Tc-99m sestamibi single photon emission computed tomography for guiding percutaneous coronary intervention in patients with multivessel disease: a comparison with quantitative coronary angiography and fractional flow reserve
- First Online:
- Cite this article as:
- Förster, S., Rieber, J., Übleis, C. et al. Int J Cardiovasc Imaging (2010) 26: 203. doi:10.1007/s10554-009-9510-x
- 175 Views
To evaluate the accuracy of myocardial perfusion SPECT (MPI) in the detection and allocation of vessel specific perfusion defects (PD) using standard distribution territories in a routine clinical procedure of patients with multivessel disease (MVD). Combined quantitative coronary angiography and fractional flow reserve (QCA/FFR) measurements were used as invasive reference standard. 216 vessels in 72 MVD patients (67 ± 10 years, 28 female) were investigated using MPI and QCA. FFR of 93 vessels with intermediate stenoses was determined. MPI detected significant stenoses according to QCA/FFR findings with a sensitivity of 85%. However, vessel-based evaluation using standard myocardial distribution territories delivered a sensitivity of only 62% (28 MPI+ out of 45 (QCA/FFR)+ findings), with specificity, PPV and NPV of 90, 62 and 90%. 7/17 false positive and 7/17 false negative findings (41%) could be attributed to incorrect allocation of reversible PD to their respective coronary arteries. 6/17 (35%) perfusion territories were classified as false negative when additional fixed PD were present. MPI had reasonable sensitivity for the detection of significant coronary artery disease in patients with multivessel disease. However, sensitivity decreased markedly, when the significance of each individual stenosis was evaluated using standard myocardial supplying territories. In this setting, 41% of false negative and false positive MPI findings resulted from incorrect allocation of reversible perfusion defects to their determining supplying vessel.
KeywordsMyocardial perfusion SPECTFractional flow reserveMultivessel diseasePercutaneous coronary interventionCoronary angiography
Based on an extensive body of data, myocardial perfusion single photon emission tomography (MPI) is widely used for risk stratification and assessment of both ischemia and viability in patients with coronary artery disease (CAD) . However, given that MPI is an imaging technique measuring flow enhancement in diverse myocardial beds based on changes in relative radiotracer uptake , there arise distinct limitations in its use in the evaluation of patients with multivessel disease (MVD) [7, 10]. In particular, the allocation of perfusion defects to their determining coronary arteries or specific coronary lesions—a precondition for performing appropriate percutaneous coronary intervention (PCI) [15, 16, 36]—is frequently hampered when morphological correlation is not available [7, 10, 35, 41, 23, 40]. Additionally, the detection of ischemic myocardial regions might be compromised in patients with fixed perfusion defects due to the presence of myocardial scaring or chronic hypo perfusion, as has been shown in a recent study .
In a clinical cardiology setting, two-dimensional quantitative coronary angiography (QCA), although often underestimating or overestimating a lesion`s functional severity, is still the standard technique for guiding PCI in patients with multivessel CAD [14, 39].
Initial results from the Fractional Flow Reserve versus Angiography for Multivessel Evaluation (FAME) study, however, reported in 1,005 patients with multivessel CAD a significant reduction of the composite end point of death, non-fatal myocardial infarction (MI), and repeated revascularization during a 1 year follow-up, when additional measurements of fractional flow reserve (FFR) were performed .
FFR is defined as the ratio of maximum achievable coronary blood flow in a stenotic coronary artery relative to the maximal blood flow in the same vessel in the absence of all epicardial obstructions . Initial studies compared FFR with MPI as a reference standard in patients both with single-vessel and multivessel disease [5, 7, 8, 12, 21, 25, 29, 43]. On the basis of various clinical studies, an FFR cut-off value <0.75 was established for the detection of flow-limiting or functionally significant coronary artery stenoses. Recent publications confirmed the validity of the cut-off value of 0.75 also in comparison to H215O positron emission tomography blood flow measurements in patients with chronic MI , as well as after revascularization therapy and during long-term follow up [3, 31].
The aim of the present study was to evaluate MPI for the detection and allocation of flow-limiting stenoses in patients with multivessel disease, compared to an invasive reference standard of QCA/FFR. We hypothesized that the accuracy of MPI is limited for vessel-based evaluation using standard myocardial distribution territories due to the uncertainty of allocating perfusion defects to particular coronary arteries without knowledge of individual coronary vascular anatomy.
Patients were included in the study if they had multivessel coronary artery disease, which was defined as coronary artery stenoses of at least 50% of the vessel diameter in at least two of the three major epicardial coronary arteries. Patients who had had a myocardial infarction were included if the infarction had occurred at least 10 days before study inclusion. Patients who had undergone previous PCI were also included in the study. Patients who had angiographically significant left main coronary artery disease, previous coronary-artery bypass surgery, or patients who were pregnant were excluded.
QCA with FFR measurements and MPI were performed in all patients.
The study was approved by the local ethics committee and written informed consent was obtained from all patients.
Quantitative Coronary Angiography (QCA) and FFR measurements
All patients were instructed to abstain from caffeine and chocolate for 12 h prior to catheterization. At least two orthogonal views were obtained, and the projection showing the most severe narrowing was used for quantitative coronary measurements (Philips DCI, The Netherlands). Using the guiding catheter as a scaling device, measurements of the minimal lumen diameter as well as proximal and distal reference diameters were made .
FFR was measured in all vessels with intermediate stenoses, i.e. in the range ≥50 and ≤75%. Vessels with severe (>75%) or low-grade stenoses (<50%) were not investigated for pressure measurements. After crossing the target lesion with a dedicated sensor-tipped 0.014-inch angioplasty guidewire (WaveWireWaveMap, Volcano Therapeutics, Rancho Cordova, CA, USA; or PressureWire, Radi Medical Uppsala, Sweden) while under angiographic guidance, the pressure sensor was positioned beyond the stenosis in the distal portion of the artery. Phasic and mean aortic pressure as well as phasic and mean coronary pressure distal to the stenoses were then measured under maximum coronary hyperemia, which was induced by intravenous administration of 140 μg/kg min−1 of adenosine (Adrecar, Sanofi, Munich, Germany). FFR was defined as the ratio of mean poststenotic pressure and mean aortic pressure measured during maximum hyperemia.
Significance of a stenosis was classified by dichotomous criteria (significant or non-significant), according to the composite of the QCA and FFR findings. A lesion was classified as significant if severe stenosis was detected in QCA, or if FFR measurement yielded a value of <0.75. Occluded vessels or those who could not be assessed by FFR due to subtotal occlusions were also rated as significant. A lesion was classified as non-significant if QCA showed no abnormality, stenosis <50%, or if FFR was ≥0.75.
Myocardial Perfusion SPECT (MPI)
When appropriate, physical or pharmacological stress/rest MPI was performed according to a one-day protocol with Tc-99m sestamibi, as follows; before the termination of the stress test, a dose of 4 MBq/kg of Tc-99m sestamibi (at least 300 MBq) was administered intravenously. For the subsequent resting study, a dose of 10 MBq/kg of Tc-99m sestamibi (at least 700 MBq) was injected. If systolic blood pressure was greater than 120 mm Hg, 0.8 mg nitroglycerine was sublingually administered to the patients before injection of the radiopharmaceutical for the rest image.
Image acquisition was performed with a triple-headed camera system (Philips [formerly Picker] Prism 3000 XP, Cleveland, Ohio). Possible attenuation artefacts were corrected by applying an attenuation correction based on a simultaneous transmission measurement with 153Gd (STEP®), with 360° rotation in continuous mode, or alternately by performing gated SPECT acquisition for wall motion analyses, as described previously . Images were reconstructed over 360° with 20 slices along the short axis, the long axis, and the four-chamber view for each study. A standardized filter (Low Pass 4th power, cut-off-frequency 0.26) was used. Quantitative analysis of MPI perfusion studies was carried out using QPS processing software (Cedars-Sinai Medical Center, Los Angeles, California).
