Prognostic Value of Coronary Computed Tomographic Angiography In Comparison With Calcium Scoring and Clinical Risk Scores

OBJECTIVES: Several studies have demonstrated a high accuracy of coronary computed tomography angiography (CCTA) for detection of obstructive coronary artery disease (CAD), whereas some studies have also shown a good prediction of cardiac events. However, it remains to be proven whether CCTA is better predictive of events than conventional risk scores or calcium scoring. Therefore, we compared CCTA with calcium scoring and clinical risk scores for the ability to predict cardiac events.

METHODS: Patients (n=2223) with suspected CAD undergoing CCTA were followed up for a median of 28 months. The end point was the occurrence of cardiac events (cardiac death, nonfatal myocardial infarction, unstable angina requiring hospitalization, and coronary revascularization later than 90 days after CCTA).

RESULTS: Patients with obstructive CAD had a significantly higher event rate (2.9% per year; 95% confidence interval, 2.1 to 4.0) than those without obstructive CAD, having an event rate 0.3% per year (95% confidence interval, 0.1 to 0.5; hazard ratio, 13.5; 95% confidence interval, 6.7 to 27.2; P<0.001). CCTA had significant incremental predictive value when compared with calcium scoring, both with scores assessing the degree of stenosis (P<0.001) and with scores assessing the number of diseased coronary segments (P=0.027).

CONCLUSIONS: In patients with suspected CAD, CCTA not only detects coronary stenosis but also improves prediction of cardiac events over and above conventional risk scores and calcium scoring. This may result in a reclassification of cardiovascular risk in a substantial proportion of patients. Marcus Davenport Womens Jersey

PMID: 20884832

Posted in * Journal Club Selections, Computed Tomography and tagged , , , , , .

3 Comments

  1. What is the purpose of this study?
    The accuracy of coronary computed tomography (CCTA), particularly the negative predictive value, has been shown in many studies. There is a relative paucity of studies addressing the value of CCTA in predicting clinical outcomes. This investigation addressed the value of CCTA in predicting future events, with comparison to calcium scoring and clinical risk scores.

    What were the endpoints? Do they raise any initial concerns?
    Endpoints were the occurrence of cardiac death, non fatal myocardial infarction, unstable angina requiring hospitalization, and coronary revascularization later than 90 days after CCTA. Including coronary revascularization as an endpoint raises the possibility that findings on CCTA such as luminal narrowing were not actually predictive of cardiac events, but rather predictive of luminal narrowing that would also be seen at time of catheterization, and eventually lead to intervention.

    What was the population evaluated?
    Consecutive patients with suspected but known CAD undergoing CCTA and calcium scoring between 2003 and 2008. Patients without stable sinus rhythm, or in acute life threatening conditions were excluded from analysis.

    What was analyzed in addition to CCTA?
    Patient age, height, weight, cardiac disease history, medications were all noted. Lab testing included total cholesterol, LDL and HDL fraction, and triglycerides. From this data, Framingham risk scores using Wilson and Morise scores were calculated.

    Was there any variation in CCTA acquisition over time?
    16-slice MDCT was initially used (12/03 – 9/04), 64-slice single source from 10/04 – 09/06, and 64-slice dual source CT system from 10/06 – 02/08. No significant discussion or analysis of the effect of this evolution of technology was made, although the improved temporal and spatial resolution, as well as shorter breath hold associated with newer generation scanners could arguably lead to improved depiction of coronary arteries.

    What were the diagnostic criteria for CCTA analysis?
    Coronary tree was segmented by AHA classification; segments > 1.5 mm were evaluated by a radiologist and cardiologist, with disagreement settled by consensus. Stenosis was rated visually: no relevant stenosis (75%). If artifacts prevented accurate assessment, the segment was classified as moderately stenotic. Obstructive CAD was defined as 1 or more segments with stenosis greater than or equal to 50%. Each segment was also assessed for the presence of calcified or non calcified plaques. Good interreader agreement (Cohen’s K = 0.84) was shown in the assessment of stenosis on a per artery basis.

    What indices were calculated?
    Coronary obstruction score, # of involved segments, presence of proximal left lesion, as well as recently published severity of CAD scores proposed by Chow and Min for CCTA analysis.

    What was the statistical approach?
    All stats were based on event-free survival for the study endpoint using the Kaplan-Meier method; hazard ratios for difference between the 75th and 25th percentile and multivariable analyses were calculated with the Cox proportional hazard model. C-indices were calculated from time-to-event data, for correlation the spearman test was used. Incremental predictive value was assessed primarily using the Integrated Discrimination Improvement method.

