Serum and urinary biomarkers of collagen type-I turnover predict prognosis in patients with heart failure Dear Editor, Previous studies indicated an association of biomarkers reflecting collagen turnover with heart failure. 1,2 However, their value in predicting unfavorable outcomes is unknown. In the myocardium, the collagen matrix forms a scaffold that maintains the anatomical integrity and function of the heart. Collagen I is dominant among the four primary types of cardiac collagen. 3 The abundance of collagen is the net result of the balance between synthesis and degradation. 4,5 Collagen degradation releases fragments to the circulation and urine. Here, we investigate if specific changes in collagen I turnover, detected based on urinary peptide fragments and circulating biomarkers, are associated with heart failure and subsequent clinical events. Our study includes all 354 participants with suspected heart failure from the HOMAGE-HULL subcohort 6,7 recruited between 2010 and 2014, whose baseline serum and urine samples were available. Table S1 shows their demographic and clinical data. We used capillary electrophoresis coupled with mass spectrometry (CE-MS) to quantify type I alpha-1 (COL1A1) fragments in urine, confining the analyses to 293 fragments present in at least 30% of subjects. For serum samples, we used ELISA Kit to quantify the collagen synthesis biomarkers carboxyterminal propeptide of procollagen type I (PICP), 8 collagen degradation biomarkers carboxy-terminal telopeptide of collagen type I (CITP), 9 and matrix metalloproteinase (MMP-1) 10 , which are known as good indicators of collagen turnover (Figure S1). The primary outcome is heart failure death. The rank-normalized biomarker abundance was regressed with the time of survival in a Cox proportional hazard model corrected for six confounders: age, sex, body mass index, history of ischemic disease, heart rate, and serum creatinine. Significant biomarkers were combined into a classifier, using the support vector machinebased MosaCluster software. 2 Cox regression was used to determine if prognosis via the classifier was independent of the six confounders plus NT-proBNP. All analyses were This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited.