OUP user menu

Diagnostic performance of intracoronary gradient-based methods by coronary computed tomography angiography for the evaluation of physiologically significant coronary artery stenoses: a validation study with fractional flow reserve

Jin-Ho Choi, Bon-Kwon Koo, Yeonyee E. Yoon, James K. Min, Young-Bin Song, Joo-Yong Hahn, Seung-Hyuk Choi, Hyeon-Cheol Gwon, Yeon Hyeon Choe
DOI: http://dx.doi.org/10.1093/ehjci/jes130 1001-1007 First published online: 16 July 2012

Abstract

Aims We validated and compared transluminal attenuation gradient (TAG) and corrected coronary opacification (CCO) of coronary computed tomography angiography (CCTA) with invasively measured fractional flow reserve (FFR).

Background One of the major limitations of CCTA is the discrepancy between angiographical stenosis and ischaemia-causing stenosis. Recently two new CCTA analysis methods, TAG and CCO, have been attempted to overcome this limitation but without physiological validation.

Methods and results We measured TAG and CCO of 97 major epicardial coronary arteries from 63 patients who underwent CCTA and followed by invasive coronary angiography and FFR. Diagnostic performance of TAG and CCO was assessed using FFR <0.80 as the reference standard. The overall diagnostic performance of TAG and CCO on a per-vessel basis was moderate and similar (c-statistic = 0.696 vs. 0.637, P = 0.29). The sensitivity, specificity, positive, and negative predictive values of TAG cut-off ≤−0.654 for FFR <0.80 were 47.5, 91.2, 79.2, and 71.2%, and those of CCO cut-off >0.063 were 65.0, 61.4, 54.2, and 71.4%. TAG showed an incremental value to the diagnostic performance of CCTA but CCO did not (c-statistic =0.726 vs. 0.809, P = 0.025; c-statistic =0.726 vs. 0.784, P = 0.09). In net reclassification improvement (NRI) analysis, addition of TAG to CCTA did not result in significant reclassification (NRI = 1.0%, P = 0.41) and addition of CCO to CCTA resulted in negative reclassification (NRI = −9.3%, P = 0.036).

Conclusion Intracoronary attenuation-based CCTA analyses, TAG and CCO, showed moderate correlation with physiological coronary artery stenosis. The incremental value of TAG or CCO to the evaluation of haemodynamically stenosis by CCTA seemed to be limited.

  • Coronary computed tomography
  • Contrast dye gradient
  • Fractional flow reserve

Introduction

Current coronary computed tomography angiography (CCTA) has very high negative predictive power for coronary artery disease (89–99%). However, positive predictive power for the detection of anatomically or functionally significant stenosis is still moderate (51–82%).15 It is not unexpected because the diameter stenosis has been known to be a poor indicator of myocardial ischaemia, even by invasive coronary angiography (CAG) with higher resolution than CCTA.6,7 Recent investigations have shown that the identification of lesion-specific ischaemia is necessary for determining future clinical benefit from coronary revascularization.8,9 Hence, an increasing specificity and a positive predictive value of CCTA would be crucial to guide or trigger revascularization or other treatment strategy.10,11

Analysis of the intracoronary luminal attenuation is available from standard CCTA data without modification of acquisition protocols or additional imaging, and has been known to reflect the intracoronary blood flow.12,13 Based on this concept, transluminal attenuation gradient (TAG) and corrected coronary opacification (CCO) have been developed to increase the diagnostic performance of CCTA, but have not validated against lesion-specific myocardial ischaemia.12,14,15 We validated and compared the diagnostic performance of TAG and CCO for the detection of lesion-specific functional significance determined by invasively acquired fractional flow reserve (FFR).

Methods

Patients

We retrospectively evaluated CCTA datasets from 65 patients who were found to have ≥50% diameter stenosis (DS) in a major coronary artery and subsequently underwent invasive CAG and FFR studies in Seoul National University Hospital from October 2009 to January 2011. All CCTA were clinically indicated and imaged to diagnose significantly obstructive coronary artery disease prior to invasive CAG. The median interval between CCTA and CAG was 35 days (inter-quartile range: 23–46). Patients with CCTA images that were determined as non-evaluable were excluded. The institutional review board committee had approved the study protocol.