    How many patients were analyzed? How many events were noted? What was the follow up period?
    2223 pts (after exclusion of 6 pts with aortic dissection, and 62 pts without stable sinus rhythm, and patient who could not be followed up) were followed for a median of 29 months. Given the fairly low amount of events observed (4 cardiac deaths, 8 non fatal MI, 4 unstable angina, and 31 revascularizations), the follow up time may represent a limitation to this study.

    What were the CCTA results?
    Normal coronary arteries were noted in 24% of pts, 47% had non obstructive CAD and 29% had obstructive CAD. Annual event rates were 0% for pts with normal coronary arteries, and 4%(2-8.8%) for pts with stenosis > 75%. Pts with obstructive CAD had a significantly higher event rate than those without (2.9% per year vs. 0.3%, hazard ratio 13.5). Annual event rate also increased with extent of coronary involvement, with 0% for no involvement and 4.9% for more than 9 segments.

    Was incremental value shown for CCTA or calcium score?
    The clinical (Morise score, c-index = 0.76) predictive value was increased by inclusion of calcium score (c-index = 0.84 combined model, p<0.001), and further addition of coronary obstruction score (c-index = 0.89, p<0.001). The extent of atherosclerotic burden did not show strong prognostic value.

    What is the take away message from this study? What are future steps?
    This study tackled an important question – how does CCTA compare to clinical prediction models and calcium scoring in the prediction of future cardiac events. This type of investigation is important in this era where close scrutiny of the clinical value of imaging studies is occurring, perhaps nowhere more intensely than for CCTA. A large number of patients were prospectively followed in this study, with calcium scoring and clinical risk models for comparison. Incremental value was shown for both calcium scoring, as well as for CCTA. However, limits to the confidence of these conclusions include a fairly low number of cardiac events, the bulk of which were composed of revascularization. A longer follow up period may allow for more confident depiction of the role of CCTA in prediction of downstream cardiac events. Discussing the actual differences in clinical management of patients based on risk restratification was likely out of the scope of this study, but the question of how a patient would be treated differently based on the clinical predictive values versus the combined clinical plus imaging approach deserves a more structured explanation.

  2. » What is the purpose of this study?
    To determine whether CCTA is a better predictor of events than conventional risk scores or calcium scoring.

    » What were the endpoints? Do they raise any initial concerns?
    Cardiac events defined as cardiac death (including any death without definitive noncardiac cause), nonfatal myocardial infarction, or unstable angina pectoris requiring hospitalization or coronary revascularization, whatever occurred first.

    » What was the population evaluated?
    The population included mainly patients with intermediate risk (pre-test) that would benefit from further stratification. We thought this was appropriate and where CCTA has the biggest impact.

    » What was analyzed in addition to CCTA?
    It was very interesting to us that the design of the study mirrored the type of analysis used in conventional angiography, mainly the ‘number of affected segments’. This is rarely analyzed in CCTA studies.
    It would’ve been very interesting to see if other findings that can be seen with CCTA but not with conventional angiography, such as positive remodeling, had an impact on the predictive value.

    » Was there any variation in CCTA acquisition over time?
    Different scanners were used. However, a recent meta-analysis done assessing outcome data after CCTA showed no significant difference among scanners. []

    » What were the diagnostic criteria for CCTA analysis?
    The criteria used was:
    1. Number of segments involved.
    2. Grade of narrowing (by segment)
    3. Obstructive vs Non-obstructive (worst segment per patient)
    4. Type of plaque.

    » What was the statistical approach?
    We enjoyed reviewing the figures presented. There were Meyer-Kaplan graphs as well as ROC comparing the ‘standard’ clinical risk factors with out and with the added value of calcium scoring or both calcium scoring and the assessed CCTA data.
    We did feel that while Figure 4 was extremely clear and very didactic. Having a similar graph using only calcium score rather than CCTA would’ve been very interesting.

    » How many patients were analyzed? How many events were noted? What was the follow up period?
    A total of 2387 patients were included and only 47 cardiac events were recorded.
    One issue that we had was with the length of follow-up. The literature available for the use of clinical risk factors as predictors uses 10-years for follow up, while this paper follow-up ranged from 7 to 64.8 months, with a median of 29months (Q1, 17 months; Q3, 42 months). This is fine, but the strong opinions voiced in the discussion should be more tempered with if this shorter follow-up period is being used.

    » Was incremental value shown for CCTA or calcium score?
    Both, probably more dramatic between calcium score and clinical factors only than between the addition of calcium score by itself or combined with CCTA (against calcium score alone). This was more pronounced when using the degree of obstruction data rather than the number of affected segments.

    » What is the take away message from this study? What are future steps?
    Although the data seems solid, it remains to be determined what is the cost-benefit and risk-benefit of performing a more elaborate, time-consuming, expensive study that uses more radiation and IV contrast.

Leave a Reply

Your email address will not be published. Required fields are marked *