Utility of 99mTc-sestamibi Heart/Liver uptake ratio in Screening Non-alcoholic Fatty Liver Disease during Myocardial Perfusion Imaging

Purpose Non-alcoholic fatty liver disease (NAFLD) is the most common chronic hepatic disease worldwide with functional impairment of the mitochondria occurring from early stages. Technetium-99m methoxy-isobutyl-isonitrile (99mTc-MIBI) is a lipophilic agent trapped in the mitochondria. This study aims to evaluate the utility of 99mTc-MIBI heart/liver uptake ratio in screening for NAFLD during myocardial perfusion imaging (MPI).


Introduction
Signi cantly correlated with metabolic syndrome, non-alcoholic fatty liver disease (NAFLD) has become a global "pandemic" with increasing prevalence, affecting about 24.1% of the population. This umbrella term covers a wide spectrum of conditions, from simple hepatic steatosis to non-alcoholic steatohepatitis (NASH), with presence of in ammation and hepatocyte damage (steatohepatitis), ultimately leading to cirrhosis +/-hepatocellular carcinoma in most severe cases [1,2]. NAFLD is associated with atherogenic dyslipidaemia and an increased cardiovascular risk [3], with cardiovascular disease being the leading cause of death in these patients [4]. While liver biopsy remains the gold standard method of diagnosis, there is growing interest in implementing non-invasive diagnostic modalities, such as liver ultrasound and magnetic resonance spectroscopy (MRS), with the former currently recommended as the rst-line screening modality in selected populations [5]. Despite all, NAFLD remains an underdiagnosed condition and is generally underappreciated by primary health care physicians [6]. Myocardial perfusion imaging (MPI) with Technetium-99m methoxy-isobutyl-isonitrile (99mTc-MIBI) is a mainstay modality in diagnosis and management of coronary artery disease, as a non-invasive, accessible, cost-effective modality with relatively low radiation burden. 99mTc-MIBI is a lipophilic cationic agent, predominantly retaining in the mitochondria, because of their signi cantly negative transmembrane potential. Therefore, it is also being implemented in assessing the mitochondrial metabolic impairment and oxidative stress in various other mitochondrial diseases and neurodegenerative disorders [7,8]. Responsible for fatty acid oxidation, structural and molecular alterations of mitochondria has a pivotal role in the pathogenesis of NAFLD. Mitochondrial function impairment occurs from the early stages of hepatic steatosis, gradually leading to oxidative stress and in ammation, eventually resulting in apoptosis of hepatocytes and brotic changes [9,10]. Considering the independent risk NAFLD poses for coronary artery disease (CAD), as well as the signi cant correlation both have with metabolic syndrome; NAFLD is expected to be more prevalent in candidates for MPI scan [3,4]. This study aims to evaluate the potential role of determining the relative hepatic uptake of 99mTc-MIBI during rest phase MPI in screening NAFLD in patients referred for ruling out myocardial ischemia; thereby, facilitating its early diagnosis and treatment.

Patient Selection And Allocation
This study was approved by the Clinical Research Ethics Committee of the Medical School of Shahid Beheshti Medical University (IR.SBMU.MSP.REC.1398.308). All procedures involving human participants were carried out in accordance with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards. All patients were interviewed by a nuclear medicine physician and were thoroughly informed about the methods and objectives of this study. Eligible patients were included only after having their oral and written informed consent documented.
This prospective, single-center study compromised 70 eligible patients who were referred for 99mTc-MIBI MPI to rule out myocardial ischemia during a period of 6 months. Patients with any documented cardiac disease, patients with high pre-test probability of ischemic heart disease (IHD), patients with history of diabetes mellitus, regular alcohol consumption (>20 g per day), benign or malignant hepatic neoplasm, any kind of viral, metabolic or autoimmune liver disease as well as patients taking any hepatic steatosis-inducing medications (such as corticosteroids, Methotrexate, Tamoxifen and insulin), were not included in the study. 197 patients with low to intermediate/intermediate pre-test probability of IHD, who were referred for MPI to rule out myocardial ischemia, underwent a two-day rest/stress 99mTc-MIBI gated single photon emission computed tomography (SPECT), from whom 127 patients with abnormal myocardial perfusion or signi cant soft tissue attenuation effects, de ned as summed stress score (SSS) or summed rest score (SRS) > 2, were excluded from the study. Acquisition details will be discussed in the following section. The remaining 70 participants had their body mass index (BMI = Height/Weight 2 ) documented and underwent peripheral venous blood sampling, after a 9-hour fasting, to check their biochemical pro le and liver function tests (LFT); including fasting blood sugar (FBS), triglyceride (TG), Total Cholesterol, aspartate aminotransferase (AST), alanine aminotransferase (ALT), Albumin (Alb) and prothrombin time (PT). All patients underwent liver ultrasound examination, according to which were allocated into group A, patients with any degree of fatty liver disease, and group B, patients with normal results having no evidence of fatty liver disease.