Image analysis was performed by agreement of two experienced observers (M.H. and S.F.) blinded to the results of QCA/FFR, the coronary distribution type, or the presence of coronary normal variants such as ramus intermedius, or the location of stenoses. A commonly used 20 segment model was employed for division of the left ventricular myocardium images . Each of the 20 segments was scored according to the guideline for semiquantitative analysis (“Semiquantitative Scoring System: The Fivepoint Model”: 0 = normal; 1 = mildly reduced—not definitely abnormal; 2 = moderate reduced—definitely abnormal; 3 = severely reduced; 4 = absent radiotracer distribution) . Segmental scores were summed for the three main coronary arteries (LAD, RCA, LCx) according to standard myocardial perfusion territories, as described elsewhere , resulting in regional perfusion scores under stress (SSSr, regional Summed Stress Score) and rest (SRSr, regional Summed Rest Score) conditions. The difference of SSSr and SRSr was defined as the regional Summed Difference Score (SDSr). On the basis of previously published results, stenoses and their respective supplying territories with an SDSr ≥ 1 were considered significant, while stenoses and their respective supplying territories with an SDSr = 0 were considered as non-significant . SRSr ≥ 1 was defined as fixed perfusion defect.
Evaluation of the allocation process
MPI suggested ischemia of a target vessel with non-significant stenosis (<50% or an FFR ≥ 0.75); and at the same time significant stenosis (between 50 and 75% with an FFR < 0.75 or stenosis > 75%) was present in another vessel, which (according to standard distribution territories) did not show ischemia on MPI ((MPI)+/(QCA/FFR)−).
MPI did not suggest ischemia, but significant stenosis was nonetheless present in the respective distribution territory, and at the same time MPI detected ischemia in another vessel with non-significant stenosis ((MPI)−/(QCA/FFR)+)).
Descriptive analysis for categorical and continuous parameters was performed using SPSS version 13.0 (SPSS, Chicago, IL, USA). Results are presented as mean ± standard deviation (SD) and range, unless stated otherwise. Paired and unpaired t-tests were used when appropriate. Statistical significance was tested on the 5% level.
Seventy two consecutive patients (28 female, mean age 67 ± 10 years) with multivessel disease (32 patients with two- and 40 patients with three-vessel disease) were eligible for the study.
QCA with FFR measurements and MPI were performed in each patient within an interval of 13 ± 43 days. QCA/FFR was performed before MPI in 13 and after MPI in 59 patients.
Clinical characteristics of study cohort (n = 72)
Age years ± SD
67 ± 8.5
Diabetes mellitus (%)
Current smoker (%)
Family predisposition (%)
First-pass LVEFrest %
53 ± 10
2-vessel disease (%)
3-vessel disease (%)
Quantitative coronary angiography and FFR measurements
Procedural characteristics of significant versus non-significant lesions as defined by QCA/FFR
Significant n = 45
Non-significant n = 171
QCA diameter stenosis %
71.2 ± 11.5**
34.8 ± 28.7
QCA widthprox (mm)
2.8 ± 1.6**
1.8 ± 1.6
QCA widthdist (mm)
2.7 ± 1.3**
1.7 ± 1.5
FFR (n = 93)
0.63 ± 0.1**
0.85 ± 0.1
4.8 ± 6.6**
1.5 ± 4.0
3.1 ± 6.2**
1.0 ± 3.4
1.7 ± 2.5**
0.4 ± 1.5
According to QCA, 14 vessels showed severe stenoses (>75%) and were rated as significant. Ten of these coronary arteries showed total occlusions, six in the proximal and four in the distal part of the artery. However, at least partial collateral filling (Rentrop grade 2 or higher) was present in seven (70%) of the occluded vessels. 31 of the 93 intermediate stenoses showed FFR <0.75, such that overall 45/216 lesions were rated as significant by definition.
