Background: Modern antiretroviral therapy (ART) regimens are widely assumed to forgive modest nonadherence, because virological suppression in plasma is common at adherence levels of >70%. Yet, it is unknown whether human immunodeficiency virus type 1 (HIV-1) replication is completely suppressed at these levels of adherence.
Methods: We longitudinally quantified levels of cell-associated HIV-1 RNA and DNA in 40 patients (median duration of successful ART before study initiation, 46 months), whose 1-week adherence to therapy prior to the sampling moments was measured electronically.
Results: Patients were constantly 100% adherent (the optimal-adherence group), demonstrated improving adherence over time (the improving-adherence group), or neither of the above (the poor-adherence group). Adherence never decreased to <70% in any patient, and no rebound in plasma virological levels was observed. Nevertheless, poor adherence but not optimal or improving adherence caused a significant longitudinal increase in cell-associated HIV RNA levels (P = .006). Time-weighted changes and regression slopes of viral RNA load for the poor-adherence group were significantly higher than those for the optimal-adherence group (P < .01).
Conclusions: Because ART only blocks infection of new cells but not viral RNA transcription in cells infected before therapy initiation, the observed effects strongly suggest that modest nonadherence can cause new cycles of HIV-1 replication that are undetectable by commercial plasma viral load assays.