Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling

Loic Yengo, Abdelilah Arredouani, Michel Marre, Ronan Roussel, Martine Vaxillaire, Mario Falchi, Abdelali Haoudi, Jean Tichet, Beverley Balkau, Amélie Bonnefond, Philippe Froguel

Research output: Contribution to journalArticle

5 Citations (Scopus)

Abstract

Objective Characterizing specific metabolites in sub-clinical phases preceding the onset of type 2 diabetes to enable efficient preventive and personalized interventions. Research design and methods We developed predictive models of type 2 diabetes using two strategies. One strategy focused on the probability of incidence only and was based on logistic regression (MRS1); the other strategy accounted for the age at diagnosis of diabetes and was based on Cox regression (MRS2). We assessed 293 metabolites using non-targeted metabolomics in fasting plasma samples of 1,044 participants (including 231 incident cases over 9 years) used as training population; and fasting serum samples of 128 participants (64 incident cases versus 64 controls) used as validation population. We applied a LASSO-based variable selection aiming at maximizing the out-of-sample area under the receiver operating characteristic curve (AROC) and integrated AROC. Results Sixteen and 17 metabolites were selected for MRS1 and MRS2, respectively, with AROC = 90% and 73% in the training and validation populations, respectively for MRS1. MRS2 had a similar performance and was significantly associated with a younger age of onset of type 2 diabetes (β = −3.44 years per MRS2 SD in the training population, p = 1.56 × 10−7; β = −4.73 years per MRS2 SD in the validation population, p = 4.04 × 10−3). Conclusions Overall, this study illustrates that metabolomics improves prediction of type 2 diabetes incidence of 4.5% on top of known clinical and biological markers, reaching 90% in total AROC, which is considered the threshold for clinical validity, suggesting it may be used in targeting interventions to prevent type 2 diabetes.

Original languageEnglish
Pages (from-to)918-925
Number of pages8
JournalMolecular Metabolism
Volume5
Issue number10
DOIs
Publication statusPublished - 1 Oct 2016

Fingerprint

Metabolomics
Statistical Models
Type 2 Diabetes Mellitus
ROC Curve
Population
Fasting
Biomarkers
Incidence
Age of Onset
Area Under Curve
Research Design
Logistic Models
Serum

Keywords

  • High dimensional regression
  • LASSO
  • Metabolomics
  • Risk prediction
  • Type 2 diabetes

ASJC Scopus subject areas

  • Molecular Biology
  • Cell Biology

Cite this

Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling. / Yengo, Loic; Arredouani, Abdelilah; Marre, Michel; Roussel, Ronan; Vaxillaire, Martine; Falchi, Mario; Haoudi, Abdelali; Tichet, Jean; Balkau, Beverley; Bonnefond, Amélie; Froguel, Philippe.

In: Molecular Metabolism, Vol. 5, No. 10, 01.10.2016, p. 918-925.

Research output: Contribution to journalArticle

Yengo, L, Arredouani, A, Marre, M, Roussel, R, Vaxillaire, M, Falchi, M, Haoudi, A, Tichet, J, Balkau, B, Bonnefond, A & Froguel, P 2016, 'Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling', Molecular Metabolism, vol. 5, no. 10, pp. 918-925. https://doi.org/10.1016/j.molmet.2016.08.011
Yengo, Loic ; Arredouani, Abdelilah ; Marre, Michel ; Roussel, Ronan ; Vaxillaire, Martine ; Falchi, Mario ; Haoudi, Abdelali ; Tichet, Jean ; Balkau, Beverley ; Bonnefond, Amélie ; Froguel, Philippe. / Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling. In: Molecular Metabolism. 2016 ; Vol. 5, No. 10. pp. 918-925.
@article{41fa62a146d74388b871bbdbaae5885d,
title = "Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling",
abstract = "Objective Characterizing specific metabolites in sub-clinical phases preceding the onset of type 2 diabetes to enable efficient preventive and personalized interventions. Research design and methods We developed predictive models of type 2 diabetes using two strategies. One strategy focused on the probability of incidence only and was based on logistic regression (MRS1); the other strategy accounted for the age at diagnosis of diabetes and was based on Cox regression (MRS2). We assessed 293 metabolites using non-targeted metabolomics in fasting plasma samples of 1,044 participants (including 231 incident cases over 9 years) used as training population; and fasting serum samples of 128 participants (64 incident cases versus 64 controls) used as validation population. We applied a LASSO-based variable selection aiming at maximizing the out-of-sample area under the receiver operating characteristic curve (AROC) and integrated AROC. Results Sixteen and 17 metabolites were selected for MRS1 and MRS2, respectively, with AROC = 90{\%} and 73{\%} in the training and validation populations, respectively for MRS1. MRS2 had a similar performance and was significantly associated with a younger age of onset of type 2 diabetes (β = −3.44 years per MRS2 SD in the training population, p = 1.56 × 10−7; β = −4.73 years per MRS2 SD in the validation population, p = 4.04 × 10−3). Conclusions Overall, this study illustrates that metabolomics improves prediction of type 2 diabetes incidence of 4.5{\%} on top of known clinical and biological markers, reaching 90{\%} in total AROC, which is considered the threshold for clinical validity, suggesting it may be used in targeting interventions to prevent type 2 diabetes.",
keywords = "High dimensional regression, LASSO, Metabolomics, Risk prediction, Type 2 diabetes",
author = "Loic Yengo and Abdelilah Arredouani and Michel Marre and Ronan Roussel and Martine Vaxillaire and Mario Falchi and Abdelali Haoudi and Jean Tichet and Beverley Balkau and Am{\'e}lie Bonnefond and Philippe Froguel",
year = "2016",
month = "10",
day = "1",
doi = "10.1016/j.molmet.2016.08.011",
language = "English",
volume = "5",
pages = "918--925",
journal = "Molecular Metabolism",
issn = "2212-8778",
publisher = "Elsevier GmbH",
number = "10",

