Correlation Between Liver Histology and Novel MRI in NAFLD
Correlation Between Liver Histology and Novel MRI in NAFLD
In this cohort of adult patients with NAFLD, MRI-determined PDFF is closely associated with histology-determined steatosis grade. The average PDFF at grades 1, 2 and 3 steatosis were 8.9%, 16.3% and 25.0% respectively, (P < 0.0001). Previous studies have shown that MRI fat fraction calculated by the Dixon/IOP method correlates well with hepatic steatosis in adult patients with liver disease consisting of various aetiologies and in paediatric patients with NAFLD. In addition, MRS has also been shown to be an accurate technique in quantifying hepatic steatosis, but remains largely a research tool. This study uses an improved, well-validated PDFF technique in a middle-aged, adult cohort of patients with biopsy-proven NAFLD. Two general approaches for MRI-determined PDFF estimation have been developed- one technique uses complex data, and the other technique uses magnitude data. Based on phantom studies and comparisons with MR spectroscopy, both approaches appear to be superior to conventional approaches and similar to one another. Our study selected the magnitude technique because our team had expertise in its use and UCSD liver imaging group is developing tools to apply this technique in the clinical setting. Using this magnitude approach, we found that that MRI-PDFF quantification significantly correlates with grade of steatosis by liver biopsy. This lends support to the potential use of MRI-determined PDFF as a non-invasive and quantitative means to quantify hepatic steatosis in patients with NAFLD.
Furthermore, the inverse relationship between advanced fibrosis and degree of hepatic steatosis as seen in our subset analysis of patients with stage 4 fibrosis and grade 1 steatosis was previously not well understood in adult patients with NAFLD. This study suggests that stage 4 fibrosis is associated with decreased hepatic steatosis by both a lower MRI-determined PDFF and histology-determined steatosis grade than patients with stage 0–3 fibrosis (7.6% vs. 17.8%, P < 0.005; and 1.4 vs. 2.2, P = 0.03 respectively). Compared to patients with histology-determined grade 3 steatosis, when patient's with grade 1 steatosis were examined further, their biochemical profile on average suggested more severe liver disease as reflected by higher GGT, AST to ALT ratio, INR, as well as hepatocellular ballooning on liver histology.
As more and more clinicians utilise non-invasive assessment of NAFLD using imaging techniques it is critical to understand the caveats associated with assessment of imaging-determined hepatic steatosis in NAFLD. This study illustrates that a lower hepatic fat content on imaging should not be confused as mild NAFLD but other parameters should be considered and a liver biopsy may be entertained if there is a suspicion of advanced NASH or fibrosis despite lower hepatic fat content on imaging. We would like to propose that less hepatic steatosis (measured either by MRI-PDFF or liver biopsy) is perhaps bi-modally distributed with respect to disease severity: with a group of mild NAFLD having less fibrosis (no fibrosis) and a group of severe NAFLD or even NASH-related cirrhosis where hepatic steatosis is replaced by collagen (more advanced fibrosis). Thus, imaging-determined hepatic steatosis may not be a reliable measure of the severity of NAFLD, and one should remain cautious about interpreting low levels of hepatic steatosis as early or mild NAFLD. The exact mechanisms underlying this association remain to be elucidated and are beyond the scope of the present study.
The major strengths of the study include the use of well-characterised adult patients with liver biopsy-confirmed NAFLD, and novel (and detailed) MRI imaging techniques, prospectively collected biochemical and MRI data, and inclusion of both genders. The study utilised a well-validated, MRI-determined PDFF technique which improves on previous technique of Dixon/IOP methods by correcting for T1 bias and bias, T2* decay, eddy currents and the spectral complexity of fat. Furthermore, the biopsy and MRI data were obtained within an average time between biopsy and MRI of 2.1 months. The radiologist and the pathologists were blinded to clinical as well as pathology and MRI data, respectively. Statistical analyses were performed after all data collections and quality control procedures were completed. Despite these strengths, the study is limited by the small number patients and lacks the power to detect differences in stage by stage changes in liver fat content. Secondly, there was no control or non-NAFLD arm, although this would not have supported the primary aim to study MRI-PDFF specifically in NAFLD patients. Lastly, we only used one of the two available MRI-PDFF techniques, and this selection of technique was due to availability of local expertise in the magnitude technique at UCSD. We recently showed that this MRI-PDFF technique has an excellent correlation with MRS (r = 0.98). Therefore, we believe that our results are valid for NAFLD patients examined in this study.
