OUP user menu

Non-invasive detection and quantification of acute myocardial infarction in rabbits using mono-[123I]iodohypericin µSPECT

Humphrey Fonge, Kathleen Vunckx, Huaijun Wang, Yuanbo Feng, Luc Mortelmans, Johan Nuyts, Guy Bormans, Alfons Verbruggen, Yicheng Ni
DOI: http://dx.doi.org/10.1093/eurheartj/ehm588 260-269 First published online: 21 December 2007


Aims Mono-[123I]iodohypericin ([123I]MIH) has been reported to have high avidity for necrosis. In the present study, by using rabbit models of acute myocardial infarction, we explored the suitability of [123I]MIH micro single photon emission computed tomography (µSPECT) for non-invasive visualization of myocardial infarcts in comparison with [13N]ammonia micro positron emission tomography (µPET) imaging, postmortem histomorphometry, and [123I]MIH autoradiography.

Methods and results Fourteen rabbits were divided into four groups. The left circumflex coronary artery was permanently occluded in group A (n = 3), reperfused by releasing the ligature after 15 min in group B (n = 3) or 90 min in group C (n = 6), or not occluded in group D (n = 2). Animals received [13N]ammonia µPET perfusion imaging 18 h after infarct induction followed by µSPECT imaging at 2–3.5, 9–11, and 22–24 h post injection (p.i.) of [123I]MIH. The cardiac images were assembled into polar maps for assessment of tracer uptake. Animals were sacrificed and the excised heart was sliced for autoradiography, triphenyl tetrazolium chloride, and haematoxylin–eosin staining. Using [123I]MIH µSPECT, infarcts were well delineated at 9 h p.i. Mean µSPECT infarct size was 38.8 and 32.7% of left ventricular area for groups A and C, respectively, whereas group B showed low uptake of [123I]MIH. Highest mean infarct/viable tissue activity ratio of 61/1 was obtained by autoradiography in group C animals at 24 h p.i.

Conclusion The study indicates the suitability of [123I]MIH for in vivo visualization of myocardial infarcts.

  • Necrosis
  • Mono-[123I]iodohypericin
  • µSPECT
  • Infarct avidity
  • Cardiac imaging


In most patients, the diagnosis of acute myocardial infarction (AMI) is usually made on the basis of typical chest pain, electrocardiographic (ECG) changes, and the pattern of cardiac enzymes release.1,2 The diagnosis is often difficult and about one-fifth of the patients is misdiagnosed on the basis of the above procedures.2 In such cases, a definitive diagnosis requires the use of other methods. Non-invasive localization of AMI and assessment of infarct size with infarct avid agents are essential to determine the severity, to estimate prognosis, and to formulate therapeutic strategy. Up to now, infarct avid scintigraphy has not been widely accepted because no such agents fulfill all required criteria: rapid localization at the infarct, high avidity and specificity, and reasonable duration of scan positivity. 99mTc-pyrophosphate,3,4 radiolabelled antimyosin,5,6 and more recently 99mTc-glucarate79 have all been used to localize and quantify infarct AMI in humans. Of all these clinically relevant infarct avid agents the most promising seems to be 99mTc-glucarate, but the agent has not yet obtained approval as a radiopharmaceutical. Even though 99mTc-glucarate has been able to overcome most of the limitations of the other infarct avid agents, it presents two major limitations:8,10 (1) scan positivity is limited to the early hours following an acute injury and (2) it is rapidly washed out from the infarcted myocardium. In clinical practice, PET11,12 and SPECT13 perfusion tracers have gained widespread use in detecting ischaemic and necrotic myocardium, although these flow tracers are unable to differentiate reversibly (ischaemic but viable) from irreversibly (necrotic/apoptotic) injured myocardium.

