We sought to compare the diagnostic performance of coronary computed tomography angiography (CCTA), computed tomography perfusion (CTP), and computed tomography (CT)-fractional flow reserve (FFR) for assessing the functional significance of coronary stenosis as defined by invasive FFR in patients with known or suspected coronary artery disease (CAD). CCTA has proved clinically useful for excluding obstructive CAD because of its high sensitivity and negative predictive value (NPV); however, the ability of CTA to identify functionally significant CAD has remained challenging. We searched PubMed/Medline for studies evaluating CCTA, CTP, or CT-FFR for the noninvasive detection of obstructive CAD compared with catheter-derived FFR as the reference standard. Pooled sensitivity, specificity, PPV, NPV, likelihood ratios, and odds ratio of all diagnostic tests were assessed. Eighteen studies involving a total of 1,535 patients were included. CTA demonstrated a pooled sensitivity of 0.92, specificity 0.43, PPV of 0.56, and NPV of 0.87 on a per-patient level. CT-FFR and CTP increased the specificity to 0.72 and 0.77, respectively (p = 0.004 and p = 0.0009) resulting in higher point estimates for PPV 0.70 and 0.83, respectively. There was no improvement in the sensitivity. The CTP protocol involved more radiation (3.5 mSv CCTA vs 9.6 mSv CTP) and a higher volume of iodinated contrast (145 ml). In conclusion, CTP and CT-FFR improve the specificity of CCTA for detecting functionally significant stenosis as defined by invasive FFR on a per-patient level; both techniques could advance the ability to noninvasively detect the functional significance of coronary lesions.
Coronary artery disease (CAD) is responsible for 17% of all death worldwide. Given that nearly 40% of patients without known CAD who undergo coronary angiography have nonobstructive disease, improved techniques for noninvasive assessment of CAD are of considerable clinical importance. Coronary computed tomography angiography (CCTA) has demonstrated high sensitivity and negative predictive value (NPV) for excluding significant CAD. However, given the known discordance between anatomic severity and functional significance of a lesion, CCTA is only modestly predictive of an abnormal invasive fractional flow reserve (FFR) that has become the clinical reference standard for defining significant lesions as the DEFER and FAME (Fractional Flow Reserve vs Angiography for Multivessel Evaluation) studies demonstrated that the strategy of revascularization based on FFR is associated with a low risk of adverse cardiovascular outcomes. Computed tomography perfusion (CTP) and CT-FFR are novel CT imaging techniques that can help determine the physiological significance of a coronary lesion detected by CCTA and could, thus, avoid unnecessary referrals to the catheterization laboratory for nonsignificant stenoses. To date, most of the studies examining stress CTP imaging have been small and single center. CT-FFR has been evaluated in a limited number of multicenter trials but has not been widely available clinically. Previous CCTA and CT-FFR meta-analyses have been published ; however, a systematic comparison among CTA, CTP, and CT-FFR to assess the diagnostic performance of a functional assessment versus an anatomic assessment by CT has not. We, thus, performed a meta-analysis of the diagnostic performance of CCTA, CTP, and CT-FFR to assess for functional ischemia of coronary lesions compared with catheter-based FFR as the gold standard.
Methods
The meta-analysis was performed using standard guidelines from the Meta-analysis of Observational Studies in Epidemiology and the Preferred Reporting Items for Systematic Reviews and Meta-analyses documents. We conducted a systematic search using MEDLINE (search last updated April 2015) for studies published in English using CCTA, CTP, and CT-FFR as diagnostic techniques. Key words used were “computed tomography” AND “fractional flow reserve” OR “FFR” OR “Perfusion.” The search was limited to studies published in peer-reviewed journals. Abstracts from meetings were excluded because of limited information regarding data. The retrieved studies were examined for potentially overlapping data. The references of these reports were evaluated and also key publications, related articles, and citations. Three investigators (JAG, MJL, and MS) independently scanned all abstracts and performed data extraction. General consensus was achieved after reviewing full-text articles. We included a study if (1) it used CTA, CTP, or CT-FFR for noninvasive evaluation of CAD and (2) it compared the noninvasive results with catheter-derived FFR. Data regarding the independent performance of CTA, CTP, and CT-FFR were used for the analysis.
