Advanced

Integrative analysis of prognostic biomarkers derived from multiomics panels helps discrimination of chronic kidney disease trajectories in people with type 2 diabetes

Kammer, Michael ; Heinzel, Andreas ; Willency, Jill A. ; Duffin, Kevin L. ; Mayer, Gert ; Simons, Kai ; Gerl, Mathias J. ; Klose, Christian ; Heinze, Georg and Reindl-Schwaighofer, Roman , et al. (2019) In Kidney International
Abstract

Clinical risk factors explain only a fraction of the variability of estimated glomerular filtration rate (eGFR) decline in people with type 2 diabetes. Cross-omics technologies by virtue of a wide spectrum screening of plasma samples have the potential to identify biomarkers for the refinement of prognosis in addition to clinical variables. Here we utilized proteomics, metabolomics and lipidomics panel assay measurements in baseline plasma samples from the multinational PROVALID study (PROspective cohort study in patients with type 2 diabetes mellitus for VALIDation of biomarkers) of patients with incident or early chronic kidney disease (median follow-up 35 months, median baseline eGFR 84 mL/min/1.73 m2, urine... (More)

Clinical risk factors explain only a fraction of the variability of estimated glomerular filtration rate (eGFR) decline in people with type 2 diabetes. Cross-omics technologies by virtue of a wide spectrum screening of plasma samples have the potential to identify biomarkers for the refinement of prognosis in addition to clinical variables. Here we utilized proteomics, metabolomics and lipidomics panel assay measurements in baseline plasma samples from the multinational PROVALID study (PROspective cohort study in patients with type 2 diabetes mellitus for VALIDation of biomarkers) of patients with incident or early chronic kidney disease (median follow-up 35 months, median baseline eGFR 84 mL/min/1.73 m2, urine albumin-to-creatinine ratio 8.1 mg/g). In an accelerated case-control study, 258 individuals with a stable eGFR course (median eGFR change 0.1 mL/min/year) were compared to 223 individuals with a rapid eGFR decline (median eGFR decline -6.75 mL/min/year) using Bayesian multivariable logistic regression models to assess the discrimination of eGFR trajectories. The analysis included 402 candidate predictors and showed two protein markers (KIM-1, NTproBNP) to be relevant predictors of the eGFR trajectory with baseline eGFR being an important clinical covariate. The inclusion of metabolomic and lipidomic platforms did not improve discrimination substantially. Predictions using all available variables were statistically indistinguishable from predictions using only KIM-1 and baseline eGFR (area under the receiver operating characteristic curve 0.63). Thus, the discrimination of eGFR trajectories in patients with incident or early diabetic kidney disease and maintained baseline eGFR was modest and the protein marker KIM-1 was the most important predictor.

(Less)
Please use this url to cite or link to this publication:
@article{b465b536-3cd9-442b-a16e-6f94c67a3b59,
  abstract     = {<p>Clinical risk factors explain only a fraction of the variability of estimated glomerular filtration rate (eGFR) decline in people with type 2 diabetes. Cross-omics technologies by virtue of a wide spectrum screening of plasma samples have the potential to identify biomarkers for the refinement of prognosis in addition to clinical variables. Here we utilized proteomics, metabolomics and lipidomics panel assay measurements in baseline plasma samples from the multinational PROVALID study (PROspective cohort study in patients with type 2 diabetes mellitus for VALIDation of biomarkers) of patients with incident or early chronic kidney disease (median follow-up 35 months, median baseline eGFR 84 mL/min/1.73 m<sup>2</sup>, urine albumin-to-creatinine ratio 8.1 mg/g). In an accelerated case-control study, 258 individuals with a stable eGFR course (median eGFR change 0.1 mL/min/year) were compared to 223 individuals with a rapid eGFR decline (median eGFR decline -6.75 mL/min/year) using Bayesian multivariable logistic regression models to assess the discrimination of eGFR trajectories. The analysis included 402 candidate predictors and showed two protein markers (KIM-1, NTproBNP) to be relevant predictors of the eGFR trajectory with baseline eGFR being an important clinical covariate. The inclusion of metabolomic and lipidomic platforms did not improve discrimination substantially. Predictions using all available variables were statistically indistinguishable from predictions using only KIM-1 and baseline eGFR (area under the receiver operating characteristic curve 0.63). Thus, the discrimination of eGFR trajectories in patients with incident or early diabetic kidney disease and maintained baseline eGFR was modest and the protein marker KIM-1 was the most important predictor.</p>},
  author       = {Kammer, Michael and Heinzel, Andreas and Willency, Jill A. and Duffin, Kevin L. and Mayer, Gert and Simons, Kai and Gerl, Mathias J. and Klose, Christian and Heinze, Georg and Reindl-Schwaighofer, Roman and Hu, Karin and Perco, Paul and Eder, Susanne and Rosivall, Laszlo and Mark, Patrick B. and Ju, Wenjun and Kretzler, Matthias and McCarthy, Mark I. and Heerspink, Hiddo L. and Wiecek, Andrzej and Gomez, Maria F. and Oberbauer, Rainer},
  issn         = {0085-2538},
  language     = {eng},
  month        = {08},
  publisher    = {Nature Publishing Group},
  series       = {Kidney International},
  title        = {Integrative analysis of prognostic biomarkers derived from multiomics panels helps discrimination of chronic kidney disease trajectories in people with type 2 diabetes},
  url          = {http://dx.doi.org/10.1016/j.kint.2019.07.025},
  doi          = {10.1016/j.kint.2019.07.025},
  year         = {2019},
}