Value of dynamic computed tomography myocardial perfusion in CAD: a systematic review and meta-analysis

Abstract

Aims

Dynamic stress computed tomography (CT) perfusion is a non-invasive method for quantifying myocardial ischaemia by assessing myocardial blood flow (MBF). In this meta-analysis, we evaluated the diagnostic accuracy of dynamic CT perfusion for the detection of significant coronary artery disease (CAD) across various CT scanners, obese patients, and its prognostic value.

Methods and results

We systematically searched PubMed, Embase, Web of Science, and Cochrane library for published studies evaluating the accuracy of CT myocardial perfusion in diagnosing functional significant ischaemia by invasive fractional flow reserve. The diagnostic performance of dynamic CT perfusion in detecting ischaemia was evaluated using a summary receiver operating characteristic (sROC) curve. A total of 23 studies underwent meta-analysis. In myocardial region without ischaemia, MBF was measured at 1.39 mL/min/g [95% confidence interval (CI) 1.25–1.54], while in region with ischaemia, it was 0.92 mL/min/g (95% CI 0.83–1.01) (P < 0.001). On the patient-based analysis, the area under the sROC curve of CT-MBF was 0.92, with a sensitivity of 0.82 and specificity of 0.86. Differences in CT type (dual source vs. single source), and body mass index did not significantly affect the diagnostic performance. The pooled hazard ratio of dynamic CT perfusion for predicting adverse events was 4.98 (95% CI 2.08–11.93, P ≤ 0.001, I² = 61%, P for heterogeneity = 0.07).

Conclusion

Dynamic CT perfusion has high diagnostic performance in the quantitative assessment of ischaemia and detection of functional myocardial ischaemia as defined by invasive FFR and may be useful in risk stratification of CAD patients.

Graphical abstract

Open in new tabDownload slide

Introduction

Coronary artery disease (CAD) stands as the third most prevalent cause of mortality globally and necessitates conquering.¹ Besides assessing the anatomical stenosis of the coronary arteries, functional evaluation, specifically myocardial ischaemia, assumes significance in formulating treatment strategies for CAD. At present, single-photon emission computed tomography (SPECT) represents the most widely employed non-invasive imaging modality for evaluating myocardial ischaemia.^2,3 However, the assessment of myocardial ischaemia via cardiac computed tomography (CT) has recently garnered considerable attention.^4,5 Dynamic perfusion CT utilizing contrast material allows for a quantitative evaluation of myocardial ischaemia in terms of myocardial blood flow (MBF).⁶ A primary advantage of dynamic perfusion CT lies in its capacity to simultaneously evaluate coronary artery anatomy, thereby enabling an accurate assessment of the culprit lesion. Moreover, dynamic perfusion CT has demonstrated its potential in predicting future cardiovascular events in cases involving severe myocardial ischaemia and is anticipated to serve as a valuable tool in risk stratification for CAD.⁷ A recent meta-analysis conducted a comprehensive evaluation of the ability of dynamic perfusion CT to identify significant coronary artery stenosis based on MBF and reported exceedingly high diagnostic performance.⁸ However, several important clinical questions remain unanswered in this literature, including the comparison of the diagnostic performance of different CT scanners, particularly single source CT (SSCT) and dual source CT (DSCT), the influence of patient body size, and the prognostic ability of CTP. To clarify these questions, we performed a comprehensive meta-analysis of dynamic perfusion CT.

Methods

Search strategy and selection criteria

We employed the methodology proposed by the Cochrane Collaboration and adhered to the reporting criteria outlined in the 2020 guideline Preferred Reporting Items for Systematic Review and Meta-analysis (PRISMA).⁹ Database searches were conducted on 2 June 2023, utilizing PubMed, Web of Science Core Collection, Cochrane advanced search, and the EMBASE electronic database. The following terms were used as keywords: dynamic myocardial perfusion CT, dynamic CT perfusion, CAD, myocardial infarction, and diagnostic accuracy etc. (see Supplementary data online, Material  S1). After screening all titles and abstracts within the search results, two reviewers (S.K. and Y.K.) thoroughly examined the potentially relevant studies for eligibility. Any disparities were resolved by a third reviewer. The study protocol was registered with the University Medical Information Network (registration number: UMIN000051223). Institutional review board approval was not required because the study was a meta-analysis and did not include clinical patient data. Both prospective and retrospective studies using invasive fractional flow reserve (FFR) as a reference criterion for ischaemia were included in this meta-analysis. Studies using significant stenosis on quantitative coronary angiography as a reference criterion were excluded. Literature including case reports, animal studies, and non-English language articles were excluded.

