A comparison between cardiac CT scanning and cardiac digital subtraction angiography (DSA)
Coronary artery disease affects nearly one tenth of the UK population and remains the leading cause of death in the western world. To investigate and provide interventions for coronary artery disease, imaging of the coronary arteries to enable visualisation of atheromatous plaque is required. This review looked at the techniques of cardiac computed tomography scanning and cardiac digital subtraction angiography, and their contribution to the investigation of coronary artery disease.
or any similar topic only for you
Comparison of the procedures considered technique, radiation exposure, contrast agent, clinical indications and efficacy in diagnosis of coronary artery stenosis. On comparison of the effectiveness of the techniques, both were shown to have been effective non-invasive procedures that may be used to rule out diagnoses and avoid inappropriate use of invasive angiography. The research shows mixed evidence for cardiac computed tomography angiography as a test of high specificity, however sensitivity and specificity of cardiac digital subtraction angiography is high, and therefore suggest that the techniques may be useful in low risk patients.
Coronary artery disease remains the main cause of death in the UK and western world (Hacker, 2013; Liu et al., 2002) and contributes a substantial disease burden, affecting 7% of men and 5% of women in the UK in 1999 (Liu et al., 2002).
Coronary artery disease results from the build up of atherosclerotic plaque within the arteries supplying the myocardium. This plaque limits the flow of blood through the arteries, and can cause ischaemia of the heart muscle. If the plaque becomes unstable and ruptures, this may lead to thrombus formation and the complete occlusion of an artery, resulting in a myocardial infarction (McClure et al., 2009).
Acute coronary syndrome (ACS) is a sub-classification of coronary artery disease and encompasses unstable angina, non-ST elevation myocardial infarction and ST elevation myocardial infarction. ACS represents a range of conditions that result from thrombus formation in coronary arteries, and if untreated has poor prognosis and high mortality (NICE, 2010).
To investigate and provide information for interventions for coronary artery disease and ACS, visualization of the coronary arteries and any lesions caused by atherosclerotic plaque is required. Advances in technology have given rise to several sophisticated perfusion analysis techniques, which provide greater prognostic value than morphological imaging (Hacker, 2013). Perfusion analysis allows the blood flow through the coronary vessels to be observed and any abnormalities in the perfusion can be interpreted as a functional consequence of atherosclerotic changes within the vessels (Hacker, 2013).
Currently, UK guidelines (NICE, 2010a) recommend coronary angiogram as first line management for patients presenting with ACS. This enables imaging of the coronary arteries to assess perfusion. It is important to assess the circulation as arteries can be affected from the earliest stages of endothelial dysfunction to high-grade coronary artery stenoses (Hacker et al., 2010; Bugiardini et al. 2004; Kaufmann et al., 2000), and this provides the information necessary for prognosis and intervention. With the advances in technology, there are now various diagnostic tests available to assess coronary artery disease, including coronary angiography and computed tomography (CT) scanning (Gorenoi, Schonermark & Hagen, 2012).
This review aims to review the literature on coronary CT scanning and digital subtraction angiography, their clinical applications, techniques and comparative value in coronary artery assessment and diagnosis.
Cardiac Digital subtraction angiography
Coronary angiography is the conventional diagnostic procedure used in coronary artery disease. It is a minimally invasive technique, whereby a catheter is placed into the radial or femoral artery and is advanced through the arterial system to the coronary arteries. A contrast agent is then injected at the aortic root and allows visualization of the arteries using x-ray in real time at up to 30 frames per second. This allows a view of the extent, location and severity of coronary obstructive lesions such as atherosclerosis and enables prognostic indication (Miller et al., 2008). Coronary angiography also enables catheter placement either side of the lesion to assess pressure changes and determines the degree of flow obstruction (Miller et al., 2008). .
Digital subtraction angiography (DSA) again works by introducing a contrast agent into the coronary arteries and taking x-rays in real time, however a pre image is taken by x-ray. This allows for the post images to be subtracted from the original mask image, eliminating bone and soft tissue images, which would otherwise overlie the artery under study (Hasegawa, 1987). Unlike conventional angiography, it is possible to conduct DSA via the venous system, through accessing the superior vena cava via the basillic vein (Myerowitz, 1982). This removes the risks associated with arterial cannulation (Mancini & Higgins, 1985). The procedure can also be performed with a lower dose of contrast agent and be done more quickly therefore eliminating constraints of using too much contrast during a procedure (Myerowitz, 1982).
