Georgia Keramida

Quantitative magnetization transfer imaging as a biomarker for effects of systemic inflammation on the brain

Background Systemic inflammation impairs brain function and is increasingly implicated in the etiology of common mental illnesses, particularly depression and Alzheimer’s disease. Immunotherapies selectively targeting proinflammatory cytokines demonstrate efficacy in a subset of patients with depression. However, efforts to identify patients most vulnerable to the central effects of inflammation are hindered by insensitivity of conventional structural magnetic resonance imaging. Methods We used quantitative magnetization transfer (qMT) imaging, a magnetic resonance imaging technique that enables quantification of changes in brain macromolecular density, together with experimentally induced inflammation to investigate effects of systemic inflammatory challenge on human brain microstructure. Imaging with qMT was performed in 20 healthy participants after typhoid vaccination and saline control injection. An additional 20 participants underwent fluorodeoxyglucose positron emission tomography following the same inflammatory challenge. Results The qMT data demonstrated that inflammation induced a rapid change in brain microstructure, reflected in increased magnetization exchange from free (water) to macromolecular-bound protons, within a discrete region of insular cortex implicated in representing internal physiologic states including inflammation. The functional significance of this change in insular microstructure was demonstrated by correlation with inflammation-induced fatigue and fluorodeoxyglucose positron emission tomography imaging, which revealed increased resting glucose metabolism within this region following the same inflammatory challenge. Conclusions Together these observations highlight a novel structural biomarker of the central physiologic and behavioral effects of mild systemic inflammation. The widespread clinical availability of magnetic resonance imaging supports the viability of qMT imaging as a clinical biomarker in trials of immunotherapeutics, both to identify patients vulnerable to the effects of systemic inflammation and to monitor neurobiological responses.

Accumulation of (18)F-FDG in the liver in hepatic steatosis.

OBJECTIVE Nonalcoholic fatty liver disease is associated with hepatic inflammation. An emerging technique to image inflammation is PET using the glucose tracer, (18)F-FDG. The purpose of this study was to determine whether in hepatic steatosis the liver accumulates FDG in excess of FDG physiologically exchanging between blood and hepatocyte. MATERIALS AND METHODS Hepatic FDG uptake, as SUV = [voxel counts / administered activity] × body weight), and CT density were measured in a liver region in images obtained 60 minutes after injection of FDG in 304 patients referred for routine PET/CT. Maximum SUV (region voxel with the highest count rate, SUVmax) and average SUV ( SUVave) were measured. Blood FDG concentration was measured as the maximum SUV over the left ventricular cavity (SUVLV). SUVave was adjusted for hepatic fat using a formula equating percentage fat to CT density. Patients were divided in subgroups on the basis of blood glucose (< 4, 4 to < 5, 5 to < 6, 6 to < 8, 8 to < 10, and > 10 mmol/L). Hepatic steatosis was defined as CT density less than 40 HU (n = 71). RESULTS The percentage of hepatic fat increased exponentially with blood glucose. SUVmax / SUVLV and fat-adjusted SUVave / SUVLV but not SUVave / SUVLV correlated with blood glucose. Fat-adjusted SUVave was higher in patients with hepatic steatosis (p < 0.001) by ~0.4 in all blood glucose groups. There was a similar difference (~0.3) in SUVmax (p < 0.005) but no difference in SUVave. SUVmax / SUVLV and fat-adjusted SUVave / SUVLV correlated with blood glucose in patients with hepatic steatosis but not in those without. SUVave / SUVLV correlated with blood glucose in neither group. CONCLUSION FDG uptake is increased in hepatic steatosis, probably resulting from irreversible uptake in inflammatory cells superimposed on reversible hepatocyte uptake.

