Diagnostic accuracy of postmortem computed tomography, magnetic resonance imaging, and computed tomography-guided biopsies for the detection of ischaemic heart disease in a hospital setting

Diagnostic accuracy of postmortem computed tomography, magnetic resonance imaging, and computed... Abstract Aims The autopsy rate worldwide is alarmingly low (0–15%). Mortality statistics are important, and it is, therefore, essential to perform autopsies in a sufficient proportion of deaths. The imaging autopsy, non-invasive, or minimally invasive autopsy (MIA) can be used as an alternative to the conventional autopsy in an attempt to improve postmortem diagnostics by increasing the number of postmortem procedures. The aim of this study was to determine the diagnostic accuracy of postmortem magnetic resonance imaging (MRI), computed tomography (CT), and CT-guided biopsy for the detection of acute and chronic myocardial ischaemia. Methods and results We included 100 consecutive adult patients who died in hospital, and for whom next-of-kin gave permission to perform both conventional autopsy and MIA. The MIA consists of unenhanced total-body MRI and CT followed by CT-guided biopsies. Conventional autopsy was used as reference standard. We calculated sensitivity and specificity and receiver operating characteristics curves for CT and MRI as the stand-alone test or combined with biopsy for detection of acute and chronic myocardial infarction (MI). Sensitivity and specificity of MRI with biopsies for acute MI was 0.97 and 0.95, respectively and 0.90 and 0.75, respectively for chronic MI. MRI without biopsies showed a high specificity (acute: 0.92; chronic: 1.00), but low sensitivity (acute: 0.50; chronic: 0.35). CT (total Agatston calcium score) had a good diagnostic value for chronic MI [area under curve (AUC) 0.74, 95% confidence interval (CI) 0.64–0.84], but not for acute MI (AUC 0.60, 95% CI 0.48–0.72). Conclusion We found that the combination of MRI with biopsies had high sensitivity and specificity for the detection of acute and chronic myocardial ischaemia. postmortem , magnetic resonance imaging , computed tomography , autopsy , ischaemic heart disease Introduction According to the World Health Organization, cardiovascular disease and stroke are the foremost contributors to worldwide mortality with ischaemic heart disease globally causing almost nine million deaths per year in 2015.1 Accurate mortality statistics are important for both policy- and decision-making regarding healthcare funding. For reliable statistics, it is essential to perform autopsies in a sufficient proportion of deaths, both in and out of hospital. Despite available modern diagnostic tests, the conventional autopsy still reveals unexpected findings related to the cause of death in 8.4–24.4% and findings that would have affected patient outcome (Class I errors) in 4.1–6.7% of cases.2–4 Unfortunately, today’s autopsy rate has dropped to alarmingly low percentages worldwide (0–15%), both for academic and non-academic hospitals.5 In the late 90s, the imaging autopsy was introduced as an alternative to conventional autopsy as a stimulus to postmortem diagnostics. Since then, a growing number of studies have evaluated the diagnostic value of postmortem computed tomography (CT) and magnetic resonance imaging (MRI) with or without image-guided biopsies. Two review articles concluded that the imaging autopsy, using a non-invasive or minimally invasive approach, can potentially serve as an alternative to conventional autopsy, but more extensive research in different settings is needed to validate these new autopsy methods.6,7 Imaging protocols designed for the living patient differ from postmortem imaging protocols, in particular for cardiac imaging. For example, wall movement abnormalities of the heart cannot be diagnosed. In living patients, contrast-enhanced imaging, either non-invasive or invasive, is the gold standard for diagnosing ischaemic heart disease. Postmortem angiographic studies are feasible and not new; since the discovery of X-rays, angiography of organs and tissues has been used as an adjunct to the autopsy procedure.8 More recently, postmortem total-body angiography, using CT or MRI, has become technically feasible, and there are a growing number of studies investigating its diagnostic value. Preliminary results are promising, especially for establishing ischaemic heart disease as the cause of death.9–11 Interestingly postmortem MRI without the use of contrast agents also shows a sufficient accuracy for detecting both acute and chronic myocardial infarction (MI). The presence and age of MI can be diagnosed by evaluating the signal changes related to morphological alterations in the infarcted myocardium, such as the presence of myocardial oedema, fibrosis, or fat.12–15 Non-enhanced cardiac CT is also useful for detecting coronary artery calcifications. In this study, we evaluate the minimally invasive autopsy (MIA) approach using both non-enhanced CT and MRI followed by CT-guided biopsies. The aim of this study was to determine the diagnostic accuracy of MRI, CT, and CT-guided biopsy for the detection of acute and chronic MI, with conventional autopsy as the reference standard. Methods Study design This study was undertaken as part of the MIA study; this is a prospective single centre cross-sectional study in a tertiary referral hospital comparing diagnostic performance of conventional autopsy and MIA. Conventional autopsy followed MIA on the same case. Pathologists involved with conventional autopsy were blinded to MIA findings as much as possible; however, biopsy sites could potentially lead to increased suspicion of the biopsied organs and tissues by the autopsy pathologist. MIA personnel were blinded to autopsy findings. Patients From January 2012 to December 2014 all hospitalized patients aged 18 years and older who died at Erasmus University Medical Center were eligible, if written informed consent was obtained from next-of-kin for MIA and conventional autopsy of at least the torso. Exclusion criteria were (suspected) unnatural cause of death, body size exceeding height of 16 inches in supine position (limitation for MRI), known or suspected ‘high-risk’ infected bodies (tuberculosis, hepatitis B and C, human immunodeficiency virus, methicillin-resistant Staphylococcus aureus, and multi-drug resistant Acinetobacter), and open abdominal wounds that could not be completely closed or taped to prevent leakage of body fluids. Clinical information All relevant clinical information including medical history and suspected cause of death was recorded and available for both the MIA and the conventional autopsy team. The treating physician decided the most likely cause of death and a differential diagnosis based on the clinical presentation. For the analysis, the population was divided into a group with and a group without clinical suspicion for ischaemic heart disease. MIA procedure MRI and CT scans were made according to standardized protocols (Tables 1 and 2, respectively). Total acquisition time was 1 h for MRI and around 5 min for CT. One radiologist (A.C.W.) with expertise in postmortem radiology, performed the initial read of the MRI and CT scans, compared these to the available premortem imaging, and marked suspected pathological lesions on CT and MRI key images that were used to plan the biopsies. Table 1 Postmortem magnetic resonance protocol Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  All scans were made in the axial orientation. FLAIR, fluid attenuated inversion recovery; FOV, field of view; fs FSPGR, fat-saturated fast spoiled gradient echo; FSE, fast spin echo; STIR, short tau inversion recovery; TE, echo time; TI, inversion time; TR, repetition time. Table 1 Postmortem magnetic resonance protocol Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  All scans were made in the axial orientation. FLAIR, fluid attenuated inversion recovery; FOV, field of view; fs FSPGR, fat-saturated fast spoiled gradient echo; FSE, fast spin echo; STIR, short tau inversion recovery; TE, echo time; TI, inversion time; TR, repetition time. Table 2 Postmortem computed tomography protocol Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Table 2 Postmortem computed tomography protocol Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  MRI was performed on a 1.5T scanner (Discovery MR450, GE Medical systems, Milwaukee, WI, USA) and consisted of scans of the brain, neck, thorax, abdomen, and pelvis. The MRI total-body protocol consisted of axially-acquired short tau inversion recovery (STIR) fast spin echo (FSE) T2w and fluid attenuated inversion recovery (FLAIR) FSE T1w from the cranium to the pelvis. An additional 2D STIR FSE T2w scan and 3D Fatsat FSPGR T1w scan with higher resolution than the total-body scans were acquired of the thorax, using an 8-channel torso array coil. All MRI scans were made in the axial orientation. After MRI was completed CT scans were acquired from head to feet (Somatom Definition, Siemens Healthcare, Forchheim, Germany). CT datasets of the head, thorax, and abdomen were reconstructed with section thickness of 1.0 mm and 5.0 mm in the axial plane and 3.0 mm in the coronal and sagittal planes, by using medium-to-smooth (H31/B31) and very sharp (H70/B70) convolution kernels. CT-guided biopsies (12 Gauge) were taken