Acquisition, reconstruction, and analysis of CCTA image

CCTA was performed with 64-detector-row scanners (Somatom Definition, Siemens, Forchheim, Germany; Brilliance 64, Philips, Best, the Netherlands). Oral metoprolol 100 mg was administered to patients with a heart rate ≥65/min. Sublingual nitroglycerin 0.6 mg was administered before image acquisition. Contrast injection was done after test bolus tracking, and the protocol consisted of 80 mL of contrast (Iomeron 400; Bracco, Milan, Italy) followed by a saline flush. The scan parameters were set to 100–120 kV, tube current 340 mA/800 mA, 64 × 0.625/0.750 mm collimation, and 0.33/0.4 s tube rotation time. A radiation dose ranged between 3 and 15 mSv. Image was reconstructed with the use of retrospective electrocardiographic (ECG) gating.

CT images were post-processed with GE Advantage Workstation and interpreted by expert interpreters blinded to clinical data and results of invasive CAG or FFR. Coronary segments were scored in a semi-quantitative manner using an 18-segment SCCT model.16 The severity of luminal stenosis was classified and expressed on an ordinal scale as 0 (none; DS = 0%), 1 (very mild; 1–29%), 2 (mild; 30–49%), 3 (moderate; 50–69%), 4 (severe; 70–99%), or 5 (total occlusion; 100%). Plaque composition was classified as non-calcified (<30% calcified plaque volume), partially calcified (30–70%), or calcified plaque (>70%) according to the volume of calcified component [>130 HU (Hounsfield Unit)] in the plaque. For the analysis of maximum stenosis in each vessel, none to mild stenoses was combined into single category (Figure 1).14

Figure 1

Concept of transluminal attenuation gradient (TAG) and corrected coronary opacification (CCO). (A) Transluminal attenuation gradient (TAG) is defined by linear regression coefficient between intraluminal radiological attenuation (HU) measured in 5-mm intervals and length from the ostium (mm). TAG is measured in HU/mm and shown in blue arrows. To measure TAG, the mean Hounsfield Unit (HU) value is assessed from each region of interest (ROI, shown as red circles). TAG of vessel with significant stenosis is lower that TAG of vessel with insignificant stenosis. (B) Corrected coronary opacification (CCO) is defined as the quotient of the mean intraluminal HU of the normal-looking coronary segment and the descending aorta in the same axial plane to minimize the lack of temporal uniformity caused by 64-detector row CT. For the assessment of CCO, the quotient of the mean intraluminal HU of the normal-looking coronary segment and the descending aorta in the same axial plane was calculated in the intracoronary segment most proximal to stenosis and most distal to the stenosis in axial slices. Then CCO was defined as the difference between two quotients. CCO of vessel with significant stenosis is higher than CCO of vessel with insignificant stenosis.

Assessment of plaque burden, transluminal attenuation gradient, and corrected coronary opacification

The stenosis and plaque characteristics were evaluated in each lesion. Vessels were classified as having non-calcified lesion if the most stenotic portion was non-calcified, and as having calcified lesion if the most stenotic portion was calcified or partially calcified. The severity of luminal stenosis was classified and expressed on an ordinal scale as 0 (none; DS = 0%), 1 (very mild; 1–29%), 2 (mild; 30–49%), 3 (moderate; 50–69%), 4 (severe; 70–99%), or 5 (total occlusion; 100%). For the analysis of maximum stenosis in each vessel, none to mild stenoses was combined into single category. The segment stenosis score (0–15), the sum of maximal stenosis grade (0–5) in each of the three major epicardial coronary arteries, and segments-at-risk score (0–15), in which proximal segments were weighted, were calculated in a manner that we have described previously to reflect the overall burden of obstructive coronary artery disease and the higher contribution to ischaemia in proximal coronary segment, respectively.17

TAG and CCO were assessed as previously described.14,15 Briefly, cross-sectional images perpendicular to the vessel centreline were reconstructed for each major coronary artery. The following variables were measured at 5-mm intervals, from the ostium to the distal level where the vessel cross-sectional area fell <2.0 mm2. The contour of the region of interest and the vessel centreline were manually corrected if necessary. TAG was determined from the change in HU per 10 mm length of the coronary artery, and defined as the linear regression coefficient between intraluminal radiological attenuation (HU) and length from the ostium (mm). CCO was assessed by the analysis of CCTA axial slices. The quotient of the mean intraluminal HU of the normal-looking coronary segment and the descending aorta in the same axial plane was calculated in the intracoronary segment most proximal to stenosis and most distal to the stenosis, respectively. CCO was defined as the difference between two quotients.