Whilst DSA is the gold standard in arterial imaging of carotid artery stenosis (Herzig et al., 2004), the application of DSA to the coronary arteries is limited due to motion artefacts associated with each heartbeat and respiration (Yamamoto et al., 2009).
There are numerous cardiac clinical applications of DSA, it can be used to assess coronary blood flow (Molloi et al., 1996), valvular regurgitation (Booth, Nissen & DeMaria, 1985), cardiac phase (Katritsis et al., 1988), congenital heart shunts (Myerowitz, Swanson, & Turnipseed, 1985), coronary bypass grafts and percutaneous coronary intervention outcomes (Katritsis et al, 1988; Guthaner, Wexler & Bradley, 1985). However, others have suggested that the coronary arteries are not visualized well due to their small size, movement, their position overlying the opacified aorta and left ventricle, and confusion with other structures such as the pulmonary veins (Myerowitz, 1982).
Cardiac CT Scanning
Development of CT scanning in the 1990s enabled an increase in temporal resolution that was sufficient to view the beating heart, and they now provide a non-invasive technique for diagnostic and prognostic purposes. Cardiac CT scans have clinical applications that go beyond perfusion investigation, and can be used to assess structure and function of the heart (for example in electrophysiology disorders or congenital heart disease) due to its ability to provide anatomical detail (Achenbach & Raggi, 2010). CT scans can be used to assess coronary artery disease with and without injection of contrast agent (Achenbach & Raggi, 2010) by calcium scan or CT angiography.
Coronary calcium CT scanning uses the evidence base that coronary artery calcium is a correlate of atherosclerosis (Burke et al., 2003) and is a strong prognostic predictor of the future development of coronary artery disease and cardiac events (Arad et al., 2000; Budoff et al., 2009; Achenbach & Raggi, 2010). Calcium is easily depicted on CT scan due to its high CT attenuation, and is classified according to the Agatson score, which considers the density and area of the calcification (Hoffman, Brady & Muller, 2003).
Coronary CT angiography (CTA) allows visualization of the coronary artery lumen to identify any atherosclerosis or stenosis within the vessels. Patients are injected intravenously with a contrast agent and then undergo a CT scan. There are limitations regarding the suitability of patients for coronary CTA due to prerequisites of sinus rhythm, low heart rate and ability to follow breath-holding commands. Additionally, obesity presents a problem for patients that cannot fit into the scanner and affects the accuracy of the procedure.
(Achenbach & Raggi, 2010).
Comparison of cardiac DSA and cardiac CT scanning
The technical differences between cardiac DSA and cardiac CT scanning give rise to differences in the clinical indications for the procedures, their diagnostic efficacy and also different risks or relative benefits to the patients.
Due to the nature of the images produced by coronary CTA and DSA, each lends itself to different indications for use. Whilst coronary DSA provides imaging of all aspects of perfusion, CTA used with contrast agent also provides this however has the additional advantage of being able to assess structure and function of the heart.
Coronary CTA has been shown to have a high accuracy at detection and exclusion of coronary artery stenoses (Achenbach & Raggi, 2010). In a multicentre trial conducted by Miller et al. (2008), patients underwent coronary calcium scoring and CT angiography prior to conventional invasive coronary angiography. The diagnostic accuracy of coronary CTA at ruling out or detecting coronary stenoses of 50% was shown to have a sensitivity of 85% and a specificity of 90%. This showed that coronary CTA was particularly effective at ruling out non-significant stenoses. Additionally, coronary CTA was shown to be of equal efficacy as conventional coronary angiography at identifying the patients that subsequently went on to have revascularisation via percutaneous intervention. This was shown by an area under the curve (AUC), a measure of accuracy of 0.84 for coronary CTA and 0.82 for coronary angiography. Miller et al.’s (2008) study included a large number of patients at different study sites, and additionally represented a large variety of clinical patient characteristics. The author’s claim that these factors contribute to the strength and validity of the study findings, and suggest that in addition to using patients with clinical indications for anatomical coronary imaging, should be used as evidence that coronary CTA is accurate at identifying disease severity in coronary artery disease.