}

TY - JOUR

T1 - Impact of statistical models on the prediction of type 2 diabetes using non-targeted metabolomics profiling

AU - Yengo, Loic

AU - Arredouani, Abdelilah

AU - Marre, Michel

AU - Roussel, Ronan

AU - Vaxillaire, Martine

AU - Falchi, Mario

AU - Haoudi, Abdelali

AU - Tichet, Jean

AU - Balkau, Beverley

AU - Bonnefond, Amélie

AU - Froguel, Philippe

PY - 2016/10/1

Y1 - 2016/10/1

N2 - Objective Characterizing specific metabolites in sub-clinical phases preceding the onset of type 2 diabetes to enable efficient preventive and personalized interventions. Research design and methods We developed predictive models of type 2 diabetes using two strategies. One strategy focused on the probability of incidence only and was based on logistic regression (MRS1); the other strategy accounted for the age at diagnosis of diabetes and was based on Cox regression (MRS2). We assessed 293 metabolites using non-targeted metabolomics in fasting plasma samples of 1,044 participants (including 231 incident cases over 9 years) used as training population; and fasting serum samples of 128 participants (64 incident cases versus 64 controls) used as validation population. We applied a LASSO-based variable selection aiming at maximizing the out-of-sample area under the receiver operating characteristic curve (AROC) and integrated AROC. Results Sixteen and 17 metabolites were selected for MRS1 and MRS2, respectively, with AROC = 90% and 73% in the training and validation populations, respectively for MRS1. MRS2 had a similar performance and was significantly associated with a younger age of onset of type 2 diabetes (β = −3.44 years per MRS2 SD in the training population, p = 1.56 × 10−7; β = −4.73 years per MRS2 SD in the validation population, p = 4.04 × 10−3). Conclusions Overall, this study illustrates that metabolomics improves prediction of type 2 diabetes incidence of 4.5% on top of known clinical and biological markers, reaching 90% in total AROC, which is considered the threshold for clinical validity, suggesting it may be used in targeting interventions to prevent type 2 diabetes.

AB - Objective Characterizing specific metabolites in sub-clinical phases preceding the onset of type 2 diabetes to enable efficient preventive and personalized interventions. Research design and methods We developed predictive models of type 2 diabetes using two strategies. One strategy focused on the probability of incidence only and was based on logistic regression (MRS1); the other strategy accounted for the age at diagnosis of diabetes and was based on Cox regression (MRS2). We assessed 293 metabolites using non-targeted metabolomics in fasting plasma samples of 1,044 participants (including 231 incident cases over 9 years) used as training population; and fasting serum samples of 128 participants (64 incident cases versus 64 controls) used as validation population. We applied a LASSO-based variable selection aiming at maximizing the out-of-sample area under the receiver operating characteristic curve (AROC) and integrated AROC. Results Sixteen and 17 metabolites were selected for MRS1 and MRS2, respectively, with AROC = 90% and 73% in the training and validation populations, respectively for MRS1. MRS2 had a similar performance and was significantly associated with a younger age of onset of type 2 diabetes (β = −3.44 years per MRS2 SD in the training population, p = 1.56 × 10−7; β = −4.73 years per MRS2 SD in the validation population, p = 4.04 × 10−3). Conclusions Overall, this study illustrates that metabolomics improves prediction of type 2 diabetes incidence of 4.5% on top of known clinical and biological markers, reaching 90% in total AROC, which is considered the threshold for clinical validity, suggesting it may be used in targeting interventions to prevent type 2 diabetes.

KW - High dimensional regression

KW - LASSO

KW - Metabolomics

KW - Risk prediction

KW - Type 2 diabetes

UR - http://www.scopus.com/inward/record.url?scp=84992183696&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84992183696&partnerID=8YFLogxK

U2 - 10.1016/j.molmet.2016.08.011

DO - 10.1016/j.molmet.2016.08.011

M3 - Article

AN - SCOPUS:84992183696

VL - 5

SP - 918

EP - 925

JO - Molecular Metabolism

JF - Molecular Metabolism

SN - 2212-8778

IS - 10

ER -