Invasive coronary angiography and fractional flow reserve

CAG and FFR were done after CCTA as described previously by us.5 In brief, the maximally stenotic lesions in each major epicardial coronary arteries were evaluated using quantitative coronary angiography (QCA). FFR was performed in clinically indicated vessels using hyperaemia attained by i.v. adenosine (140 mcg/kg/min). A pressure-sensing guidewire (PressureWire Certus, St Jude Medical Systems, Uppsala, Sweden; ComboWire, Volcano Corporation, San Diego, CA, USA) was used. FFR was considered diagnostic of ischaemia at a threshold of ≤0.80.

Statistical analysis

Data were shown as mean ± SD, mean (95% confidence interval), or % (n) unless specified. All analysis was done per-vessel basis, and no adjustments were made for multiple vessels or segments within individuals. Relationships between two continuous variables and continuous and categorical variables were tested by the Kruskal–Wallis test and the Jonckheere–Terpstra test for trend. Diagnostic performance was compared using receiver-operating characteristic (ROC) analysis with DeLong's method and net reclassification improvement index.18 Optimal cut-off of measured variables was determined by values with highest Youden's J statistics. SPSS version 13 and Medcalc version 11 were used. A two-tailed P < 0.05 was considered statistically significant.

Results

Clinical characteristics

CCTA studies from 65 patients were evaluated. After excluding two patients with non-evaluable CCTA, 63 patients (63.1 ± 8.4 years; 69.8% male gender) and 97 coronary arteries were included in the final analysis. Baseline characteristics are listed in Table 1.

View this table:
Table 1

Clinical characteristics

Age (year)63.1 ± 8.4
Male gender69.8 (44)
Body mass index (kg/m2)24.1 ± 2.6
Cardiovascular history and risk factors
 Hypertension61.9 (39)
 Diabetes33.3 (21)
 Hyperlipidaemia63.5 (40)
 Smoking46.0 (29)
 Previous history of MI3.2 (2)
 Previous history of PCI3.2 (2)
 Left ventricular ejection fraction (%)61.5 ± 5.0
Vital signa
 Systolic blood pressure (mmHg)125 ± 17
 Diastolic blood pressure (mmHg)75 ± 10
 Heart rate (/min)70 ± 10
Laboratory and imaging test
 Haemoglobin (g/dL)13.5 ± 1.7
 Creatinine (g/dL)1.0 ± 0.7
Medication
 Aspirin100 (63)
 Clopidogrel85.7 (54)
 Statin73.0 (46)
 Beta-blocker41.3 (26)
 ACE inhibitor or ARB33.3 (21)
 Calcium channel blocker34.9 (22)
 Nitrate54.0 (34)
  • MI, myocardial infarction; PCI, percutaneous coronary intervention; ACE, angiotensin-converting enzyme; ARB, angiotensin II receptor blocker.

  • aMeasured during cardiac catheterization.

Assessment of coronary artery lesion

A total of 55 lesions (56.7%) were located in left anterior coronary artery (LAD), and 42 lesions (43.4%) were in non-LAD. The frequency of lesion with DS ≥50% was comparable between CCTA and CAG, 66.0% (64/97) and in 61.9% (60/97), respectively. Functionally significant lesion defined by FFR <0.8 was found in 41.2% (40/97) (Table 2).

View this table:
Table 2

Assessment of coronary artery lesion

n97
LAD56.7 (55)
LCX19.6 (19)
RCA23.7 (23)
CCTA stenosis and plaque burden scores
 CCTA DS ≥50%66.0 (64)
 CCTA DS ≥70%18.6 (18)
 Segment stenosis score3.6 ± 2.0
 Segment-at-risk score6.7 ± 3.3
 Non-calcified plaque37.6 (35)
 Partially calcified plaque34.4 (32)
 Calcified plaque32.4 (30)
CAG stenosis by QCA
 QCA DS%54.8 ± 16.4%
 QCA DS ≥50%61.9 (60)
 QCA DS ≥70%25.8 (25)
Functional measurement
 TAG−0.404 ± 0.611
 CCO0.106 ± 0.187
 FFR0.81 ± 0.12
 FFR <0.841.2 (40)
  • LAD, left anterior coronary artery; LCX, left circumflex artery; RCA, right coronary artery; CCTA, coronary computed tomography angiography; DS, diameter stenosis; CAG, invasive coronary angiography; QCA, quantitative coronary angiography; TAG, transluminal attenuation gradient; CCO, corrected coronary opacification; FFR, fractional flow reserve.