MPI in the detection of significant coronary artery lesions using standard myocardial distribution territories
Vessel based evaluation
A total of 216 perfusion territories were analyzed for MPI (3 territories per patient). 64 territories were identified as showing abnormal perfusion (any PD), 45 with reversible perfusion defects (SDSr ≥ 1) and 19 with fixed perfusion defect (PD) (SRS ≥ 1). 15 perfusion territories showed a combination of both, reversible and fixed perfusion defects.
No PD present
Positive predictive value
Negative predictive value
When the presence of any perfusion defect was taken into account, 33 of 64 myocardial supplying territories were rated significant by QCA/FFR and 12 of 45 significant lesions showed neither reversible nor fixed perfusion defects on MPI (Table 3). This indicates sensitivity of 73%, specificity of 82%, ppV of 52%, npV of 92%, and an accuracy of 80%.
Considering only the 93 vessels with intermediate stenoses (≥50 and ≤75%) with all individually available FFR measurements, sensitivity was 77%, specificity was 82%, ppV was 69%, npV was 88% and accuracy was 81% when reversible perfusion defects in MPI were evaluated (data not shown in detail).
Patient based evaluation
No PD present
Positive predictive value
Negative predictive value
No PD present
Positive predictive value
Negative predictive value
MPI and allocation of reversible perfusion defects
Overall, 14 of 34 (MPI)+/(QCA/FFR)− or (MPI)−/(QCA/FFR)+ findings (41%) occurred due to wrong allocation of MPI reversible perfusion defects to their determining and supplying vessel, according to standard myocardial distribution territories.
Numerous studies have shown high overall sensitivity, up to 90%, for MPI to identify patients with two- or three vessel disease [9, 28, 30, 42]. Particularly in these patients, the detection of functionally significant coronary artery stenoses is an important precondition for adequate treatment and improved outcome. However, as compared to intracoronary pressure measurements, which are performed directly in the target vessels, MPI alone has limited capacity to detect functionally significant stenoses, and to allocate correctly the perfusion defects to specific vessels or even coronary lesions in patients with MVD, as shown in the present study.
The most important result for planning and guiding individual therapies is the limited utility of MPI alone for allocating correctly reversible and fixed perfusion defects to their respective coronary artery, or even to their determining coronary artery lesion. Indeed, 41% of our disagreements (41% of (MPI)−/(QCA/FFR)+ and 41% of (MPI)+/(QCA/FFR)−) resulted from just such allocation problems. These limitations of MPI are well known and previously documented in comparison studies between MPI and invasive coronary angiography  and can be explained by previously published post mortem analysis, reporting only 50–60% accordance between standard myocardial perfusion territories and supplying areas of the three main coronary arteries, a discrepancy arising from the extensive inter-individual variability of the coronary tree .
However, the diagnostic accuracy of MPI was rarely investigated using a combined morphological and functional reference standard, which was recently shown superior to QCA alone in patients with MVD . To ensure an objective comparison in the present study, MPI perfusion defects were systematically assigned to one of the three main coronary arteries (RCA, LAD and LCX).
As such, our procedure does not necessarily reflect clinical routine diagnostics in patients with MVD. One possible clinical scenario is to determine whether a demonstrated anatomical abnormality is causing flow limitation requiring intervention in the setting of an intermediate or no diagnostic finding on QCA, especially when revascularization may not be straightforward for the interventionist like in the presence of long segment of disease, nearby branches or in the presence of poor visualization on QCA, in- or peri-stent restenosis.
Indeed, specific limitations of the MPI method without knowledge of patient`s coronary anatomy were evident in the present study, even if early studies have demonstrated how well this method can demonstrate a culprit lesion or draw attention to one which may have been initially overlooked at QCA.
In a recent study of 36 patients (88 vessels) suffering from MVD, Ragosta et al.  reported that 36% of all vascular zones lacking fixed or reversible perfusion abnormalities on MPI showed either pathological FFR (<0.75), or total occlusions in QCA. From this observation, it was concluded that numerous hemodynamically significant stenoses would be overlooked if clinical judgment were based only upon MPI. In their interpretation, most cases of discordance between MPI and FFR measurements were primarily due to perfusion imaging correctly identifying the most severe stenosis, but not identifying other zones subtended by lesser, but still significant, lesions. In seven of 22 patients (32%), MPI was completely normal in all perfusion territories, despite the occurrence of pathologically FFR in one or more territories.