Hepatic glucose utilisation in hepatic steatosis and obesity

Hepatic steatosis is associated with obesity and insulin resistance. Whether hepatic glucose utilization rate (glucose phosphorylation rate; MRglu) is increased in steatosis and/or obesity is uncertain. Our aim was to determine the separate relationships of steatosis and obesity with MRglu. Sixty patients referred for routine PET/CT had dynamic PET imaging over the abdomen for 30 min post-injection of F-18-fluorodeoxyglucose (FDG), followed by Patlak–Rutland graphical analysis of the liver using abdominal aorta for arterial input signal. The plot gradient was divided by the intercept to give hepatic FDG clearance normalized to hepatic FDG distribution volume (ml/min per 100 ml) and multiplied by blood glucose to give hepatic MRglu (μmol/min per 100 ml). Hepatic steatosis was defined as CT density of ≤40 HU measured from the 60 min whole body routine PET/CT and obesity as body mass index of ≥30 kg/m2. Hepatic MRglu was higher in patients with steatosis (3.3±1.3 μmol/min per 100 ml) than those without (1.7±1.2 μmol/min per 100 ml; P<0.001) but there was no significant difference between obese (2.5±1.6 μmol/min per 100 ml) and non-obese patients (2.1±1.3 μmol/min per 100 ml). MRglu was increased in obese patients only if they had steatosis. Non-obese patients with steatosis still had increased MRglu. There was no association between MRglu and chemotherapy history. We conclude that MRglu is increased in hepatic steatosis probably through insulin resistance, hyperinsulinaemia and up-regulation of hepatic hexokinase, irrespective of obesity.

Fallacy of Quantifying Lymphoma Activity by Scaling to the Liver in [18F]Fluorodeoxyglucose Positron Emission Tomography (Deauville criteria)

TO THE EDITOR: It is recommended that uptake of [F]fluorodeoxyglucose (FDG) in lymphoma, as determined from positron emission tomography, should be quantified by comparison with liver uptake (the Deauville criteria). We believe, however, that this is flawed because of the contrasting effects of blood glucose level on tumor and hepatic FDG concentrations. We recently investigated methods for scaling tumor FDG uptake, according to blood glucose, using the brain as a surrogate for tumors, arguing that brain and tumor FDG handling can be regarded as similar because they both metabolically trap FDG. Because FDG competes with blood glucose for tissue uptake in general, tumors and brain both have FDG concentrations that bear inverse relationships with blood glucose. Expressing FDG accumulation as the standardized uptake value (SUV), we found this relationship for the brain to be exponential, with exponential constants for several different brain regions all of approximately 0.1 mmol/L . In other words, SUV decreases by 10% in response to an increase in blood glucose of 1 mmol/L. A difficulty of studying the relationship between tumor FDG uptake and blood glucose in patients with cancer is that the glucose use rates of tumors are highly variable, depending, for example, on the histologic type and treatment stage, explaining why we used the brain as a surrogate for tumors. As far as we could discern from the limited literature on the quantitative acute effect of glucose administration on tumor FDG accumulation, the corresponding exponential constants for tumors are broadly similar to those for the brain. For example, the data of Lindholm et al give a constant between 0.1 and 0.15 mmol/L 1 for head and neck cancers, whereas those of Ishizu et al give a value of approximately 0.09 mmol/L 1 for gliomas. Lindholm et al found that the tumor glucose use rate (measured as FDG clearance multiplied by blood glucose) changed little after glucose loading, similarly to the brain, but different from muscle, which increases its glucose use rate in hyperglycemia. No data are available for lymphoma. Because of rapid exchange of FDG between liver and blood, hepatic FDG concentration, in contrast to tumor FDG concentration, closely parallels blood FDG concentration. Indeed, the liver has been suggested as an alternative blood pool region for FDG kinetic studies in small mammals. In other words, liver and brain show directionally opposite relationships with blood glucose. As shown from intrahepatic FDG kinetic studies before and after glucose loading or hyperinsulinemic clamp, the effect of hyperinsulinemia on the hepatic FDG signal is small compared with the signal from unphosphorylated FDG in dynamic exchange with blood FDG. Not unexpectedly, therefore, we found a strong inverse relationship between the ratio of brain-to-liver SUV and blood glucose. This relationship turned out to be hyperbolic. In tissues that maintain a constant rate of glucose use, the relationship between tissue FDG clearance and blood glucose would be hyperbolic (ie, y constant/x), indicating that the brain-to-liver SUV ratio is a closer reflection of brain FDG clearance than brain SUV alone. The same is likely to apply to tumors. In hyperbolic relationships in general, the greatest change in the y values occurs over the lower-to-mid range of x values. Accordingly, the greatest change in the brain-to-liver SUV ratio in relation to blood glucose occurred over the normoglycemic range. Thus, we found that a change in blood glucose from 3 to 6 mmol/L decreased the brain-toliver SUV ratio from approximately 5 to 2.5. The ratio of lymphomato-hepatic SUV values, therefore, is likely to be equally sensitive to blood glucose if, as in other tumors, the glucose use rate is independent of blood glucose. European Association of Nuclear Medicine guidelines recommend that tumor SUV should be normalized to blood glucose of 5 mmol/L; in other words, multiply the SUV by blood glucose and divide by 5. We found, however, that after this procedure, brain SUV was still dependent on blood glucose. When the normalized value was then divided by liver SUV, brain SUV showed a negligible relationship with blood glucose. We provided a theoretical explanation for this finding. These arguments apply equally to the mediastinum as a reference region for tumor uptake because the mediastinum largely gives a blood pool signal, which, as explained above, closely follows the hepatic signal. So, we believe that the Deauville criteria should be modified and that before scaling to liver SUV, lymphoma SUV should first be normalized to blood glucose, as per European Association of Nuclear Medicine guidelines.