from heart, lungs, liver, kidneys, spleen, and radiologically suspected pathology as indicated. In the heart, standard biopsies (5–10 samples) were taken from the lateral wall (mid- and basal-parts) and an apex of the left ventricle. Additional biopsies were taken from MRI signal abnormalities within the myocardium. In those cases where there was a clinical suspicion of MI and the MRI showed no signal abnormalities additional biopsies were taken from the septum, anterior, and posterior wall. The MIA pathologist (J.W.O.) and researcher (B.M.B.) examined the microscopic slides of the biopsies; when in doubt, they consulted pathologists with specific expertise, not involved in the matching conventional autopsy, to reach a conclusion. Cardiac imaging evaluation CT For each case, CT calcium score was calculated by one observer (I.M.W.) using dedicated software (Syngo.via 3.0 Calcium Scoring®, Siemens Healthcare, Forchheim, Germany) and expressed as total Agatston scores. MRI MR images were reconstructed and evaluated in the short-axis view. Two radiologists with expertise in cardiac radiology (A.C.W. and A.P.P.) independently evaluated MR images and in case of disagreement, consensus was reached in joint sessions. MIs were classified according to a modified classification by Jackowski et al.12,16,17 Peracute infarction (within 6 h after onset) is characterized by T2 hypointense signal in the necrotic centre, caused by a state of hypoperfusion. In the acute phase (within 6 h–1 week after onset), the marginal areas become oedematous and show T2 hyperintense signal, T1 signal in the centre is isointense, and the oedematous marginal regions can show T1 hypointense signal. Subacute infarction (>1 week after onset) shows T2 hyperintense signal in the infarcted area when the area becomes reperfused, while the marginal areas show normal T1 and T2 signal. Chronic infarction (>2 months after onset) shows wall thinning and scar tissue reflected by T1 and T2 hypointense signal and foci of T1 hyperintense signal can be seen due to fatty infiltration. In our analysis, peracute and acute infarctions were grouped into one category and defined as acute MI (<1-week old infarction). Subacute and chronic infarctions were grouped into one category and defined as chronic MI (>1-week old infarction). MRI criteria for determining infarction age are detailed in Table 3. Table 3 MRI criteria for determining infarction agea   Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =    Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =  a Criteria based on Jackowski et al.12,16,17 Table 3 MRI criteria for determining infarction agea   Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =    Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =  a Criteria based on Jackowski et al.12,16,17 Conventional autopsy The day after MIA, a resident in pathology, supervised by the attending pathologist, performed conventional autopsy according to the departmental protocol. The autopsy report included medical history, postmortem diagnoses, a presumed cause of death, and answers to specific clinical questions, and was authorized by the pathologist. Macroscopic evaluation consisted of sectioning of the heart in slices of 0.5 to 1 cm and visual inspection of the myocardium. Lactate dehydrogenase (LDH) staining was performed on a mid-ventricular slice. Haematoxylin and eosin were used for histological staining.18 When there was a discrepancy between the histology of the MIA and conventional autopsy, a pathologist with expertise in cardiac pathology (JHT) reviewed the histology. The following criteria for MI age were used (separately or in combination): acute MI: hypereosinophilia and loss of cross striation within myocardial fibres, contraction band necrosis, coagulation necrosis with or without granulocyte infiltration or haemorrhage, and various degrees of nuclear pyknosis, karyolysis, granulocyte infiltration, and myocardial oedema; chronic MI: fibroblasts with loose connective tissue formation, angiogenesis (subacute), and paucicellular collagenous fibrosis (chronic).12 Statistical analysis Analyses were performed on the patient level. Conventional autopsy was used as the reference standard. We calculated sensitivity and specificity and 95% confidence intervals (CIs) for the detection of acute and chronic MI for MRI and for MIA (MRI, CT, and biopsies). CIs for sensitivity and specificity are Clopper–Pearson CIs.19 Inter-observer agreement was calculated using kappa statistics. Calculations were performed using IBM® SPSS® Statistics version 21. Receiver operating characteristics (ROC) curves We calculated ROC curves to investigate diagnostic value of CT (total Agatston calcium score), MRI, and biopsies for diagnosing acute and chronic MI. In the analysis, the diagnostic value (sensitivity and specificity and ROC curves) of biopsies was combined with MRI, because biopsies were taken from radiologically suspect areas, identified at MRI. Results Case recruitment From January 2012 to December 2014, 100 consecutive cases (62 men and 38 women) were included in the study. One case was excluded because autopsy findings warranted a forensic autopsy. The mean interval between death and start of imaging was 23.2 ± 15.6 h (range 3.2–71.6). Mean age at the time of death was 62.5 years (range 25–92). In the group with clinical suspicion of ischaemic heart disease, 14/30 (46%) patients were admitted to the hospital with out-of-hospital cardiac arrest vs. 3/69 (4%) in the group without clinical suspicion of cardiac death. Agreement between clinical suspicion and autopsy findings In the group with a clinical suspicion of ischaemic heart disease as the cause of death, acute MI was found in 16/30 (53.3%) of cases by conventional autopsy. In the group without clinical suspicion of ischaemic heart disease as the cause of death, acute MI was found in 18/69 (26.1%) cases. Twenty-two of the 34 cases with acute MI found with conventional autopsy had no known ischaemic heart disease during life. Twenty cases had clinically known ischaemic heart disease during life (at least one ischaemic episode during life, determined by clinicians). MI, either chronic or acute, was confirmed by conventional autopsy and MIA in 16/20 of these cases (the same cases were identified with MIA and conventional autopsy). Diagnostic performance Acute MI was found in 34/99 cases on conventional autopsy and 36/99 cases on MIA. Chronic MI was found in 40/99 cases on conventional autopsy and 51/99 cases on MIA. The diagnostic accuracy of MRI and biopsies for the detection of acute and chronic MI is shown in Table 4 and for CT (Total Agatston calcium score) in Table 5, and the ROC curves for acute and chronic MI are shown in Figures 1 and 2, respectively. Table 4 Diagnostic accuracy of MRI and MIA   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; MIA, minimally invasive autopsy; MRI, magnetic resonance imaging; n, number; TN, true negatives; TP, true positives. Table 4 Diagnostic accuracy of MRI and MIA   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; MIA, minimally invasive autopsy; MRI, magnetic resonance imaging; n, number; TN, true negatives; TP, true positives. Table 5 Diagnostic accuracy of CT (total Agatston calcium score)   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; n, number; TN, true negatives; TP, true positives. Table 5 Diagnostic accuracy of CT (total Agatston calcium score)   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; n, number; TN, true negatives; TP, true positives. Figure 1 View largeDownload slide ROC curves for CT, MRI, and biopsy combined with MRI for the detection of acute myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 1 View largeDownload slide ROC curves for CT, MRI, and biopsy combined with MRI for the detection of acute myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 2 View largeDownload slide ROC curve for CT, MRI, and biopsy combined with MRI for the detection of chronic myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 2 View largeDownload slide ROC curve for CT, MRI, and biopsy combined with MRI for the detection of chronic myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Acute MI Sensitivity of MRI for acute MI was 0.50 (95% CI 0.32–0.68) and specificity was 0.92 (0.83–0.97). Sensitivity of MIA for acute MI was 0.97 (95% CI 0.85–1.00) and specificity was 0.95 (0.87–0.99). Figure 3 shows a case with acute MI. Figure 3 View largeDownload slide A 45-year-old man who died shortly after requiring cardiopulmonary resuscitation following a period of acute chest pain during sports. T2w MRI short axis (A–C) and axial (D) views show diffuse hypointense signal (asterisks) on LAD territory (the entire septum, and anterior and posterior apical wall). T1w MRI did not show any abnormalities. Macroscopy (E) appeared normal. These areas were biopsied and microscopy (haematoxylin and eosin stain) shows contraction band necrosis (arrow heads) (F) and infiltration of granulocytes (arrow) (G) confirming acute myocardial infarction. Figure 3 View largeDownload slide A 45-year-old man who died shortly after requiring cardiopulmonary resuscitation following a period of acute chest pain during sports. T2w MRI short axis (A–C) and axial (D) views show diffuse hypointense signal (asterisks) on LAD territory (the entire septum, and anterior