Acquisition Protocols And Image Processing
All patients underwent a two-day rest/stress 99mTc-MIBI gated SPECT MPI, according to the European Association of Nuclear Medicine guidelines for MPI scan [7]. All 99mTc-MIBI kits underwent thin layer chromatography and radiochemical purity > 95% was deemed acceptable for utilization. Following IV injection of 555 MBq 99mTcsestamibi, a 2-minute planar image of the thoraco-abdominal region was acquired at 30, 60 and 120 minutes, in anterior view, with the patient lying supine. The standard rest SPECT was acquired, as per usual, at 60 minutes postinjection. On the day after, myocardial stress was achieved by exercise (Bruce protocol) or Dipyridamole injection (0.14mg/kg/min over 4 minutes), with 740 MBq 99mTc-sestamibi injected at peak stress and acquisition performed as per routine. All patients took 120 cc of whole milk 10 minutes after each tracer injection. Images were acquired with a variable angle dual head Evo-Voxel Siemens gamma camera with low-energy high resolution collimator, 140 Kev photopeak setting, symmetrical width energy window of 10%, matrix 64 * 64 and zoom of 1.46; including the rest planar image (2 minute duration, anterior view) and standard SPECT images (64 projections; 25 seconds/projection, 8-frame gated study, non-circular orbit). Images were processed by a nuclear medicine physician, blinded to patient allocations, using quantitative perfusion SPECT (QPS) (Fig. 1) and quantitative gated SPECT (QGS) software (Cedars-Sinai Medical Center Cardiac Suite). As described above, patients with abnormal myocardial perfusion or signi cant soft tissue attenuation effects, de ned as summed stress score (SSS) or summed rest score (SRS) > 2, were excluded from the study. Hand-drawn regions of interest (ROI) were set on the each planar rest image (3 images per person at 30, 60 and 120 minutes), with identical size and shape ROIs placed on the upper segments of right liver lobe as well as anterolateral wall of the left ventricle, to minimize the interfering effects (such as scatter) of these adjacent organs (Fig. 2). The mean count per pixel values were obtained and the heart to liver ratios calculated, which considering that we had excluded patients with any cardiac disease or abnormal MPI results, would be an indicator of relative hepatic uptake.