Hypericin, a polycyclic polyaromatic quinone, recently gained much attention because of its antiviral and antineoplastic activities.14 In our pursuit for infarct avid tracer agents with optimal properties, we found in preliminary experiments that mono-[123I]iodohypericin ([123I]MIH), a radioiodinated derivative of hypericin (Figure 1), localizes in necrotic tissue in animal models of reperfused hepatic and occlusive MI.15 This necrosis avid property is being shared by similar compounds that are also currently being evaluated in magnetic resonance imaging (MRI) as contrast enhancing agents.16 The preliminary results with [123I]MIH in localizing infarcts stimulated us to undertake the present study for exploring the potential of [123I]MIH SPECT in diagnosis of AMI using rabbit models of occlusive and reperfused AMI and myocardial ischaemia. The goal was also to compare infarct size estimated with three techniques: in vivo [123I]MIH µSPECT, [13N]ammonia µPET (perfusion defect), and ex vivo triphenyl tetrazolium chloride (TTC) histochemical staining. To confirm specificity of [123I]MIH for necrotic myocardium, autoradiography of myocardial slices was performed and compared with the results of TTC staining, haematoxylin–eosin (H&E) stained microscopy, and gamma counting techniques.

Figure 1

Chemical structure of hypericin (R = H) and mono-[123I]iodohypericin (R = 123I)


Myocardial infarction

Fourteen white male New Zealand rabbits (named rabbits 1–14) weighing 3–4.5 kg were premedicated with intramuscular (im) injection of a 1:1 v/v solution of Ketalar® (50 mg/mL ketamine hydrochloride) and Rompun® (2% xylazine hydrochloride) at 0.5 mL/kg each. Anaesthesia was induced and maintained with intravenous (iv) injection of sodium pentobarbital (Nembutal®) at 20–40 mg/kg/h. The rabbits were intubated and mechanically ventilated with oxygen-enriched room air. Electrocardiography (ECG) was continuously monitored. A left thoracotomy was performed through the fifth intercostal space and the heart was suspended in a pericardial cradle. The rabbits were divided into four groups: group A of occlusive AMI with left circumflex coronary artery (LCx) permanently ligated (n = 3, rabbits 1–3); group B of myocardial ischaemia with LCx occluded for 15 min (n = 3, rabbits 4–6); group C of reperfused AMI with LCx occluded for 90 min (n = 6, rabbits 7–12); and group D of sham operation without coronary intervention (n = 2, rabbits 13 and 14). For inducing reperfusion models, the LCx was encircled by a snare of 2-0 suture, and occluded by sliding a piece of polyethylene tubing over the encircling snare, and both were introduced outside the chest through an incision for releasing the occlusion at desired times. The pericardium was replaced loosely and the chest was closed in three layers, after the evacuation of pleural air. Arrhythmia was treated with IV xylocaine (0.8 mg/kg) when necessary. Temgesic® was administered im at 0.1 mg/kg for pain relief to all rabbits with an interval of 12 h starting after closure of the chest. All animals received care in compliance with the principles of laboratory care formulated by the institutional ethics committee.