and posterior apical wall). T1w MRI did not show any abnormalities. Macroscopy (E) appeared normal. These areas were biopsied and microscopy (haematoxylin and eosin stain) shows contraction band necrosis (arrow heads) (F) and infiltration of granulocytes (arrow) (G) confirming acute myocardial infarction. The area under curve (AUC) for the detection of acute MI was 0.60 (95% CI 0.48–0.72) for CT (total Agatston calcium score), 0.71 (95% CI 0.60–0.83) for MRI, and 0.96 (95% CI 0.92–1.00) for MRI with biopsy (MIA). Five cases were classified as peracute infarction on MRI, in two of these cases, no evidence for an infarction was found on conventional autopsy. Chronic MI Sensitivity of MRI for chronic MI was 0.35 (95% CI 0.21–0.52) and specificity was 1.00 (95% CI 0.94–1.00). Sensitivity of MIA for chronic MI was 0.90 (95% CI 0.76–0.97) and specificity was 0.75 (95% CI 0.62–0.85). Figures 4 and 5 are examples of cases with chronic MI. Figure 4 View largeDownload slide A 66-year-old man with a medical history of ischaemic heart disease, who died after a period of dyspnoea and anaemia. MRI STIR FSE T2w MRI (A) axial and (B) short axis showing septal mid- and apical-wall thinning of the left ventricle with hyperintense T2 signal (black arrows), indicative of a chronic infarction with subacute infarction overlying. (C and D) Mid-ventricular slice stained with LDH, showing discolouration of the corresponding areas of the myocardium (white arrows) indicative of acute infarction. (E and F) haematoxylin and eosin staining of CT-guided biopsy-taken from the suspected area in the interventricular septum showing uninflamed replacement fibrosis with viable adjacent myocardium (asterisk). Figure 4 View largeDownload slide A 66-year-old man with a medical history of ischaemic heart disease, who died after a period of dyspnoea and anaemia. MRI STIR FSE T2w MRI (A) axial and (B) short axis showing septal mid- and apical-wall thinning of the left ventricle with hyperintense T2 signal (black arrows), indicative of a chronic infarction with subacute infarction overlying. (C and D) Mid-ventricular slice stained with LDH, showing discolouration of the corresponding areas of the myocardium (white arrows) indicative of acute infarction. (E and F) haematoxylin and eosin staining of CT-guided biopsy-taken from the suspected area in the interventricular septum showing uninflamed replacement fibrosis with viable adjacent myocardium (asterisk). Figure 5 View largeDownload slide A 54-year-old man with no medical history of ischaemic heart disease. CT curved multiplanar reconstruction of the right coronary artery (A), left anterior descending artery (B), and left circumflex artery (C), showing severely calcified coronary arteries (total Agatston score: 409). T2w MRI (D) diffuse T2 hyperintense signal indicative of myocardial oedema as a sign of acute myocardial infarction. These areas were biopsied and microscopy (haematoxylin and eosin) shows contraction band necrosis (E: arrow heads) and hypereosinophilia fitting with the diagnosis of acute myocardial infarction and connective tissue (F: asterisk) within a region of chronic myocardial infarction. Figure 5 View largeDownload slide A 54-year-old man with no medical history of ischaemic heart disease. CT curved multiplanar reconstruction of the right coronary artery (A), left anterior descending artery (B), and left circumflex artery (C), showing severely calcified coronary arteries (total Agatston score: 409). T2w MRI (D) diffuse T2 hyperintense signal indicative of myocardial oedema as a sign of acute myocardial infarction. These areas were biopsied and microscopy (haematoxylin and eosin) shows contraction band necrosis (E: arrow heads) and hypereosinophilia fitting with the diagnosis of acute myocardial infarction and connective tissue (F: asterisk) within a region of chronic myocardial infarction. The AUC for the detection of chronic MI was 0.74 (95% CI 0.64–0.84) for CT (Total Agatston calcium score), 0.68 (95% CI 0.56–0.79) for MRI and 0.82 (95% CI 0.74–0.91) for MRI with biopsy (MIA). Inter-observer agreement MRI The two radiologists were in agreement in 82 of 99 cases, in the remaining 17 cases consensus was reached in joint sessions. The kappa score of inter-observer agreement was 0.85. Discussion In this study, we investigated the diagnostic accuracy of an MIA consisting of MRI, CT, and CT-guided biopsy for detection of ischaemic heart disease in a hospital setting. We found that the combination of MRI and biopsies had the highest accuracy for detecting acute and chronic MI with conventional autopsy as reference standard. MRI without biopsies showed a high specificity, but low sensitivity for acute and chronic MI. High CT Agatston calcium score (>400) was a good predictor for chronic MI, but not for acute MI. We found a lower sensitivity of MRI as a stand-alone test for acute MI (0.50) compared with other studies investigating MRI. Ruder et al.20 reported that with MRI acute MI (within 3 h after onset) could be detected in ex vivo porcine hearts in which they correctly detected acute infarctions in all 21 cases. Forensic studies showed that with MRI acute and chronic MI (up to 100% sensitivity) could be accurately diagnosed in human subjects. Importantly, MRI could diagnose peracute MI (onset within 3 h) in cases not yet showing histological changes, but with a matching coronary stenosis at conventional autopsy.12,13,17 The differences in sensitivity and specificity among studies can be explained by the differences in studied population and clinical setting; most are forensic studies that investigated subjects who died under the suspicion of an out-of-hospital cardiac-arrest and as such had a high pre-test probability. Also, these studies often involve high-resolution cardiac imaging at 3T scanners using surface coils and relatively long scan time for imaging only the heart (∼1 h).16 Conversely, we scanned in a hospital setting and performed total-body imaging to diagnose both cardiac and non-cardiac cause of death. So as not to interfere with the patient workflow at the MR scanner, we were restricted to 1-h scan time for imaging the entire body. The addition of biopsies to MRI increased the sensitivity substantially. This highlights the importance of extensive sampling, even when no changes are visible yet on MRI. The big difference between sensitivity of MRI and MRI combined with biopsies can be explained by the quantity of sampling. From each biopsy location, at least five samples were taken, e.g. from the lateral wall, also the mid- and posterior-segments were biopsied. Furthermore, in those cases where there was a clinical suspicion of myocardial ischaemia and the MRI showed no signal abnormalities, extra biopsies were taken from the septum, anterior, and posterior wall (both mid- and posterior-segments). The non-invasive approach (CT and/or MRI) is less expensive than the minimally invasive approach (imaging plus biopsy). CT is now widely used as a stand-alone modality because of its high accessibility, short examination time, and robust performance. CT can provide better mortality statistics than the cause of death determined by the clinician, and is useful for excluding certain diagnoses. However, for diagnosing acute MI, our results show that the diagnostic accuracy of CT as stand-alone test is insufficient.21,22 To improve CT performance, in particular for ischaemic heart disease, more recent studies report on the diagnostic value of CT angiography. Grabherr et al. extensively performed feasibility studies on CTA using different contrast agents and perfusion techniques. There are different CTA approaches; it can be targeted at the coronary arteries by selective placement of the catheter at the level of the coronary ostia, or total-body CTA can be performed including multiphase scanning.9,23–25 Wichman et al. applied total-body CTA in 50 ICU patients who died unexpectedly or within 48 h of an event requiring resuscitation and found that CTA confirmed 93% of the clinical diagnoses, and autopsy confirmed 80%. In addition, CTA and CA identified 16 new major and 238 new minor diagnoses. They concluded that in cases of unexpected death CTA was a valuable addition to autopsy. Rutty et al. performed total-body CTA in 210 cases of natural and non-suspicious unnatural death and found that CTA established a cause of death in 92% of cases. The number of discrepancies with the final cause of death was not significantly different between autopsy and CTA, suggesting that total-body CTA is a feasible alternative to autopsy.26,27 A drawback of CTA is that it requires specific training, technical equipment, and contrast agents and is time-consuming due to extensive preparation of the body (e.g. intra-arterial and/or intravenous femoral access for catheter placement) prior to scanning. At the time of the study, the equipment was not available and professional expertise and scanner availability to perform CTA was lacking. Another non-invasive approach is the use of stand-alone MRI. Diffusion tensor imaging (DTI) is showing promising results in diagnosing myocardial ischaemia in situ, correctly predicting MI (either acute or chronic) with an accuracy of 0.73, using fractional anisotropy and mean diffusivity.28 First studies show that quantitative MRI can detect and differentiate between early and