The quality of included studies was assessed by 3 investigators (JAG, MS, and PS) using the Quality Assessment of Diagnostic Accuracy Studies (QUADAS) instrument. It consists of a list of 14 questions with closed-ended questions (yes, no, or unclear). The items included in this instrument covered patient spectrum, reference standard, disease progression bias, verification bias, review bias, clinical review bias, incorporation bias, test execution, study withdrawal, and indeterminate results. Publication bias was assessed using the Peter’s and Egger’s methods.
Categorical data are presented as percentages and continuous variables as mean values. The analysis of diagnostic performance was carried out both at the per-patient and per-vessel levels. Sensitivity, specificity, PPV, and NPV and their 95% confidence intervals were calculated using an exact method for binomial proportions using the F-distribution method. Pooled estimates were determined by weighting the studies by the inverse of their sample size. Likelihood ratios and diagnostic odds ratios were pooled using a random-effects model using the DerSimonanian-Laird method. Symmetric receiver-operating curves were created. Statistical analysis was performed using MetaDiSc, version 1.4 freeware package (Universidad Complutense, Madrid, Spain) with statistical significance for hypothesis testing for a 2-tailed test set at the 0.05 2-tailed level. We assessed heterogeneity between studies visually from Forest plots of the individual parameters and using the Cochran Q index and the inconsistency index (I 2 ). Bivariate comparison of sensitivity and specificity between the diagnostic techniques (CCTA, CTP, and CT-FFR) was performed as described by Reitsmaa et al and Van Houwelingen et al using SAS/STAT software, version 9.4, of the SAS System for Windows (SAS Institute Inc., Cary, North Carolina).
Results
Our literature search identified 1,292 relevant abstracts of full-text articles; of these, 43 unique reports were extracted for review. Twenty-four studies were excluded for various reasons, including overlapping data with other reports, lack of FFR catheter-derived data, and insufficient data to calculate sensitivity and specificity. Figure 1 shows the details of our literature search. A total of 18 studies were included in the study for analysis ( Table 1 ). The 18 included studies had a total of 1,535 patients. The mean age was 62 years, 68% of subjects were men, 68% had hypertension, 21% had diabetes, 25% were smokers, 33% had a family history of CAD, and the mean body mass index (BMI) was 27 kg/m 2 ( Table 2 ). All studies used scanners with a minimum of 64 detectors, tube voltage between 100 and 120 kVp depending on the patient’s BMI, and tube current between 200 and 500 mA. Protocols used a variety of techniques including single acquisition and retrospective or prospective triggering. Perfusion studies typically used a 3- to 5-minute infusion of adenosine at a dose of 140 μg/kg/min for the vasodilator protocol. Protocols typically included stress and rest CCTA images using retrospective triggering. In 1 study, delayed imaging for scar was performed, but the information from the delayed imaging was not used in our meta-analysis or for estimation of radiation dose.