Outcome assessment

Two reviewers, Y.K. and S.K., conducted the extraction of the following study characteristics from the articles: author name, year of publication, country, patient disease, age, gender, BMI, type of CT scanner (SSCT vs. DSCT), CT-MBF values, and other relevant CT parameters. The ability to detect functionally significant coronary stenosis, defined as FFR <0.75 or <0.80, was evaluated in patient- and vessel-based analysis. A meta-analysis was performed using summary receiver operating characteristic (sROC) analysis for (i) CT angiography (CTA) alone, (ii) dynamic CTP alone, and (iii) CTA + CTP combination. Pooled sensitivity and specificity were also calculated. In the vessel-based analysis, sub-analysis was performed for differences in the type of CT device used (SSCT vs. DSCT) and BMI (<25 kg/m² vs. ≥25 kg/m²). The Quality Assessment of Diagnostic Accuracy Studies-2 (QUADAS-2) tool was employed to evaluate the risk of bias¹⁰ (see Supplementary data online, Material S2). In terms of prognostic evaluation, the hazard ratios predicting cardiovascular events during the follow-up period after CT imaging were examined.

Data integration and statistical analysis

Meta-analysis was performed utilizing RevMan 5.41 (Cochrane Collaboration, London, UK) and R Statistical Software (v3.5.1, Boston, MA, USA). A random-effects model was employed to estimate pooled CT-MBF values, and the inverse variance method was utilized to assign weights to each study in the meta-analysis. A summary ROC curve was constructed to evaluate the diagnostic value of dynamic CT perfusion in detecting significant coronary artery stenosis.¹¹ Inverse variance and random-effects models were also employed for integrating hazard ratios. Heterogeneity was assessed using the I² statistic, where 0% indicates no heterogeneity and 100% suggests strong heterogeneity.¹² A significance level of P < 0.05 was considered statistically significant.

Results

A flowchart of the study selection process is shown in Figure 1. Finally, a total of 23 studies were analysed. Characteristics of included studies were summarized in Tables 1 and 2. Twenty studies presented diagnostic performance using FFR as the reference standard,^5,7,13–30 and three studies presented data on prognostic value of dynamic CTP.^31–33 The total number of patients included in the analysis was 2,190, which included patients with known CAD as well as those with suspected CAD. The countries of publication were as follows: five from China,^{20,21,28–30} five from Japan,^{7,19,23,27,32} four from the USA,^5,13,31,33 three from the UK^16,25,26 and the Netherlands,^14,15,18 two from Germany,^17,22 and one from Italy.²⁴ Study designs included 12 prospective single-centre studies,^{13,14,16–20,24,26,28–30} 7 prospective multicentre studies,^{5,7,15,22,25,31,33} and 4 retrospective single-centre studies.^21,23,27,32 The type of CT used was SSCT in 4 studies^17,19,23,27 and DSCT in 19 studies.^{5,7,13–16,18,20–22,24–26,28–33}

Comparison of CT-MBF between regions with and without functional myocardial ischaemia

Figure 2 shows the forest plot of MBF in the region without myocardial ischaemia. The pooled MBF in the region without myocardial ischaemia was 1.39 mL/min/g [95% confidence interval (CI) 1.25–1.54, I² = 99%, P for heterogeneity < 0.001]. Figure 3 shows a forest plot of MBF in the region where functional myocardial ischaemia was observed in the FFR. In the region of myocardial ischaemia, MBF was 0.92 mL/min/g (95% CI 0.83–1.01, I² = 98%, P for heterogeneity < 0.001) (Figure 3). There was a statistically significant difference in pooled MBF between areas with and without functional myocardial ischaemia (P < 0.001).

Diagnostic accuracy of dynamic CTP for detection of functionally significant coronary artery stenosis

In the detection of functionally significant coronary artery stenosis as defined by FFR, the area under sROC with dynamic CT perfusion alone was 0.92 (95% CI 0.90–0.94) at the vessel level, with a sensitivity of 0.82 (95% CI 0.79–0.85) and a specificity of 0.86 (95% CI 0.84–0.88) (Figure 4). The area under sROC for diagnostic performance of coronary CTA in the analysed literature was 0.85 (95% CI 0.81–0.89) at the vessel level, with a sensitivity of 0.78 (95% CI 0.74–0.81) and specificity of 0.70 (95% CI 0.68–0.72) (see Supplementary data online, Figure S1). Area under sROC for diagnostic performance of the combination of CTA and CTP was 0.90 (95% CI 0.86–0.94) at the vessel level, sensitivity was 0.78 (95% CI 0.74–0.81) and specificity was 0.85 (95% CI 0.83–0.87) (Figure 5). The addition of CTP on CTA increased the AUC at the vessel level from 0.85 to 0.90 but was not statistically significant (P = 0.07 by χ² test). The patient-level diagnostic performance of the combination of CTA and CTP is depicted in Supplementary data online, Figure S2, with an AUC of 0.88 (95% CI 0.82–0.94) and a pooled sensitivity of 0.83 (95% CI 0.76–0.90). The pooled sensitivity was 0.83 (95% CI 0.76–0.90), and the pooled specificity was 0.77 (95% CI 0.71–0.83).

Influence of CT scanner type and patient BMI on diagnostic accuracy of dynamic CT perfusion

Figure 6 compares the sROC of dynamic CTP alone with DSCT and SSCT. The AUC of DSCT was 0.92 (95% CI 0.88–0.96) and that of SSCT was 0.95 (95%CI: 0.89–1.00), and the area under sROC was similar between the two groups. The area under sROC was similar between the two groups.