following stages of myocardial ischaemia based on T1, T2, and proton density values.29–31 They concluded that temperature-corrected quantitative MRI can diagnose early acute, acute and chronic MI, but histological confirmation is required. In the hospital setting, an important part of every postmortem examination should be a thorough evaluation of the medical history and clinical circumstances prior to death. The patients with known obstructive coronary artery disease, chronic, and acute MI was confirmed in 16 out of 20 cases by postmortem examination (both by MIA and conventional autopsy). The conventional autopsy found 22 new cases of acute MI highlighting the lasting need for postmortem examinations. Today the use of postmortem imaging, mostly CT, is widely accepted as adjunct to the medicolegal autopsy.7,26,27,32,33 In the hospital setting, there is a growing interest in postmortem imaging, however, expertise and logistics (e.g. access to scanners) are still important limitations. Conclusion We evaluated the diagnostic accuracy of MIA for the detection of ischaemic heart disease in a hospital setting. We found that the combination of MRI with biopsies had high sensitivity and specificity for the detection of acute and chronic MI. Funding The following funders supported this work: Erasmus MC health care efficiency, 2010-10112; Stichting Coolsingel (NL), 255; Erasmus MC vriendenfonds, 104117. Conflict of interest: None declared. References 1 World Health Organization. The Top 10 Causes of Death. 2017. http://www.who.int/mediacentre/factsheets/fs310/en/ (25 January 2018, date last accessed). 2 Shojania KG, Burton EC, McDonald KM, Goldman L. Changes in rates of autopsy-detected diagnostic errors over time: a systematic review. JAMA  2003; 289: 2849– 56. Google Scholar CrossRef Search ADS PubMed  3 van den Tweel JG, Wittekind C. The medical autopsy as quality assurance tool in clinical medicine: dreams and realities. Virchows Arch  2016; 468: 75– 81. Google Scholar CrossRef Search ADS PubMed  4 Burton JL, Underwood J. Clinical, educational, and epidemiological value of autopsy. Lancet  2007; 369: 1471– 80. Google Scholar CrossRef Search ADS PubMed  5 Turnbull A, Osborn M, Nicholas N. Hospital autopsy: endangered or extinct? J Clin Pathol  2015; 68: 601– 4. Google Scholar CrossRef Search ADS PubMed  6 Blokker BM, Wagensveld IM, Weustink AC, Oosterhuis JW, Hunink MG. Non-invasive or minimally invasive autopsy compared to conventional autopsy of suspected natural deaths in adults: a systematic review. Eur Radiol  2016; 26: 1159– 79. Google Scholar CrossRef Search ADS PubMed  7 Eriksson A, Gustafsson T, Hoistad M, Hultcrantz M, Jacobson S, Mejare I et al.   Diagnostic accuracy of postmortem imaging vs autopsy-A systematic review. Eur J Radiol  2017; 89: 249– 69. Google Scholar CrossRef Search ADS PubMed  8 Grabherr S, Djonov V, Yen K, Thali MJ, Dirnhofer R. Postmortem angiography: review of former and current methods. AJR Am J Roentgenol  2007; 188: 832– 8. Google Scholar CrossRef Search ADS PubMed  9 Turillazzi E, Frati P, Pascale N, Pomara C, Grilli G, Viola RV et al.   Multi-phase post-mortem CT-angiography: a pathologic correlation study on cardiovascular sudden death. J Geriatr Cardiol  2016; 13: 855– 65. Google Scholar PubMed  10 Grabherr S, Grimm J, Dominguez A, Vanhaebost J, Mangin P. Advances in post-mortem CT-angiography. Br J Radiol  2014; 87: 20130488. Google Scholar CrossRef Search ADS PubMed  11 Grabherr S, Grimm J, Baumann P, Mangin P. Application of contrast media in post-mortem imaging (CT and MRI). Radiol Med  2015; 120: 824– 34. Google Scholar CrossRef Search ADS PubMed  12 Jackowski C, Christe A, Sonnenschein M, Aghayev E, Thali MJ. Postmortem unenhanced magnetic resonance imaging of myocardial infarction in correlation to histological infarction age characterization. Eur Heart J  2006; 27: 2459– 67. Google Scholar CrossRef Search ADS PubMed  13 Jackowski C. Special issue on postmortem imaging 2013. Forensic Sci Int  2013; 225: 1– 2. Google Scholar CrossRef Search ADS PubMed  14 Jackowski C, Sonnenschein M, Thali MJ, Aghayev E, von Allmen G, Yen K et al.   Virtopsy: postmortem minimally invasive angiography using cross section techniques–implementation and preliminary results. J Forensic Sci  2005; 50: 1175– 86. Google Scholar CrossRef Search ADS PubMed  15 Michaud K, Grabherr S, Jackowski C, Bollmann MD, Doenz F, Mangin P. Postmortem imaging of sudden cardiac death. Int J Legal Med  2014; 128: 127– 37. Google Scholar CrossRef Search ADS PubMed  16 Jackowski C, Schwendener N, Grabherr S, Persson A. Post-mortem cardiac 3-T magnetic resonance imaging: visualization of sudden cardiac death? J Am Coll Cardiol  2013; 62: 617– 29. Google Scholar CrossRef Search ADS PubMed  17 Jackowski C, Warntjes MJ, Berge J, Bar W, Persson A. Magnetic resonance imaging goes postmortem: noninvasive detection and assessment of myocardial infarction by postmortem MRI. Eur Radiol  2011; 21: 70– 8. Google Scholar CrossRef Search ADS PubMed  18 Basso C, Aguilera B, Banner J, Cohle S, D'Amati G, de Gouveia RH et al.   Guidelines for autopsy investigation of sudden cardiac death: 2017 update from the Association for European Cardiovascular Pathology. Virchows Arch  2017; 471: 691– 705. Google Scholar CrossRef Search ADS PubMed  19 Ross D. Accurate confidence intervals for binomial proportion and Poisson rate estimation. Comput Biol Med  2003; 33: 509– 31. Google Scholar CrossRef Search ADS PubMed  20 Ruder TD, Ebert LC, Khattab AA, Rieben R, Thali MJ, Kamat P. Edema is a sign of early acute myocardial infarction on post-mortem magnetic resonance imaging. Forensic Sci Med Pathol  2013; 9: 501– 5. Google Scholar CrossRef Search ADS PubMed  21 Oyake Y, Aoki T, Shiotani S, Kohno M, Ohashi N, Akutsu H et al.   Postmortem computed tomography for detecting causes of sudden death in infants and children: retrospective review of cases. Radiat Med  2006; 24: 493– 502. Google Scholar CrossRef Search ADS PubMed  22 Inai K, Noriki S, Kinoshita K, Sakai T, Kimura H, Nishijima A et al.   Postmortem CT is more accurate than clinical diagnosis for identifying the immediate cause of death in hospitalized patients: a prospective autopsy-based study. Virchows Arch  2016; 469: 101– 9. Google Scholar CrossRef Search ADS PubMed  23 Roberts IS, Benamore RE, Peebles C, Roobottom C, Traill ZC. Technical report: diagnosis of coronary artery disease using minimally invasive autopsy: evaluation of a novel method of post-mortem coronary CT angiography. Clin Radiol  2011; 66: 645– 50. Google Scholar CrossRef Search ADS PubMed  24 Saunders SL, Morgan B, Raj V, Robinson CE, Rutty GN. Targeted post-mortem computed tomography cardiac angiography: proof of concept. Int J Legal Med  2011; 125: 609– 16. Google Scholar CrossRef Search ADS PubMed  25 Rutty G, Saunders S, Morgan B, Raj V. Targeted cardiac post-mortem computed tomography angiography: a pictorial review. Forensic Sci Med Pathol  2012; 8: 40– 7. Google Scholar CrossRef Search ADS PubMed  26 Guy N, Rutty BM,C, Robinson V, Raj M, Pakkal J, Amoroso T, Visser S et al.   Diagnostic accuracy of post-mortem CT with targeted coronary angiography versus autopsy for coroner-requested post-mortem investigations: a prospective, masked, comparison study. Lancet  2017; 390: 145– 54. Google Scholar CrossRef Search ADS PubMed  27 Wichmann D, Heinemann A, Weinberg C, Vogel H, Hoepker WW, Grabherr S et al.   Virtual autopsy with multiphase postmortem computed tomographic angiography versus traditional medical autopsy to investigate unexpected deaths of hospitalized patients. Ann Intern Med  2014; 160: 534– 41. Google Scholar CrossRef Search ADS PubMed  28 Winklhofer S, Stoeck CT, Berger N, Thali M, Manka R, Kozerke S et al.   Post-mortem cardiac diffusion tensor imaging: detection of myocardial infarction and remodeling of myofiber architecture. Eur Radiol  2014; 24: 2810– 8. Google Scholar CrossRef Search ADS PubMed  29 Schwendener N, Jackowski C, Persson A, Warntjes MJ, Schuster F, Riva F et al.   Detection and differentiation of early acute and following age stages of myocardial infarction with quantitative post-mortem cardiac 1.5T MR. Forensic Sci Int  2017; 270: 248– 54. Google Scholar CrossRef Search ADS PubMed  30 Zech WD, Schwendener N, Persson A, Warntjes MJ, Jackowski C. Postmortem MR quantification of the heart for characterization and differentiation of ischaemic myocardial lesions. Eur Radiol  2015; 25: 2067– 73. Google Scholar CrossRef Search ADS PubMed  31 Persson A, Baeckmann J, Berge J, Jackowski C, Warntjes M, Zech WD. Temperature-corrected postmortem 3-T MR quantification of histopathological early acute and chronic myocardial infarction: a feasibility study. Int J Legal Med  2017; https://doi.org/10.1007/s00414-017-1614-6. 32 Dirnhofer R, Jackowski C, Vock P, Potter K, Thali MJ. VIRTOPSY: minimally invasive, imaging-guided virtual autopsy. Radiographics  2006; 26: 1305– 33. Google Scholar CrossRef Search ADS PubMed  33 Flach PM, Thali MJ, Germerott T. Times have changed! Forensic radiology–a new challenge for radiology and forensic pathology. AJR Am J Roentgenol  2014; 202: W325– 34. Google Scholar CrossRef Search ADS PubMed  Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com. http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Heart Journal – Cardiovascular Imaging Oxford University Press