| First Author | Year Published | Patients (n) | Study Design | Population | Modality | FFR Cut-off | FFR Procedural Criteria | Criteria for positive CTA |
|---|---|---|---|---|---|---|---|---|
| Bettencourt | 2013 | 105 | Prospective | Suspected CAD | CCTA, CTP | 0.80 | 50-90% | >50% stenosis |
| Choo | 2013 | 37 | Prospective | Suspected CAD | CCTA, CTP | 0.75 | 50-85% | >50% stenosis |
| Greif | 2013 | 65 | Prospective | CP with known CAD or suspected CAD | CCTA, CTP | 0.80 | 50-85% | >50% stenosis |
| Kim | 2013 | 44 | Prospective | Suspected or known CAD with + CAD on CCTA | CCTA, CTFFR | 0.80 | >30% | >50% stenosis |
| Ko (A) | 2012 | 42 | Prospective | Known CAD by CA scheduled for revascularization | CCTA, CTP | 0.80 | >50% | >50% stenosis |
| Ko (B) | 2012 | 40 | Prospective | Suspected CAD (High Risk Patients) | CCTA, CTP | 0.80 | >30% | >50% stenosis |
| Koo (DISCOVER-FLOW Study) | 2011 | 103 | Prospective | Suspected or known CAD | CCTA, CTFFR | 0.80 | Not Specified | >50% stenosis |
| Kristensen | 2009 | 42 | Prospective | Intermediate lesions on CCTA | CCTA | 0.75 | Not Specified | >50% stenosis |
| Meijboom | 2008 | 79 | Retrospective | Suspected CAD | CCTA | 0.75 | Not Specified | >50% stenosis |
| Min (DeFacto Study) | 2012 | 252 | Prospective | Suspected or known CAD | CCTA, CTFFR | 0.80 | 30-90% | >50% stenosis |
| Norgaard (NXT Trial) | 2014 | 254 | Prospective | Suspected CAD | CCTA, CTFFR | 0.80 | Not Specified | >50% stenosis |
| Opolski | 2013 | 61 | Prospective | Intermediate lesions on CCTA | CCTA | 0.80 | Not Specified | >50% stenosis |
| Renker | 2014 | 53 | Retrospective | Suspected or known CAD | CCTA, CTFFR | 0.80 | >30% | >50% stenosis |
| Rossi | 2014 | 80 | Prospective | Suspected CAD | CCTA,CTP | 0.75 | 30-90% | >50% stenosis |
| Stuijfzand | 2014 | 85 | Prospective | Suspected CAD | CCTA | 0.80 | >30% | >50% stenosis |
| Van Werkhoven | 2009 | 33 | Prospective | Suspected or known CAD | CCTA | 0.75 | >50% | >50% stenosis |
| Voros (ATLANTA Study) | 2014 | 85 | Prospective | Known CAD by CA or CCTA | CCTA | 0.75 | 40-90% | >50% stenosis |
| Wong | 2014 | 75 | Retrospective | Suspected or known CAD | CCTA, CTP | 0.80 | >30% | >50% stenosis |
| First Author | Age (yrs) | Age (SD) | Male (%) | HTN (%) | Smoking Hx (%) | HLD (%) | Diabetes (%) | Prior MI (%) | Fam Hx CAD (%) | BMI (kg/m 2 ) | BMI SD | Known CAD (%) |
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bettencourt | 62 | 8 | 67 | 71 | 32 | 79 | 38 | 0 | 20 | 27.9 | 4.43 | 0 |
| Choo | 61.7 | 20.5 | 75.7 | 56.7 | 37.8 | 18.9 | 24.3 | 0 | n/a | n/a | n/a | n/a |
| Greif | 70.4 | 9 | 42 | 67.3 | 25.4 | 47.6 | 17.9 | 0 | 33.4 | n/a | n/a | 74.3 |
| Kim | 65 | 9.1 | 80 | 81 | n/a | 63 | 29 | 10 | n/a | 24.4 | 2.6 | n/a |
| Ko 2012 (A) | 65.1 | 8.3 | 64.3 | 88.1 | 16.7 | 69 | 21.4 | 11.9 | 40.5 | 27.9 | 6.5 | n/a |
| Ko 2012 (B) | 62.1 | 9.9 | 67.5 | 75 | 15 | 80 | 12.5 | 0 | 27.5 | 28.2 | 4.9 | n/a |
| Koo | 62.7 | 8.5 | 72 | 65 | 36 | 65 | 26 | 17 | n/a | 25.8 | 3.5 | 32 |
| Kristensen | 61 | 10 | 76 | n/a | n/a | n/a | n/a | 19 | n/a | 29 | 4 | n/a |
| Meijboom | 60 | 9 | 81 | n/a | n/a | n/a | n/a | 12.7 | n/a | 26.6 | 3.9 | n/a |
