The Value Of Targeted Metabolomics In Patients With Heart Failure

ObjectivesWe identify the distinctive makers in heart failure patients using a targeted metabolomics approach,which is considered diagnostic and prognosis.MethodsMass spectrometry-based profiling of plasma metabolites was performed in 428 participants consisted of 327 patients in training set and 101 patients in validation set.The training set included 105 non-HF control patients,97 HFp EF subjects,51 HFm EF and 74 HFr EF;there were 46 non-HF control patients,27 HFp EF subjects,7 HFm EF and 21 HFr EF in validation set.Validation set was used to assess the credibility of training set.A supervised multivariate approach,orthogonal partial least-squares discriminant analysis(OPLS-DA)was used to discriminate the four groups,non-HF,HFp EF,HFm EF and HFr EF,respectively.The quality of the OPLS-DA was assessed with the 200-iteration permutation test.Combination of the variable importance in the projection(VIP)values and logistic regression,we identified the distinctive biomarkers between non-HF and HFr EF.A new biomarker,W2,was calculated by the logistical equation.The accordance of W2 and NT-pro BNP were demonstrated with Person ananlysis.The receiver operating characteristic(ROC)and Kaplan-Meier curve were used to evaluate the diagnostic and prognostic values of metabolomics in HF.ResultsThe score plots of OPLS-DA models showed considerable separation between non-HF controls and HFp EF,HFm EF,HFr EF,respectively.A 200-time permutation test indicated the corresponding models were not random and overfitted as original R2 and Q2 values(top right)were significantly higher than the corresponding permuted values,the OPLS-DA models were well discrimination and forecast.In training set,HFp EF and HFr EF were significantly separated in the 3D PLS-DA score plot,but HFm EF were overlap with HFp EF and HFr EF.The similar plot was evidence in validation set.W2,a combination of the 4 markers,was hightly associated with LN(NT-pro BNP)with a significant r of 0.6628.W2 were deemed to ascend with ejection fraction decreasing in heart failure.W2 were-9.66±6.698 in contorls,-1.82±6.97 in HFp EF,2.94±9.65 in HFm EF and 11.29±8.99 in HFr EF with a significant P,respectively.A panel of metabolites,including C16,C18:1 and LN(NT-pro BNP)had the similar trend with W2,otherwise,the ratio of valine to phenylalanine,ratio of C5 to C2 showed the opposite way.W2 had a diagnostic value similar to NT-pro BNP(area under the curve of 0.995 vs.1.0 for NT-pro BNP).The prognostic value of the metabolite panel was better than that of NT-pro BNP(area under the curve of 0.653 vs 0.589 for NT-pro BNP,P< 0.01)and Kaplan-Meier curves(log rank: 20.13 vs 13.59 for NT-pro BNP,P< 0.01).These findings were approved in the validation set which promote the results in training set were reliable.ConclusionsWith ejection fraction decreasing in heart failure,long-chain acyl-carnitine and NT-pro BNP were accumulated in vivo,otherwise,the ratio of valine to phenylalanine and short-chain acyl-carnitine decreased.The biomarkers may play an important role on heart failure,and heart failure may deal with diverse metabolic pathways.Studying the metabolism process may provide an novel thinkings for diagnosing the heart failure.