Diagnostic accuracy of postmortem computed tomography, magnetic resonance imaging, and computed tomography-guided biopsies for the detection of ischaemic heart disease in a hospital setting

Loading next page...
 
/lp/ou_press/diagnostic-accuracy-of-postmortem-computed-tomography-magnetic-bgcGSmuljR
Publisher
Oxford University Press
Copyright
Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com.
ISSN
2047-2404
D.O.I.
10.1093/ehjci/jey015
Publisher site
See Article on Publisher Site

Abstract

Abstract Aims The autopsy rate worldwide is alarmingly low (0–15%). Mortality statistics are important, and it is, therefore, essential to perform autopsies in a sufficient proportion of deaths. The imaging autopsy, non-invasive, or minimally invasive autopsy (MIA) can be used as an alternative to the conventional autopsy in an attempt to improve postmortem diagnostics by increasing the number of postmortem procedures. The aim of this study was to determine the diagnostic accuracy of postmortem magnetic resonance imaging (MRI), computed tomography (CT), and CT-guided biopsy for the detection of acute and chronic myocardial ischaemia. Methods and results We included 100 consecutive adult patients who died in hospital, and for whom next-of-kin gave permission to perform both conventional autopsy and MIA. The MIA consists of unenhanced total-body MRI and CT followed by CT-guided biopsies. Conventional autopsy was used as reference standard. We calculated sensitivity and specificity and receiver operating characteristics curves for CT and MRI as the stand-alone test or combined with biopsy for detection of acute and chronic myocardial infarction (MI). Sensitivity and specificity of MRI with biopsies for acute MI was 0.97 and 0.95, respectively and 0.90 and 0.75, respectively for chronic MI. MRI without biopsies showed a high specificity (acute: 0.92; chronic: 1.00), but low sensitivity (acute: 0.50; chronic: 0.35). CT (total Agatston calcium score) had a good diagnostic value for chronic MI [area under curve (AUC) 0.74, 95% confidence interval (CI) 0.64–0.84], but not for acute MI (AUC 0.60, 95% CI 0.48–0.72). Conclusion We found that the combination of MRI with biopsies had high sensitivity and specificity for the detection of acute and chronic myocardial ischaemia. postmortem , magnetic resonance imaging , computed tomography , autopsy , ischaemic heart disease Introduction According to the World Health Organization, cardiovascular disease and stroke are the foremost contributors to worldwide mortality with ischaemic heart disease globally causing almost nine million deaths per year in 2015.1 Accurate mortality statistics are important for both policy- and decision-making regarding healthcare funding. For reliable statistics, it is essential to perform autopsies in a sufficient proportion of deaths, both in and out of hospital. Despite available modern diagnostic tests, the conventional autopsy still reveals unexpected findings related to the cause of death in 8.4–24.4% and findings that would have affected patient outcome (Class I errors) in 4.1–6.7% of cases.2–4 Unfortunately, today’s autopsy rate has dropped to alarmingly low percentages worldwide (0–15%), both for academic and non-academic hospitals.5 In the late 90s, the imaging autopsy was introduced as an alternative to conventional autopsy as a stimulus to postmortem diagnostics. Since then, a growing number of studies have evaluated the diagnostic value of postmortem computed tomography (CT) and magnetic resonance imaging (MRI) with or without image-guided biopsies. Two review articles concluded that the imaging autopsy, using a non-invasive or minimally invasive approach, can potentially serve as an alternative to conventional autopsy, but more extensive research in different settings is needed to validate these new autopsy methods.6,7 Imaging protocols designed for the living patient differ from postmortem imaging protocols, in particular for cardiac imaging. For example, wall movement abnormalities of the heart cannot be diagnosed. In living patients, contrast-enhanced imaging, either non-invasive or invasive, is the gold standard for diagnosing ischaemic heart disease. Postmortem angiographic studies are feasible and not new; since the discovery of X-rays, angiography of organs and tissues has been used as an adjunct to the autopsy procedure.8 More recently, postmortem total-body angiography, using CT or MRI, has become technically feasible, and there are a growing number of studies investigating its diagnostic value. Preliminary results are promising, especially for establishing ischaemic heart disease as the cause of death.9–11 Interestingly postmortem MRI without the use of contrast agents also shows a sufficient accuracy for detecting both acute and chronic myocardial infarction (MI). The presence and age of MI can be diagnosed by evaluating the signal changes related to morphological alterations in the infarcted myocardium, such as the presence of myocardial oedema, fibrosis, or fat.12–15 Non-enhanced cardiac CT is also useful for detecting coronary artery calcifications. In this study, we evaluate the minimally invasive autopsy (MIA) approach using both non-enhanced CT and MRI followed by CT-guided biopsies. The aim of this study was to determine the diagnostic accuracy of MRI, CT, and CT-guided biopsy for the detection of acute and chronic MI, with conventional autopsy as the reference standard. Methods Study design This study was undertaken as part of the MIA study; this is a prospective single centre cross-sectional study in a tertiary referral hospital comparing diagnostic performance of conventional autopsy and MIA. Conventional autopsy followed MIA on the same case. Pathologists involved with conventional autopsy were blinded to MIA findings as much as possible; however, biopsy sites could potentially lead to increased suspicion of the biopsied organs and tissues by the autopsy pathologist. MIA personnel were blinded to autopsy findings. Patients From January 2012 to December 2014 all hospitalized patients aged 18 years and older who died at Erasmus University Medical Center were eligible, if written informed consent was obtained from next-of-kin for MIA and conventional autopsy of at least the torso. Exclusion criteria were (suspected) unnatural cause of death, body size exceeding height of 16 inches in supine position (limitation for MRI), known or suspected ‘high-risk’ infected bodies (tuberculosis, hepatitis B and C, human immunodeficiency virus, methicillin-resistant Staphylococcus aureus, and multi-drug resistant Acinetobacter), and open abdominal wounds that could not be completely closed or taped to prevent leakage of body fluids. Clinical information All relevant clinical information including medical history and suspected cause of death was recorded and available for both the MIA and the conventional autopsy team. The treating physician decided the most likely cause of death and a differential diagnosis based on the clinical presentation. For the analysis, the population was divided into a group with and a group without clinical suspicion for ischaemic heart disease. MIA procedure MRI and CT scans were made according to standardized protocols (Tables 1 and 2, respectively). Total acquisition time was 1 h for MRI and around 5 min for CT. One radiologist (A.C.W.) with expertise in postmortem radiology, performed the initial read of the MRI and CT scans, compared these to the available premortem imaging, and marked suspected pathological lesions on CT and MRI key images that were used to plan the biopsies. Table 1 Postmortem magnetic resonance protocol Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  All scans were made in the axial orientation. FLAIR, fluid attenuated inversion recovery; FOV, field of view; fs FSPGR, fat-saturated fast spoiled gradient echo; FSE, fast spin echo; STIR, short tau inversion recovery; TE, echo time; TI, inversion time; TR, repetition time. Table 1 Postmortem magnetic resonance protocol Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  Scan area  Coil  Sequence  TR/TE/TI (ms)  Slice width (mm)  FOV (cm)  Matrix  Number of slices  Coverage per section (cm)  Number of sections  Scan time per section (s)  Head–pelvis  Body  FLAIR FSE T1w  2320/9.5/963  4.0/no gap  48 × 48  384 × 320  50  20.0  5–8  174  Head–pelvis  Body  STIR FSE T2w  12 000/41/120  4.0/no gap  48 × 48  288 × 224  50  20.0  5–8  168  Thorax  8-channel torso array  3D fs FSPGR T1w  3.3/1.2/14  1.6  40 × 40  256 × 256  212  33.9  1  153  Thorax  8-channel torso array  2D STIR FSE T2w  11 200/94/120  2.0/no gap  40 × 40  256 × 256  170  34.0  1  359  All scans were made in the axial orientation. FLAIR, fluid attenuated inversion recovery; FOV, field of view; fs FSPGR, fat-saturated fast spoiled gradient echo; FSE, fast spin echo; STIR, short tau inversion recovery; TE, echo time; TI, inversion time; TR, repetition time. Table 2 Postmortem computed tomography protocol Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Table 2 Postmortem computed tomography protocol Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  Scan area  Rotation time (s)  Tube voltage (kV)  Tube current (eff. mAs)  Slice collimation (mm)  Pitch  Scan time (s)  Reconstruction  Head–neck  1.0  100  750  2 × 64 × 0.6  0.35  21  Filtered back-projection  Thorax–pelvis  1.0  120  600  2 × 64 × 0.6  0.6  32  Filtered back-projection  Pelvis–lower extremities  1.0  120  600  2 × 64 × 0.6  0.6  57  Filtered back-projection  MRI was performed on a 1.5T scanner (Discovery MR450, GE Medical systems, Milwaukee, WI, USA) and consisted of scans of the brain, neck, thorax, abdomen, and pelvis. The MRI total-body protocol consisted of axially-acquired short tau inversion recovery (STIR) fast spin echo (FSE) T2w and fluid attenuated inversion recovery (FLAIR) FSE T1w from the cranium to the pelvis. An additional 2D STIR FSE T2w scan and 3D Fatsat FSPGR T1w scan with higher resolution than the total-body scans were acquired of the thorax, using an 8-channel torso array coil. All MRI scans were made in the axial orientation. After MRI was completed CT scans were acquired from head to feet (Somatom Definition, Siemens Healthcare, Forchheim, Germany). CT datasets of the head, thorax, and abdomen