A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis

Natalie N. Atabaki; Daniel E. Coral; Hugo Pomares-Millan; Kieran Smith; Harry H. Behjat; Robert W. Koivula; Andrea Tura; Hamish Miller; Katherine Pinnick; Leandro Agudelo; Kristine H. Allin; Andrew A. Brown; Elizaveta Chabanova; Piotr J. Chmura; Ulrik P. Jacobsen; Adem Y. Dawed; Petra J. M. Elders; Juan J. Fernandez-Tajes; Ian M. Forgie; Mark Haid; Tue H. Hansen; Angus G Jones; Tarja Kokkola; Sebastian Kalamajski; Anubha Mahajan; Timothy J. McDonald; Donna McEvoy; Mirthe Muilwijk; Konstantinos D Tsirigos; Jagadish Vangipurapu; Sabine van Oort; Henrik Vestergaard; Jerzy Adamski; Joline W Beulens; Søren Brunak; Emmanouil T Dermitzakis; Giuseppe N Giordano; Ramneek Gupta; Torben Hansen; Leen M. t'Hart; Andrew T. Hattersley; Leanne Hodson; Markku Laakso; Ruth J. F. Loos; Jordi Merino; Mattias Ohlsson; Oluf Pedersen; Martin Ridderstråle; Hartmut Ruetten; Femke Rutters; Jochen M. Schwenk; Jeremy Tomlinson; Mark Walker; Hanieh Yaghootkar; Fredrik Karpe; Mark I McCarthy; Elizabeth Louise Thomas; Jimmy D. Bell; Andrea Mari; Imre Pavo; Ewan R. Pearson; Ana Viñuela; Paul W. Franks

doi:10.1016/j.metabol.2026.156552

A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis

Natalie N. Atabaki (Lead / Corresponding author)
, Daniel E. Coral
, Hugo Pomares-Millan
, Kieran Smith
, Harry H. Behjat
, Robert W. Koivula
, Andrea Tura
, Hamish Miller
, Katherine Pinnick
, Leandro Agudelo
, Kristine H. Allin
, Andrew A. Brown
, Elizaveta Chabanova
, Piotr J. Chmura
, Ulrik P. Jacobsen
, Adem Y. Dawed
, Petra J. M. Elders
, Juan J. Fernandez-Tajes
, Ian M. Forgie
, Mark Haid

Tue H. Hansen, Angus G Jones, Tarja Kokkola, Sebastian Kalamajski, Anubha Mahajan, Timothy J. McDonald, Donna McEvoy, Mirthe Muilwijk, Konstantinos D Tsirigos, Jagadish Vangipurapu, Sabine van Oort, Henrik Vestergaard, Jerzy Adamski, Joline W Beulens, Søren Brunak, Emmanouil T Dermitzakis, Giuseppe N Giordano, Ramneek Gupta, Torben Hansen, Leen M. t'Hart, Andrew T. Hattersley, Leanne Hodson, Markku Laakso, Ruth J. F. Loos, Jordi Merino, Mattias Ohlsson, Oluf Pedersen, Martin Ridderstråle, Hartmut Ruetten, Femke Rutters, Jochen M. Schwenk, Jeremy Tomlinson, Mark Walker, Hanieh Yaghootkar, Fredrik Karpe, Mark I McCarthy, Elizabeth Louise Thomas, Jimmy D. Bell, Andrea Mari, Imre Pavo, Ewan R. Pearson, Ana Viñuela, Paul W. Franks (Lead / Corresponding author)

Research output: Contribution to journal › Article › peer-review

4 Downloads (Pure)

Abstract

Objective: To delineate organ-specific and systemic drivers of metabolic dysfunction-associated steatotic liver disease (MASLD), we applied integrative causal inference across clinical, imaging, and proteomic domains in individuals with and without type 2 diabetes (T2D).

