BACKGROUND: Metabolites such as creatinine and urea are established kidney function markers. High-throughput metabolomic studies have not been reported in large general population samples spanning normal kidney function and chronic kidney disease (CKD). STUDY DESIGN: Cross-sectional observational studies of the general population. SETTING AND PARTICIPANTS: 2 independent samples: KORA F4 (discovery sample, n = 3,011) and Twins UK (validation sample, n = 984). EXPOSURE FACTORS: 151 serum metabolites, quantified by targeted mass spectrometry. OUTCOMES AND MEASUREMENTS: Metabolites and their 22,650 ratios were analyzed by multivariable-adjusted linear regression for their association with glomerular filtration rate (eGFR), estimated separately from creatinine and cystatin C levels by CKD-EPI (CKD Epidemiology Collaboration) equations. After correction for multiple testing, significant metabolites (P < 3.3 × 10(-4) for single metabolites; P < 2.2 × 10(-6) for ratios) were meta-analyzed with independent data from the TwinsUK Study. RESULTS: Replicated associations with eGFR were observed for 22 metabolites and 516 metabolite ratios. Pooled P values ranged from 7.1 × 10(-7) to 1.8 × 10(-69) for the replicated single metabolites. Acylcarnitines such as glutarylcarnitine were associated inversely with eGFR (-3.73 mL/min/1.73 m(2) per standard deviation [SD] increase, pooled P = 1.8 × 10(-69)). The replicated ratio with the strongest association was the ratio of serine to glutarylcarnitine (P = 3.6 × 10(-81)). Almost all replicated phenotypes associated with decreased eGFR (<60 mL/min/1.73 m(2); n = 172 cases) in KORA F4: per 1-SD increment, ORs ranged from 0.29-2.06. Across categories of a metabolic score consisting of 3 uncorrelated metabolites, the prevalence of decreased eGFR increased from 3% to 53%. LIMITATIONS: Cross-sectional study design, GFR was estimated, limited number of metabolites. CONCLUSIONS: Distinct metabolic phenotypes were reproducibly associated with eGFR in 2 separate population studies. They may provide novel insights into renal metabolite handling, improve understanding of pathophysiology, or aid in the diagnosis of kidney disease. Longitudinal studies are needed to clarify whether changes in metabolic phenotypes precede or result from kidney function impairment.
BACKGROUND: Metabolites such as creatinine and urea are established kidney function markers. High-throughput metabolomic studies have not been reported in large general population samples spanning normal kidney function and chronic kidney disease (CKD). STUDY DESIGN: Cross-sectional observational studies of the general population. SETTING AND PARTICIPANTS: 2 independent samples: KORA F4 (discovery sample, n = 3,011) and Twins UK (validation sample, n = 984). EXPOSURE FACTORS: 151 serum metabolites, quantified by targeted mass spectrometry. OUTCOMES AND MEASUREMENTS: Metabolites and their 22,650 ratios were analyzed by multivariable-adjusted linear regression for their association with glomerular filtration rate (eGFR), estimated separately from creatinine and cystatin C levels by CKD-EPI (CKD Epidemiology Collaboration) equations. After correction for multiple testing, significant metabolites (P < 3.3 × 10(-4) for single metabolites; P < 2.2 × 10(-6) for ratios) were meta-analyzed with independent data from the TwinsUK Study. RESULTS: Replicated associations with eGFR were observed for 22 metabolites and 516 metabolite ratios. Pooled P values ranged from 7.1 × 10(-7) to 1.8 × 10(-69) for the replicated single metabolites. Acylcarnitines such as glutarylcarnitine were associated inversely with eGFR (-3.73 mL/min/1.73 m(2) per standard deviation [SD] increase, pooled P = 1.8 × 10(-69)). The replicated ratio with the strongest association was the ratio of serine to glutarylcarnitine (P = 3.6 × 10(-81)). Almost all replicated phenotypes associated with decreased eGFR (<60 mL/min/1.73 m(2); n = 172 cases) in KORA F4: per 1-SD increment, ORs ranged from 0.29-2.06. Across categories of a metabolic score consisting of 3 uncorrelated metabolites, the prevalence of decreased eGFR increased from 3% to 53%. LIMITATIONS: Cross-sectional study design, GFR was estimated, limited number of metabolites. CONCLUSIONS: Distinct metabolic phenotypes were reproducibly associated with eGFR in 2 separate population studies. They may provide novel insights into renal metabolite handling, improve understanding of pathophysiology, or aid in the diagnosis of kidney disease. Longitudinal studies are needed to clarify whether changes in metabolic phenotypes precede or result from kidney function impairment.
Authors: Karsten Suhre; Joseph E Schwartz; Vijay K Sharma; Qiuying Chen; John R Lee; Thangamani Muthukumar; Darshana M Dadhania; Ruchuang Ding; David N Ikle; Nancy D Bridges; Nikki M Williams; Gabi Kastenmüller; Edward D Karoly; Robert P Mohney; Michael Abecassis; John Friedewald; Stuart J Knechtle; Yolanda T Becker; Benjamin Samstein; Abraham Shaked; Steven S Gross; Manikkam Suthanthiran Journal: J Am Soc Nephrol Date: 2015-06-05 Impact factor: 10.121
Authors: Gesiane Tavares; Gabriela Venturini; Kallyandra Padilha; Roberto Zatz; Alexandre C Pereira; Ravi I Thadhani; Eugene P Rhee; Silvia M O Titan Journal: Metabolomics Date: 2018-02-27 Impact factor: 4.290
Authors: Eugene P Rhee; Clary B Clish; Anahita Ghorbani; Martin G Larson; Sammy Elmariah; Elizabeth McCabe; Qiong Yang; Susan Cheng; Kerry Pierce; Amy Deik; Amanda L Souza; Laurie Farrell; Carly Domos; Robert W Yeh; Igor Palacios; Kenneth Rosenfield; Ramachandran S Vasan; Jose C Florez; Thomas J Wang; Caroline S Fox; Robert E Gerszten Journal: J Am Soc Nephrol Date: 2013-05-16 Impact factor: 10.121