Marked difference in liver fat measured by histology vs. magnetic resonance-proton density fat fraction: A meta-analysis

S. U. Qadri, Emilia Vartiainen, Mari Lahelma, Kimmo Porthan, An Tang, Ilkay S. Idılman, Jurgen H. Runge, Anne Juuti, Anne Penttilä, Juhani Dabek, Tiina E. Lehtimäki, Wenla Seppänen, Johanna Arola, Perttu Arkkila, Jaap Stoker, Muşturay Karçaaltıncaba, Michael Pavlides, Rohit Loomba, Claude B. Sirlin, Taru Tukiainen, Hannele Yki‐JärvinenJHEP 2023 · DOI

TL;DR

Histology and MRI-based PDFF are fundamentally different methods for measuring liver fat, with histology often reporting significantly higher values, especially at higher fat concentrations. This meta-analysis provides crucial conversion equations and thresholds to bridge this gap, enabling more accurate interpretation of PDFF measurements in clinical practice.

What Problem Does This Paper Address?

Diagnosing and monitoring non-alcoholic fatty liver disease (NAFLD) relies on quantifying liver fat. Traditionally, this has been done through liver biopsy, where pathologists estimate the percentage of hepatocytes containing fat droplets (histology). However, liver biopsy is invasive and subject to inter-observer variability. Magnetic Resonance Imaging (MRI)-based proton density fat fraction (PDFF) is emerging as a non-invasive alternative. While PDFF is increasingly used, its direct comparison to histological steatosis has been challenging due to their different underlying principles. This paper addresses the critical need to understand the quantitative relationship between these two methods to prevent misinterpretations of clinical status and treatment effects.

Methods

The researchers conducted a systematic review and meta-analysis of nine studies involving 597 individuals. They pooled individual patient-level data to compare histological steatosis (assessed by pathologists) with PDFF measurements obtained via MRI or magnetic resonance spectroscopy (MRS).

Key methodological aspects include:

  • Meta-analysis of correlation coefficients: To assess the linearity of the relationship between histology and PDFF.
  • Linear mixed modeling: To model the pooled data and derive a best-fit curve for the relationship.
  • Bland-Altman analysis: To evaluate the agreement between the two methods.
  • Receiver Operating Characteristic (ROC) analysis: To determine optimal PDFF thresholds for classifying different grades of histological steatosis.
  • RNA-seq analysis: To determine the proportion of hepatocytes in human liver biopsies, providing a potential explanation for observed differences.

Datasets

The study utilized data from nine previously published studies and included unpublished data from the Helsinki MRS-PDFF cohort. The combined dataset comprised 597 unique subjects. No new benchmark datasets were introduced; instead, existing data were pooled and analyzed.

Key Insights and Ideas

  • Fundamental Differences: Histology and PDFF are not directly interchangeable. Histology estimates the proportion of affected cells (hepatocytes), while PDFF measures the tissue concentration of triacylglycerol.
  • Non-Linear Discrepancy: While the relationship is largely linear, PDFF values are consistently lower than histological steatosis, especially as liver fat content increases. This discrepancy widens significantly above 5% steatosis.
  • Hepatocyte Proportion Matters: The fact that hepatocytes constitute only about 58% of liver cells (as shown by RNA-seq) provides a biological explanation for why PDFF, which measures fat in the entire sampled liver volume, would be lower than histology, which focuses solely on hepatocytes.
  • Thresholds are Essential: Standard histological grading thresholds are not directly applicable to PDFF. The study provides specific, cross-validated PDFF thresholds to accurately predict histological steatosis grades, which are crucial for clinical decision-making.

Key Results

  • Linearity: The relationship between histological steatosis and PDFF was predominantly linear, with a meta-analytic correlation coefficient (r) of 0.85.
  • Discrepancy: At 5% histological steatosis, PDFF and histology approximately coincided. However, above 5%, histology values diverged markedly, with PDFF being up to 3.4-fold lower.
  • Average Difference: 100% histological steatosis corresponded to an average PDFF of 33.0%.
  • Hepatocyte Fraction: Hepatocytes comprised 58.5 ± 5.2% of liver cells, and this fraction negatively correlated with liver fat content.
  • PDFF Thresholds (90% specificity):
    • ≥5.75% PDFF for predicting ≥S1 (NAFLD)
    • ≥15.50% PDFF for predicting ≥S2 (moderate-to-severe steatosis)
    • ≥21.35% PDFF for predicting S3 (severe steatosis)

Limitations and Future Work

  • Methodological Variability: The study acknowledges variability in MR protocols and, more significantly, the inherent subjectivity and potential sampling error of histological assessment, which likely contributed to between-study heterogeneity.
  • Digital Histology: The authors suggest that future studies should investigate the relationship between PDFF and digital image analysis of histology, as this method might offer more standardized and less variable steatosis quantification.
  • Clinical Outcomes: While the study provides thresholds for predicting histological grades, it suggests that future research could also focus on determining PDFF levels associated with clinically significant increases in adverse liver-related outcomes.
  • Non-Parenchymal Cells: While the RNA-seq analysis provided insight into the hepatocyte fraction, the study acknowledges that it discounted volume differences between cells and did not measure extracellular water, which could also influence PDFF.

Summary updated

Abstract

Magnetic resonance-proton density fat fraction (PDFF) is increasingly being used to measure liver fat in place of the invasive liver biopsy. Understanding the relationship between PDFF and histological steatosis fraction is important for preventing misjudgement of clinical status or treatment effects in patient care. Our analysis revealed that histological steatosis fraction is often significantly higher than PDFF, and their association varies across the spectrum of fatty liver severity. These findings are particularly important for physicians and clinical researchers, who may use these data to interpret PDFF measurements in the context of histologically evaluated liver fat content.
BibTeX
@inproceedings{qadri2023marked,
  author = {S. U. Qadri and Emilia Vartiainen and Mari Lahelma and Kimmo Porthan and An Tang and Ilkay S. Idılman and Jurgen H. Runge and Anne Juuti and Anne Penttilä and Juhani Dabek and Tiina E. Lehtimäki and Wenla Seppänen and Johanna Arola and Perttu Arkkila and Jaap Stoker and Muşturay Karçaaltıncaba and Michael Pavlides and Rohit Loomba and Claude B. Sirlin and Taru Tukiainen and Hannele Yki‐Järvinen},
  title = {{Marked difference in liver fat measured by histology vs. magnetic resonance-proton density fat fraction: A meta-analysis}},
  booktitle = {JHEP},
  year = {2023},
  doi = {10.1016/j.jhepr.2023.100928},
}
Figure