| Min | 62.9 | 8.7 | 70.6 | 71.2 | 17.5 | 79.8 | 21.2 | 6 | 19.9 | n/a | n/a | 12.3 |
| Norgaard | 64 | 10 | 64 | 69 | 18 | 79 | 23 | 2 | n/a | 26 | 3 | n/a |
| Opolski | 63 | 9 | 64 | 79 | 25 | 95 | 10 | 15 | n/a | 28 | 4 | 100 |
| Renker | 61.2 | 12 | 64 | 54 | 14 | 54 | 32 | n/a | n/a | 28.9 | 6.5 | 16 |
| Rossi | 60 | 10 | 79 | 60 | 33 | 66 | 20 | 0 | 44 | 27 | 4 | n/a |
| Stuijfzand | 57.3 | 9.7 | 60 | 37 | 45 | 38 | 16 | 0 | 46 | 27.1 | 4.1 | n/a |
| Van Werkhoven | 57 | 11 | n/a | 42 | 21 | 36 | 9 | n/a | 36 | n/a | n/a | 91 |
| Voros | 61.3 | 7.8 | 62 | 78 | 20 | 91 | 21 | n/a | n/a | n/a | n/a | 100 |
| Wong | 64 | 10.8 | 69.3 | 83 | 16 | 73 | 19 | 7 | 33 | n/a | n/a | 51 |
The per-patient and per-vessel analysis results are included in Figures 2 and 3 and Tables 3 and 4 . The bivariate analysis for comparing the sensitivity and specificity across the included studies did not show a significant difference in a per-vessel analysis for either sensitivity or specificity among CCTA, CTP or CT-FFR. However, in analysis by patient, there was a significantly higher specificity of both CTP (p = 0.004) and CT-FFR (p = 0.0009) compared with CCTA. The specificity of CTP and CT-FFR was not different. There was no difference in sensitivity among the 3 different techniques.
| Technique | # Studies | # Patients | Sensitivity | Specificity | PPV | NPV | Positive LR | Negative LR | Diagnostic OR |
|---|---|---|---|---|---|---|---|---|---|
| CCTA | 9 | 1039 | 0.92 [0.88-0.98] | 0.43 [0.38-0.47] | 0.57 [0.51-0.64] | 0.87 [0.78-0.94] | 1.64 [1.38-1.93] | 0.19 [0.10-0.35] | 9.17 [4.54-18.52] |
| CTP | 3 | 187 | 0.94 [0.88-0.98] | 0.77 [0.66-0.85] | 0.83 [0.75-0.92] | 0.92 [0.88-0.95] | 3.85 [2.16-6.84] | 0.09 [0.04-0.19] | 63.42 [22.41-179.5] |
| CT-FFR | 4 | 662 | 0.90 [0.85-0.93] | 0.72 [0.67-0.76] | 0.70 [0.58-0.82] | 0.90 [0.84-0.95] | 3.70 [2.11-6.49] | 0.16 [0.11-0.23] | 24.34 [1.84-54.65] |
| Technique | # Studies | # Patients | Sensitivity | Specificity | PPV | NPV | Positive LR | Negative LR | Diagnostic OR |
|---|---|---|---|---|---|---|---|---|---|
| CCTA | 16 | 1239 | 0.89 [0.86-0.91] | 0.65 [0.62-0.67] | 0.48 [0.38-0.58] | 0.94 [0.82-0.94] | 2.66 [2.13-3.31] | 0.17 [0.11-0.26] | 19.78 [11.98-32.66] |
| CTP | 5 | 264 | 0.83 [0.77-0.88] | 0.76 [0.72-0.80] | 0.61 [0.46-0.75] | 0.91 [0.84-0.99] | 3.68 [2.60-5.21] | 0.22 [0.12-0.39] | 20.10 [7.89-51.2] |
| CT-FFR | 5 | 714 | 0.83 [0.79-0.87] | 0.77 [0.74-0.80] | 0.63 [0.52-0.72] | 0.91 [0.79-1.03] | 3.76 [2.17-6.54] | 0.23 [0.16-0.35] | 18.21 [7.45-44.52] |
To assess the impact of which invasive FFR cut-point was used to define a physiologically significant obstructive coronary lesion on per-vessel diagnostic CCTA test performance, we abstracted data from studies using both FFR cut-point of 0.75 and 0.80. Per-vessel CCTA test sensitivity was similar when using the 0.75 or 0.80 FFR cut-point (0.850 [0.802 to 0.890] vs 0.845 [0.800 to 0.884], respectively). Furthermore, per-vessel CCTA specificity was also similar when using the 0.75 or the 0.80 FFR cut-point (0.591 [0.557 to 0.624] vs 0.602 [0.568 to 0.636], respectively).