Miller et al. (2008) did however,, find that positive predictive and negative predictive values of coronary CTA were 91% and 83% respectively and therefore suggested that coronary CTA should not be used in place of the more accurate conventional coronary angiography. A low positive predictive value (in relation to the prevalence of disease) was proposed to be due to a tendency to overestimate stenosis degree as well as the presence of artefacts leading to false positive interpretation (Achenbach & Raggi, 2010).
Other research providing comparison between coronary CTA and conventional coronary angiogram has highlighted variability in results. A meta-analysis conducted by Gorenoi, Schonermark and Hagen (2012) investigated the diagnostic capabilities of coronary CTA and invasive coronary angiography using intracoronary pressure measurement as the reference standard. The authors found that CT coronary angiography had a greater sensitivity than invasive coronary angiography (80% vs 67%), meaning that coronary CTA was more likely to identify functionally relevant coronary artery stenoses in patients. Despite this,, specificity of coronary CTA was 67%, compared to 75% in invasive coronary angiography, meaning that the technique was less effective at correctly excluding non-diagnoses than invasive coronary angiogram. This research appears to contradict the power of cardiac CTA at excluding diagnoses of coronary artery stenosis as suggested by Miller et al. (2008), he study did combine evidence from over 44 studies to provide their results and therefore had a large statistical power. The authors interpret the results in light of the clinical relevance of cardiac imaging, suggesting that patients with a higher pretest possibility of coronary heart disease will likely require invasive coronary angiography for revascularisation indicating that coronary CTA may be a helpful technique in those patients with an intermediate pre-test probability of coronary heart disease that will therefore not require invasive angiography.
Goldberg et al. (1986) investigated the efficacy of DSA in comparison to conventional coronary angiography in 77 patients. They found that the two angiograms agreed within one grade of severity in 84% of single cases and 90% of multiple cases, identifying both patent and lesioned arteries. The results led the authors to conclude that there was no significant difference between the two methods and that DSA could be used in selective coronary angiography to find results comparable to that of conventional angiography. In addition to being a small study into the efficacy of DSA, the study also had several sources of inherent variability that should be considered when interpreting the results. These included differing sizes of digital imaging screen and non-use of calipers, meaning that the interpretation of the images could vary throughout the study. The authors also suggest that whilst showing strong support for the use of DSA in coronary artery disease, the technique may not actually permit better prognostic determinations or clinical judgements that are better than conventional angiography, and therefore the further implementation of the techniques may not be founded or necessitated.
More recently, there has been further research looking at the effectiveness of DSA as a way of measuring coronary blood flow. Whilst motion artefacts have proven a problem in lots of past research (Marinus, Buis & Benthem, 1990; Hangiandreou, 1990), recent research has developed methods to minimise these. Moilloi and colleaues (1996) showed that using a motion-immune dual-energy digital subtraction angiography, absolute volumetric coronary blood flow could be measured accurately and thus provide an indication of the severity of any arterial stenosis.This may provide further suggestion for clinical implementation of DSA.
Although these studies provide evidence for the efficacy of cardiac DSA and CTA, they often make comparisons to conventional angiography. This is useful as a baseline comparison, however it is difficult to make comparisons between the two procedures directly due to less available evidence making direct comparisons.
Lupon-Roses et al. (1985) conducted a study investigating both coronary CTA and venous DSA. The study looked at the efficacy of both techniques at diagnosing patency of coronary artery grafts compared to the control conventional angiography. CT was shown to diagnose 93% of the patent grafts and 67% of the occluded grafts whereas DSA correctly diagnosed 98% and 100% of patent and occluded grafts respectively. Interestingly, the DSA picked up the 11 grafts that were misdiagnosed by CTA and the CTA picked up the 2 grafts misdiagnosed by the DSA. This data may suggest that individually, DSA has a better profile for diagnosis of coronary artery occlusion, however if the two procedures are used in combination exclusion of patent arteries and diagnosis of occluded arteries would be effective (Lupon-Roses et al., 1985).