Statistical Analysis
Quantitative and categorical or qualitative data were represented by the mean, standard deviation (SD) and numbers/percentages, respectively. We used the Kolmogorov-Smirnov test for the assessment of the normality of distribution. Mann-Whitney U-test was utilized to determine the statistical differences between two groups of normal and NAFLD patients. P-value < 0.05 was considered signi cant. Random forest classi er with 5-fold cross validation was used to develop a model based on clinical and scintigraphy ndings to predict NAFLD, by implementing three different strategies (Scheme: weka.classi ers.trees.RandomForest -P 100 -I 100 -num-slots 1 -K 0 -M 1.0 -V 0.001 -S 1). Area under the receiver operating characteristic curve (AUROC) was selected as the assessment criterion.
Results 70 patients, consisting of 24 males (34.3%) and 46 females (65.7%), with the mean age of 51.0 ± 9.0 years; were categorized into groups A, having evidence of fatty liver disease (n = 30); and B, with normal liver (n = 40). The demographic, clinical and laboratory characteristics of the participants are presented in Table 1. Age, gender and liver function test results (including AST, ALT, serum Alb and PT) had no statistically signi cant difference among the groups (P-value > 0.05). As expected, BMI, TG and total cholesterol levels were signi cantly higher in group A (Pvalue = 0.000). Regarding the intrahepatic 99mTc-MIBI uptake, the mean count per pixel heart/liver ratios gradually increased over time in either groups; nonetheless the values were signi cantly higher in group A, regardless of acquisition timing; with the P-value equal to 0.007, 0.014 and 0.010 at 30, 60 and 120 minutes, respectively ( Table 2). As for random forest classi er results (Table 3) Repeating the third strategy by including single time point scintigraphy data at 1 hr. resulted in AUROC of 0.85, comparable to the accuracy of multiple time point measurement. Heart/liver uptake ratios higher than the threshold of 1.0, at 1 hour post radiotracer injection, warrant further evaluation for potential NAFLD (Fig. 3). and preventing further progression remains the most appropriate measure. Therefore, developing non-invasive, accessible, simple methods of screening and diagnosis is of utmost importance. Liver biopsy remains the gold standard of diagnosis, unsuitable for screening due its invasive nature, potential complications and cost. Serum biomarkers have proven to signi cantly underestimate presence of NAFLD. The most commonly implemented radiologic methods are ultrasound and magnetic resonance spectroscopy (MRS); however, the cost-effectiveness of screening the general population by them is still under debate [11]. Our study indicates that rest phase heart/liver uptake ratio of 99mTc-MIBI is signi cantly higher in patients with NAFLD; proposing it as a simple screening method in patients already undergoing MPI to rule out myocardial ischemia, who are at a higher risk for NAFLD as well, with no additional cost, radiation burden or complication.
As expected we observed a gradual rise in Heart/Liver ratios in either groups, which is explained by the physiologic clearance of 99mTc-MIBI from the liver [12]. NAFLD patients demonstrated signi cantly higher Heart/Liver ratios, regardless of measuring time, including 1 hour post injection, which is the usual time point when rest phase acquisition is performed. Our ndings are in accordance with ndings of Masuda K et al, who evaluated 26 patients with biopsy-proven NAFLD by performing laboratory tests and a 99mTc-MIBI liver scintigraphy. They acquired a 2 minute planar image of thoraco-abdominal region in anterior view, 10 minutes after bolus injection of 600 MBq 99mTc-MIBI in rest phase. Mean count per pixel liver/heart ratios were calculated by de ning ROIs in the right upper lobe of liver and the heart. They categorized the patients into non-NASH/simple hepatic steatosis (n=4), borderline NASH (n=11) and NASH (n=11) and reported that the relative hepatic uptake was signi cantly lower in NASH and borderline NASH compared to the non-NASH group, also compared to the healthy population (based on existing literature) and observed a signi cant correlation between the liver/heart ratio and degree of NASH [13]. Aside the small number of patients and not having a control group with the same exclusion criteria, a prominent limitation of this study is that the effect of cardiac uptake, which may be signi cantly reduced in patients with CAD and other cardiomyopathies, is not accounted for, as the patients did not undergo a stress phase MPI scan [13,14]. 99mTc-MIBI uptake portrays the mitochondrial function of cells [8] and NAFLD has been established as a disease of the mitochondria, with functional alterations of the mitochondria present even in the early stages of hepatic steatosis.
These functional alterations lead to gradual impairment of fatty acid oxidation, resulting in oxidative stress and in turn leading to in ammation (NASH), hepatocyte apoptosis and brosis in advanced stages [9,10], justifying the observed decreased hepatic uptake of 99mTc-MIBI in NAFLD patients compared to healthy individuals and possibly a correlation between relative uptake ratio and disease severity.
We developed a model to predict NAFLD based on patient's demographic, laboratory and scintigraphy (at 0.5, 1 and 2 hrs.) data. The model's strength while excluding laboratory data, which may not be always available, decreases to 0.84 AUROC, which is still su cient enough for a screening tool. Employing single time point scintigraphy measurement results at 1 hour post tracer injection, the time rest phase MPI acquisitions usually take place, yields a comparable accuracy (AUROC = 0.85), alleviating the need for multiple acquisitions, with heart/liver uptake ratios higher than the threshold of 1.0 warranting evaluation for potential NAFLD (Fig. 3). Naturally, implementing the predicting NAFLD, with their AUROC determined 0.874, 0.886, and 0.884, respectively [19]. Considering all these data, the accuracy of our scintigraphy-based model for predicting NAFLD (AUROC of 0.85) is acceptable and comparable to other commonly used US and laboratory-based indexes. Of course, we are not implying it as an independent screening method competing with liver ultrasound; but when applied in patients already undergoing MPI to rule out myocardial ischemia, it has the prominent advantage of it being a secondary nding of MPI, with no additional cost, radiation burden or side-effect.
Our study demonstrated no signi cant difference in age, gender distribution, AST, ALT, serum Albumin and PT levels among the groups; while BMI, TG and total cholesterol levels were expectedly higher in NAFLD patients. These ndings are in line with existing literature as liver function tests are proven to be neither sensitive nor speci c in screening for NAFLD and are normal in about 50% of NAFLD patients [20,21]. On the other hand hyperlipidemia is reported to be signi cantly associated with NAFLD. Nakahara et al, reported hyper-LDL cholesterolemia and hypo-HDL cholesterolemia in 37.5 and 19.5% of biopsy-proven NAFLD patients [22] and another study has determined TG as the strongest predictor of NAFLD among other parameters [23].

Study Limitations
This study has some limitations that should be considered. While liver ultrasound is currently recommended as the rst line screening method for NAFLD; liver biopsy remains the gold standard of diagnosis, despite being rarely performed for this sole purpose. Regarding its invasive nature, possible complications and signi cant cost, acquiring pathological correlation for all included patients in this study was not feasible, nor ethical. The relatively small number of patients as well as limited accuracy of ultrasound did not allow us to further categorize NAFLD patients based grade of disease. To prevent the interfering effect of possibly reduced hepatic uptake and rest phase myocardial activity on relative hepatic uptake ratios, both due to micro-vascular disease, diabetic patients were excluded from the study. Whether similar results would be obtained in these excluded patients should be assessed in future studies. To ensure better signal to noise ratios, the relative hepatic uptake was calculated on an additional 2-minute planar acquisition, future studies are required to assess the feasibility of using the routine SPECT projections, omitting the need for additional acquisitions, for this measurement.

New Knowledge Gained
The main novelties of our manuscript are the following: Permission to Reproduce Material from Other Sources: Not applicable.
Author Contribution: All authors have made substantial contributions to conception, design as well as acquisition of patients data; have been involved in drafting and revising the manuscript; have given nal approval of the version to be published with each of them having participated su ciently in the work to take public responsibility for appropriate portions of the content; and have agreed to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.