Scholarlink

Background <p>Hepatic steatosis grades derived from magnetic resonance imaging proton density fat fraction (MRI-PDFF) might disagree with those determined by histology. We investigated whether the size distribution of lipid droplets (LDs) assessed with digital image analysis (DIA) explains the discrepancies between histology and MRI-PDFF.</p> Materials and methods <p>Multicentric, prospective study of 355 patients with chronic liver disease, having paired biopsy and MRI. Using conventional microscopy, steatosis was graded by pathologists (S0-S3), based on the proportion of hepatocytes containing large LDs. MRI-PDFF graded steatosis using validated thresholds (PDFF-S1 ≥ 5.75%, PDFF-S2 ≥ 15.5%, PDFF-S3 ≥ 21.35%). DIA categorized LDs into tiny (&lt; 1 μm<sup>2</sup>), small (1–100 μm<sup>2</sup>), and large (≥ 100 μm<sup>2</sup>) subtypes. Multivariable modeling was performed to identify predictors of discordances.</p> Results <p>Histology- and PDFF-derived steatosis grades were discordant in 36%. Disagreement was associated with higher proportions of tiny and small LDs, and lower content of large LDs. Within histology-derived S0, disagreements were due to MRI overestimation having a higher content of tiny and small LDs. Within histology-derived S2-S3, disagreements were due to MRI underestimation, having a lower content of large and total LDs. Total LD proportionate area strongly correlated with MRI-PDFF (<i>r</i> = 0.89). DIA showed that as steatosis accumulates, the size of LDs progressively increases.</p> Conclusion <p>Taking DIA as ground truth, the size distribution of LDs explains the discrepancies between histology and MRI-PDFF steatosis gradings. Compared to histology, MRI-PDFF overestimation is due to abundant tiny-small LDs, and underestimation is due to lower content of large LDs. These results indicate that MRI captures the whole spectrum of LD size, avoiding the conventional histology subjective assessment bias.</p> Key Points <p><Emphasis Type="BoldItalic">Question</Emphasis><i> Can the size distribution of hepatic lipid droplets (LDs), determined with digital pathology, provide insights into the discordances between histology and MRI-PDFF steatosis grading?</i></p> <p><Emphasis Type="BoldItalic">Findings</Emphasis><i> There is a 36% disagreement rate between histology and MRI-PDFF. Tiny and small LDs explain discordance, while large LDs favor concordance between both techniques</i>.</p> <p><Emphasis Type="BoldItalic">Clinical relevance</Emphasis><i> MRI-PDFF captures the whole spectrum of LDs size beyond pathologists scoring system and strongly correlates with LDs total content avoiding conventional histology bias</i><i>.</i></p> Graphical Abstract <p></p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Digital pathology and lipid droplet size as a key determinant of discrepancies between histology and MRI gradings in steatotic liver disease

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年8月8日

查看全文
  • David Marti-Aguado,
  • Clara Alfaro-Cervello,
  • Matías Fernández-Patón,
  • Amadeo Ten-Esteve,
  • Leonor Cerdá-Alberich,
  • Ana Crespo,
  • Irene Navarrete-Pérez,
  • María Pilar Ballester,
  • Alexandre Perez-Girbes,
  • Cristina Montón,
  • Judith Pérez-Rojas,
  • Víctor Puglia,
  • Antonio Ferrández,
  • Victoria Aguilera,
  • Desamparados Escudero-García,
  • Salvador Benlloch,
  • Ana Jimenez-Pastor,
  • Ángel Alberich-Bayarri,
  • Claude B. Sirlin,
  • Luis Marti-Bonmati

摘要

Background

Hepatic steatosis grades derived from magnetic resonance imaging proton density fat fraction (MRI-PDFF) might disagree with those determined by histology. We investigated whether the size distribution of lipid droplets (LDs) assessed with digital image analysis (DIA) explains the discrepancies between histology and MRI-PDFF.

Materials and methods

Multicentric, prospective study of 355 patients with chronic liver disease, having paired biopsy and MRI. Using conventional microscopy, steatosis was graded by pathologists (S0-S3), based on the proportion of hepatocytes containing large LDs. MRI-PDFF graded steatosis using validated thresholds (PDFF-S1 ≥ 5.75%, PDFF-S2 ≥ 15.5%, PDFF-S3 ≥ 21.35%). DIA categorized LDs into tiny (< 1 μm2), small (1–100 μm2), and large (≥ 100 μm2) subtypes. Multivariable modeling was performed to identify predictors of discordances.

Results

Histology- and PDFF-derived steatosis grades were discordant in 36%. Disagreement was associated with higher proportions of tiny and small LDs, and lower content of large LDs. Within histology-derived S0, disagreements were due to MRI overestimation having a higher content of tiny and small LDs. Within histology-derived S2-S3, disagreements were due to MRI underestimation, having a lower content of large and total LDs. Total LD proportionate area strongly correlated with MRI-PDFF (r = 0.89). DIA showed that as steatosis accumulates, the size of LDs progressively increases.

Conclusion

Taking DIA as ground truth, the size distribution of LDs explains the discrepancies between histology and MRI-PDFF steatosis gradings. Compared to histology, MRI-PDFF overestimation is due to abundant tiny-small LDs, and underestimation is due to lower content of large LDs. These results indicate that MRI captures the whole spectrum of LD size, avoiding the conventional histology subjective assessment bias.

Key Points

Question Can the size distribution of hepatic lipid droplets (LDs), determined with digital pathology, provide insights into the discordances between histology and MRI-PDFF steatosis grading?

Findings There is a 36% disagreement rate between histology and MRI-PDFF. Tiny and small LDs explain discordance, while large LDs favor concordance between both techniques.

Clinical relevance MRI-PDFF captures the whole spectrum of LDs size beyond pathologists scoring system and strongly correlates with LDs total content avoiding conventional histology bias.

Graphical Abstract