The currently accepted patient preparation protocols for ¹⁸F-FDG PET involve fasting for 4–12 hours. The aim of fasting is to produce low levels of glucose and insulin so that FDG, a glucose analog, shows optimal uptake. Competitive inhibition of uptake by serum glucose and of insulin-dependent uptake by muscle and fat, mediated by glucose transporter (GLUT)-4 receptors, is minimized. However, this also results in variable physiologic myocardial uptake, which has been reported to result in false-positive findings in PET studies [1, 2]. It would therefore be advantageous to have a preparation method that reproducibly decreased myocardial FDG uptake. Minimizing myocardial FDG uptake should decrease false-positive studies because a decrease in the background uptake should facilitate the definition of mediastinal FDG uptake.

Factors that might influence myocardial uptake of FDG that have been examined include patient age, fasting time, blood glucose levels, and use of a low-carbohydrate diet. Age and fasting time do not influence FDG physiologic uptake in the myocardium. Changes induced in blood glucose levels may have a nonlinear effect on myocardial uptake [2]. The authors of one report described their finding that a low-carbohydrate diet decreased artifactual myocardial uptake [3].

The Randle cycle (fatty acid–glucose cycle) has established that fatty acid loading suppresses glucose metabolism [4] and that glucose loading suppresses fatty acid utilization by the myocardium. Our patient preparation protocol initially involved a fast of at least 4 hours after a low-carbohydrate diet. We elected to further optimize insulin and glucose levels by having patients eat a very high-fat, low-carbohydrate, protein-permitted (VHFLCPP) meal before fasting. We reasoned that the Randle effect would assist in lowering glucose and insulin levels further. Soon after the protocol change, we noticed that myocardial uptake was apparently decreased using this new protocol; therefore, a retrospective analysis was undertaken of the studies performed before and after the change in patient preparation to determine whether this impression was correct.

The study was a HIPAA-compliant retrospective study and was approved by the Beth Israel Deaconess Medical Center's institutional review board. The “fasting” group comprised patients who presented to our institution for PET/CT before September 2005, when the patient preparation protocol was fasting overnight or for 4 hours before scanning with instructions to cease sugar intake the evening before scanning. The VHFLCPP group comprised patients who presented after September 2005, when the protocol was changed to a VHFLCPP meal before fasting prior to scanning.

The indications for scanning in both groups were primarily lymphoma. However, patients also included those with lung cancer, breast cancer, melanoma, and other cancers. In the VHFLCPP group, 35 patients (58%) had lymphoma, 13 (22%) had lung cancer, four (7%) had breast cancer, three (5%) had melanoma, and five (8%) had other cancers. In the fasting group, 50 (50%) had lymphoma, 25 (25%) had lung cancer, six (6%) had breast cancer, four (4%) had melanoma, and 16 (16%) had other cancers.

After IV injection of 740 MBq (20 mCi) of FDG, patients stayed in a semidarkened quiet room for 45–60 minutes. Plasma glucose measurements before injection were less than 200 mg/dL in all patients. The patients were then scanned using a 4-MDCT scanner (Discovery/LightSpeed PET/CT, GE Healthcare). An initial CT scout image (30 mA, 120 kVp) was obtained. The body was then scanned from the base of the skull to the mid thighs using helical CT at 0.8 second per rotation at 100 mA and 149 kVp with a section thickness of 5 mm and a 4.25-mm interval. Thin oral, but no IV, barium contrast material was used in all studies. Patients were instructed to breathe normally during the acquisition. Images were then obtained with 5 minutes per bed position and were iteratively reconstructed using CT-based attenuation correction.

Table 1 shows patient age, blood glucose level, sex, diagnosis, and prevalence of diabetes and renal disease in the control and test groups. There was no statistically significant difference in any of these measures between the two groups.

Figure 1 shows the myocardial SUV_min and SUV_max for the fasting protocol patients whose myocardial FDG uptake was qualitatively (i.e., visually) graded as 0 (homogeneously minimal), 1 (mostly minimal or mild uptake), 2 (mostly intense or moderate uptake), or 3 (homogeneously intense). Qualitatively, visual grading (a categoric value) correlated with FDG uptake as determined by SUV_max measurements (a continuous variable).

Figure 2 shows the qualitative FDG uptake for nonadherent patients who ate a VHFLCPP meal at different times before FDG injection.

Figure 3 compares the SUVs_max for the 60 test patients who completed questionnaires indicating adherence to the VHFLCPP diet (from 83 total) with those for the 101 fasting protocol patients. The mean and SD for the SUV_max was 3.9 ± 3.6 and 8.8 ± 5.7, respectively, with a one-tail Student's t test indicating significance (p < 0.000001). No effect was observed with SUV_min.

Figure 4A, 4B shows lateral and frontal views of a maximum-intensity-projection scan from a PET study in a patient with several metastases close to the heart that are clearly delineated as a result of suppression of myocardial FDG uptake.

To test the hypothesis that the VHFLCPP diet has no effect on other sites of FDG uptake, we determined FDG uptake in sites of chronic inflammation. We retrospectively analyzed FDG uptake data, as determined by SUVs, in extracardiac sites of chronic inflammation in patients who had been scanned within 8 weeks, first prepared by the fasting method then by the new protocol. No PET studies were excluded. In eight patients and 16 sites of extracardiac inflammation (osteoarthritis, n = 10; tendinitis, n = 5; dental, n = 1), there was no statistical difference between pairs using the paired Student's t test (p = 0.30; SUV_max range, 1.8–5.6).

