Figure 2.

CMV‐specific antibody levels followed over time. (a) Antibody levels of all CMV⁺ individuals (n = 155). The trend line is based on local polynomial regression through the data of long‐term (LT) CMV⁺ participants (n = 136). (b) Fold increase in CMV‐specific antibody levels over 25 years for each individual. Dashed line shows no increase or decrease. (c) Individual CMV‐specific antibody levels during time of the CMV seroconverters (n = 19) (blue lines) compared to the average trend line of the long‐term CMV⁺ individuals. The first CMV⁺ time point of CMV seroconverters after CMV seroconversion (<5.5 years after CMV conversion) is highlighted by a larger blue dot. (d) Duration of CMV infection: CMV‐specific antibody levels of recently seroconverted individuals (max < 5.5 years after CMV conversion, n = 19) compared with those of long‐term CMV⁺ individuals (> 25 years CMV⁺, n = 136). (e) Age at seroconversion: Antibody levels of individuals that seroconverted at younger age (≤ 38 years of age, n = 26) or older age (≥ 45 years of age, n = 18, mean age 58.5 ± 8.1, shortly after CMV seroconversion (max < 5.5 years)). (f) Antibody levels associated with age at CMV seroconversion. For the selection of LT CMV⁺ individuals, CMV⁺ individuals of round 1 were included that were ≤ 38 years of age and age of seroconversion was set at 38 years. (g) Antibody levels of all CMV⁺ individuals at endpoint. Interval represents geometric mean level ± geometric standard deviation. (h) Variable importance when predicting CMV antibody levels at study endpoint with a random forest algorithm. % increase in MSE: proportion increase in mean squared error in the model when a single variable is randomly shuffled. Slope of Ab level: log‐linear variation in CMV‐specific antibody levels after first CMV⁺ measurement, until time point 6. Boxplots show median with interquartile range. * P < 0.05, **P < 0.01. The dotted line in a and c is the upper boundary of the cut‐off for CMV seropositivity.