significant coronary artery disease to unnecessary and potentially harmful invasive procedures. Acknowledgments Funding: GE Healthcare, Norway, funded the costs of the contrast medium for the MSCT examinations in the study. Financial support was received from the Research Committee, Helse Sunnmøre (Regional Health Authority). The authors’ work was independent of the funders. Conflicts of interest: The main author received a lecture fee and reimbursement for attending symposia arranged by GE Healthcare. For the other authors, there are no conflicts of interest. References 1. Achenbach S, Giesler T, Ropers D, Ulzheimer S, Derlien H, Schulte C, et al. Detection of coronary artery stenoses by contrast-enhanced, retrospectively electrocardiographically-gated, multislice spiral computed tomography. Circulation 2001;103:2523–38. 2. Agatston AS, Janowitz WR, Hildner FJ, Zusmer NR, Viamonte M Jr, Detrano R. Quantification of coronary artery calcium using ultrafast computed tomography. J Am Coll Cardiol 1990;15:827–32. 3. Austen WG, Edwards JE, Frye RL, Gensini GG, Gott VL, Griffith LS, et al. A reporting system on patients evaluated for coronary artery disease. Report of the Ad Hoc Committee for Grading of Coronary Artery Disease, Council on Cardiovascular Surgery, American Heart Association. Circulation 1975;51 Suppl 4:5–40. 4. Becker CR. Estimation of cardiac event risk by MDCT. Eur Radiol 2005;15 Suppl 2:B17–22. 5. Budoff MJ, Cohen MC, Garcia MJ, Hodgson JM, Hundley WG, Lima JA, et al., American College of Cardiology Foundation; American Heart Association; American College of Physicians Task Force on Clinical Competence and Training; American Society of Echocardiography; American Society of Nuclear Cardiology; Society of Atherosclerosis Imaging; Society for Cardiovascular Angiography and Interventions, ACCF/AHA clinical competence statement on cardiac imaging with computed tomography and magnetic resonance: a report of the American College of Cardiology Foundation/American Heart Association/American College of Physicians Task Force on Clinical Competence and Training. J Am Coll Cardiol 2005;46:383–402. 6. Flohr TG, Schoepf UJ, Kuettner A, Halliburton S, Bruder H, Suess C, et al. Advances in cardiac imaging with 16-section CT systems. Acad Radiol 2003;10: 386–401. 7. Heuschmid M, Kuettner A, Schroeder S, Trabold T, Feyer A, Seemann MD, et al. ECG-gated 16-MDCT of Acta Radiol 2006 (7)

686

8.

9.

10.

11.

12.

13.

K. Kolnes et al. the coronary arteries: assessment of image quality and accuracy in detecting stenoses. Am J Roentgenol 2005;184:1413–9. Hoffmann MH, Shi H, Schmitz BL, Schmid FT, Lieberknecht M, Schulze R, et al. Noninvasive coronary angiography with multislice computed tomography. JAMA 2005;250:2471–8. Hunold P, Vogt FM, Schmermund A, Debatin JF, Kerkhoff G, Budde T, et al. Radiation exposure during cardiac CT: effective doses at multi-detector row CT and electron-beam CT. Radiology 2003;226: 145–52. Kuettner A, Beck T, Drosch T, Kettering K, Heuschmid M, Burgstahler C, et al. Diagnostic accuracy of noninvasive coronary imaging using 16-detector slice spiral computed tomography with 188 ms temporal resolution. J Am Coll Cardiol 2005;45:123–7. Kuettner A, Kopp AF, Schroeder S, Rieger T, Brunn J, Meisner C, et al. Diagnostic accuracy of multidetector computed tomography coronary angiography in patients with angiographically proven coronary artery disease. J Am Coll Cardiol 2004;43:831–9. Mollet N, Cademartiri F, Nieman K, Saia F, Lemos PA, McFadden EP, et al. Multislice spiral computed tomography coronary angiography in patients with stable angina pectoris. J Am Coll Cardiol 2004;43: 2265–70. Morin R, Gerber T, McCollough C. Radiation dose in computed tomography of the heart. Circulation 2003;107:917–22.

Acta Radiol 2006 (7)

14. Nakanishi T, Kayashima Y, Inoue R, Sumii K, Gomyo Y. Pitfalls in 16-detector row CT of the coronary arteries. Radiographics 2005;25:425–40. 15. Nieman K, Oudkerk M, Rensing BJ, van Ooijen P, Munne A, van Geuns RJ, et al. Coronary angiography with multi-slice computed tomography. Lancet 2001;357:599–603. 16. Nieman K, Rensing BJ, van Geuns RJ, Munne A, Ligthart JM, Pattynama PM, et al. Usefulness of multislice computed tomography for detecting obstructive coronary artery disease. Am J Cardiol 2002;89:913–8. 17. Raff GL, Gallagher MJ, O’Neill WW, Goldstein JA. Diagnostic accuracy of noninvasive coronary angiography using 64-slice spiral computed tomography. J Am Coll Cardiol 2005;46:552–7. 18. Ropers D, Baum U, Pohle K, Anders K, Ulzheimer S, Ohnesorge B, et al. Detection of coronary artery stenoses with thin-slice multi-detector row spiral computed tomography and multiplanar reconstruction. Circulation 2003;107:664–6. 19. Secthem U, Vo¨hringer M. The clinical role of ‘‘noninvasive’’ coronary angiography by multidetector spiral computed tomography: yet to be defined. Eur Heart J 2005;26:1942–4. 20. Weinreb JC, Larson PA, Woodard PK, Stanford W, Rubin GD, Stillman AE, et al. American College of Radiology. American College of Radiology clinical statement on noninvasive cardiac imaging. Radiology 2005;235:723–7.