Our results show that a VHFLCPP meal 3–6 hours before FDG injection suppresses myocardial FDG uptake. The mechanism is likely the result of the Randle cycle, which has established that fatty acid loading suppresses glucose metabolism [4]. Further, myocardial tissue prefers free fatty acids (FFAs). Therefore, it appears flooding the myocardium with its preferred substrate and waiting until after the peak of postprandial hyperinsulinemia suppresses myocardial FDG uptake. During postprandial hyperinsulinemia, GLUT-4–mediated myocardial FDG uptake is high. After 6–8 hours, myocardial FDG uptake becomes variable, emulating studies with the fasting preparation.

A basis for our observations is suggested by a report that elevated blood FFAs decreased myocardial glucose uptake. This change was effected by infusing heparin and triglycerides [5]. Heparin displaces lipoprotein lipase in capillaries so triglycerides are cleaved to yield FFAs, which are the myocardium's preferred substrate. There is also a report that FFAs inhibit GLUT-4 expression in cardiac muscle [6].

Although sucralose (Splenda, Johnson & Johnson) is not known to have carbohydrate properties, whether it would affect myocardial FDG uptake was unknown because its chemical structure, unlike the chemical structure of other artificial sweeteners, is similar to that of glucose. We therefore instructed patients not to eat or drink foods containing Splenda before scanning.

There are two potential uses of a VHFLCPP preparation protocol. First, regarding more accurate definition of mediastinal pathology, the variable myocardial uptake due to the usual fasting patient protocols can give rise to false-positive studies, especially in PET-only studies [1]. This new method should facilitate increased accuracy, particularly in the definition of mediastinal pathology (Fig. 4A, 4B). Empirically, tumor FDG uptake did not appear to be affected. Selected patients with prior FDG-avid scans showed essentially unchanged uptake when other measures of tumor change were absent when prepared with the VHFLCPP diet. However, a prospective study, now in preparation, rather than this retrospective study, is necessary to confirm this finding. Uptake in sites of chronic inflammation, such as arthritic joints, in the same patient was compared before and after the change in the patient preparation protocol. Sixteen sites in eight patients analyzed showed no significant change in patients prepared by either method (p = 0.30).

Second, this method of patient preparation may permit the detection of biologically active coronary artery disease (CAD) [7]. Intravascular inflammation can be detected in extracardiac areas where background FDG uptake is minimal. Vasculitides not only can-be detected by FDG PET but also have begun to be used to monitor treatment [8]. CAD is proposed to progress from active inflammation to chronic calcification. The detection of inflamed, soft, vulnerable plaque could be clinically advantageous. Because noncoronary artery atherosclerosis is detectable in humans with FDG PET [9–11], it is reasonable to suggest that the inflammation accompanying biologically active CAD might also be detected by FDG PET. Preliminary data suggest that using this new method to suppress myocardial FDG and glucose uptake, biologically active CAD can be detected [7].

Glycolysis is outside normal physiologic control in cancer cells (the Warburg effect). Macrophages associated with inflammation in arthritis, tendinitis, infection, and coronary artery plaque use GLUT-1 and high-affinity GLUT-3 receptors that are insulin-independent [12] and are not expected to show decreased FDG uptake in patients prepared with a high-fat, low-carbohydrate diet. The patient preparation protocol used here mainly affects insulin-dependent GLUT-4 receptors that are induced in muscle (including cardiac muscle) and adipose tissue [13]. Although theoretically and by empiric observation tumor uptake appears unaffected, matched studies of the same patients prepared by each of the two protocols and scanned 24 hours apart have not been performed. Preparations are now under way to perform such a prospective study to complement this retrospective study.

Theoretically, the decrease in myocardial FDG uptake could result in higher uptake in tumors. A prospective study, now in preparation, comparing uptake in tumors in patients prepared by fasting and uptake in the same sites in patients prepared by a high-fat diet in scans obtained 24 hours later will be necessary to determine whether tumor SUVs are indeed higher in patients prepared by this method. However, inflammatory FDG uptake does not appear to be affected by the new patient preparation protocol, as illustrated by the matched studies presented here.

Despite the decrease in myocardial FDG uptake, minimal uptake may be seen in some regions. Although improvement in definition of mediastinal pathology is evident, further prospective studies are necessary to document this improvement.

The optimal composition of the preparatory diet has not been defined. In this study, the diet was high in protein and fat and very low in carbohydrate, with the assumption that the high fat and very low carbohydrate, via the Randle cycle, minimized myocardial FDG uptake. Studies are being planned to evaluate the optimal composition and a more standardized delivery system, so compliance is increased and exact intake is known. In this study, the time of the last meal and meal description were reported by patients and were not independently verified. This reliance on self-reporting may explain the outliers seen in Figure 3 using the VHFLCPP protocol. In addition, because this study was retrospective, we do not know fatty acid or insulin levels. However, we do know that analysis of blood glucose levels in a large number of fasting versus VHFLCPP diet patients showed that the VHFLCPP diet patients had lower blood glucose levels than the fasting group, which is suggestive of the Randle cycle effect. A prospective study analyzing insulin, glucose, and FFA levels would be useful and is planned.

Other limitations of this study include the definition of sites of chronic inflammation (osteoarthritis, tendinitis, and dental) that, although empirically reasonable, lack a clear pathologic gold standard. Also, medication profiles were not compared, and theoretically some medications may alter FDG uptake, although the specific effect of most medications on myocardial FDG uptake is not known.

A VHFLCPP meal eaten 3–6 hours before FDG injection suppresses myocardial FDG uptake. This should facilitate definition of mediastinal pathology by FDG PET, particularly with stand-alone PET, and may permit the detection of biologically active CAD [7].

Address correspondence to G. M. Kolodny ([email protected]).

This is a Web exclusive article.