Importance of accurate ilio-inguinal quantification in lower extremity lymphoscintigraphy

Aims The aims of this study were to improve the quantification of lower extremity lymphoscintigraphy, determine its value and lower limit of normal, and determine whether intermediate postinjection time imaging is necessary. Patients and methods This was a study of 102 consecutive patients undergoing routine lower extremity lymphoscintigraphy using subcutaneous 99mTc-99m-nanocolloid with imaging at 5, 45 and 150 min after injection. Abnormal imaging criteria were delay (no activity in ilio-inguinal nodes at 45 min or negligible activity at 150 min), lymph diversion (through skin or deep system) and focal accumulation suggesting cellulitis. Lymphatic function was quantified as % injected activity in ilio-inguinal nodes at 150 min (IIQ) using a standard placed, by image guidance, exactly over the nodes. Results Forty-one patients had bilateral normal scintigraphy. IIQ was normally distributed in 15 limbs, with IIQ of 1–7.5%. In contrast, it was log-normally distributed in 68 limbs, with IIQ of at least 7.5%, suggesting 8% as the lower limit of normal. In 57 limbs, delay was the only scintigraphic abnormality at 45 min. Of these, 33 were abnormal at 150 min. Of the remaining 24 limbs, 17 had reduced IIQ; thus, 50 of these 57 (88%) limbs had lymphatic dysfunction. The seven limbs that remained normal at 150 min were in six patients. The contralateral limb was abnormal in five of these six patients; hence, lymphatic dysfunction would have been missed in only one patient without 45 min imaging. Conclusion IIQ is strongly recommended. Isolated delay at 45 min is abnormal. However, 45 min imaging is not necessary if IIQ is performed.

Lymph proteins may access peripheral blood without entering thoracic duct in patients with lymphatic dysfunction.

OBJECTIVE The objective was to investigate the hypothesis that lymphovenous communications, which allow lymph proteins to access peripheral blood without first entering the thoracic duct, open in patients with abnormal lymphatic function. METHODS Routine lymphoscintigraphy of 182 patients, including 27 without clinical evidence of lymphedema (controls), was performed immediately and 45 and 150 minutes after subcutaneous injection of technetium Tc 99m nanocolloid into both feet. Counts per pixel in a region of interest over the liver (L) were divided by total counts in bilateral ilioinguinal nodes (N) at 45 minutes (L/N45) and 150 minutes (L/N150). If all activity leaving ilioinguinal lymph nodes entered the thoracic duct, these L/N ratios would be similar from patient to patient. RESULTS Eight patients were excluded because of immediate liver activity suggesting inadvertent intravascular injection of tracer. In controls (group 1), L/N150 displayed a normal distribution with mean (± standard deviation) of 0.16 (0.09) × 10(-4) pixels(-1). Patients with L/N150 >0.34 × 10(-4) pixels(-1) (ie, 0.16 + 2 standard deviations) were assumed to have lymphovenous communications. Of 34 patients with clinical evidence of lymphedema but with normal findings on lymphoscintigraphy (group 2), 3 (9%) had lymphovenous communications; of 114 with abnormalities on lymphoscintigraphy (group 3), 43 (38%) had lymphovenous communications (P = .001). N45/150 was significantly higher than L45/150 in all four groups, indicating arrival of activity in nodes before the liver. Abnormal features of lymphoscintigraphy-lymph transport delay, popliteal node visualization, and diversion of lymph through the skin-showed no association with L/N ratios. CONCLUSIONS Lymphovenous communications exist in about one-third of patients with abnormalities detected on lymphoscintigraphy. The timings of tracer arrival in the liver and lymph nodes is consistent with lymphovenous communication within lymph nodes themselves.