Coronary DSA and CTA are both non-invasive procedures (unlike the conventional coronary angiography where a wire is placed in the coronary vasculature). With the only invasive part of the procedure being the injection of the contrast material into a vein. This presents a significant advantage to both procedures over that of conventional angiography, and may even permit investigation on an outpatient basis (Meaney et al., 1980). Similarly, both DSA and coronary CTA are favoured because of their intravenous approach, eliminating the risks of bleeding or arterial injury from an intra-arterial catheterization and being able to be used in those with limited arterial access. However, although the intravenous approach used in cardiac DSA makes it favourable, it does lead to difficulty with visualisation of the coronary arteries due to the overlying iodinated pulmonary and cardiac structures (Mancini & Higgins, 1985). Therefore,, intra-arterial DSA is also sometimes used (Yamamoto et al., 2009).
As with all CT scanning, coronary CTA carries with it a dose of ionizing radiation (Brenner & Hall, 2007). Studies have estimated that for diagnostic CT scanning, patients are on average exposed to 12mSv of radiation during the procedure, the equivalent of 600 x-rays (Hausleiter et al., 2009). Estimates of radiation doses associated with conventional coronary angiography are lower than that of coronary CTA at 7mSv (Einstein et al., 2007). Additionally, DSA technique reduces the radiation dose from that of conventional coronary angiography as the vessels are visualised more clearly (Yamamoto et al., 2008). The dangers of radiation exposure are increased risk of developing cancer, skin injuries and cateracts (Einstein et al., 2007). It is therefore important that the benefits of conducting the procedure greatly outweigh the risk of radiation exposure. CT calcium scanning provides a low radiation dose at around 1mSv (Hunold et al., 2003).
Cardiac CT calcium scanning does not require administration of a contrast agent, unlike in coronary CTA and DSA that use iodine based contrast agents. The risks associated with contrast agent include nephrotoxicity and risks of hives, allergic reactions and anaphylaxis (Maddox, 2002). The amount of contrast agent used is partly dependent on the length of the procedure and how clearly the arteries can be visualised. For this reason, both cardiac CTA and DSA use less contrast agent that conventional coronary angiography (Brant-Zawadzki, et al., 1983). CT calcium scanning of the coronary arteries is therefore recommended in those with less likelihood of coronary artery disease (NICE, 2010).
Both coronary CTA and DSA require interpretation by trained physicians, and the importance of training and achieving intra-rater reliability should not be underestimated (Pugliese et al., 2009).
Overall, both coronary CT and DSA have been demonstrated as effective procedures for the imaging of the coronary arteries in CAD (Achenbach & Raggi, 2010; Miller et al., 2008; Moilloi et al., 1996; Goldberg et al., 1986). Whilst cardiac CT scanning does provide a wider range of clinical applications, allowing assessment of perfusion as well as cardiac structure and function (Achenbach & Raggi, 2010), coronary DSA has many applications that allow assessment of coronary blood flow (Molloi et al., 1996; Katritsis et al, 1988; Booth, Nissen & DeMaria, 1985; Guthaner, Wexler & Bradley, 1985; Myerowitz, Swanson, & Turnipseed, 198).
Both cardiac DSA and CTA procedures have their advantages. As non-invasive procedures, these techniques pose less risk to patients, and enable the possibility of outpatient investigation, to be used to rule out diagnoses and to avoid inappropriate invasive coronary angiogram (Gorenori et al., 2012). Additionally, intravenous access is preferential to arterial cannulation for the contrast infusion, removing the risks associated with bleeding or intra-arterial injury. Cardiac DSA exposes the patient to a lower dose of radiation that coronary CTA (Hausleiter et al., 2009; Yamamoto et al., 2008; Einstein et al., 2007), which is beneficial at reducing the risk of genetic mutations and cancer.
Cardiac CTA and DSA also have their common disadvantages. The use of contrast agent may present side effects for the patient including kidney damage and risk of allergic reactions and anaphylaxis (Maddox, 2002). For this reason, calcium CT scanning can be useful in patients that are not at high likelihood of coronary artery disease (NICE, 2010b). Additionally, both cardiac DSA and CTA are subject to motion artefacts from respiration and heart beats, which can cause difficulties with interpretation (Achenbach & Raggi, 2010; Yamamoto et al., 2009). In the case of cardiac CTA, this excludes a subset of patients that are unable to follow commands and those who have high heart rates.
Overall, cardiac CTA and cardiac DSA are effective, non-invasive imaging techniques for assessment of coronary artery disease. Whilst they are not the gold standards in cardiac monitoring, they can provide important diagnostic information without exposing patients to the risks of invasive angiography. Due to this, their use should be weighted against clinical need, the risks of the procedures, and the suitability of the patient. Interpretation of cardiac CTA and DSA imaging should be by trained individuals.