Imaging protocol and reconstruction

Under anaesthesia similar to premedication for open chest modelling, the rabbit was fixed in a half open cylindrical tube with the heart positioned at the centre of the field of view of a µPET or µSPECT camera. Figure 2 is a scheme of the experimental protocol. The rabbits were injected via an ear vein with 99.1 ± 20.8 MBq of [13N]ammonia 18 h after open-chest intervention to detect the myocardial zone with perfusion defect.17 µPET imaging was started immediately after tracer injection using a Siemens µPET Focus 220 LSO scanner (Knoxville, TN, USA).18 An 8.5 min cobalt-57 transmission scan was done prior to all µPET scans. Dynamic µPET images were acquired over a period of 30 min. The images were reconstructed using Fourier rebinning and 2D ordered subsets expectation maximum algorithm. The list-mode data were summed over two time frames (0–600 s and 600–1800 s). Two hours after µPET imaging, animals were iv injected with 98.8 ± 16.9 MBq of [123I]MIH for infarct delineation. In vivo µSPECT imaging experiments were performed, at 2–3.5 and 9–11 h post injection (p.i.), comprising two scans of 35 min each, using one detector head of a dual-head gamma camera (E-cam, Siemens Medical Systems, Hoffman Estates, IL, USA), equipped with a single pinhole collimator (aperture diameter of 3 mm, Nuclear fields International, Vortum-Mullem, The Netherlands). For group C, additional delayed µSPECT scans (35 min each) were acquired at 22–24 h p.i. (44–46 h post infarction) to investigate the duration of positivity of [123I]MIH images for the infarcted myocardium. The µSPECT projection images were collected using the following acquisition parameters: 64 projections over 360° in step-and-shoot mode, 30 s per projection angle, a matrix size of 256 × 196 and a pixel size of 1.95 mm. The detector had an intrinsic resolution of 4 mm. A combination of two acquisitions of each 64 projections was used for the image reconstruction. The reconstruction was done with the aid of the maximum likelihood ordered subsets expectation-maximum method19 using 2 × 16, 2 × 8, 2 × 4, 3 × 2, 4 × 1 iterations and subsets. The reconstructed images consisted of 140 × 140 × 140 cubic voxels with a size of 1.0 mm and were post-smoothed with an isotropic Gaussian filter with a full width at half maximum of 2–5 mm, dependent on the level of noise in the image. A phantom consisting of three point sources with 1.85 MBq 99mTc per point source was scanned after each series of acquisitions to calibrate the scanner.20 Each series of µSPECT acquisitions was immediately followed by a CT scan. The animal holder contained five fiducial markers, which enable co-registration between the µSPECT and CT images. The CT and µPET transmission images can be aligned with affine registration software to correct the µSPECT images for attenuation. The final registration between µPET and µSPECT images of the heart was obtained with manual fine tuning. Animals were sacrificed immediately after µSPECT imaging by overdose injection of pentobarbital. The heart was excised and subjected to ex vivo studies.

Figure 2

Scheme of the experimental protocol

Ex vivo studies

The heart was washed with saline (4°C) to remove blood pool activity and placed in a matrix filled with agar gel (Acros Organics, Geel, Belgium), which was allowed to solidify at −20°C for 15 min, and then cut perpendicular to the long axis into 7–8 slices of 3 mm from the apex to the base (Figure 3A). The slices were then stained separately in TTC solution (1% solution in normal saline) at 37°C for 15 min. Calibrated digital photographs were taken prior to and after staining (Figure 3B).

Figure 3

Example of postmortem procedures. (A) Digital photographs of 3 mm myocardial slices from apex to base. An area of haemorrhage is visible on the left ventricle, and this area increases from base to apex in all cases. (B) Digital photographs of same slices as above, after TTC staining. Infarcted areas are clearly visible as TTC-negative (pale) while viable myocardial areas are stained brick red (TTC-positive). (C) Corresponding autoradiograms of serial 50 µm sections with regions of high [123I]MIH uptake (red), perfectly matching the TTC-negative areas, and low uptake in viable myocardial regions. (D) Scanned photographs after haematoxylin–eosin staining of the same myocardial sections showing visible evidence of necrosis. Severe infarct is seen at all regions supplied by left circumflex coronary artery

The slices were then frozen to −80°C and cut with a cryotome into 5–50 µm thick serial sections mounted on slides. Autoradiograms were obtained by exposing the slides for 2–3 days to a high-performance phosphor screen (super resolution screen; Canberra-Packard, Meriden, CT, USA). The images were analysed with Optiquant software (Canberra-Packard), and the activity concentration in the autoradiograms was expressed in digital light units (DLU)/mm2 (Figure 3C). Relative tracer concentration in the necrotic and viable sections of the left ventricle was estimated by regions of interest analysis for the necrotic and the viable regions of all autoradiograms. After autoradiographic exposure, the slides were stained with H&E to examine microscopic evidence of myocardial necrosis (Figure 3D).

Guided by TTC staining, the necrotic portions of the remaining unsliced parts of the myocardium were separated from the normal myocardium, weighed separately, and counted for radioactivity using a 3 in. NaI(Tl) scintillation detector mounted on a sample changer (Wallac Wizard, Turku, Finland). The activity in normal and necrotic parts was expressed as counts per minute per gram tissue.

Blood was withdrawn at different time points from group D rabbits (n = 2), and activity in blood was counted and expressed as percentage of injected dose (% I.D., assuming blood constitutes 7% of body weight), with which a blood clearance curve was generated.

Tracer synthesis

[13N]ammonia was prepared in an on-site cyclotron (IBA 18/9, Louvain la Neuve, Belgium), and purity was checked by cation exchange HPLC to be >95%. A sterile formulation in saline was prepared for injection.

[123I]MIH was prepared by a classical electrophilic substitution method using peracetic acid to generate 123I+ in situ as described earlier.21 It was purified on RP-HPLC (LaChrom Elite, Hitachi, Darmstadt, Germany) using an XTerra C18 column (5 µm, 4.6 mm × 250 mm) with EtOH/0.05 M NH4OAc (80/20 v/v) as mobile phase after which the ethanol was evaporated with a gentle N2 flush. The tracer was then formulated in water/PEG 400 80/20 v/v immediately before injection.21

Infarct size estimation

Perfusion defect size on [13N]ammonia µPET and hot spots on [123I]MIH µSPECT images were quantified on polar maps according to a method described earlier.22,23 The polar maps were divided in 17 segments. The mean tracer uptake and the volume of the corresponding portion of the left ventricular (LV) wall were computed for each segment. On the [13N]ammonia polar map, perfusion defect size was obtained by summing all the volumes of the segments with a mean of <60% of maximum flow. A threshold of ≥70% of maximum activity concentration on [123I]MIH polar maps was used to define the infarcted zones.

Histomorphometric infarct size was estimated on digital photographs of TTC staining by outlining the left ventricular area (LVA) and TTC negative infarct area (IA) using ImageJ software (NIH, Bethesda) in squared millimeters. Infarct volume (mL) and left ventricular volume (LVV, mL) were calculated using the following formula: ∑ (summed area of each slice [mm2] × slice thickness [3 mm]). Mean infarct volume and LVV for each animal were then converted to mass unit (g) assuming a specific gravity of the myocardium of 1.05 g/mL. Infarct size was then expressed as % LVA.

Statistical analysis

In vivo and ex vivo measured infarct size was expressed as mean % LVA ± SD. Correlations between in vivo (µPET and µSPECT) and ex vivo (histomorphometry) measured infarct sizes were made by Pearson product moment correlation coefficient (r2) and the significance tested by a paired t-test. Uptake ratio at the different myocardial regions from autoradiograms was submitted to a paired t-test. P < 0.05 was considered statistically significant.


Qualitative image analysis

In the two control animals from group D, [13N]ammonia was distributed homogeneously throughout the LV wall, while all animals in groups A–C with LCx intervention showed LV perfusion defects of varying extent on µPET. Infarcts were indiscernible with [123I]MIH µSPECT at 2–3.5 h p.i., due to the high residual blood pool activity of the tracer at this time. In all rabbits with postmortem evidence of AMI, infarcts were unequivocally visualized at 9 h p.i. and remained well delineated even at 24 h p.i. Figure 4 shows representative long and short axes images from rabbits 1 (large infarct) and 2 (small infarct) of group A with occlusive AMI at 9 h p.i. A perfect match between perfusion defect area of [13N]ammonia µPET and area of high [123I]MIH uptake on µSPECT is seen. Figure 5 shows in vivo µPET and µSPECT images of rabbits 5 and 6 in group B with 15 min of ischaemia. The µSPECT images show low uptake of [123I]MIH as only blood pool activity is evident, and this was confirmed by autoradiography. Figure 6 shows sets of long and short axes tomographic images from rabbits 8 and 12 of group C obtained at 9–11 h p.i. Figure 7 shows representative images of rabbits 8 and 10 from group C at 22–24 h p.i. As can be seen from these images, perfusion defect zones correlated very well with the hot spots on [123I]MIH µSPECT. Excellent correlations (r2 > 0.92) were found in these comparisons.