The appropriate whole-body index on which to base standardized uptake value in 2-deoxy-2-[18F]fludeoxyglucose PET

OBJECTIVE Tissue uptake of 2-deoxy-2-fluorine-18 fludeoxyglucose ((18)F-FDG) is routinely quantified as standardized uptake value (SUV), which in general is the fraction (F) of administered activity per millilitre of tissue multiplied by an index of body size, usually weight (W), i.e. F/ml × W = SUV or F/ml = SUV × (1/W). Other indices have been suggested as preferable to W, especially lean body mass (LBM) and body surface area (BSA). The second equation mentioned above shows that the reciprocal of the ideal index should correlate closely with F/ml and give a regression line through the origin. The purpose of this study was to determine which of these three indices best meets these criteria. METHODS Data were evaluated from 49 males and 51 females undergoing routine (18)F-FDG positron emission tomography/CT. A 3 cm diameter region of interest was drawn over the liver and F/ml recorded. LBM and BSA were estimated from height and weight. RESULTS Based on all patients, the reciprocals of the three indices gave similar correlation coefficients with F/ml, but only 1/LBM gave regressions close to the origin. Intercepts were significantly higher for females for 1/W and 1/BSA, consistent with females having more body fat, but there was no significant difference with 1/LBM. CONCLUSION LBM is the best index on which to base SUV because adipose tissue accumulates less (18)F-FDG than other soft tissues. ADVANCES IN KNOWLEDGE The value of this study lies in its use of a novel, more rational approach than previously to confirm that SUV should be based on LBM.

FAMILIAL COMBINED HYPERLIPIDEMIA IS CHARACTERIZED BY HIGHER HEPATIC FDG UPTAKE AND VISCERAL ADIPOSE TISSUE VOLUME COMPARED TO HETEROZYGOUS FAMILIAL HYPERCHOLESTEROLAEMIA

Familial combined hyperlipidaemia (FCH) phenotype is associated with increased prevalence of non-alcoholic fatty liver disease and obesity. No data exists regarding the impact of heterozygous familial hypercholesterolaemia (heFH), another type of familial dyslipidaemia, on liver function and

Assessment of alteration in liver 18 F–FDG uptake due to steatosis in lymphoma patients and its impact on the Deauville score