Achenbach, S., & Raggi, P. (2010) Imaging of coronary atherosclerosis by
computed tomography. European Heart Journal. 31:1442
Arad, Y., Spadaro, L. A., Goodman, K., Newstein, D., & Guerci, A. D. (2000). Prediction of coronary events with electron beam computed tomography.Journal of the American College of Cardiology, 36(4), 1253-1260.
Booth, D. C., Nissen, S., & DeMaria, A. N. (1985). Assessment of the severity of valvular regurgitation by digital subtraction angiography compared to cineangiography. American heart journal, 110(2), 409-416.
Brenner D. J., & Hall EJ. Computed tomography: an increasing source of radiation exposure. N Engl J Med. 2007;357(22):2277-2284
Budoff, M. J., McClelland, R. L., Nasir, K., Greenland, P., Kronmal, R. A., Kondos, G. T., … & Blumenthal, R. S. (2009). Cardiovascular events with absent or minimal coronary calcification: the Multi-Ethnic Study of Atherosclerosis (MESA). American heart journal, 158(4), 554-561.
Bugiardini, R., Manfrini, O., Pizzi, C., Fontana, F., & Morgagni, G. (2004). Endothelial function predicts future development of coronary artery disease a study of women with chest pain and normal coronary angiograms.Circulation, 109(21), 2518-2523.
Burke, A. P., Virmani, R., Galis, Z., Haudenschild, C. C., & Muller, J. E. (2003). Task force# 2—what is the pathologic basis for new atherosclerosis imaging techniques?. Journal of the American College of Cardiology,41(11), 1874-1886.
Einstein, A. J., Moser, K. W., Thompson, R. C., Cerqueira, M. D., & Henzlova, M. J. (2007). Radiation dose to patients from cardiac diagnostic imaging.Circulation, 116(11), 1290-1305.
Goldberg HL, Moses JW, Fisher J, Tamari I, Borer JS (1986). Diagnostic accuracy of coronary angiography utilizing computer-based digital subtraction methods; Comparison to conventional cineangiography. Chest 90, 793–797,
Gorenoi, V., Schonermark, M. P., & Hagen, A. (2012). CT coronary angiography vs. invasive coronary angiography in CHD. GMS health technology assessment, 8, Doc02-Doc02..
Guthaner, D. F., Wexler, L., & Bradley, B. (1985). Digital subtraction angiography of coronary grafts: optimization of technique. American journal of roentgenology, 145(6), 1185-1190.
Hacker, M., Jakobs, T., Hack, N., Nikolaou, K., Becker, C., von Ziegler, F., … & Tiling, R. (2007). Sixty-four slice spiral CT angiography does not predict the functional relevance of coronary artery stenoses in patients with stable angina. European journal of nuclear medicine and molecular imaging,34(1), 4-10.
Hangiandreou N. J. (1990) Coronary Blood Flow Measurement Using Digital Subtraction Angiography and First Pass Distribution Analysis. Madison, Wis: University of Wisconsin-Madison; Thesis.
Hasegawa, B. (1987). Physics of Medical X-Ray Imaging 2nd Edition. Medical Physics Publishing Corporation.
Hausleiter, J., Meyer, T., Hermann, F., Hadamitzky, M., Krebs, M., Gerber, T. C., … & Achenbach, S. (2009). Estimated radiation dose associated with cardiac CT angiography. Jama, 301(5), 500-507.
Herzig, R., Burval, S., Krupka, B., Vlachova, I., Urbanek, K., & Mares, J. (2004). Comparison of ultrasonography, CT angiography, and digital subtraction angiography in severe carotid stenoses. European Journal of Neurology, 11(11), 774-781.
Brant-Zawadzki, M., Gould, R., Norman, D., Newton, T. H., & Lane, B. (1983). Digital subtraction cerebral angiography by intraarterial injection: comparison with conventional angiography. American Journal of Roentgenology, 140(2), 347-353.
Hoffmann, U., Brady, T.J., & Muller, J. (2003). Cardiology patient page. Use of new imaging techniques to screen for coronary artery disease. Circulation 108 (8): e50–3.