were reconstructed with section thickness of 1.0 mm and 5.0 mm in the axial plane and 3.0 mm in the coronal and sagittal planes, by using medium-to-smooth (H31/B31) and very sharp (H70/B70) convolution kernels. CT-guided biopsies (12 Gauge) were taken from heart, lungs, liver, kidneys, spleen, and radiologically suspected pathology as indicated. In the heart, standard biopsies (5–10 samples) were taken from the lateral wall (mid- and basal-parts) and an apex of the left ventricle. Additional biopsies were taken from MRI signal abnormalities within the myocardium. In those cases where there was a clinical suspicion of MI and the MRI showed no signal abnormalities additional biopsies were taken from the septum, anterior, and posterior wall. The MIA pathologist (J.W.O.) and researcher (B.M.B.) examined the microscopic slides of the biopsies; when in doubt, they consulted pathologists with specific expertise, not involved in the matching conventional autopsy, to reach a conclusion. Cardiac imaging evaluation CT For each case, CT calcium score was calculated by one observer (I.M.W.) using dedicated software (Syngo.via 3.0 Calcium Scoring®, Siemens Healthcare, Forchheim, Germany) and expressed as total Agatston scores. MRI MR images were reconstructed and evaluated in the short-axis view. Two radiologists with expertise in cardiac radiology (A.C.W. and A.P.P.) independently evaluated MR images and in case of disagreement, consensus was reached in joint sessions. MIs were classified according to a modified classification by Jackowski et al.12,16,17 Peracute infarction (within 6 h after onset) is characterized by T2 hypointense signal in the necrotic centre, caused by a state of hypoperfusion. In the acute phase (within 6 h–1 week after onset), the marginal areas become oedematous and show T2 hyperintense signal, T1 signal in the centre is isointense, and the oedematous marginal regions can show T1 hypointense signal. Subacute infarction (>1 week after onset) shows T2 hyperintense signal in the infarcted area when the area becomes reperfused, while the marginal areas show normal T1 and T2 signal. Chronic infarction (>2 months after onset) shows wall thinning and scar tissue reflected by T1 and T2 hypointense signal and foci of T1 hyperintense signal can be seen due to fatty infiltration. In our analysis, peracute and acute infarctions were grouped into one category and defined as acute MI (<1-week old infarction). Subacute and chronic infarctions were grouped into one category and defined as chronic MI (>1-week old infarction). MRI criteria for determining infarction age are detailed in Table 3. Table 3 MRI criteria for determining infarction agea   Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =    Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =  a Criteria based on Jackowski et al.12,16,17 Table 3 MRI criteria for determining infarction agea   Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =    Necrotic centre  Marginal regions          T1  T2  T1  T2  Acute   Peracute (<6 h)  =  ↓  =  =   Acute (6 h–1 week)  =/↑  ↓  ↓  ↑  Chronic           Subacute (1 week–2 months)  =  ↑  =  =   Chronic (>2 months)  ↓/↑ (fat)  ↓  =  =  a Criteria based on Jackowski et al.12,16,17 Conventional autopsy The day after MIA, a resident in pathology, supervised by the attending pathologist, performed conventional autopsy according to the departmental protocol. The autopsy report included medical history, postmortem diagnoses, a presumed cause of death, and answers to specific clinical questions, and was authorized by the pathologist. Macroscopic evaluation consisted of sectioning of the heart in slices of 0.5 to 1 cm and visual inspection of the myocardium. Lactate dehydrogenase (LDH) staining was performed on a mid-ventricular slice. Haematoxylin and eosin were used for histological staining.18 When there was a discrepancy between the histology of the MIA and conventional autopsy, a pathologist with expertise in cardiac pathology (JHT) reviewed the histology. The following criteria for MI age were used (separately or in combination): acute MI: hypereosinophilia and loss of cross striation within myocardial fibres, contraction band necrosis, coagulation necrosis with or without granulocyte infiltration or haemorrhage, and various degrees of nuclear pyknosis, karyolysis, granulocyte infiltration, and myocardial oedema; chronic MI: fibroblasts with loose connective tissue formation, angiogenesis (subacute), and paucicellular collagenous fibrosis (chronic).12 Statistical analysis Analyses were performed on the patient level. Conventional autopsy was used as the reference standard. We calculated sensitivity and specificity and 95% confidence intervals (CIs) for the detection of acute and chronic MI for MRI and for MIA (MRI, CT, and biopsies). CIs for sensitivity and specificity are Clopper–Pearson CIs.19 Inter-observer agreement was calculated using kappa statistics. Calculations were performed using IBM® SPSS® Statistics version 21. Receiver operating characteristics (ROC) curves We calculated ROC curves to investigate diagnostic value of CT (total Agatston calcium score), MRI, and biopsies for diagnosing acute and chronic MI. In the analysis, the diagnostic value (sensitivity and specificity and ROC curves) of biopsies was combined with MRI, because biopsies were taken from radiologically suspect areas, identified at MRI. Results Case recruitment From January 2012 to December 2014, 100 consecutive cases (62 men and 38 women) were included in the study. One case was excluded because autopsy findings warranted a forensic autopsy. The mean interval between death and start of imaging was 23.2 ± 15.6 h (range 3.2–71.6). Mean age at the time of death was 62.5 years (range 25–92). In the group with clinical suspicion of ischaemic heart disease, 14/30 (46%) patients were admitted to the hospital with out-of-hospital cardiac arrest vs. 3/69 (4%) in the group without clinical suspicion of cardiac death. Agreement between clinical suspicion and autopsy findings In the group with a clinical suspicion of ischaemic heart disease as the cause of death, acute MI was found in 16/30 (53.3%) of cases by conventional autopsy. In the group without clinical suspicion of ischaemic heart disease as the cause of death, acute MI was found in 18/69 (26.1%) cases. Twenty-two of the 34 cases with acute MI found with conventional autopsy had no known ischaemic heart disease during life. Twenty cases had clinically known ischaemic heart disease during life (at least one ischaemic episode during life, determined by clinicians). MI, either chronic or acute, was confirmed by conventional autopsy and MIA in 16/20 of these cases (the same cases were identified with MIA and conventional autopsy). Diagnostic performance Acute MI was found in 34/99 cases on conventional autopsy and 36/99 cases on MIA. Chronic MI was found in 40/99 cases on conventional autopsy and 51/99 cases on MIA. The diagnostic accuracy of MRI and biopsies for the detection of acute and chronic MI is shown in Table 4 and for CT (Total Agatston calcium score) in Table 5, and the ROC curves for acute and chronic MI are shown in Figures 1 and 2, respectively. Table 4 Diagnostic accuracy of MRI and MIA   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; MIA, minimally invasive autopsy; MRI, magnetic resonance imaging; n, number; TN, true negatives; TP, true positives. Table 4 Diagnostic accuracy of MRI and MIA   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  MRI                     Acute MI  34  99  17  60  5  17  0.46  50 (32–68)  92 (83–97)   Chronic MI  40  99  14  59  0  26  0.39  35 (21–52)  100 (94–100)  MIA                     Acute MI  34  99  33  62  3  1  0.91  97 (85–100)  95 (87–99)   Chronic MI  40  99  36  44  15  4  0.62  90 (76–97)  75 (62–85)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; MIA, minimally invasive autopsy; MRI, magnetic resonance imaging; n, number; TN, true negatives; TP, true positives. Table 5 Diagnostic accuracy of CT (total Agatston calcium score)   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; n, number; TN, true negatives; TP, true positives. Table 5 Diagnostic accuracy of CT (total Agatston calcium score)   Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)    Prevalence of disease (%)  n  TP  TN  FP  FN  Κ  Sensitivity (%)  Specificity (%)  Calcium score >0                     Acute MI  34  99  29  13  52  5  0.04  85 (68–95)  20 (11–32)   Chronic MI  40  99  38  16  43  2  0.19  95 (83–99)  27 (16–40)  Calcium score >100                     Acute MI  34  99  24  30  35  10  0.14  71 (53–85)  46 (34–59)   Chronic MI  40  99  32  32  27  8  0.32  80 (64–91)  54 (41–67)  Calcium score >400                     Acute MI  34  99  17  41  24  17  0.13  50 (32–68)  63 (50–75)   Chronic MI  40  99  25  43  16  15  0.35  63 (46–77)  73 (59–84)  Values in parentheses represent upper and lower bound for 95% CI. FN, false negatives; FP, false positives; K, kappa measure of agreement; MI, myocardial infarction; n, number; TN, true negatives; TP, true positives. Figure 1 View largeDownload slide ROC curves for CT, MRI, and biopsy combined with MRI for the detection of acute myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 1 View largeDownload slide ROC curves for CT, MRI, and biopsy combined with MRI for the detection of acute myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 2 View largeDownload slide ROC curve for CT, MRI, and biopsy combined with MRI for the detection of chronic myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Figure 2 View largeDownload slide ROC curve for CT, MRI, and biopsy combined with MRI for the detection of chronic myocardial infarction. The table details area under the curve for the different diagnostic tests and their corresponding 95% CIs. Acute MI Sensitivity of MRI for acute MI was 0.50 (95% CI 0.32–0.68) and specificity was 0.92 (0.83–0.97). Sensitivity of MIA for acute MI was 0.97 (95% CI 0.85–1.00) and specificity was 0.95 (0.87–0.99). Figure 3 shows a case with acute MI. Figure 3 View largeDownload slide A 45-year-old man who died shortly after requiring cardiopulmonary resuscitation following a period of acute chest pain during sports. T2w MRI short axis (A–C) and axial (D) views show