Variation in HIV-1 set-point viral load: Epidemiological analysis and an evolutionary hypothesis

Fraser et al. 10.1073/pnas.0708559104.

Supporting Information

Files in this Data Supplement:

SI Text
SI Figure 6
SI Figure 7
SI Figure 8
SI Figure 9
SI Figure 10

SI Figure 6

Fig. 6. the relation between set-point viral load and survival in different populations. The survival distribution (time from infection to death) in two seroconverters cohorts: the Amsterdam seroconverter cohort (A) and a cohort of female sex workers in Nairobi (B) (40). We have matched the set-point viral load groups: thick line, less than or equal to 10,000 copies; thin line, >10,000 copies and less than or equal to 100,000 copies, and dashed line, >100,000 copies. The survival curves are clearly similar, with a tendency for longer survival in the female cohort (statistical comparison is not possible without knowledge of the censoring events in B).

SI Figure 7

Fig. 7. Sensitivity analysis for the transmission potential. We explore the sensitivity of our analysis to different assumptions regarding the distribution of durations of the incubation period (see SI Methods for details). Different statistical distributions for the duration for a given viral load result in similar predictions (black lines), including the generalized Gamma distribution (solid line) and Weibull distribution (dashed lines). Including the possibility of a lower bound for this duration at high viral loads (corresponding to D_min > 0, see Methods) does not fit the data significantly better, but results in a somewhat different shape (red lines). These curves fall within the 95% confidence intervals of the simpler model for the range of viral loads where we have data and differ most when extrapolating to very high viral loads. Different distributions are explored (solid, generalized g; long-dash, g; dotted, Weibull). The figure also shows the separate contribution of primary infection and late-stage preAIDS/AIDS infection to the transmission potential. These estimates are summarized from T.D.H., R.M. Anderson, and C.F. (unpublished work) (see SI Methods).

SI Figure 8

Fig. 8. the distribution of viral loads. The distributions of viral loads from the Amsterdam serconverters and the Zambian cohort are shown (as in Fig. 1, but as points) alongside the best-fit skew-Normal distributions (see SI Methods).

SI Figure 9

Fig. 9. effect of varying the partner change rate. The predictions in the main manuscript are made for a model with random mixing, i.e., corresponding to a very high partner change rate. The effect of varying the partner change rate in a simple model of serial-monogamous partnerships is illustrated. The mean duration of partnerships is defined as S and is varied from 0.25 to 2 years. (A) The transmission potential of early (primary) and late (preAIDS/AIDS) stages of infection are most affected because of their short duration relative to the asymptomatic stage of infection. (B) The transmission potential. (C) The basic reproduction number R0. (D) the initial exponential growth rate r0 are all reduced and shifted slightly to the left as partnerships are formed and reformed less frequently.

SI Figure 10

Fig. 10. Effect of varying the partner change rate (revisited). Because our model of transmission is calibrated from cohorts where counseling was given and condoms were actively distributed, we repeated the estimates of SI Fig. 9 but with a 75% boost to the rate of transmission within partnerships, which possibly better represents transmission rates within the general population. For Fig. 5, this value was chosen to produce plausible epidemic growth rates r0 with a mean partner change rate of 1.25 per year (based on data from ref. 41).

40. Lavreys L, Baeten JM, Chohan V, McClelland RS, Hassan WM, Richardson BA, Mandaliya K, Ndinya-Achola JO, Overbaugh J (2006) Clin Infect Dis 42:1333-1339.

41. Gray RH, Li XB, Wawer MJ, Gange SJ, Serwadda D, Sewandambo NK, Moore R, Wabwire-Manger F, Lutalo T, Quinn TC (2003) AIDS 17:1941-1951.