Dear Sir, We read with interest the article recently published in EJNMMI by Salomon et al. [1] that described a negative relationship between hepatic steatosis and SUV, and recommended the use of SULmax rather than SUVmax as the comparator in the Deauville scoring (DS) system because it minimises the influence of body mass index (BMI). They found, nevertheless, that it made little difference to DS whether SULmax or SUVmax was used. It is stated (but only in the Abstract) that in their scoring system, SULmax was applied to the lymphoma. Provided lymphoma and liver are compared either on the basis of SUVor SUL, body weight will cancel out and remove any influence of BMI on DS. It makes no sense to compare an SULwith an SUV, or vice versa, so normalising any tumour to liver removes the need to choose between SUL and SUV, provided whichever one is used is applied to both tissues. In any event, in our view, the liver is just about the least appropriate tissue against which to normalise tumour SUV because of the opposing relationships that tumour and liver SUV respectively have with blood glucose level. The wellknown inverse relationship between tumour SUV and blood glucose [2] implies that tumours are insensitive to insulin. This opposing relationship with blood glucose makes the ratio of tumour-to-liver SUVexquisitely sensitive to blood glucose. Using brain, which is also insensitive to insulin [3], as a surrogate for tumour, we previously demonstrated a hyperbolic relationship between brain-to-liver SUV ratio and blood glucose. If lymphoma behaves like brain, then a similar hyperbolic relationship would exist for lymphoma. Liver and brain SUValso display opposing relationships with time from FDG injection. Time course of tissue activity depends on the ratio of tissue blood clearance of FDG (to phosphorylation) and tissue FDG distribution volume. This ratio is much higher for brain, and almost certainly also for tumours, than for liver [4]. Even at 60 min post-injection, about 50% of administered FDG is still in the circulation [4]. The positive relationship between liver SUV and blood glucose then arises because at 60 min about 75% of intrahepatic FDG is unphosphorylated and in rapid dynamic exchange with blood FDG [4, 5]. Because of competition for tissue uptake, blood FDG level increases with increasing blood glucose level [6]. The liver is sensitive to insulin [7] so increasing blood glucose level increases the intrahepatic concentration of phosphorylated FDG, accentuating the positive relationship between blood glucose and hepatic SUV, although this is probably less important than unphosphorylated FDG. The dependence of liver SUVon blood FDG concentration is well shown in Fig. 4 of the paper of Salomon et al. in which in (b) there is a prominent blood pool signal matching the high liver activity, in contrast to (a) in which there is lower liver activity and a much less prominent blood pool signal. The old literature even went as far as suggesting the use of the liver as a blood pool region in small animal dynamic FDG studies in which placement of ROI over left ventricular cavity or major artery was not considered feasible [8]. Mediastinum, which is also widely used as a comparator to derive the DS, is also predominantly blood pool, and therefore similarly unsuitable for quantification of FDG uptake in lymphoma. Theoretically, FDG is ‘shared’ around the body such that whole body SUV is unity. Regional SUV is influenced by many factors, including the tumour burden, which determines how much FDG is available for uptake elsewhere (likewise treatment-activated bone marrow). Paradoxically, therefore, a response to therapy with reduction of tumour burden could result in an increase in tumour SUV. Duration of patient fasting, even when blood glucose is normal, is probably also a factor as it affects insulin levels. As it is intuitively likely that lymphoma and brain respond similarly to all these factors, it * A. Michael Peters a.m.peters@bsms.ac.uk

Fasting hepatic glucose uptake is higher in men than women

Differences in glucose metabolism between men and women have previously been reported. Our purpose was to determine if there is a gender difference in fasting hepatic glucose uptake (MRglu). Fifty‐five patients (44 men, 11 women) referred for routine PET/CT using the glucose tracer 2‐deoxy‐2‐[F‐18]fluoro‐D‐glucose (FDG), mainly for cancer, had dynamic imaging for 30 min immediately following injection. Hepatic FDG clearance (mL/min/100 mL) was measured as gradient divided by intercept from Patlak–Rutland graphical analysis using a volume of interest over the abdominal aorta to record input function. Hepatic MRglu was obtained by multiplication of clearance by blood glucose concentration. Hepatic steatosis was diagnosed as CT density ≤40 HU. Mean (standard deviation) hepatic MRglu in 44 men was 2.30 (1.14) μmol/min/100 mL, significantly higher than in 11 women in whom it was 1.07 (1.35) μmol/min/100 mL (P = 0.003). CT density was 52 (12) HU in women compared with 45 (9) HU in men (P = 0.04), but there was no significant difference in blood glucose, BMI, or prevalence of recent chemotherapy (within 6 months preceding PET/CT). When patients were subdivided into those without hepatic steatosis (31 men/9 women), those without evidence of FDG‐avid malignancy on PET/CT (15/6), and those without either (11/5), gender differences in hepatic MRglu remained highly significant, but there were no significant differences in CT density, blood glucose, BMI, or recent chemotherapy history. Despite this being a population of clinically referred patients, the results strongly suggest that fasting hepatic MRglu is higher in men than in women.