Discussion
Main Findings
In this cohort of adult patients with NAFLD, MRI-determined PDFF is closely associated with histology-determined steatosis grade. The average PDFF at grades 1, 2 and 3 steatosis were 8.9%, 16.3% and 25.0% respectively, (P < 0.0001). Previous studies have shown that MRI fat fraction calculated by the Dixon/IOP method correlates well with hepatic steatosis in adult patients with liver disease consisting of various aetiologies and in paediatric patients with NAFLD. In addition, MRS has also been shown to be an accurate technique in quantifying hepatic steatosis, but remains largely a research tool. This study uses an improved, well-validated PDFF technique in a middle-aged, adult cohort of patients with biopsy-proven NAFLD. Two general approaches for MRI-determined PDFF estimation have been developed- one technique uses complex data, and the other technique uses magnitude data. Based on phantom studies and comparisons with MR spectroscopy, both approaches appear to be superior to conventional approaches and similar to one another. Our study selected the magnitude technique because our team had expertise in its use and UCSD liver imaging group is developing tools to apply this technique in the clinical setting. Using this magnitude approach, we found that that MRI-PDFF quantification significantly correlates with grade of steatosis by liver biopsy. This lends support to the potential use of MRI-determined PDFF as a non-invasive and quantitative means to quantify hepatic steatosis in patients with NAFLD.
Furthermore, the inverse relationship between advanced fibrosis and degree of hepatic steatosis as seen in our subset analysis of patients with stage 4 fibrosis and grade 1 steatosis was previously not well understood in adult patients with NAFLD. This study suggests that stage 4 fibrosis is associated with decreased hepatic steatosis by both a lower MRI-determined PDFF and histology-determined steatosis grade than patients with stage 0–3 fibrosis (7.6% vs. 17.8%, P < 0.005; and 1.4 vs. 2.2, P = 0.03 respectively). Compared to patients with histology-determined grade 3 steatosis, when patient's with grade 1 steatosis were examined further, their biochemical profile on average suggested more severe liver disease as reflected by higher GGT, AST to ALT ratio, INR, as well as hepatocellular ballooning on liver histology.
As more and more clinicians utilise non-invasive assessment of NAFLD using imaging techniques it is critical to understand the caveats associated with assessment of imaging-determined hepatic steatosis in NAFLD. This study illustrates that a lower hepatic fat content on imaging should not be confused as mild NAFLD but other parameters should be considered and a liver biopsy may be entertained if there is a suspicion of advanced NASH or fibrosis despite lower hepatic fat content on imaging. We would like to propose that less hepatic steatosis (measured either by MRI-PDFF or liver biopsy) is perhaps bi-modally distributed with respect to disease severity: with a group of mild NAFLD having less fibrosis (no fibrosis) and a group of severe NAFLD or even NASH-related cirrhosis where hepatic steatosis is replaced by collagen (more advanced fibrosis). Thus, imaging-determined hepatic steatosis may not be a reliable measure of the severity of NAFLD, and one should remain cautious about interpreting low levels of hepatic steatosis as early or mild NAFLD. The exact mechanisms underlying this association remain to be elucidated and are beyond the scope of the present study.
Strengths and Limitations of the Study
The major strengths of the study include the use of well-characterised adult patients with liver biopsy-confirmed NAFLD, and novel (and detailed) MRI imaging techniques, prospectively collected biochemical and MRI data, and inclusion of both genders. The study utilised a well-validated, MRI-determined PDFF technique which improves on previous technique of Dixon/IOP methods by correcting for T1 bias and bias, T2* decay, eddy currents and the spectral complexity of fat. Furthermore, the biopsy and MRI data were obtained within an average time between biopsy and MRI of 2.1 months. The radiologist and the pathologists were blinded to clinical as well as pathology and MRI data, respectively. Statistical analyses were performed after all data collections and quality control procedures were completed. Despite these strengths, the study is limited by the small number patients and lacks the power to detect differences in stage by stage changes in liver fat content. Secondly, there was no control or non-NAFLD arm, although this would not have supported the primary aim to study MRI-PDFF specifically in NAFLD patients. Lastly, we only used one of the two available MRI-PDFF techniques, and this selection of technique was due to availability of local expertise in the magnitude technique at UCSD. We recently showed that this MRI-PDFF technique has an excellent correlation with MRS (r = 0.98). Therefore, we believe that our results are valid for NAFLD patients examined in this study.