The observed high rate of improper classification of an individual vascular territory based on MPI using standard myocardial distribution territories would lead to errors in management that are clinically unacceptable. This well known limitation has led to the proposed use of 3-D image fusion of the coronary arteries visualized by coronary angiography, with myocardial perfusion maps , an elegant approach that has not yet found widespread use in routine diagnostic practice. It has to be emphasized, that decisions regarding revascularization of specific arteries in patients with MVD cannot be based on MPI alone and that there is a need of proper anatomical information (i.e. one should better rely on incorporation of the angiogram with adjunctive use of FFR as appropriate). Certainly, non-invasive methods combining morphological and functional imaging strategies using SPECT-CT or PET-CT hybrid scanners will improve MPI as a tool for the selection of vessel regions that are candidates for intervention. There are already promising studies using hybrid imaging technology [17, 27], which were able to show improved allocation of perfusion defects to specific coronary artery lesions and, thus, potentially will improve therapy planning.
Assessing the diagnostic performance of MPI was a further aspect in the present study, we found that 4/17 patients with significant stenosis in QCA/FFR (24%) had no perfusion defect in MPI, reflecting “true” false negative ((MPI)−/(QCA/FFR)+) results. This reflects a well-known limitation for MPI, which can be attributed to the occurrence of balanced ischemia in the absence of valid myocardial reference areas [22, 41]. A further reason for discrepant results in which MPI was negative while QCA/FFR positive in the present study (6/17 patients, 35%), was the occasional presence of fixed perfusion defects masking ischemia, which is also a well-known limitation of MPI arising from an underestimation of the true extent of myocardial viability in the standard resting images ; see example Fig. 1. Some of the fixed perfusion defects might have reflected not MI, but rather stunned or hibernating myocardium, particularly in this selective patient group with more advanced stages of disease. For this reason, we applied nitroglycerin before the tracer injection at rest, so as to enhance tracer uptake in the ischemic myocardium compared with that seen in the nonviable and nonischemic myocardium. This approach has earlier been shown to improve MPI viability detection .
Our study has several limitations. First, FFR measurement was not performed in all three coronary arteries; therefore, the reference standard consists of a combination of angiographic diameter stenosis measurements with and without corresponding FFR values, reflecting an approach already published by our group . A separate comparison of MPI with FFR in intermediate stenoses delivered a sensitivity of 77%, which was in line with previous studies, confirming that these assumptions were appropriate. In general, it is at present difficult to control for the effects of perfusion derived from collateral vessels, competitive flow, differential ischemia and other factors . Additionally, functionally relevant stenoses <50% as well as intramyocardial vessel affections and peripheral stenoses, which both are inaccessible for FFR measurements, could have lead to false positive MPI findings. However, analysis in the subgroup of patients with diabetes mellitus (and therefore higher possibility of microvascular disease) revealed similar ability for MPI in the detection of significant stenoses when compared to the whole patient cohort, such that the rate of microvasculature related false positives in the current study is deemed to be low.
Another limitation is presented by the lack of gated SPECT in all patients, such that regional wall motion and thickening patterns have not been implemented for optimal validation of MPI for the identification of significant stenoses.
Myocardial perfusion SPECT had reasonable sensitivity for the detection of significant coronary artery disease in patients with multivessel disease relative to quantitative coronary angiography with/without additional FFR measurements. However, sensitivity decreased markedly, when the significance of each individual stenosis was evaluated using standard myocardial supplying territories. In this setting, 41% of false negative and false positive MPI findings resulted from incorrect allocation of reversible perfusion defects to their determining supplying vessel.
We are grateful for the support and superb technical assistance of the staff in the departments of Nuclear Medicine and Cardiology at the University of Munich.