Sara van den Berg

143 6 T-cell dynamics in CMV where L represents the fraction of labelled DNA in the cell population, here the granulocyte population. All T-cell deuterium labelling curves were fitted using the kinetic heterogeneity model introduced by Asquith et al. [28]. where L again represents the fraction of labelled DNA, here in the T-cell population of interest, yielding the average production rate p of the T-cell population, and the average loss rate of the labelled T-cells d* for each cell subset. As CMV-specific cell numbers may not be constant over time, we did not make a steady‐state assumption and therefore interpreted p as the average production rate of a cell population. We then calculated the average loss rate d of each cell subset using the cell number data, by fitting an exponential function to the number or frequency of CMV-specific T-cells for each individual: where X 0 is the CMV-specific T-cell number or frequency at the time of inclusion in the study and p was fixed to the estimated values from the deuterium analyses. In this paper, we present death rates d based on the fits to the frequencies of CMV-specific T-cells, because T-cell numbers including CMV-specific T-cell numbers fluctuated considerably between visits, as a direct consequence of fluctuations in total leukocyte counts. Although the kinetic heterogeneity model has the disadvantage that the average production rate p is somewhat dependent on the length of the labelling period, we nevertheless used it because the multi‐ exponential model that we previously proposed [29], frequently led to overfitting of the data. The estimated production rates p are presented in this paper either as rates per day (in tables and figures), or as percentages per year (where percentage production per year = production rate per day*365*100). For cell populations in steady state, the estimated production rate p can be translated into the average lifespan of the cells in the population by taking the inverse of the production rate. Statistical analysis Differences between groups were assessed using Mann-Whitney U test, and comparisons within the same individuals with the Wilcoxon singed-rank test. Correlations were tested with Spearman’s rank correlation coefficient. Principal component analysis (PCA) was performed in SPSS, factors were included of an eigenvalue >1, iteration was set at a maximum of 25. For all analyses p -values <0.05 were considered significant. Data were analyzed using GraphPad Prism 8.3 and SPSS statistics 22 for Windows (SPSS Inc., Chicago, IL, USA). Deuterium-enrichment data were fitted using R (version 3.6.1, R Core Team (2019). R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria ( https://www.R-project.org/) ), parameters were estimated using a maximum