Between 06/2018 and 02/2019, 100 consecutive patients with severe aortic stenosis undergoing pre-TAVI CT at the Swiss Cardiovascular Center, Bern University Hospital, who fulfill the inclusion and exclusion criteria were analyzed for the present study. In detail, of the 211 patients with severe aortic stenosis who underwent a pre-TAVI CT, 45 patients with prior percutaneous coronary interventions with stent implantation, 15 patients with coronary artery bypass grafting, 7 patients with valve- in valve procedure, 5 patients with pacemakers, and 39 patients with external CTAs were excluded, resulting in 100 patients for the purpose of the present analysis.
Baseline clinical characteristics
The baseline clinical characteristics of the cohort are shown in Table 1. The average age of patients was 82.3 ± 6.5 years, 70% were female and the mean Euro SCORE II Value of 4.9 ± 6.1. 30 patients had CAD and thereof 16 patients with 1-vessel-disease, 13 with 2-vessel-disease and 1 patient with 3-vessel-disease. TAVI patients with CAD had significantly higher CACS as compared with those with no CAD (1019 [542–1907] versus 386 [75–765]; p < 0.001).
Table 1
|
|
Invasive coronary angiography (QCA)
|
|
|
Total/Mean ± SD
|
No obstructive
CAD
|
Obstructive
CAD
|
p value
|
Number of patients
|
100
|
70
|
30
|
|
Female gender
|
70
|
50 (71.4 %)
|
20 (66.7 %)
|
0.812
|
BMI (kg/m2)
|
25.5 ± 5.6
|
24.9 ± 5.4
|
25.1 ± 3.7
|
0.116
|
Age (years)
|
82.3 ± 6.5
|
81.6 ± 6.7
|
83.8 ± 5.8
|
0.254
|
Arterial hypertension
|
84
|
56 (80.0 %)
|
28 (93.3 %)
|
0.138
|
Diabetes mellitus
|
20
|
15 (21.4 %)
|
5 (16.7 %)
|
0.786
|
Peripheral artery disease
|
5
|
5 (7.2 %)
|
0 (0 %)
|
0.318
|
Stroke/TIA
|
15
|
7 (9.1 %)
|
8 (24.2 %)
|
1.000
|
Pacemaker
|
3
|
2 (2.9 %)
|
1 (3.3 %)
|
1.000
|
Atrial fibrillation
|
14
|
9 (12.9 %)
|
5 (16.7 %)
|
0.754
|
Hemoglobin (g/l)
|
114.7 ± 17.8
|
116.9 ± 17.8
|
109.6 ± 17.0
|
0.120
|
COPD
|
6
|
6 (8.6 %)
|
0 (0 %)
|
0.174
|
GFR (Cockcroft Gault; ml/min/1.73 m2)
|
52.0 ± 21.8
|
53.9 ± 22.6
|
47.5 ± 19.3
|
0.218
|
STS score
|
4.6 ± 2.9
|
4.2 ± 2.6
|
5.2 ± 3.0
|
0.080
|
Aortic valve area (cm2)
|
0.60 ± 0.23
|
0.61 ± 0.22
|
0.58 ± 0.27
|
0.420
|
Mean aortic gradient (mmHg)
|
41 ± 16
|
40 ± 17
|
43 ± 14
|
0.336
|
LVEF (%)
|
57 ± 13
|
56 ± 13
|
58 ± 12
|
0.319
|
Reduced LVEF < 45%
|
17
|
14 (20.0 %)
|
3 (10.0 %)
|
0.262
|
Total CACS (Median IQR [25–75])
|
564
[160–1045]
|
386
[75–765]
|
1019
[542–1907]
|
0.001
|
BMI = Body mass index; CAD = Coronary artery disease; COPD = chronic obstructive pulmonary disease; CACS = Coronary artery calcium score; GFR = Glomerular filtration rate; LVEF = Left ventricular ejection fraction; IQR = Interquartile range; QCA = Quantitative coronary angiography; |
TIA = transient ischemic attack; |
CTA and image quality
In 16 (16%) out of 100 patients, adequate visualization of the entire coronary tree (all 16 segments) was possible with CTA, while 84 (84%) of 100 patients had at least one segment which was not evaluable for CAD analysis due to impaired image quality (e.g. motion artefacts). Out of the 84 patients with non-evaluable segments, a total of 578 segments (578/1533 = 38%) showed a CTA image quality that was insufficient for quantitative CAD evaluation. When focusing only on the left main and proximal segments, 54 out of 100 patients could be assessed, whereas 46% of patients showed at least one segment (i.e. left main or proximal segment) which depicted impaired image quality. On a segment analysis, 301 (75%) out of 400 left main and proximal segments showed good image quality.
