These figures show model predictions of the impact of therapeutic interventions on infected cell numbers in the drug sanctuaries (red lines) and main compartment (black lines). In each figure, prior to time 0 (grey shaded area), the host is only taking antiretroviral therapy (ART). At this stage, the effectiveness of antiretroviral therapy is high in the main compartment of the body (z1 = 0.97), but lower in the drug sanctuaries (z0 = 0.6). Furthermore, the per cell rate that cells traffic between compartments (governed by parameter τ0 = 0.5 day-1) and the per cell rate that infected cells are cleared from the drug sanctuaries (δ0 = 0.5 day-1) are slow enough to allow ongoing cycles of replication to persist in the drug sanctuaries. At time 0, additional therapy that either increases the cell trafficking rate (a and b), increases the cell clearance rate of infected cells from the drug sanctuaries (c), or both (d) is applied. In a), the trafficking rate is increased sufficiently (κ = 5) that ongoing viral replication is no longer sustainable. In b), the trafficking rate is increased to a level (κ = 3.5) just below the critical threshold and infected cell numbers decline to a new, lower equilibrium in the drug sanctuaries, but in the main compartment they increase. In c), the per cell clearance rate of infected cells in the drug sanctuaries is increased to the same level as in the main compartment (δ0 = 1 day-1). Here, the assumption that the effective drug concentration is highly impaired in the drug sanctuaries results in infected cell numbers declining to a new equilibrium. In d) both the per cell trafficking rate and the per cell clearance rate are increased to levels (κ = 3.5 and δ0 = 1 day-1) which independently would not stop ongoing replication (b and c), but, in combination, do so. All parameter values used in these calculations are provided in S1 Table.