Here, we show that there is a significant effect of suppression “within” infarcted tissue suggesting either, (a) an effect of suppression on macrophages and/or other inflammatory cells or alternatively, (b) the presence of a significant volume of viable myocytes in the infarcted region, that are subject to the same metabolic effects of suppression on the cardiomyocytes within the non-infarcted regions. Myocardial glucose suppression has previously been thought to only affect cardiomyocytes but not inflammatory cells such as macrophages [22]. This has been largely an unquestioned assumption in publications dealing with protocols associated with the needed suppression of FDG uptake in normal myocytes in order to visualize inflammation in diseases like cardiac sarcoidosis. It should be noted that there is a large amount of variability between animals in terms of infarct uptake vs remote and blood pool. This could be due to a number of factors including differences in infarct size, cardiac function or the number of macrophages in the infarct. The relationship between RT and blood pool is also not the same in each animal which may be due to some of the same factors, but most importantly a difference in effectiveness of fasting in suppressing myocardial glucose uptake.

As shown in Fig. 5, MRGlu in the infarct decreased 75% (from mean value of 0.109 to 0.027) while it dropped 91% (from mean value 0.083 to 0.007) in the remote tissue. For this decrease to be due to viable myocytes within the infarct zone, ~ 80% of the cells would have to remain alive. In the normal left myocardium, myocytes can occupy approximately 75% of the volume [23], while the extracellular volume (ECV) occupies approximately 0.2 mL/mL (Fig. 6), and fibrosis only about 1.2% or 0.01 mL/mL (Fig. 9). This leaves approximately 4% for other cells which are primarily fibroblasts. By day five post-acute MI, the ECV has increased to 48% as seen in Fig. 6. If we assume that fibrotic tissue (e.g., collagen) has increased to 4% from 1% (Fig. 6) and that the 4% fibroblasts remain viable as they initiate the fibroblast activity known to occur in acute MI[24], then the maximum space for viable myocytes in the INOT is only 0.44 or 44% (i.e., 1.0–0.48–0.04–0.04; see Fig. 13). This maximum component of 0.48/0.75, or 64%, ignores the space to be occupied by the inflammatory cells, which must occupy significant volume as these cells take up as much glucose as the myocytes in the remote tissue. This 44% (upper limit of maximum possible) is short of the 80% viable myocytes needed to explain the entire drop in glucose metabolism seen in the infarcted tissue in response to glucose suppression.

Approximate percentage volume of myocardial tissue occupied by different tissue types. Baseline—values expected in normal myocardium; Post-MI – predicted values. Note that this post-MI estimate of values occupied by viable myocytes is a maximum as other cell types such as inflammatory and dead cells are present and would diminish the myocyte volume

However, it is possible that there is a contribution from myocytes in the remote tissue next to the INOT due to partial volume effects. Although unlikely, it may be possible that the activated fibroblasts in the INOT would reduce FDG uptake on glucose suppression. Given that it is unlikely that 64% of the INOT is composed of viable myocytes, it is more likely that glucose suppression has affected the inflammatory cells that are predominantly present at five days post-acute MI. Note as well, that the number of GLUT1 receptors is not different between the INOT and the RT. Since the cellular volume has decreased (ECV increased from 0.2 to 0.48) and there are far less cells in the INOT compared to the RT, the average number of GLUT1 receptors per cell must be greater in the INOT compared to the RT. This again argues against the cells in the INOT being predominantly myocytes.