Methods: Bayesian network analyses and complementary two-sample Mendelian randomization were used to quantify causal pathways linking adipose distribution, glycemia, and insulin dynamics with liver fat in the IMI-DIRECT prospective cohort study. Data included frequently sampled metabolic challenge tests, MRI-derived abdominal and hepatic fat content, serological biomarkers, and Olink plasma proteomics from 331 adults with new-onset T2D and 964 adults without diabetes, with harmonized protocols enabling replication.

Results: High basal insulin secretion rate (BasalISR), estimated via C-peptide deconvolution, emerged as the primary potential causal driver of liver fat accumulation in both cohorts. BasalISR, a clearance-independent measure of β-cell insulin output distinct from peripheral insulin levels, was independently linked to hepatic steatosis. Visceral adipose tissue exhibited bidirectional associations with liver fat, suggesting a self-reinforcing metabolic loop. Of 446 analyzed proteins, 34 mapped to these metabolic networks (27 in the non-diabetes network, 18 in the T2D network, and 11 shared). Key proteins directly associated with liver fat included GUSB, ALDH1A1, LPL, IGFBP1/2, CTSD, HMOX1, FGF21, AGRP, and ACE2. Sex-stratified analyses identified GUSB in females and LEP in males as the strongest protein predictors of liver fat.

Conclusions: BasalISR may better capture early β-cell-driven disturbances contributing to MASLD. These findings outline a multifactorial, sex- and disease stage-specific proteo-metabolic architecture of hepatic steatosis and identify potential biomarkers or therapeutic targets.

Original language	English
Article number	156552
Number of pages	15
Journal	Metabolism: Clinical and Experimental
Volume	178
Early online date	6 Feb 2026
DOIs	https://doi.org/10.1016/j.metabol.2026.156552
Publication status	E-pub ahead of print - 6 Feb 2026

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

SDG 3 Good Health and Well-being

Keywords

Basal insulin secretion
Bayesian networks
Hepatic steatosis
MASLD
Mendelian randomization
Proteomics
Type 2 diabetes

ASJC Scopus subject areas

Endocrinology, Diabetes and Metabolism
Endocrinology

Access to Document

10.1016/j.metabol.2026.156552Licence: CC BY

Final Published VersionFinal published version, 3.59 MBLicence: CC BY

DIRECT: Diabetes Research on Patient Stratification (joint with 25 other partners)
Colhoun, H. (Investigator), Houston, G. (Investigator), Morris, A. (Investigator), Palmer, C. (Investigator) & Pearson, E. (Investigator)
COMMISSION OF THE EUROPEAN COMMUNITIES
1/02/12 → 28/02/27
Project: Research

Cite this

Atabaki, N. N., Coral, D. E., Pomares-Millan, H., Smith, K., Behjat, H. H., Koivula, R. W., Tura, A., Miller, H., Pinnick, K., Agudelo, L., Allin, K. H., Brown, A. A., Chabanova, E., Chmura, P. J., Jacobsen, U. P., Dawed, A. Y., Elders, P. J. M., Fernandez-Tajes, J. J., Forgie, I. M., ... Franks, P. W. (2026). A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis. Metabolism: Clinical and Experimental, 178, Article 156552. Advance online publication. https://doi.org/10.1016/j.metabol.2026.156552