Six studies using CTP included radiation dosages in millisieverts for both the CTA and CTP components of the examination ( Table 5 ). Data were available for a total of 407 patients. The effective radiation dose was calculated by multiplying the dose-length product by the same constant (k = 0.014 mSv/mGy/cm) in all studies. The CCTA and CTP protocols delivered a pooled average effective radiation dose of 3.5 mSv and 6.1 mSv, respectively, and 9.6 mSv for the total study protocol. The amount in milliliters (ml) of iodinated contrast material is listed in Table 5 . The average use of contrast volume among the 6 studies that used a combined protocol of CCTA and CTP was 145 ml.
| Author | Patients (n) | CCTA Radiation Dose (mSv) | CTP Radiation Dose (mSv) | CCTA + CTP Combined Radiation Dose (mSv) | Contrast Used (mL) |
|---|---|---|---|---|---|
| Bettencourt | 105 | 1.5 | 3.3 | 4.8 | 160 |
| Greif | 65 | 2.9 | 9.7 | 12.6 | 130 |
| Ko 2012 (A) | 42 | 4.8 | 5.3 | 10.1 | 178 |
| Ko 2012 (B) | 40 | 4.7 | 4.5 | 9.2 | 178 |
| Rossi | 80 | 4.2 | 9.4 | 13.6 | 115-135 |
| Wong | 75 | 4.6 | 4.8 | 9.4 | 122 |
| Weighted Avg | 407 | 3.5 | 6.1 | 9.6 | 145 |
The selected studies showed overall high-quality scores in all the 14 items of the QUADAS questionnaire as listed in Table 6 . There is no indication of publication bias when using the Egger’s test for any of the diagnostic techniques (p >0.05 for all analyses). Likewise, the Peter’s test did not suggest presence of publication bias (p >0.05 for all analyses).
| Article | 1 | 2 | 3 | 4 | 5 | 6 | 7 | 8 | 9 | 10 | 11 | 12 | 13 | 14 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Bettencort 2013 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | No | YES | YES |
| Choo 2013 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| Greif 2013 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | YES | YES |
| Kim 2013 | YES | NO | YES | YES | YES | YES | YES | YES | NO | YES | Unclear | NO | NO | NO |
| Ko EHJ 2012 (A) | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | Unclear | YES |
| Ko JACC 2012 (B) | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | NO | YES |
| Koo 2011 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | NO | Unclear |
| Kristensen 2010 | YES | NO | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | NO | NO |
| Meijboom 2008 | YES | YES | YES | Unclear | YES | YES | YES | YES | YES | YES | YES | Unclear | NO | NO |
| Min 2012 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear | NO | NO |
| Norgaard 2014 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| Opolski 2014 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear |
| Renker 2014 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| Rossi 2014 | YES | YES | YES | Unclear | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| Stuijfzand 2014 | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
| Van Werkhoven 2009 | YES | NO | YES | Unclear | YES | YES | YES | YES | YES | YES | YES | YES | YES | Unclear |
| Voros 2014 | YES | NO | YES | Unclear | YES | YES | YES | YES | YES | Unclear | Unclear | Unclear | YES | NO |
| Wong 2014 | YES | YES | YES | Unclear | YES | YES | YES | YES | YES | YES | YES | YES | YES | YES |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
