Background:  We investigated the evolution of mutations associated with resistance to protease inhibitors (PI) during treatment interruption, and their ability to affect virus replication.

Methods: 195 patients (pts) having a genotype at last HAART-virological failure and during treatment interruption, were selected. Treatment interruption lasted 1 to 12 months (median, 4 months). The following major, mino,r and recently associated PI-mutations (with prevalence > 2%) were analyzed: L10F/I/V, K20R/I/T, L24I, D30N, V32I, L33F, M36I, K43T, M46I/L, I47V, G48V, F53L, I54V, K55R, Q58E, I62V, L63P, A71V/T, G73S, T74S, V77I, V82A, I84V, I85V, N88D, L89M, L90M, I93L, and C95F. A logistic regression analysis was used to define the mean decrease of mutations over time. The association between mutations and viremia was assessed by t-tests; Fisher’s exact test was used to detect whether specific mutations, at the time of treatment interruption, affect the dynamics of disappearance of other mutations.

Results:  At logistic regression, the probability of monthly disappearance was significant for 73S, 58E, 33F, and 90M (91%, 89%, 87%, and 66%, respectively, p < 0.05). L90M disappeared in 75% of patients without 36I, but only in 22% of patients carrying also 36I (p = 0.04), confirming the important role of 36I for the appearance and maintenance of 90M at virologic failure. As expected, all major mutations progressively disappeared during treatment interruption, although at different rate. Generally, complete disappearance occurred between 6 and 12 months for mutations 30N 32I, 33F, 46L, 47V, 53L, 73S, 88D, 46I, and 90M. Minor mutations had a discordant behavior:  while 71V completely disappeared at rate similar to major mutations, others (10V, 36I, 63P, 77I, and 93L) remained at > 80% frequency even after treatment interruption, suggesting their limited interference with viral fitness. Other mutations, recently associated with resistance (20T, 43T, 58E, 71I, 85V, and 95F) disappeared at rate similar to major mutations from treatment interruption, thus supporting their involvement in drug resistance. Higher viral load increase was detected at disappearance of mutations 46I and 82A, compared with viral load in pts that maintained such mutations (p < 0.001 and p < 0.04, respectively).

Conclusions:  The different dynamics of disappearance of mutations (major, minor, and novel) during treatment interruption led to the identification of those mutations more detrimental for viral fitness. They can be used in designing clinical strategies aimed to the characterization of virus related damage in highly drug experienced patients.

Keywords: HIV-1 protease; treatment interruption ; resistance mutations