To this end, we compared the viral decay and the time to viral suppression between the early treated patients who reinitiated cART and the patients in whom treatment was deferred until conventional criteria to start long-term cART had been reached. Temporary cART during PHI did not select for clinically relevant drug resistance mutations and was associated with a durable and persisting virologic response after subsequent reinitiation of long-term cART. Our data thereby support the use of early treatment during PHI. Of note, the slower decline of the pVL from 8 weeks onward in the second treatment period of the early treated patients is probably an artefact of the model, because patients had a lower baseline pVL at restart than during early treatment and their pVL was usually already undetectable by week 8. 
We were not able to perform resistance testing after TI in 7/63 of the early treated patients. This is in particular relevant for patients who interrupted an NNRTI-based regimen, because of the long half-life of NNRTIs. However, in all seven patients, irrespective of the regimen, the pVL was resuppressed upon restart, which virtually excludes clinically important mutations. The early treated patients reinitiated long-term cART more often with a boosted PI than naive patients. Many early treated patients preferred not to restart an NNRTI because of side-effects they had experienced previously during the early cART period, and therefore favoured a PI-containing regimen. Our study is supported by another study in which 37 PHIpatients were treated with temporary early cART and no drug resistance was observed after TI. However, in this study the NNRTI was stopped 96 h before the NRTI-backbone. Because NNRTIs have a slower metabolism and a low genetic barrier to resistance, simultaneous TI of an NNRTI-containing Albaspidin-AA regimen may result in a period of NNRTI-monotherapy, which may select for drug resistance mutations. NNRTI-drug resistance mutations that were selected after intrapartum exposure to single-dose nevirapine in HIV-infected women have been associated with decreased virologic response after subsequent treatment with an NNRTI-containing regimen. Noteworthy, the pVL in these women exposed to single-dose nevirapine was much higher than the pVL in a controlled TI-se ing in which patients have an undetectable pVL at TI. In the SMART trial, NNRTI-drug resistance mutations were more common in case of simultaneous TI than in case of a staggered or a switched interruption, in which the NNRTI is replaced by a boosted PI. However, in SMART most drug combinations included a zidovudine/lamivudine-backbone in combination with an NNRTI, whereas in our trial half of the patients were using a tenofovircontaining regimen, which has a longer half-life, and 20(S)-Notoginsenoside-R2 together with an NNRTI forms a more balanced regimen that is less prone to development of drug resistance when treatment is discontinued simultaneously. To date, there is no clear consensus how to stop cART regimens. In our study we found no indication for selection of drug resistance mutations after interrupting all drugs simultaneously once an undetectable pVL had been reached.