Figure 3.

Modest yet statistically significant increase in the average normalized T-cell response breadth over the course of the first year of HIV infection. We divided the observations into different time bins ((A) 50-day intervals; (B) 100-day intervals) and calculated the relative breadth for the corresponding interval. The relative breadth was calculated as the number of HIV-specific CD8${}^{+}$ T-cell responses detected in a given time period divided by the number of all responses measured for that patient in all time periods; data were averaged to simplify presentation. Averaging did not influence the statistical significance of conclusions. Colors and symbols represent the data from different patients as shown in Figure S5 in Supplementary Material. Black horizontal bars denote the mean relative breadth for that time interval for all patients. There was a statistically significant increase in relative breadth (Spearman’s rank correlation coefficient $\rho$ and p values indicated on panels). There was no change in the average total immune response in all patients (Figure S6). Detailed analysis of the relative number of CD8${}^{+}$ T-cell responses in individual patients revealed variable patterns: constant breadth, increasing breadth, decreasing breadth, and breadth changing non-monotonically over time (Figure S7). Also, no overall change in the average breadth (un-normalized) was observed (Figure S5). We observed a similarly modest but significant increase in $SE$ and $EI$ of HIV-specific CD8${}^{+}$ T-cell response with time (Figure S8).