Figure 4

Longitudinal and transverse cardiac µPET and µSPECT images of rabbits 1 and 2 from group A with occlusive acute myocardial infarction. µSPECT images were obtained at 9 h post injection. Good match between perfusion defect and areas of high [123I]MIH uptake. Note the high sensitivity of [123I]MIH to detect very small infarcts in rabbit 2

Figure 5

Cardiac µPET and µSPECT images of rabbits 5 and 6 from group B of 15 min ischaemia. µSPECT images were recorded at 9 h post injection. µPET images show minimal perfusion abnormality. Only blood pool activity is evident on the µSPECT images

Figure 6

Tomographic images of rabbits 8 and 12 from group C of reperfused acute myocardial infarction. [123I]MIH µSPECT images were recorded at 9 h post injection. Note the perfect match between perfusion defect on µPET and regions of high [123I]MIH uptake on µSPECT

Figure 7

Longitudinal and transverse µPET and µSPECT images of heart of rabbits 8 and 10 from group C. [123I]MIH µSPECT images were obtained at 24 h post injection (44–46 h post infarct induction). Perfusion defect on the left ventricle is clearly delineated on 13NH3 µPET images, and the infarcts are unequivocally delineated on the corresponding [123I]MIH µSPECT images

Autoradiographic findings

Figure 8 shows the patterns of [123I]MIH uptake in groups A–C rabbits on the autoradiograms. Group A animals show a characteristic rim uptake due to a lack of perfusion but only diffusion from endocardium and epicardium (Figure 8A), group B animals show very low uptake due to absence of infarction (Figure 8B), whereas group C animals show a strong homogeneous uptake of [123I]MIH (Figure 8C). Myocardial slices from group D of sham-operated rabbits revealed very low [123I]MIH uptake similar to that from group B. [123I]MIH turned out to be the most avid infarct agent in occlusive and reperfused AMI models ever reported so far. Infarct:viable tissue activity ratio obtained by autoradiography was maximally 57 (mean of 38) in occlusive AMI and 81 (mean of 61) in reperfused AMI at the infarct centre, while the mean overall ratio was ∼30 and 50 for occlusive and reperfused AMI, respectively. Uptake at the infarct centre was significantly higher than the mean uptake in the rest of the infarcted myocardium (P < 0.05) in group C of reperfused AMI.

Figure 8

Patterns of [123I]MIH uptake in groups A, B, and C rabbits on autoradiograms. (A) Rabbit 1 from group A shows a characteristic rim uptake due to a lack of perfusion but only diffusion from endocardium and epicardium. (B) Rabbit 4 from group B animals shows very low uptake due to absence of infarction. (C) Rabbit 8 from group C animals shows a strong homogeneous uptake of [123I]MIH

Histologic findings

In contrast to sham-operated controls in group D without evidence of infarction on TTC and H&E staining, all rabbits in groups A–C that underwent coronary intervention had histopathological evidence of myocardial injury or damage. Rabbits in group A of permanent LCx occlusion showed evidence of transmural AMI extending from the endocardium to epicardium of LV wall without evidence of haemorrhage as suggested in Figure 8A. Rabbits in group B of 15 min LCx occlusion showed mainly microscopic evidence of myocardial injury without tissue necrosis. As shown in Figure 3 of five consecutive myocardial slices from a rabbit in group C, all animals with reperfused AMI showed severe haemorrhage. Microscopy confirms the presence of severe necrosis characterized by thinning and wavy myocardial fibres, contraction bands at the periphery, disintegrated nuclei, and absence of cytoplasm, intermingled with extravasated erythrocytes. These were the areas that showed high uptake of [123I]MIH as confirmed on autoradiograms.