@article{06a3b6b2272c454194cc8f0eb90d56c3,

title = "A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis",

abstract = "Objective: To delineate organ-specific and systemic drivers of metabolic dysfunction-associated steatotic liver disease (MASLD), we applied integrative causal inference across clinical, imaging, and proteomic domains in individuals with and without type 2 diabetes (T2D).Methods: Bayesian network analyses and complementary two-sample Mendelian randomization were used to quantify causal pathways linking adipose distribution, glycemia, and insulin dynamics with liver fat in the IMI-DIRECT prospective cohort study. Data included frequently sampled metabolic challenge tests, MRI-derived abdominal and hepatic fat content, serological biomarkers, and Olink plasma proteomics from 331 adults with new-onset T2D and 964 adults without diabetes, with harmonized protocols enabling replication.Results: High basal insulin secretion rate (BasalISR), estimated via C-peptide deconvolution, emerged as the primary potential causal driver of liver fat accumulation in both cohorts. BasalISR, a clearance-independent measure of β-cell insulin output distinct from peripheral insulin levels, was independently linked to hepatic steatosis. Visceral adipose tissue exhibited bidirectional associations with liver fat, suggesting a self-reinforcing metabolic loop. Of 446 analyzed proteins, 34 mapped to these metabolic networks (27 in the non-diabetes network, 18 in the T2D network, and 11 shared). Key proteins directly associated with liver fat included GUSB, ALDH1A1, LPL, IGFBP1/2, CTSD, HMOX1, FGF21, AGRP, and ACE2. Sex-stratified analyses identified GUSB in females and LEP in males as the strongest protein predictors of liver fat.Conclusions: BasalISR may better capture early β-cell-driven disturbances contributing to MASLD. These findings outline a multifactorial, sex- and disease stage-specific proteo-metabolic architecture of hepatic steatosis and identify potential biomarkers or therapeutic targets.",

keywords = "Basal insulin secretion, Bayesian networks, Hepatic steatosis, MASLD, Mendelian randomization, Proteomics, Type 2 diabetes",

author = "Atabaki, \{Natalie N.\} and Coral, \{Daniel E.\} and Hugo Pomares-Millan and Kieran Smith and Behjat, \{Harry H.\} and Koivula, \{Robert W.\} and Andrea Tura and Hamish Miller and Katherine Pinnick and Leandro Agudelo and Allin, \{Kristine H.\} and Brown, \{Andrew A.\} and Elizaveta Chabanova and Chmura, \{Piotr J.\} and Jacobsen, \{Ulrik P.\} and Dawed, \{Adem Y.\} and Elders, \{Petra J. M.\} and Fernandez-Tajes, \{Juan J.\} and Forgie, \{Ian M.\} and Mark Haid and Hansen, \{Tue H.\} and Jones, \{Angus G\} and Tarja Kokkola and Sebastian Kalamajski and Anubha Mahajan and McDonald, \{Timothy J.\} and Donna McEvoy and Mirthe Muilwijk and Tsirigos, \{Konstantinos D\} and Jagadish Vangipurapu and \{van Oort\}, Sabine and Henrik Vestergaard and Jerzy Adamski and Beulens, \{Joline W\} and S{\o}ren Brunak and Dermitzakis, \{Emmanouil T\} and Giordano, \{Giuseppe N\} and Ramneek Gupta and Torben Hansen and t'Hart, \{Leen M.\} and Hattersley, \{Andrew T.\} and Leanne Hodson and Markku Laakso and Loos, \{Ruth J. F.\} and Jordi Merino and Mattias Ohlsson and Oluf Pedersen and Martin Ridderstr{\aa}le and Hartmut Ruetten and Femke Rutters and Schwenk, \{Jochen M.\} and Jeremy Tomlinson and Mark Walker and Hanieh Yaghootkar and Fredrik Karpe and McCarthy, \{Mark I\} and Thomas, \{Elizabeth Louise\} and Bell, \{Jimmy D.\} and Andrea Mari and Imre Pavo and Pearson, \{Ewan R.\} and Ana Vi{\~n}uela and Franks, \{Paul W.\}",

note = "Copyright: Crown Copyright {\textcopyright} 2026 Published by Elsevier Inc. ",