diffuse hypointense signal (asterisks) on LAD territory (the entire septum, and anterior and posterior apical wall). T1w MRI did not show any abnormalities. Macroscopy (E) appeared normal. These areas were biopsied and microscopy (haematoxylin and eosin stain) shows contraction band necrosis (arrow heads) (F) and infiltration of granulocytes (arrow) (G) confirming acute myocardial infarction. Figure 3 View largeDownload slide A 45-year-old man who died shortly after requiring cardiopulmonary resuscitation following a period of acute chest pain during sports. T2w MRI short axis (A–C) and axial (D) views show diffuse hypointense signal (asterisks) on LAD territory (the entire septum, and anterior and posterior apical wall). T1w MRI did not show any abnormalities. Macroscopy (E) appeared normal. These areas were biopsied and microscopy (haematoxylin and eosin stain) shows contraction band necrosis (arrow heads) (F) and infiltration of granulocytes (arrow) (G) confirming acute myocardial infarction. The area under curve (AUC) for the detection of acute MI was 0.60 (95% CI 0.48–0.72) for CT (total Agatston calcium score), 0.71 (95% CI 0.60–0.83) for MRI, and 0.96 (95% CI 0.92–1.00) for MRI with biopsy (MIA). Five cases were classified as peracute infarction on MRI, in two of these cases, no evidence for an infarction was found on conventional autopsy. Chronic MI Sensitivity of MRI for chronic MI was 0.35 (95% CI 0.21–0.52) and specificity was 1.00 (95% CI 0.94–1.00). Sensitivity of MIA for chronic MI was 0.90 (95% CI 0.76–0.97) and specificity was 0.75 (95% CI 0.62–0.85). Figures 4 and 5 are examples of cases with chronic MI. Figure 4 View largeDownload slide A 66-year-old man with a medical history of ischaemic heart disease, who died after a period of dyspnoea and anaemia. MRI STIR FSE T2w MRI (A) axial and (B) short axis showing septal mid- and apical-wall thinning of the left ventricle with hyperintense T2 signal (black arrows), indicative of a chronic infarction with subacute infarction overlying. (C and D) Mid-ventricular slice stained with LDH, showing discolouration of the corresponding areas of the myocardium (white arrows) indicative of acute infarction. (E and F) haematoxylin and eosin staining of CT-guided biopsy-taken from the suspected area in the interventricular septum showing uninflamed replacement fibrosis with viable adjacent myocardium (asterisk). Figure 4 View largeDownload slide A 66-year-old man with a medical history of ischaemic heart disease, who died after a period of dyspnoea and anaemia. MRI STIR FSE T2w MRI (A) axial and (B) short axis showing septal mid- and apical-wall thinning of the left ventricle with hyperintense T2 signal (black arrows), indicative of a chronic infarction with subacute infarction overlying. (C and D) Mid-ventricular slice stained with LDH, showing discolouration of the corresponding areas of the myocardium (white arrows) indicative of acute infarction. (E and F) haematoxylin and eosin staining of CT-guided biopsy-taken from the suspected area in the interventricular septum showing uninflamed replacement fibrosis with viable adjacent myocardium (asterisk). Figure 5 View largeDownload slide A 54-year-old man with no medical history of ischaemic heart disease. CT curved multiplanar reconstruction of the right coronary artery (A), left anterior descending artery (B), and left circumflex artery (C), showing severely calcified coronary arteries (total Agatston score: 409). T2w MRI (D) diffuse T2 hyperintense signal indicative of myocardial oedema as a sign of acute myocardial infarction. These areas were biopsied and microscopy (haematoxylin and eosin) shows contraction band necrosis (E: arrow heads) and hypereosinophilia fitting with the diagnosis of acute myocardial infarction and connective tissue (F: asterisk) within a region of chronic myocardial infarction. Figure 5 View largeDownload slide A 54-year-old man with no medical history of ischaemic heart disease. CT curved multiplanar reconstruction of the right coronary artery (A), left anterior descending artery (B), and left circumflex artery (C), showing severely calcified coronary arteries (total Agatston score: 409). T2w MRI (D) diffuse T2 hyperintense signal indicative of myocardial oedema as a sign of acute myocardial infarction. These areas were biopsied and microscopy (haematoxylin and eosin) shows contraction band necrosis (E: arrow heads) and hypereosinophilia fitting with the diagnosis of acute myocardial infarction and connective tissue (F: asterisk) within a region of chronic myocardial infarction. The AUC for the detection of chronic MI was 0.74 (95% CI 0.64–0.84) for CT (Total Agatston calcium score), 0.68 (95% CI 0.56–0.79) for MRI and 0.82 (95% CI 0.74–0.91) for MRI with biopsy (MIA). Inter-observer agreement MRI The two radiologists were in agreement in 82 of 99 cases, in the remaining 17 cases consensus was reached in joint sessions. The kappa score of inter-observer agreement was 0.85. Discussion In this study, we investigated the diagnostic accuracy of an MIA consisting of MRI, CT, and CT-guided biopsy for detection of ischaemic heart disease in a hospital setting. We found that the combination of MRI and biopsies had the highest accuracy for detecting acute and chronic MI with conventional autopsy as reference standard. MRI without biopsies showed a high specificity, but low sensitivity for acute and chronic MI. High CT Agatston calcium score (>400) was a good predictor for chronic MI, but not for acute MI. We found a lower sensitivity of MRI as a stand-alone test for acute MI (0.50) compared with other studies investigating MRI. Ruder et al.20 reported that with MRI acute MI (within 3 h after onset) could be detected in ex vivo porcine hearts in which they correctly detected acute infarctions in all 21 cases. Forensic studies showed that with MRI acute and chronic MI (up to 100% sensitivity) could be accurately diagnosed in human subjects. Importantly, MRI could diagnose peracute MI (onset within 3 h) in cases not yet showing histological changes, but with a matching coronary stenosis at conventional autopsy.12,13,17 The differences in sensitivity and specificity among studies can be explained by the differences in studied population and clinical setting; most are forensic studies that investigated subjects who died under the suspicion of an out-of-hospital cardiac-arrest and as such had a high pre-test probability. Also, these studies often involve high-resolution cardiac imaging at 3T scanners using surface coils and relatively long scan time for imaging only the heart (∼1 h).16 Conversely, we scanned in a hospital setting and performed total-body imaging to diagnose both cardiac and non-cardiac cause of death. So as not to interfere with the patient workflow at the MR scanner, we were restricted to 1-h scan time for imaging the entire body. The addition of biopsies to MRI increased the sensitivity substantially. This highlights the importance of extensive sampling, even when no changes are visible yet on MRI. The big difference between sensitivity of MRI and MRI combined with biopsies can be explained by the quantity of sampling. From each biopsy location, at least five samples were taken, e.g. from the lateral wall, also the mid- and posterior-segments were biopsied. Furthermore, in those cases where there was a clinical suspicion of myocardial ischaemia and the MRI showed no signal abnormalities, extra biopsies were taken from the septum, anterior, and posterior wall (both mid- and posterior-segments). The non-invasive approach (CT and/or MRI) is less expensive than the minimally invasive approach (imaging plus biopsy). CT is now widely used as a stand-alone modality because of its high accessibility, short examination time, and robust performance. CT can provide better mortality statistics than the cause of death determined by the clinician, and is useful for excluding certain diagnoses. However, for diagnosing acute MI, our results show that the diagnostic accuracy of CT as stand-alone test is insufficient.21,22 To improve CT performance, in particular for ischaemic heart disease, more recent studies report on the diagnostic value of CT angiography. Grabherr et al. extensively performed feasibility studies on CTA using different contrast agents and perfusion techniques. There are different CTA approaches; it can be targeted at the coronary arteries by selective placement of the catheter at the level of the coronary ostia, or total-body CTA can be performed including multiphase scanning.9,23–25 Wichman et al. applied total-body CTA in 50 ICU patients who died unexpectedly or within 48 h of an event requiring resuscitation and found that CTA confirmed 93% of the clinical diagnoses, and autopsy confirmed 80%. In addition, CTA and CA identified 16 new major and 238 new minor diagnoses. They concluded that in cases of unexpected death CTA was a valuable addition to autopsy. Rutty et al. performed total-body CTA in 210 cases of natural and non-suspicious unnatural death and found that CTA established a cause of death in 92% of cases. The number of discrepancies with the final cause of death was not significantly different between autopsy and CTA, suggesting that total-body CTA is a feasible alternative to autopsy.26,27 A drawback of CTA is that it requires specific training, technical equipment, and contrast agents and is time-consuming due to extensive preparation of the body (e.g. intra-arterial and/or intravenous femoral access for catheter placement) prior to scanning. At the time of the study, the equipment was not available and professional expertise and scanner availability to perform CTA was lacking. Another non-invasive approach is the use of stand-alone MRI. Diffusion tensor imaging (DTI) is showing promising results in diagnosing myocardial ischaemia in situ, correctly predicting MI (either acute or chronic) with an accuracy of 0.73, using fractional anisotropy and mean diffusivity.28 First studies show that quantitative MRI can detect and differentiate between early and following stages of myocardial ischaemia based on T1, T2, and proton density values.29–31 They concluded that temperature-corrected quantitative MRI can diagnose early acute, acute and chronic MI, but histological confirmation is required. In the hospital setting, an important part of every postmortem examination should be a thorough evaluation of the medical history and clinical circumstances prior to death. The patients with known obstructive coronary artery disease, chronic, and acute MI was confirmed in 16 out of 20 cases by postmortem examination (both by MIA and conventional autopsy). The conventional autopsy found 22 new cases of acute MI highlighting the lasting need for postmortem examinations. Today the use of postmortem imaging, mostly CT, is widely accepted as adjunct to the medicolegal autopsy.7,26,27,32,33 In the hospital setting, there is a growing interest in postmortem imaging, however, expertise and logistics (e.g. access to scanners) are still important limitations. Conclusion We evaluated the diagnostic accuracy of MIA for the detection of ischaemic heart disease in a hospital setting. We found that the combination of MRI with biopsies had high sensitivity and specificity for the detection of acute and chronic MI. Funding The following funders supported this work: Erasmus MC health care efficiency, 2010-10112; Stichting Coolsingel (NL), 255; Erasmus MC vriendenfonds, 104117. Conflict of interest: None declared. References 1 World Health Organization. The Top 10 Causes of Death. 