Blood clearance of [123I]MIH in rabbits during 24 h is presented in Figure 9. Blood clearance was fitted into a two-phase exponential decay (r2 = 0.99) with half-life values of 2.36 and 120.2 h.

Figure 9

Blood clearance of [123I]MIH in rabbits. Clearance follows a two-phase exponential decay with y = 15.39e−0.0057× + 73.09e−0.294× − 11.35 with t1/2(1) = 2.36 h and t1/2(2) = 120.2 h, r2 = 0.99

Infarct size estimation

On [13N]ammonia µPET polar maps, the area with <60% of maximum activity concentration was defined as at-risk regions. For group A animals, mean defect size (% LVA ± SD) on µPET was 31.3 ± 18.6% LVA, whereas infarct size on µSPECT and histomorphometry was 38.8 ± 23.6% LVA and 31.0 ± 26.3% LVA, respectively (µSPECT–µPET P = 0.2, µSPECT–histomorphometry P = 0.5, µPET–histomorphometry P = 1). Group B animals had no evidence of infarction on µPET, µSPECT, and TTC staining. For group C animals, mean defect size on µPET was 29.4 ± 11.9% LVA whereas mean infarct size on µSPECT and histomorphometry was 32.5 ± 10.5% LVA and 32.4 ± 13.6%, respectively. There was no significant difference (P > 0.05) in infarct size measured by the three techniques (µSPECT-µPET P = 0.08, µSPECT-histomorphometry P = 0.96 and histomorphometry-µPET P = 0.08). Figure 10 demonstrates the linear correlations between µPET and µSPECT, µSPECT and histomorphometry, and µPET and histomorphometry infarct sizes for group C animals.

Figure 10

Correlation between in vivo and ex vivo measured infarct sizes (% LVA) in group C. (A) µSPECT vs. µPET: y = 0.85x + 7.87; r2 = 0.92; standard error of estimate (SEE) = 3.2; P < 0.01. (B) µSPECT vs. TTC: y = 0.74x + 8.43; r2 = 0.92; SEE = 3.3; P < 0.01. (C) TTC vs. µPET: y = 1.11x − 0.39; r2 = 0.95; SEE = 3.4; P < 0.01


Mono-[123I]iodohypericin as an infarct avid tracer

Hypericin can be efficiently and reproducibly labelled with 123I by electrophilic substitution of the phenolic ring in the ‘bay’ region (Figure 1). Structural analysis of the iodinated derivative confirmed the position of the radioiodine substituent.21 The so-formed [123I]MIH was purified on reversed phase HPLC and obtained with >99% purity in no carrier added form (specific activity of 935 MBq/µmole) in a yield >80% relative to the starting 123I activity.

[123I]MIH is predominantly excreted via the hepatobiliary pathway and to a lesser extent via the kidneys.21 Normal mouse myocardium accumulates ∼0.5 and 0.2% I.D./g at 30 min and 24 h p.i., respectively. In a previous study, preliminary results showed high avidity of [123I]MIH for necrosis.15 The present study was therefore undertaken to explore the suitability of [123I]MIH µSPECT for in vivo detection of MI.