This potential effect of suppression on inflammatory cells may make FDG-based PET imaging less sensitive to inflammatory activity, highlighting the need for tracers that are not only specific for the presence of inflammation but whose uptake is not confounded by the metabolic state present at the time of injection and during the period of imaging. Several PET tracers are currently under investigation; one potential class of PET tracers on which investigation has begun are those that target the 18 kDa translocator protein (TSPO) receptors in activated mitochondria [25, 26]. Thackeray et al. 2018 showed increased myocardial uptake of the TSPO tracer [18F] GE180 three to seven days after induction of inflammation post-MI in mice and increased uptake of the TSPO tracer [11C] PK11195 in humans four to six days post-MI [27]. However, these TSPO PET probes have their own limitations as some do not work in humans with the low affinity allele for the TSPO receptor [28]. In recent work by MacAskill et al. 2021, they published a novel TSPO tracer, [18F] LW223 can detect macrophage-driven inflammation 7 days post-MI in mice and included an estimated radiation dose for Humans that is acceptable for future clinical use [28]. This TSPO receptor has been reported to be Independent of the rs6971 Human Polymorphism. In addition, CXCR4 tracers have been used to investigate inflammation in the heart and PET probes that are fibroblast activation protein inhibitors have recently shown promise in detecting fibroblast activity initiated by inflammation [24, 29].

The difference in MRGlu results at baseline supports our previous preliminary work[16]that, in the canine model, to achieve adequate suppression of myocardial glucose uptake, combining a heparin injection with lipid infusion performs better than a heparin injection alone. However, the variability of FDG uptake in normal myocardium due to fasting alone, prior to anesthesia, makes it difficult to compare values before glucose uptake suppression to after suppression as evident in BL1 data shown in Fig. 5.

ECV measurements, on the other hand, do not appear to be affected by heparin or lipid infusion with no significant difference seen within each baseline scan but also not between BL1 and BL2. In a previous study, we have found a non-significant trend toward higher ECV at baseline with the use of heparin and lipid infusion when compared to fasting alone with different injection methods [9]. The current results have clarified that ECV in fact does not change significantly.

While only three animals had a visible IOT, in these animals the apparent IOT volume decreased by 69%, 93% and 93% after a 150 min constant infusion of Gd-DTPA. Not only does this show that Gd-DTPA is slowly penetrating into the IOT (the concentration will slowly continue to increase until the extravascular volume is “filled”), but it also suggests that measurement of inflammation in this region using PET imaging is possible if corrections are developed for the impact of this slow delivery of FDG. Due to the resolution of PET, the IOT also presents a major partial volume problem for the INOT and penetrating it with the tracer would make quantification of inflammation within the INOT less affected by partial volume and therefore more accurate.

Histology shows that GLUT1 expression is the same in infarcted tissue as in remote tissue; this similarity lines up well with the MRGlu before suppression in RT and INOT, which is also not significantly different (Fig. 5). While histology shows GLUT1 expression to be the same in the infarct as in the remote tissue, the CD68 expression is more than twice as high in the infarct as in the remote tissue. Since ECV has gone up, there must be more GLUT1 receptors expressed per cell in macrophages at five days post-MI compared to the number per cell in RT. Previously, GLUT1 has been shown to be overexpressed in macrophages, but this has not been compared to myocytes [30]. Alternatively, fasting may have reduced expression in the myocytes. It would be interesting to compare the numbers of GLUT1 receptors at later times post-MI when the number of M2 macrophages are dominant over the M1’s that are predominantly present at five days post-MI. The high ECV also explains the high CD68 expression as more cellular debris would lead to more macrophages. CD68 has previously been shown to be increased in infarcted and remote tissue post-myocardial infarction [31].

Quantifying the percentage of fibrosis in the tissue regions by way of Masson’s Trichome Stain allowed visualization of the extent of clinically important levels of myocardial fibrosis between the four tissue regions identified [32] (infarct center, infarct edge, remote tissue, and right ventricle). The histology confirms that the protocol was successful at inducing an MI and shows the correlation between ECV measured by MRI and Fibrosis percentage (Fig. 11). Tissue from the right ventricle was used as a control being the furthest away from the induced MI. Tissue collected from the right ventricle had a mean percent fibrosis value of 1.18 ± 0.48% (Fig. 9), compared to a mean percent fibrosis value of 2.14 ± 2.77% in the right ventricle after sham surgery shown by Wang et al. [33], while Tanaka et al. [34] determined the mean percent fibrosis in the right ventricular free wall to be in healthy canines 0.359% (95% CI = 0.039–0.302; P = 0.011). With our mean percent fibrosis values from the right ventricle falling within a similar range, we have concluded that the tissue we collected from the right ventricle served as an appropriate control for our study, as validated by these published results.