year = "2026",

month = feb,

day = "6",

doi = "10.1016/j.metabol.2026.156552",

language = "English",

volume = "178",

journal = "Metabolism: Clinical and Experimental",

issn = "0026-0495",

publisher = "W.B. Saunders",

}

Atabaki, NN, Coral, DE, Pomares-Millan, H, Smith, K, Behjat, HH, Koivula, RW, Tura, A, Miller, H, Pinnick, K, Agudelo, L, Allin, KH, Brown, AA, Chabanova, E, Chmura, PJ, Jacobsen, UP, Dawed, AY, Elders, PJM, Fernandez-Tajes, JJ, Forgie, IM, Haid, M, Hansen, TH, Jones, AG, Kokkola, T, Kalamajski, S, Mahajan, A, McDonald, TJ, McEvoy, D, Muilwijk, M, Tsirigos, KD, Vangipurapu, J, van Oort, S, Vestergaard, H, Adamski, J, Beulens, JW, Brunak, S, Dermitzakis, ET, Giordano, GN, Gupta, R, Hansen, T, t'Hart, LM, Hattersley, AT, Hodson, L, Laakso, M, Loos, RJF, Merino, J, Ohlsson, M, Pedersen, O, Ridderstråle, M, Ruetten, H, Rutters, F, Schwenk, JM, Tomlinson, J, Walker, M, Yaghootkar, H, Karpe, F, McCarthy, MI, Thomas, EL, Bell, JD, Mari, A, Pavo, I, Pearson, ER , Viñuela, A & Franks, PW 2026, 'A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis', Metabolism: Clinical and Experimental, vol. 178, 156552. https://doi.org/10.1016/j.metabol.2026.156552

A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis. / Atabaki, Natalie N. (Lead / Corresponding author); Coral, Daniel E.; Pomares-Millan, Hugo et al.
In: Metabolism: Clinical and Experimental, Vol. 178, 156552, 05.2026.

Research output: Contribution to journal › Article › peer-review

TY - JOUR

T1 - A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis

AU - Atabaki, Natalie N.

AU - Coral, Daniel E.

AU - Pomares-Millan, Hugo

AU - Smith, Kieran

AU - Behjat, Harry H.

AU - Koivula, Robert W.

AU - Tura, Andrea

AU - Miller, Hamish

AU - Pinnick, Katherine

AU - Agudelo, Leandro

AU - Allin, Kristine H.

AU - Brown, Andrew A.

AU - Chabanova, Elizaveta

AU - Chmura, Piotr J.

AU - Jacobsen, Ulrik P.

AU - Dawed, Adem Y.

AU - Elders, Petra J. M.

AU - Fernandez-Tajes, Juan J.

AU - Forgie, Ian M.

AU - Haid, Mark

AU - Hansen, Tue H.

AU - Jones, Angus G

AU - Kokkola, Tarja

AU - Kalamajski, Sebastian

AU - Mahajan, Anubha

AU - McDonald, Timothy J.

AU - McEvoy, Donna

AU - Muilwijk, Mirthe

AU - Tsirigos, Konstantinos D

AU - Vangipurapu, Jagadish

AU - van Oort, Sabine

AU - Vestergaard, Henrik

AU - Adamski, Jerzy

AU - Beulens, Joline W

AU - Brunak, Søren

AU - Dermitzakis, Emmanouil T

AU - Giordano, Giuseppe N

AU - Gupta, Ramneek

AU - Hansen, Torben

AU - t'Hart, Leen M.

AU - Hattersley, Andrew T.

AU - Hodson, Leanne

AU - Laakso, Markku

AU - Loos, Ruth J. F.

AU - Merino, Jordi

AU - Ohlsson, Mattias

AU - Pedersen, Oluf

AU - Ridderstråle, Martin

AU - Ruetten, Hartmut

AU - Rutters, Femke

AU - Schwenk, Jochen M.

AU - Tomlinson, Jeremy

AU - Walker, Mark

AU - Yaghootkar, Hanieh

AU - Karpe, Fredrik

AU - McCarthy, Mark I

AU - Thomas, Elizabeth Louise

AU - Bell, Jimmy D.

AU - Mari, Andrea

AU - Pavo, Imre

AU - Pearson, Ewan R.

AU - Viñuela, Ana

AU - Franks, Paul W.