The burden of the coronary atherosclerosis plaque was assessed by the segment stenosis score or the segment-at-risk score and compared with functional measurements assessed by TAG, CCO, or FFR (Figure 2). TAG and FFR decreased consistently according to the quartiles of the segment stenosis score or the segment-at-risk score (P < 0.001 by Jonckheere–Terpstra test). CCO increased according to the quartiles of the segment-at-risk score (P < 0.001) but not to the quartiles of the segment stenosis score (P = 0.09). Significant correlations between TAG and CCO against CAT QCA and FFR were shown in Figure 3 (P < 0.05, all).

Figure 2

Comparison of plaque burden evaluated by CCTA semi-quantitative scores with functional measurements assessed by TAG, CCO, and FFR. (A and B) TAG decreased consistently according to the quartiles of the segment stenosis score (0–2, 3, 4–5, 6–8) and of the segment-at-risk score (0–4, 5–6, 7–8, 9–12) (P < 0.001 by the Jonckheere–Terpstra test). For comparison of each group, P = 0.020, P = 0.002 by the Kruskal–Wallis test, respectively. (C and D) CCO did not increased consistently according to the quartiles of the segment-at-risk score (P = 0.09) and increased according to the quartiles of the segment-at-risk score (P < 0.001). For comparison of each group, P = 0.033, P = 0.017. (E and F) FFR decreased consistently according to the quartiles of the segment stenosis score (0–2, 3, 4–5, 6–8) and the quartiles of the segment-at-risk score (0–4, 5–6, 7–8, 9–12) (P < 0.001). For comparison of each group, P < 0.001. TAG, transluminal attenuation gradient; CCO, corrected coronary opacification; FFR, fractional flow reserve.

Figure 3

Correlations between TAG and CCO against invasive QCA and FFR. (A and B) Correlation between TAG and CCTO against invasively acquired FFR. There was no statistical difference between absolute values of correlation coefficients (P = 0.28). (C and D) Correlation between TAG and CCTO against invasively acquired QCA. P = 0.93 between absolute values of correlation coefficients.

Diagnostic performance of TAG and CCO compared with FFR

Diagnostic performance of TAG and CCO for FFR <0.8 were analysed on a vessel basis and shown in Figure 4 and Table 3. In ROC analysis, there was no difference of c-statistic between CCTA DS ≥50%, TAG, and CCO (c-statistic = 0.637–0.726, P = NS). CCTA showed high sensitivity [92.5%, 95% confidence interval (CI) = 79.6–98.4%] and low specificity (52.6%, 95% CI = 39.0–66.0%). Contrarily, TAG showed insufficient sensitivity (47.5%, 95% CI = 31.5–63.9%) but rather high specificity (91.2%, 95% CI = 90.7–97.1%). The sensitivity and specificity of CCO was moderate (65.0%, 95% CI = 48.3–79.4%; 61.4%, 95% CI = 47.6–74.0%, respectively).

View this table:
Table 3

Diagnostic performance of CCTA diameter stenosis, TAG, and CCO for functionally significant stenosis defined by FFR <0.8

Optimal cut-offSensitivitySpecificityPPVNPVAccuracy
CCTA DS ≥50%92.5 (79.6–98.4)52.6 (39.0–66.0)57.8 (44.8–70.1)90.9 (75.3–98.1)69.0 (55.7–79.3)
TAG ≤−0.65447.5 (31.5–63.9)91.2 (90.7–97.1)79.2 (57.3–93.1)71.2 (59.4–81.2)73.2 (60.4–83.4)
CCO >0.06365.0 (48.3–79.4)61.4 (47.6–74.0)54.2 (39.2–68.6)71.4 (56.7–83.4)62.9 (47.9–76.2)
TAG + CT DS ≥50%90.0 (76.3–97.2)63.2 (49.3–75.6)63.2 (49.3–75.6)90.0 (76.1–97.3)74.2 (60.7–84.5)
CCO + CT DS ≥50%90.0 (76.3–97.2)63.2 (49.3–75.6)63.2 (49.3–75.6)90.0 (76.1–97.3)74.2 (60.4–84.5)
  • PPV, positive predictive value; NPV, negative predictive value. TAG was measured in HU/mm. The optimal cut-off of TAG + CT DS ≥50% and CCO + CT DS ≥50% were defined as logistic regression probability >0.416 and >0.441, respectively.

