Search

Targeted proteomics utilizing antibody-based proximity extension assays provides sensitive and highly specific quantifications of plasma protein levels. Multivariate analysis of this data is hampered by frequent missing values (random or left censored), calling for imputation approaches. While appropriate missing-value imputation methods exist, benchmarks of their performance in targeted proteomics data are lacking. Here, we assessed the performance of two methods for imputation of values missing completely at random, the previously top-benchmarked 'missForest' and the recently published 'GSimp' method. Evaluation was accomplished by comparing imputed with remeasured relative concentrations of 91 inflammation related circulating proteins in 86 samples from a cohort of 645 patients with venous thromboembolism. The median Pearson correlation between imputed and remeasured protein expression values was 69.0% for missForest and 71.6% for GSimp (p = 5.8e-4). Imputation with missForest resulted in stronger reduction of variance compared to GSimp (median relative variance of 25.3% vs. 68.6%, p = 2.4e-16) and undesired larger bias in downstream analyses. Irrespective of the imputation method used, the 91 imputed proteins revealed large variations in imputation accuracy, driven by differences in signal to noise ratio and information overlap between proteins. In summary, GSimp outperformed missForest, while both methods show good overall imputation accuracy with large variations between proteins., Competing Interests: Michael Lenz, Andreas Schulz, Thomas Koeck, Steffen Rapp, Markus Nagler, Lisa Eggebrecht, Vincent Ten Cate, Karl J. Lackner, Thomas Münzel and Miguel A. Andrade-Navarro declare no conflict of interest. Madeleine Sauer and Kirsten Leineweber are employees of Bayer AG. Marina Panova-Noeva, Jürgen H. Prochaska, and Philipp S. Wild received funding from the Center for Thrombosis and Hemostasis Mainz. Philipp S. Wild reports grants from Bayer AG and from the German Federal Ministry of Education and Research, during the conduct of the study; grants and personal fees from Boehringer Ingelheim, grants from Philips Medical Systems, grants and personal fees from Sanofi-Aventis, grants and personal fees from Bayer Vital, grants from Daiichi Sankyo Europe, personal fees from Bayer Health Care, personal fees from Astra Zeneca, personal fees and non-financial support from Diasorin and non-financial support from I.E.M., outside the submitted work. This does not alter our adherence to PLOS ONE policies on sharing data and materials.

Context and Objective: Postoperative infections should be detected earlier. We investigated the efficacy of the National Nosocomial Infection Surveillance (NNIS) score, interleukin-6 (IL-6) and various acute-phase proteins for predicting postoperative infections., Design and Setting: Case series study at the Júlio Müller University Hospital., Methods: Thirty-two patients who underwent major gastrointestinal procedures between June 2004 and February 2005 were studied. The NNIS score and the evolution of serum IL-6 and various acute-phase proteins (C-reactive protein [CRP], albumin, prealbumin and transferrin) were correlated with postoperative infections and length of hospital stay (LOS)., Results: NNIS > 1 (p = 0.01) and low preoperative albumin (p = 0.02) significantly correlated with infection. IL-6 and CRP increased significantly more in patients with infections. Multivariate analysis showed greater risk of infection when NNIS > 1 (odds ratio, OR = 10.66; 95% confidence interval, CI: 1.1-102.0; p = 0.04); preoperative albumin < 3 g/dl (OR = 8.77; 95% CI: 1.13-67.86; p = 0.03); CRP > 30 mg/l on the second postoperative day (OR = 8.27; 95% CI: 1.05-64.79; p = 0.03) and > 12 mg/l on the fifth postoperative day (OR = 25.92; 95% CI: 2.17-332.71; p < 0.01); and IL-6 > 25 pg/ml on the fifth postoperative day (OR = 15.46; 95% CI: 1.19-230.30; p = 0.03). Longer LOS was associated with cancer, transferrin, IL-6 and albumin (p < 0.05)., Conclusions: NNIS, albumin, CRP and IL-6 may be useful as predictive markers for postoperative infections. For predicting LOS, malignant condition, transferrin, albumin and IL-6 are useful.