PY - 2026/2/6

Y1 - 2026/2/6

N2 - Objective: To delineate organ-specific and systemic drivers of metabolic dysfunction-associated steatotic liver disease (MASLD), we applied integrative causal inference across clinical, imaging, and proteomic domains in individuals with and without type 2 diabetes (T2D).Methods: Bayesian network analyses and complementary two-sample Mendelian randomization were used to quantify causal pathways linking adipose distribution, glycemia, and insulin dynamics with liver fat in the IMI-DIRECT prospective cohort study. Data included frequently sampled metabolic challenge tests, MRI-derived abdominal and hepatic fat content, serological biomarkers, and Olink plasma proteomics from 331 adults with new-onset T2D and 964 adults without diabetes, with harmonized protocols enabling replication.Results: High basal insulin secretion rate (BasalISR), estimated via C-peptide deconvolution, emerged as the primary potential causal driver of liver fat accumulation in both cohorts. BasalISR, a clearance-independent measure of β-cell insulin output distinct from peripheral insulin levels, was independently linked to hepatic steatosis. Visceral adipose tissue exhibited bidirectional associations with liver fat, suggesting a self-reinforcing metabolic loop. Of 446 analyzed proteins, 34 mapped to these metabolic networks (27 in the non-diabetes network, 18 in the T2D network, and 11 shared). Key proteins directly associated with liver fat included GUSB, ALDH1A1, LPL, IGFBP1/2, CTSD, HMOX1, FGF21, AGRP, and ACE2. Sex-stratified analyses identified GUSB in females and LEP in males as the strongest protein predictors of liver fat.Conclusions: BasalISR may better capture early β-cell-driven disturbances contributing to MASLD. These findings outline a multifactorial, sex- and disease stage-specific proteo-metabolic architecture of hepatic steatosis and identify potential biomarkers or therapeutic targets.

AB - Objective: To delineate organ-specific and systemic drivers of metabolic dysfunction-associated steatotic liver disease (MASLD), we applied integrative causal inference across clinical, imaging, and proteomic domains in individuals with and without type 2 diabetes (T2D).Methods: Bayesian network analyses and complementary two-sample Mendelian randomization were used to quantify causal pathways linking adipose distribution, glycemia, and insulin dynamics with liver fat in the IMI-DIRECT prospective cohort study. Data included frequently sampled metabolic challenge tests, MRI-derived abdominal and hepatic fat content, serological biomarkers, and Olink plasma proteomics from 331 adults with new-onset T2D and 964 adults without diabetes, with harmonized protocols enabling replication.Results: High basal insulin secretion rate (BasalISR), estimated via C-peptide deconvolution, emerged as the primary potential causal driver of liver fat accumulation in both cohorts. BasalISR, a clearance-independent measure of β-cell insulin output distinct from peripheral insulin levels, was independently linked to hepatic steatosis. Visceral adipose tissue exhibited bidirectional associations with liver fat, suggesting a self-reinforcing metabolic loop. Of 446 analyzed proteins, 34 mapped to these metabolic networks (27 in the non-diabetes network, 18 in the T2D network, and 11 shared). Key proteins directly associated with liver fat included GUSB, ALDH1A1, LPL, IGFBP1/2, CTSD, HMOX1, FGF21, AGRP, and ACE2. Sex-stratified analyses identified GUSB in females and LEP in males as the strongest protein predictors of liver fat.Conclusions: BasalISR may better capture early β-cell-driven disturbances contributing to MASLD. These findings outline a multifactorial, sex- and disease stage-specific proteo-metabolic architecture of hepatic steatosis and identify potential biomarkers or therapeutic targets.

KW - Basal insulin secretion

KW - Bayesian networks

KW - Hepatic steatosis

KW - MASLD

KW - Mendelian randomization

KW - Proteomics

KW - Type 2 diabetes

U2 - 10.1016/j.metabol.2026.156552

DO - 10.1016/j.metabol.2026.156552

M3 - Article

C2 - 41655955

SN - 0026-0495

VL - 178

JO - Metabolism: Clinical and Experimental

JF - Metabolism: Clinical and Experimental

M1 - 156552

ER -

A biological-systems-based analysis using proteomic and metabolic network inference reveals mechanistic insights into hepatic steatosis

Abstract

UN SDGs

Keywords

ASJC Scopus subject areas

Access to Document

Fingerprint

Projects

DIRECT: Diabetes Research on Patient Stratification (joint with 25 other partners)

Cite this