Figure 4

Diagnostic performance of CCTA diameter stenosis, TAG, and CCO for functionally significant stenosis defined by FFR < 0.8. TAG, transluminal attenuation gradient; CCO, corrected coronary opacification; FFR, fractional flow reserve; DS, diameter stenosis. Area under the ROC curve is shown as mean ± SE. See text for details.

Addition of TAG to the CCTA DS (DS ≥50%) significantly augmented the overall diagnostic performance of CCTA for ischaemia-causing stenosis (c-statistic = 0.726 ± 0.056 vs. 0.809 ± 0.044, P = 0.025), but addition of CCO to CCTA did not (0.726 ± 0.056 vs. 0.784 ± 0.048, P = 0.09). The increased diagnostic performance was reflected to the increased diagnostic specificity (52.6%, 95% CI = 39.0–66.0%, to 63.2%, 95% CI = 49.3–75.6%). However, the net proportion of patients reclassified after addition of TAG was statistically non-significant 1.0% (P = 0.41). The net reclassification improvement index after addition of CCO was significantly impaired −9.3% (P = 0.036).

Discussion

Our study is the first that validated TAG and CCO, newly introduced intraluminal attenuation-based methods, with invasively measured FFR.1215 Although both TAG and CCO were related to the presence of functionally significant plaque as well as the burden of plaque, the diagnostic performance was moderate. Additional analyses of TAG or CCO did not result in improved reclassification of CCTA stenosis. In our study, the discrimination performance of intraluminal attenuation-based methods for the lesion-specific ischaemia was limited.

As it is widely recognized, anatomic measure of stenosis is not a good predictor of functionally significant stenosis. Moreover, a reliable assessment of stenosis severity can be challenging in heavily calcified or complex coronary artery lesions. High sensitivity of CCTA has been validated in prospective multicentre studies, but specificity has had much room for improvement.19 Intraluminal attenuation-based methods were expected to show not only anatomical stenosis, but also functional data from CCTA. However, the benefit of additional analysis of intraluminal attenuation to the standard CCTA evaluation was not clear as shown in the lack of significant reclassification with the use of TAG and negative reclassification with the use of CCO. This result can be explained by the following. Unlikely TAG, CCO is not adjusted with the length of the coronary artery which might represent the amount of myocardium perfused.20 CCO is derived from calculation involving the intraluminal density of the thoracic aorta, in which the blood flow is pulsatile and faster than the coronary artery.21

Several limitations of our study should be described here. Our study is not free from selection bias because we enrolled patients underwent both CCTA and CAG in a retrospective manner. Our result, which was derived from the analysis of patients having clinical indication of coronary artery disease evaluation, would not be applied low-risk group or general population. TAG and CCO were measured by the gradient of intraluminal attenuation generated by contrast dye inflow that depends on the coronary arterial flow velocity. As the DS, the velocity of the coronary flow or the coronary flow reserve has also been known to be moderately correlated to FFR.22 As far as the functional significance measured by FFR is the criteria, any measurements not highly correlated to FFR will fail. Comparison of TAG or CCO with CCTA-based FFR or adenosine-stress perfusion CCTA would be warranted to evaluate the value of TAG or CCO.5,23 TAG or CCO was measured from resting state, whereas FFR was measured with adenosine-induced hyperaemia. Therefore, TAG or CCO may be not analogous to FFR physiologically. TAG and CCO can be affected by non-flat nature of the intraluminal attenuation time-density curve and the individual intensity of intraluminal attenuation.13 Numerous patient-, contrast agent-, and CT scanning-related factors related to the contrast enhancement of CCTA might affect our results.24 These factors may be too complex to be adjusted. The lack of temporal uniformity caused by the use of 64-detector row CT scanners might limit the value of TAG or CCO further. Single-beat imaging with the use of scanners capable of entire coronary tree coverage may improve the diagnostic performance of TAG or CCO.12

In conclusion, TAG and CCO, which are newly introduced intracoronary attenuation-based CCTA analyses, showed moderate correlation with physiological coronary artery stenosis and limited value when added to the assessment of anatomical stenosis.

Conflict of interest: none declared.

References

View Abstract