Per patient analysis entire coronary artery tree: The diagnostic performance of CTA and combined CTA/CACS
The findings of the performance of CTA in diagnosing CAD using 50% diameter stenosis in invasive QCA as a reference are summarized in Table 2 (and for 70% diameter stenosis in supplemental table 1). On a per-patient analysis, when non-evaluable segments were rated as possible obstructive CAD, CTA showed a sensitivity of 100% (95% CI: 88.4–100), specificity of 11.4% (95% CI: 5.1–21.3), positive predictive value of 32.6% (95% CI: 30.8–34.5), negative predictive value of 100% and diagnostic accuracy of 38% (95% CI: 28.5–48.3) for obstructive CAD. On a per-patient analysis excluding patients with non-evaluable segments, (n = 16 included, n = 84 excluded), CTA had a sensitivity of 100% (95% CI: 29.2–100), specificity of 61.5% (95% CI: 31.6–86.1), positive predictive value 37.5% (95% CI: 23.2–54.4), negative predictive value of 100% and diagnostic accuracy of 68.8% (95% CI: 41.3–89). On a per-patient analysis including those with non-evaluable segments (n = 100; note that non-evaluable segments are counted as being positive for stenosis) with incorporation of CACS (cut-off ≥ 400), combined CTA/CACS had a sensitivity of 83.3% (95% CI: 65.3–94.4), specificity of 50% (95% CI: 37.8–62.2), positive predictive value of 41.7% (95% CI: 35.0-48.7), negative predictive value of 87.5% (95% CI: 75.3–94.2) and diagnostic accuracy of 60% (95% CI: 49.7–69.7), see Figs. 1 and 2. Thirty-five percent of the patients were true negative and 5% were false negative. When applying a combined CTA/CACS approach with a CACS cut-off of 100 for patients with non-evaluable segments, the sensitivity and negative predictive value increased to 100% and obstructive CAD could be ruled out in 20% of the patients versus 8%, using CTA alone.
Table 2
CTA and combined CTA/CACS versus QCA as the reference standard (≥ 50% stenosis)
|
True positive
|
False positive
|
False negative
|
True negative
|
Sensitivity
(%)
|
Specificity
(%)
|
PPV
(%)
|
NPV
(%)
|
Positive LLR
|
Negative LLR
|
Accuracy
|
Patient based analysis (total n = 100 patients)
|
|
|
|
|
|
|
|
|
|
|
|
Including patients with non-evaluable segments (n = 100 included)
|
30
|
62
|
0
|
8
|
100
|
11.4
|
32.6
|
100
|
1.13
|
0
|
38
|
(95% CI)
|
|
|
|
|
88.4–100
|
5.1–21.3
|
30.8–34.5
|
nc
|
1.04–1.23
|
nc
|
28.5–48.3
|
Excluding patients with non-evaluable segments (n = 16 included, n = 84 excluded)
|
3
|
5
|
0
|
8
|
100
|
61.5
|
37.5
|
100
|
2.60
|
0
|
68.8
|
(95% CI)
|
|
|
|
|
29.2–100
|
31.6–86.1
|
23.2–54.4
|
nc
|
1.31–5.17
|
nc
|
41.3–89.0
|
Combined CTA/CACS (n = 100 included)
(If non-evaluable segments present CACS cut-off ≥ 100 = CAD)
|
30
|
50
|
0
|
20
|
100
|
28.6
|
37.5
|
100
|
1.4
|
0
|
50
|
(95% CI)
|
|
|
|
|
88.4–100
|
18.4–40.6
|
34.1–41.0
|
nc
|
1.21–1.62
|
nc
|
39.8–60.2
|
Combined CTA/CACS (n = 100 included)
(If non-evaluable segments present CACS cut-off ≥ 400 = CAD)
|
25
|
35
|
5
|
35
|
83.3
|
50.0
|
41.7
|
87.5
|
1.67
|
0.33