Radiation exposure was 10.05 mSV (95% CI 8.35–11.75, P < 0.001, I² = 0%, P = 0.63 for heterogeneity) in the SSCT group and 5.34 mSV (95% CI 4.14–6.55, P < 0.001, I² = 83%, P < 0.001 for heterogeneity) in the DSCT group. DSCT was significantly less exposed than SSCT (P < 0.001). Figure 7 compares the sROC of dynamic CTP alone classified by low (<25 kg/m²) and high (≥25 kg/m²) BMI, with an AUC of 0.93 (95% CI 0.88–0.98) for low BMI and 0.92 (95% CI 0.88–0.96) for high BMI. The area under sROC was similar between the two groups.

Prognostic value of dynamic CT perfusion for the prediction of adverse cardiac events

Three studies showed the predictive ability of dynamic CT perfusion for cardiovascular events (Table 3). The definitions of abnormalities in dynamic CT perfusion are summarized in Table 3. Figure 8 shows the pooled heart rate (HR) of dynamic CT perfusion. A meta-analysis calculated a pooled HR of 4.98 (95% CI 2.08–11.93, P < 0.001, I² = 61%, P for heterogeneity = 0.07).

Discussion

This meta-analysis illustrated the high diagnostic accuracy of dynamic CT perfusion for the detection of functional myocardial ischaemia evaluated by invasive FFR, regardless of the CT scanner, and independent of the patient's body size. Dynamic CT perfusion also exhibits significant prognostic value for patients with known or suspected CAD, making it suitable for risk stratification. These results suggest that dynamic CT perfusion might be a useful non-invasive modality for diagnosing CAD and determining treatment strategies.

Coronary CTA has been employed for non-invasive evaluation of coronary artery stenosis, demonstrating commendable diagnostic performance in excluding coronary artery conditions with acceptable radiation exposure, particularly on CT systems with 64 or more rows.³⁴ With the introduction of a greater number of CT systems, such as 320-row systems, volumetric scanning becomes feasible, leading to a notable reduction in radiation and the potential utilization of CT for ischaemia assessment.³⁵ Initially, the evaluation of myocardial ischaemia using CT revolved around static CT perfusion, but technological advancements have led to the adoption of dynamic CT perfusion, which offers higher accuracy. Dynamic CT perfusion surpasses myocardial SPECT in assessing myocardial ischaemia. Moreover, the simultaneous capture of coronary artery CTA and CT dynamic perfusion allows for the fusion of coronary artery anatomy and myocardial ischaemia, resulting in a more precise diagnosis of ischaemia. Numerous clinical studies and meta-analyses have been conducted in this regard. However, the recent development of DSCT demonstrates the potential to enhance the image quality of cardiac CT scans and improve quantification.³⁶ Additionally, the gold standard for assessing ischaemia has shifted from the degree of stenosis in CAG to functional ischaemia evaluation via FFR.³⁷ Therefore, the accuracy of MBF for ischaemia as assessed by FFR is also a matter of concern. Furthermore, obesity significantly hampers the quality of cardiac CT images,^38,39 making it vital to ascertain whether the accuracy of MBF remains consistent in obese patients. With this in mind, we performed a meta-analysis to address the aforementioned questions. The results showed that dynamic CT perfusion has high diagnostic accuracy for functional myocardial ischaemia defined by invasive FFR, regardless of type of CT scanner and patients’ BMI.

Another critical role of cardiovascular imaging is to identify high-risk patients and guide them toward appropriate treatment, known as risk stratification. Our meta-analysis was conducted due to the presence of three articles reporting hazard ratios for dynamic CT perfusion in relation to cardiovascular events. The results revealed that dynamic CT perfusion holds significant prognostic value, as indicated by a pooled hazard ratio of 4.98. When combined with coronary CTA findings, this risk stratification can be further strengthened. Thus, intensive preventive measures may be recommended not only for patients with coronary plaque identified on coronary CTA but also for individuals exhibiting minimal atherosclerotic changes on coronary CTA yet demonstrating high-risk indicators on myocardial perfusion. These aspects should be clarified through future studies.

Several limitations exist within this study. Firstly, heterogeneity in CT-MBF values has been observed. This disparity could stem from variations in imaging methodologies or patient characteristics, although the precise reasons remain unclear and necessitate careful interpretation. Secondly, only a limited number of papers have presented prognostic predictions, and the definition of events varies across studies. Therefore, it is desirable to conduct more diverse prospective studies in the future.

Conclusions

Dynamic CT perfusion exhibits a high diagnostic capability in the quantitative assessment of ischaemia and the identification of functionally significant coronary artery stenosis. These observations remain consistent regardless of the specific CT scanner employed or the patient's BMI. Moreover, dynamic CT perfusion holds promise as a valuable tool for risk stratification among individuals diagnosed with CAD. These results suggest that dynamic CT perfusion might be a useful non-invasive modality for diagnosing CAD and determining treatment strategies.

Supplementary data

Supplementary data are available at European Heart Journal - Cardiovascular Imaging online.

Funding

None declared.

Data availability

The data underlying this article will be shared upon reasonable request from the corresponding authors.

References

Author notes