In the past, radiolabelled antimyosin Fab antibody imaging was successful in targeting myosin chains which became accessible following sarcolemma loss in damaged cardiomyocytes. Its long blood half life also implied that in vivo images of sufficient contrast could only be obtained at 24–48 h p.i.6 99mTc-glucarate seems promising but it is still in clinical development.8 The simplicity of preparation and high stability of [123I]MIH already make this tracer agent advantageous over other infarct avid agents such as 111In-labelled antimyosin Fab antibody and 99mTc-pyrophosphate. More importantly, our present study shows a high potential of [123I]MIH in the delineation of both reperfused and non-reperfused AMI. None of group B animals (n = 3, 15 min ischaemia) showed high uptake of [123I]MIH. It is known that brief periods of ischaemia-reperfusion may lead to minimal (usually reversible) cell injury by apoptosis and oncosis (early necrosis).24,25 It has been shown that [123I]MIH is not taken up by cells undergoing apoptosis (unpublished data). In the present study, [123I]MIH was administered when infarcts were ∼20 h old, and the infarcts were well delineated at 9 h p.i. and remained positive at 24 h p.i. Delineation is not however limited to old infarcts as the tracer agent unequivocally detected MIs after administration to animals having infarcts that were 2 h old (unpublished data). Tracer uptake was also seen on the wound (area of surgery). Such uptake has been encountered with other infarct avid agents.

Cardiac nuclear imaging

[123I]MIH µSPECT images presented in this study were a sum of two 35 min scans using a single-head detector with a pinhole collimator. With a dual-head detector, a single scan would generate images of similar diagnostic quality within a shorter time. Infarct delineation at early time points p.i. was not possible because of initial high blood pool. In the early time point investigated (2–3.5 h p.i.) blood pool activity contained ∼40% of I.D. but rapidly clears in a two-phase exponential decay to 1.5% I.D. at 24 h p.i. (Figure 9). The initial high blood pool activity has been attributed to the high affinity of [123I]MIH for blood lipoproteins.26 Nonetheless, the 9 h time window presented in this study is clinically relevant as thrombolysis is routinely performed prior to delineation and sizing of infarcts after admission of these patients to the emergency unit.

Infarct avidity of mono-[123I]iodohypericin

The high infarct avidity of [123I]MIH means that a low tracer dose can be administered to obtain good quality images, in this way reducing the radiation burden to the patient. Infarct:viable tissue activity ratio was maximally 57 in occlusive AMI and 81 in reperfused AMI at the infarct centre, wheras the mean overall ratio was ∼30 and 50 for occlusive and reperfused AMI, respectively, as measured by autoradiography. The avidity was also confirmed by radioactivity counting where the ratio of necrotic:viable tissue was more than 8, similarly for occlusive and reperfused AMI even though this did not yield similar ratios as found on autoradiograms. This lower ratio is attributed to the limited accuracy in tissue sampling by separation of TTC negative from TTC positive areas, resulting in mixing of infarcted tissue with viable tissue. Absolute uptake values of [123I]MIH in reperfused AMI, as obtained by ex vivo gamma counting, was higher (0.15 and 0.02% I.D./g in necrotic and viable myocardium, respectively) than in occlusive AMI (0.08 and 0.007% I.D./g in necrotic and viable myocardium, respectively). These values were similar to those obtained for 99mTc-glucarate in rabbits.9

Clinical relevance

While different forms of acute reperfusion therapies are widely used to salvage non-infarcted myocardium, reperfusion is often viewed as a double-edged sword.27 Necropsy findings in a study of 19 patients showed that haemorrhagic MI was present in 74% undergoing different types of reperfusion therapy,28 whereas cases without any interventional therapy showed only 2–5% of haemorrhagic MI.29 Whether the duration of reperfusion (ligature release) aggravated the infarct and enhanced tracer uptake has not been investigated in our study, since [123I]MIH was injected when infarcts were of similar ages. Haemorrhagic MI was visible in all animal hearts in group C characterized by visible haemorrhage within the myocardium as proved by microscopic findings.

In PET and SPECT, infarct size estimation from polar maps requires setting an arbitrary threshold (usually regions with <50–60% of the maximum flow rate values are considered to be at risk regions) and calculating the volume of infarct with values below the chosen threshold. Regions with [123I]MIH concentration of at least 70% of the maximum had a corresponding low perfusion activity of <60% of the [13N]ammonia maximum, again confirming that high [123I]MIH uptake in vivo was restricted to myocardial zones at risk. The resulting in vivo measured infarct size (µPET and µSPECT, P > 0.05) did not differ from the infarct size measured by planimetry after TTC staining (P > 0.05) in all animals with AMI).