|
60
|
(95% CI)
|
|
|
|
|
65.3–94.4
|
37.8–62.2
|
35.0-48.7
|
75.3–94.2
|
1.25–2.21
|
0.14–0.77
|
49.7–69.7
|
Segment based analysis (all coronary segments, n = 1533 segments)
|
|
|
|
|
Including non-evaluable segments (n = 1533 included)
|
41
|
668
|
15
|
809
|
73.2
|
54.8
|
5.8
|
98.2
|
1.62
|
0.49
|
55.5
|
(95% CI)
|
|
|
|
|
59.7–84.2
|
52.2–57.3
|
4.9–6.8
|
97.2–98.8
|
1.37–1.92
|
0.32–0.76
|
52.9–58.0
|
Excluding non-evaluable segments (n = 955 included, n = 578 excluded)
|
19
|
112
|
15
|
809
|
55.9
|
87.8
|
14.5
|
98.2
|
4.60
|
0.50
|
86.7
|
(95% CI)
|
|
|
|
|
37.9–72.8
|
85.6–89.9
|
10.7–19.3
|
97.4–98.8
|
3.25–6.49
|
0.34–0.73
|
84.4–88.8
|
Proximal segment based analysis (left main stem + proximal segment of LAD, LCX, RCA)
|
Including non-evaluable segments (n = 400 included)
|
15
|
140
|
4
|
241
|
78.9
|
63.3
|
9.7
|
98.4
|
2.15
|
0.33
|
64.0
|
(95% CI)
|
|
|
|
|
54.4–94.0
|
58.2–68.1
|
7.6–12.3
|
96.2–99.3
|
1.65–2.81
|
0.14–0.80
|
59.1–68.7
|
Excluding non-evaluable segments (n = 301 included, n = 99 excluded)
|
8
|
48
|
4
|
241
|
66.7
|
83.4
|
14.3
|
98.4
|
4.01
|
0.40
|
82.7
|
(95% CI)
|
|
|
|
|
34.9–90.1
|
78.6–87.5
|
9.4–21.2
|
96.4–99.3
|
2.49–6.46
|
0.18–0.89
|
78.0-86.8
|
Left coronary artery proximal segment based analysis (left main stem + proximal segment of the LAD)
|
|
|
|
Including all non-evaluable segments (n = 200 included)
|
3
|
56
|
1
|
140
|
75.0
|
71.4
|
5.1
|
99.3
|
2.62
|
0.35
|
71.5
|
(95% CI)
|
|
|
|
|
19.4–99.4
|
64.6–77.6
|
2.8-9.0
|
96.2–99.9
|
1.43–4.82
|
0.06–1.92
|
64.7–77.6
|
Excluding non-evaluable segments (n = 165 included, n = 35 excluded)
|
3
|
21
|
1
|
140
|
75.0
|
87.0
|
12.5
|
99.3
|
5.75
|
0.29
|
86.7
|
(95% CI)
|
|
|
|
|
19.4–99.4
|
80.8–91.7
|
6.7–22.2
|
99.29–99.9
|
2.88–11.49
|
0.05–1.57
|
80.5–91.5
|
LAD vessel based analysis
|
|
|
|
|
|
|
|
|
Including all non- evaluable segments (n = 500 included)
|
16
|
207
|
8
|
269
|
66.7
|
56.5
|
7.2
|
97.1
|
1.53
|
0.59
|
57
|
(95% CI)
|
|
|
|
|
44.7–84.4
|
51.9–61.0
|
5.4–9.5
|
95.0-98.4
|
1.13–2.07
|
0.33–1.04
|
52.5–61.4
|
Excluding non-evaluable segments (n = 338 included, n = 162 excluded)
|
8
|
53
|
8
|
269
|
50.0
|
83.5
|
13.1
|
97.1
|
3.04
|
0.60
|
82.0
|
(95% CI)
|
|
|
|
|
24.7–75.4
|
79.0-87.4
|
8.0-20.7
|
95.4–98.2
|
1.76–5.26
|
0.37–0.98
|
77.4–85.9
|
LCX vessel based analysis
|
|
|
|
|
|
|
|
|
Including non-evaluable segments (n = 500 included)
|
15
|
218
|
4
|
263
|
78.9
|
54.7
|
6.4
|
98.5
|
1.74
|
0.39
|
55.6
|
(95% CI)
|
|
|
|
|
54.4–94.0
|
50.1–59.2
|
5.1–8.1
|
96.5–99.4
|
1.35–2.24
|
0.16–0.92
|
51.1–60.0
|
Excluding non-evaluable segments (n = 299 included, n = 201 excluded)
|
6
|
26
|
4
|
263
|
60.0
|
91.0
|
18.8
|
98.5
|
6.67
|
0.44
|
90.0
|
(95% CI)
|
|
|
|
|
26.2–87.8
|
87.1–94.0
|
11.0-30.1
|
96.9–99.3
|
3.57–12.46
|
0.21–0.94
|
86.0-93.1
|
RCA vessel based analysis
|
|
|
|
|
|
|
|
|
Including all non-evaluable segments (n = 400 included)
|
10
|
214
|
3
|
173
|
76.9
|
44.7
|
4.5
|
98.3
|
1.39
|
0.52
|
45.8
|
(95% CI)
|
|
|
|
|
46.2–95.0
|