Hunold, P., Vogt, F. M., Schmermund, A., Debatin, J. F., Kerkhoff, G., Budde, T., … & Barkhausen, J. (2003). Radiation Exposure during Cardiac CT: Effective Doses at Multi–Detector Row CT and Electron-Beam CT 1.Radiology, 226(1), 145-152.
Katritsis, D., Lythall, D.A., Cooper, I.C., Crowther, A., & Webb-Peploe, M.M. (1988) Assessment, of coronary angioplasty: Comparison of visual assessment, hand?held caliper measurement and automated digital quantitation. Catheterization and cardiovascular diagnosis, 15(4), 237-242.
Kaufmann, P. A., Gnecchi-Ruscone, T., Schafers, K. P., Luscher, T. F., & Camici, P. G. (2000). Low density lipoprotein cholesterol and coronary microvascular dysfunction in hypercholesterolemia. Journal of the American College of Cardiology, 36(1), 103-109.
Liu, J. L. Y., Maniadakis, N., Gray, A., & Rayner, M. (2002). The economic burden of coronary heart disease in the UK. Heart, 88(6), 597-603.
Lupon-Roses, J., Domingo, E., Marinez-Vazquez, J. M., Lopez-Moreno, J. L., Montana, J., Permanyer-Miralda, G., … & Soler-Soler, J. (1985). Direct non-invasive techniques for assessing coronary bypass graft patency. The International Journal of Cardiac Imaging, 1(3), 181-188.
Maddox, T. G. (2002). Adverse reactions to contrast material: recognition, prevention, and treatment. American family physician, 66(7), 1229.
Mancini, J. G. B., & Higgins, C. B. (1985). Digital subtraction angiography: a review of cardiac applications. Progress in cardiovascular diseases, 28(2), 111-141.
Marinus, H., Buis, B., & Van Benthem, A. (1990) Pulsatile coronary flow determination by digital angiography. International Journal of Cardiac Imaging, 5, 173-182
McClure, K. H., McGivern, J. P., Stultz, M. R., & Whitehurst, T. K. (2009). U.S. Patent No. 7,481,759. Washington, DC: U.S. Patent and Trademark Office.
Meaney, T. F., Weinstein, M. A., Buonocore, E., Pavlicek, W., Borkowski, G. P., Gallagher, J. H., … & Maclntyre, W. J. (1980, August). Digital subtraction angiography of the human cardiovascular system. In Application of Optical Instrumentation in Medicine VIII (pp. 272-278). International Society for Optics and Photonics.
Miller, J. M., Rochitte, C. E., Dewey, M., Arbab-Zadeh, A., Niinuma, H., Gottlieb, I., … & Lima, J. A. (2008). Diagnostic performance of coronary angiography by 64-row CT. New England Journal of Medicine, 359(22), 2324-2336.
Molloi, S., Ersahin, A., Tang, J., Hicks, J., & Leung, C. Y. (1996). Quantification of volumetric coronary blood flow with dual-energy digital subtraction angiography. Circulation, 93(10), 1919-1927.
Myerowitz, P. D. (1982). Digital subtraction angiography: present and future uses in cardiovascular diagnosis. Clinical cardiology, 5(12), 623-629.
Myerowitz, P. D., Swanson, D. K., & Turnipseed, W. D. (1985). Applications of digital subtraction angiography in cardiovascular diagnosis. The Surgical clinics of North America, 65(3), 423-437.
National Institute for Health and Care Excellence. (2010a). Unstable angina and NSTEMI: The early management of unstable angina and non-ST-segment-elevation myocardial infarction. CG94. London: National Institute for Health and Care Excellence.
National Institute for Health and Care Excellence. (2010b). Unstable angina and NSTEMI: Chest pain of recent onset: assessment and diagnosis of recent onset chest pain or discomfort of suspected cardiac origin. CG95. London: National Institute for Health and Care Excellence.
Pugliese, F., Hunink, M. M., Gruszczynska, K., Alberghina, F., Malago, R., van Pelt, N., … & Krestin, G. P. (2009). Learning curve for coronary CT angiography: what constitutes sufficient training?. Radiology, 251(2), 359.
Yamamoto, M., Okura, Y., Ishihara, M., Kagemoto, M., Harada, K., & Ishida, T. (2009). Development of digital subtraction angiography for coronary artery. Journal of digital imaging, 22(3), 319-325.