Nonalcoholic fatty liver disease (NAFLD) is a common cause of liver disease worldwide and is associated with systemic diseases including cardiovascular disease, type 2 diabetes, and obesity (1). Thus, it can be viewed as an early indicator of systemic disease. It is present in 17-33% of the population in Western countries and in 34-74% of diabetic patients (2). Magnetic resonance spectroscopy is currently one of the most accurate imaging methods for the evaluation of fatty liver (3) by the non-invasive measure of intrahepatic triglyceride levels. Betaine is a natural supplement that stimulates the biosynthesis of glutathione, an important antioxidant in the liver (4). Animal studies have shown betaine has a hepatoprotective effect in liver diseases and therefore it is anticipated to reduce fatty liver levels (5). It is therefore our aim to measure intrahepatic triglyceride levels before and after 3 months of betaine supplement in obese insulin resistant subjects. Furthermore, to determine if the methods used are reliable and reproducible, we analyzed the covariances of the average and the standard deviation of lipid concentrations in two different voxels.Four subjects with NAFLD (1 female age 46 years and 3 males ages 60-65 years) were recruited and consented under local IRB approval. Subjects were scanned before beginning treatment with betaine and three months after. Breath-held single voxel PRESS spectroscopy (TE: 35 ms, TR: 2000 s, voxel size: 20x20x20, 8 averages with unsuppressed water spectrum) was acquired on at 3T MRI (Siemens Skyra) using 8 channel array abdominal surface coil. Two voxel locations were selected at each scan and manually shimmed to achieve optimal linewidths. In order to determine reproducibility of the IHTG measures, eight repetitions were acquired in each voxel. PRESS data was analyzed using LCModel and the primary lipid resonance comprised of β-carboxyl, methylene, and methyl resonances at 1.6, 1.3, and 0.9 ppm respectively. Secondary lipid resonance comprised of the diacyl, α-carboxyl, and α-olenic resonances at 2.8, 2.3, and 2.1 ppm respectively. A chemical shift was applied to the voxel spectra when necessary to center the water peak at 4.7 ppm. The ratios of the lipid peaks to the unsuppressed water are reported.Representative voxel locations and spectra are shown in Figure 1 and summarized in Figure 2. Results show that there is a significant difference found between baseline scan and post-betaine supplement scan in all four subjects. However, in two subjects, there was a significant decrease whereas in the other two subjects, there was a significant increase (Table 1). It is interesting to note that those subjects with baseline lower levels of IHTG showed improvement after betaine supplementation whereas those with high levels of liver steatosis showed no improvement and significant worsening of lipid levels. It is interesting to note however that in the one case shown in Figure 1 that the second voxel location showed improvement, demonstrating that there may be heterogeneity within the liver in terms of outcome. However, the second voxel was not measured in subjects 3 and 4 due to artifacts in the water reference. This may be due to difficulty with adequate breath-holds in these two subjects. Reproducibility of the repeated measures showed relatively low covariance (<10%) in most subjects, although Subject 4 had higher covariance at 27% as described in Table 1 which is also likely due to the lack of consistency of breath-holds. The results demonstrate that betaine appears to have a limited effect in patients that may be dependent on basal levels of IHTG. Betaine may only improve at the earlier stages of NAFLD progression thereby allowing for the hepatoprotective effects whereas in those individuals whose fatty liver disease is too advanced, there is no effect of the treatment. While the results are very preliminary in the limited number of subjects, it helps to guide which candidate patients could benefit most from the treatment. The results also indicate that there appears to be some variability in the effect of the treatment in different voxel locations. This requires a greater number of subjects as it is likely due to individual differences in subjects and additional locations should be measured. Finally, the reproducibility of the IHTG measurements is quite good although it also appears to be subject dependent. This is most likely due to the subject’s ability to consistently hold their breaths. This is a common issue to both breath-hold and prospective motion-corrected MRS and should highlight the need for better coaching at the time of the exam.