39.7–49.8
|
3.3-6.0
|
95.5–99.4
|
1.02–1.90
|
0.19–1.40
|
40.8–50.8
|
Excluding non-evaluable segments (n = 209 included, n = 191 excluded)
|
5
|
28
|
3
|
173
|
62.5
|
86.1
|
15.2
|
98.3
|
4.49
|
0.44
|
85.2
|
(95% CI)
|
|
|
|
|
24.5–91.5
|
80.5–90.5
|
8.6–25.3
|
95.9–99.3
|
2.37–8.49
|
0.18–1.07
|
79.6–89.7
|
CACS = Coronary artery calcium score; CAD = Coronary artery disease; CI = Confidence interval; CTA = Computed tomography angiography; LAD = Left anterior descending artery; LCX = Left circumflex artery; LLR = Likelihood ratio; nc = not computable/calculable; NPV = Negative predictive value; PPV = Positive predictive value; QCA = Quantitative coronary angiography; RCA = Right coronary artery; |
Per patient analysis assessing only left main and proximal segments of LAD, RCA and RCX
When assessing only the left main and proximal segments, 15% showed at least a 50% stenosis in invasive QCA. On a per-patient analysis, when non-evaluable segments were rated as possible obstructive CAD, CTA showed a sensitivity of 80% (95% CI: 51.9–95.7), specificity of 29.4% (95% CI: 20.0-40.03), positive predictive value of 16.7% (95% CI: 13.0-21.1), negative predictive value of 89.3% and diagnostic accuracy of 37% (95% CI: 27.6–47.2) for obstructive CAD. On a per-patient analysis excluding patients with non-evaluable segments, (n = 54 included, n = 46 excluded), CTA had a sensitivity of 57.1% (95% CI: 18.4–90.1), specificity of 53.2% (95% CI: 38.1–68.9), positive predictive value 15.4% (95% CI: 8.2–27.0), negative predictive value of 89.3% and diagnostic accuracy of 53.7% (95% CI: 39.6–67.4).
Per segment analysis: The diagnostic performance of CTA
The findings on a per segment analysis for 50% diameter stenosis are depicted in Table 2 (and for 70% stenosis in supplemental table 1). The per segment analysis, where non-evaluable segments were rated as obstructive CAD, the sensitivity was 73.2% (95%CI: 59.7–84.2), specificity 54.8% (95%CI: 52.2–57.3), positive predictive value 5.8% (95%CI: 4.9–6.8), negative predictive value 98.2% (95%CI: 97.2–98.8) and accuracy of 55.5% (95%CI: 52.9–58.0).
Diagnostic performance of CACS versus QCA
The sensitivities and specificities of different CACS thresholds to diagnose a ≥ 50% diameter stenosis on invasive coronary angiography are shown in Fig. 4. In our cohort, no patient with CACS of less than or equal to 114 was found to have obstructive CAD (≥ 50% stenosis on invasive QCA). Similarly, no patient with a CACS of less than or equal to 187 had ≥ 70% stenosis on invasive coronary angiography. Of the patients who had CACSs ≥ 400, the majority (25/30 or 83%) had at least one ≥ 50% diameter stenosis on QCA while 9/11 (82%) had at least one ≥ 70% diameter stenosis. Figure 3 summarizes the ROC of CACS with different cut-offs in detecting ≥ 50% diameter stenosis by using QCA as the reference standard in patients being assessed for TAVI. An area under the curve of 0.75 was observed with a standard deviation of 0.051 (95% CI 0.651–0.850; p < 0.001).