2017. http://www.who.int/mediacentre/factsheets/fs310/en/ (25 January 2018, date last accessed). 2 Shojania KG, Burton EC, McDonald KM, Goldman L. Changes in rates of autopsy-detected diagnostic errors over time: a systematic review. JAMA  2003; 289: 2849– 56. Google Scholar CrossRef Search ADS PubMed  3 van den Tweel JG, Wittekind C. The medical autopsy as quality assurance tool in clinical medicine: dreams and realities. Virchows Arch  2016; 468: 75– 81. Google Scholar CrossRef Search ADS PubMed  4 Burton JL, Underwood J. Clinical, educational, and epidemiological value of autopsy. Lancet  2007; 369: 1471– 80. Google Scholar CrossRef Search ADS PubMed  5 Turnbull A, Osborn M, Nicholas N. Hospital autopsy: endangered or extinct? J Clin Pathol  2015; 68: 601– 4. Google Scholar CrossRef Search ADS PubMed  6 Blokker BM, Wagensveld IM, Weustink AC, Oosterhuis JW, Hunink MG. Non-invasive or minimally invasive autopsy compared to conventional autopsy of suspected natural deaths in adults: a systematic review. Eur Radiol  2016; 26: 1159– 79. Google Scholar CrossRef Search ADS PubMed  7 Eriksson A, Gustafsson T, Hoistad M, Hultcrantz M, Jacobson S, Mejare I et al.   Diagnostic accuracy of postmortem imaging vs autopsy-A systematic review. Eur J Radiol  2017; 89: 249– 69. Google Scholar CrossRef Search ADS PubMed  8 Grabherr S, Djonov V, Yen K, Thali MJ, Dirnhofer R. Postmortem angiography: review of former and current methods. AJR Am J Roentgenol  2007; 188: 832– 8. Google Scholar CrossRef Search ADS PubMed  9 Turillazzi E, Frati P, Pascale N, Pomara C, Grilli G, Viola RV et al.   Multi-phase post-mortem CT-angiography: a pathologic correlation study on cardiovascular sudden death. J Geriatr Cardiol  2016; 13: 855– 65. Google Scholar PubMed  10 Grabherr S, Grimm J, Dominguez A, Vanhaebost J, Mangin P. Advances in post-mortem CT-angiography. Br J Radiol  2014; 87: 20130488. Google Scholar CrossRef Search ADS PubMed  11 Grabherr S, Grimm J, Baumann P, Mangin P. Application of contrast media in post-mortem imaging (CT and MRI). Radiol Med  2015; 120: 824– 34. Google Scholar CrossRef Search ADS PubMed  12 Jackowski C, Christe A, Sonnenschein M, Aghayev E, Thali MJ. Postmortem unenhanced magnetic resonance imaging of myocardial infarction in correlation to histological infarction age characterization. Eur Heart J  2006; 27: 2459– 67. Google Scholar CrossRef Search ADS PubMed  13 Jackowski C. Special issue on postmortem imaging 2013. Forensic Sci Int  2013; 225: 1– 2. Google Scholar CrossRef Search ADS PubMed  14 Jackowski C, Sonnenschein M, Thali MJ, Aghayev E, von Allmen G, Yen K et al.   Virtopsy: postmortem minimally invasive angiography using cross section techniques–implementation and preliminary results. J Forensic Sci  2005; 50: 1175– 86. Google Scholar CrossRef Search ADS PubMed  15 Michaud K, Grabherr S, Jackowski C, Bollmann MD, Doenz F, Mangin P. Postmortem imaging of sudden cardiac death. Int J Legal Med  2014; 128: 127– 37. Google Scholar CrossRef Search ADS PubMed  16 Jackowski C, Schwendener N, Grabherr S, Persson A. Post-mortem cardiac 3-T magnetic resonance imaging: visualization of sudden cardiac death? J Am Coll Cardiol  2013; 62: 617– 29. Google Scholar CrossRef Search ADS PubMed  17 Jackowski C, Warntjes MJ, Berge J, Bar W, Persson A. Magnetic resonance imaging goes postmortem: noninvasive detection and assessment of myocardial infarction by postmortem MRI. Eur Radiol  2011; 21: 70– 8. Google Scholar CrossRef Search ADS PubMed  18 Basso C, Aguilera B, Banner J, Cohle S, D'Amati G, de Gouveia RH et al.   Guidelines for autopsy investigation of sudden cardiac death: 2017 update from the Association for European Cardiovascular Pathology. Virchows Arch  2017; 471: 691– 705. Google Scholar CrossRef Search ADS PubMed  19 Ross D. Accurate confidence intervals for binomial proportion and Poisson rate estimation. Comput Biol Med  2003; 33: 509– 31. Google Scholar CrossRef Search ADS PubMed  20 Ruder TD, Ebert LC, Khattab AA, Rieben R, Thali MJ, Kamat P. Edema is a sign of early acute myocardial infarction on post-mortem magnetic resonance imaging. Forensic Sci Med Pathol  2013; 9: 501– 5. Google Scholar CrossRef Search ADS PubMed  21 Oyake Y, Aoki T, Shiotani S, Kohno M, Ohashi N, Akutsu H et al.   Postmortem computed tomography for detecting causes of sudden death in infants and children: retrospective review of cases. Radiat Med  2006; 24: 493– 502. Google Scholar CrossRef Search ADS PubMed  22 Inai K, Noriki S, Kinoshita K, Sakai T, Kimura H, Nishijima A et al.   Postmortem CT is more accurate than clinical diagnosis for identifying the immediate cause of death in hospitalized patients: a prospective autopsy-based study. Virchows Arch  2016; 469: 101– 9. Google Scholar CrossRef Search ADS PubMed  23 Roberts IS, Benamore RE, Peebles C, Roobottom C, Traill ZC. Technical report: diagnosis of coronary artery disease using minimally invasive autopsy: evaluation of a novel method of post-mortem coronary CT angiography. Clin Radiol  2011; 66: 645– 50. Google Scholar CrossRef Search ADS PubMed  24 Saunders SL, Morgan B, Raj V, Robinson CE, Rutty GN. Targeted post-mortem computed tomography cardiac angiography: proof of concept. Int J Legal Med  2011; 125: 609– 16. Google Scholar CrossRef Search ADS PubMed  25 Rutty G, Saunders S, Morgan B, Raj V. Targeted cardiac post-mortem computed tomography angiography: a pictorial review. Forensic Sci Med Pathol  2012; 8: 40– 7. Google Scholar CrossRef Search ADS PubMed  26 Guy N, Rutty BM,C, Robinson V, Raj M, Pakkal J, Amoroso T, Visser S et al.   Diagnostic accuracy of post-mortem CT with targeted coronary angiography versus autopsy for coroner-requested post-mortem investigations: a prospective, masked, comparison study. Lancet  2017; 390: 145– 54. Google Scholar CrossRef Search ADS PubMed  27 Wichmann D, Heinemann A, Weinberg C, Vogel H, Hoepker WW, Grabherr S et al.   Virtual autopsy with multiphase postmortem computed tomographic angiography versus traditional medical autopsy to investigate unexpected deaths of hospitalized patients. Ann Intern Med  2014; 160: 534– 41. Google Scholar CrossRef Search ADS PubMed  28 Winklhofer S, Stoeck CT, Berger N, Thali M, Manka R, Kozerke S et al.   Post-mortem cardiac diffusion tensor imaging: detection of myocardial infarction and remodeling of myofiber architecture. Eur Radiol  2014; 24: 2810– 8. Google Scholar CrossRef Search ADS PubMed  29 Schwendener N, Jackowski C, Persson A, Warntjes MJ, Schuster F, Riva F et al.   Detection and differentiation of early acute and following age stages of myocardial infarction with quantitative post-mortem cardiac 1.5T MR. Forensic Sci Int  2017; 270: 248– 54. Google Scholar CrossRef Search ADS PubMed  30 Zech WD, Schwendener N, Persson A, Warntjes MJ, Jackowski C. Postmortem MR quantification of the heart for characterization and differentiation of ischaemic myocardial lesions. Eur Radiol  2015; 25: 2067– 73. Google Scholar CrossRef Search ADS PubMed  31 Persson A, Baeckmann J, Berge J, Jackowski C, Warntjes M, Zech WD. Temperature-corrected postmortem 3-T MR quantification of histopathological early acute and chronic myocardial infarction: a feasibility study. Int J Legal Med  2017; https://doi.org/10.1007/s00414-017-1614-6. 32 Dirnhofer R, Jackowski C, Vock P, Potter K, Thali MJ. VIRTOPSY: minimally invasive, imaging-guided virtual autopsy. Radiographics  2006; 26: 1305– 33. Google Scholar CrossRef Search ADS PubMed  33 Flach PM, Thali MJ, Germerott T. Times have changed! Forensic radiology–a new challenge for radiology and forensic pathology. AJR Am J Roentgenol  2014; 202: W325– 34. Google Scholar CrossRef Search ADS PubMed  Published on behalf of the European Society of Cardiology. All rights reserved. © The Author(s) 2018. For permissions, please email: journals.permissions@oup.com.

Journal

European Heart Journal – Cardiovascular ImagingOxford University Press

Published: Feb 20, 2018

There are no references for this article.

You’re reading a free preview. Subscribe to read the entire article.


DeepDyve is your
personal research library

It’s your single place to instantly
discover and read the research
that matters to you.

Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.

All for just $49/month

Explore the DeepDyve Library

Search

Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly

Organize

Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.

Access

Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.

Your journals are on DeepDyve

Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.

All the latest content is available, no embargo periods.

See the journals in your area

DeepDyve

Freelancer

DeepDyve

Pro

Price

FREE

$49/month
$360/year

Save searches from
Google Scholar,
PubMed

Create lists to
organize your research

Export lists, citations

Read DeepDyve articles

Abstract access only

Unlimited access to over
18 million full-text articles

Print

20 pages / month

PDF Discount

20% off