In our canine model of acute MI, we have simulated the methods used in humans to suppress uptake in myocytes in the left myocardium. Besides fasting and a high-fat low-carbohydrate diet (HLFCD), many groups also inject unfractionated heparin, which suppresses glucose metabolism through the Randle Cycle. Heparin induces hydrolysis of triglycerides increasing the plasma levels of free fatty acids and glycerol [35], which inhibits glucose uptake by myocardium due to the increased availability for fatty acid oxidation [36]. However, these various techniques still result in up to 25% of FDG PET studies failing due to poor suppression [13]. As well, Dietz et al. have recently shown that a 100-mL lipid emulsion infusion (Intralipid 10%) in addition to an HFLCD will further improve suppression [12]. Note that in our animal model we also used an intralipid infusion as well as injection of unfractionated heparin, which gives the greatest reliability in suppression maximizing the success and reducing the number of failures resulting in greater success in repeat experiments on the same animal.

The intralipid injected into our canines was a 20% I.V. Fat Emulsion from Baxter Healthcare Corporation that contained 20% Soybean Oil, 1.2% Egg Yolk Phospholipids, 2.25% Glycerin, and Water for injection (Intralipid 20%; Baxter Healthcare Corporation, 2022). As the major component, the Soybean Oil consists of neutral triglycerides of predominantly unsaturated fatty acids. These fatty acids are linoleic acid (44–62%), oleic acid (19–30%), palmitic acid (7–14%), α-linolenic acid (4–11%) and stearic acid (1.4–5.5%); which are the same free fatty acids (FFA) that comprise approximately 95% of FFA in human plasma [37]. Additionally, Goodfriend et al. measured changes in lipid concentrations after injection of heparin, finding that derivatives of linoleic acids, the major fatty acid within Soybean Oil of the intralipid injection, displayed the largest increase in arterial plasma [38]. Furthermore, in the recent paper by Dietz et al., the intralipid infusion used to successfully suppress glucose metabolism in human patients for FDG PET, also mainly contained pure soybean oil [12]. These strong similarities between our canine and human protocols, along with the similar results in blood lipids, support the translational relevance of this investigation.

Limitations of this study include that the effect of fasting needed prior to anesthesia may be variable as we have shown previously [9]. However, the approach we have taken with the 150 min constant infusion allows us to overcome this limitation as each dog is compared to itself eliminating this diet confound. The small number of animals with an IOT limited statistical analysis on the effect of the comparison on apparent IOT size. With this small number of animals with IOT, it is impossible to determine whether the volume reduction is significant. It also means that IOT glucose metabolism could not confidently be determined. To use a similar protocol with patients, the timeline would have to be adjusted for a much shorter scan time. For example, here we look at “before”, “during” and “after” suppression but as we demonstrate here, the “during” and “after” results are effectively the same. Thus, only “before” and “during” are necessary to look at this in patients, reducing the scan time to 90 min.

Another limitation of this research is that more histology is needed to identify the cell types present that are associated with the uptake of glucose/fatty acids in the infarcted zone. Specifically, measuring the presence of different glucose transporters and inflammatory cell types would have been very useful. In the same vein, autoradiography would have been an excellent addition as it could have validated the presence or absence of partial volume effects. Blood sampling for inflammatory markers may have also provided insight on the mechanism suppressing FDG uptake in the infarcted tissue.

Future work will be to improve and include motion correction to allow for voxel-wise comparisons between pre- and post-suppression. It may also be possible and could be investigated if subtracting the pre-suppression image from the post-suppression image would make clinical translation of the methodology presented here more viable.