Putative mechanisms

In healthy cells hypericin is located transiently within lysosomes and golgi bodies and is excluded from the nucleus, even in very high doses of 10 mg/kg, and it is not genotoxic.30 Although the mechanism of [123I]MIH binding to necrotic tissue is still to be fully elucidated, a few possibilities can be eliminated based on the present results: binding to sequestered Ca2+ and/or positively charged histone, two strong signals of necrotic cell death. Histones are positively charged proteins released from disintegrating nuclei and are only present in the acute phase (usually within 6 h) of AMI and are subsequently sequestered rapidly by macrophages. The fact that [123I]MIH (negatively charged at physiological pH) shows high avidity at 44–46 h post infarct induction therefore excludes the possibility of histone binding. By localizing mainly to nuclear fractions of necrotic tissue, 99mTc-glucarate has been hypothesized to bind to histone bodies,9 and this explains why it shows high avidity only during the early phase of AMI. Low skeletal uptake of [123I]MIH also eliminates tracer binding to sequestered Ca2+. One plausible mechanism of tracer uptake is the facile access by diffusion to necrotic tissue following lipoprotein-mediated delivery of the tracer to vascular compartments of adjacent tissues and subsequent binding to one or more denatured tissue components released by necrosis.16


The present study demonstrates the potential clinical usefulness of [123I]MIH SPECT in delineating and sizing myocardial infarcts of both occlusive and reperfused types. Visualization of AMI was not possible at 2 h p.i. due to high blood pool activity. In the present study, scans of good diagnostic quality were obtained at 9 h p.i. and were still positive at 24 h p.i. However, in an ongoing study, images of good diagnostic quality have been obtained at 7 h p.i., indicating that successful imaging after [123I]MIH injection can be done at earlier time points than 9 h p.i. and probably also earlier than 7 h p.i. Further studies are necessary to demonstrate the earliest time point after [123I]MIH administration at which diagnostically useful images can be obtained. Fine tuning of the structure may allow for improved blood clearance and possibly very early visualization, which is an ongoing effort. The potential clinical usefulness of the tracer however remains to be fully explored.

Study limitation

In the present study, infarcts were indiscernible at early time points (2–3.5 h p.i.), probably due to high blood pool activity. Based on the present results, this may present some limitations in the selection of patients for revascularization therapies. However, in ongoing experiments we have found that using [123I]MIH AMI was unequivocally delineated by the tracer agent at 7 h p.i. (unpublished data). This indicates that successful imaging after [123I]MIH injection can most likely be done at earlier time points than 7 h p.i., and imaging at 5–6 h p.i. would be within the time window for thrombolytic therapy. Further studies are necessary to demonstrate the earliest time point after [123I]MIH administration at which diagnostically useful images can be obtained.

In addition to the possibility of imaging within the time window for thrombolytic therapy, [123I]MIH gives useful information about the amount of salvageable myocardial tissue which is critical for the quality of life of patients even after thrombolytic therapy.


This work was supported by a grant from Geconcerteerde Onderzoeksactie (GOA) of the Flemish Government, FWO-Vlaanderen grant G. 0257.05, and in part by the EC-FP6-project DiMI, LSHB-CT-2005-512146 and Asia-Link CfP 2006 – EuropeAid/123738/C/ACT/Multi – Proposal No. 128-498/111.


The authors would like to thank Peter Vermaelen for the µPET imaging experiments and Tjibbe de Groot, Dominique Vanderghinste, Marva Bex, Hubert Vanbilloen for practical help.

Conflict of interest: none declared.


  • This paper was guest edited by Prof. Dr Markus Schwaiger, Klinik r.d. Isar der TU Muenchen, Nuklearmed. Klinik u. Poliklinik, Munich, Germany


View Abstract