Persistence of transmitted drug resistance mutations suggests source partners may be treatment-naive

Matt Sharp, HIV i-Base

A recent analysis from the UK HIV drug resistance database on the persistence of transmitted drug resistance (TDR) over time suggests that poor adherence by people on ART is unlikely to be linked to most cases of TDR.

The results may also help interpret individual resistance tests in the context of TDR. [1, 2]

The study, published online on the 31 July 2013 in the Journal of Infectious Diseases, estimated the persistence of transmitted drug resistant virus in 313 treatment naive patients who had at least one drug-related mutation in their first resistance test (from 1997 – 2009) and at least one subsequent resistance test result prior to starting treatment. Recent infection could only be confirmed for the 15% of patients who had a previous HIV negative test result in the previous 18 months.

Previous studies looking at the loss and persistence of transmitted mutated virus have generally been small and this is the first large study to provide estimates of what occurs to viral mutations over time.

Population sequencing genotype testing (sensitive to variants present at >15% of the viral populations) was used and longitudinal samples were compared to check that samples were from the same patient and to exclude potential cases of HIV reinfection.

The researchers used an analysis model that ensured an accurate rate at which mutations became undetectable, enabling them to estimate the average rate loss of mutations as soon as they were identified in treatment-naive patients during chronic infection. Patient characteristics included CD4 count, viral subtype, number of mutations at the first test, and whether the mutation was pure or mixed.

A total of 717 mutations were detected in the first tests, with 1, 2, 3 and 4 or more mutations present in 59%, 19%, 7% and 15% of patients, respectively. Similarly, the percentage of patients with resistance to one, two or three ARV classes was 68%, 27% and 6%, respectively. By drug class, 65%, 38% and 24% of people had resistance to NRTIs, NNRTIs and PIs respectively.

Out of the 717 TDR mutations detected during the first test, 21% were a mixture (92 wild type, 37 with a non-TDR mutation alone, and 18 mixed). Most people (89%) only had one additional resistance test prior to starting treatment.

The overall rate of loss of TDR mutations was 18 (95% CI: 14-23) per 100 person-years of follow-up (PYFU) but there was a wide variability for individual mutations. Within drug classes, NRTIs showed the most variation in persistence. M184 was rapidly lost at 71 per 100 PYFU (95% CI: 34-149; median time to loss 1.0 years (95%CI 0.5 – 2.0 years); M41L was highly persistent with a rate of loss of only 8 per 100 PYFU (95% CI: 4-15; median time to loss 8,6 years (95%CI 4.6 – 16.0 years), and was similar to other TAMs (D67N, L210W and K219Q/N). However, K70R was lost more quickly.

Although there was a rapid transition of T215F and T215Y to one of the T215 revertants, these were then highly stable, with a rate of loss of only 5 mutations per 100 PYFU (95% CI 3-11; median time to loss 13.0 years (95%CI 6.6-25.7 years).

There was no statistical difference in the rate of loss with NNRTI variants (median time to loss 2.7 (95% CI: 1.8-4.1) years), with K103N being the most common with a rate loss of 18 per 100 PYFU (95% CI: 10-34; median time to loss: 3.7 (95% CI: 2.0-6.8) years). L90M was the most common PI variant, with a rate loss of 12 per 100 PYFU (95% CI 5-31; median time to loss: 5.8 (95% CI 2,2-15.3) years). There was little variation among the rate loss with the other PI mutations.

In the multivariate analysis there was no effect on the rate of TDR mutations when looking at CD4 count (p=0.5) or viral load (p=0.2), at the initial test, recent infection (p=0.3) or number of mutations detected at the first test (p=1.0). There was a statistically significant rate of loss higher loss with non-subtype B compared to subtype-B. (adj. HR 2.8; 95% CI 1.26-6.3, p=0.01). TDR mixtures were also associated with a significant higher rate of loss.

The authors concluded that the long persistence of certain mutations suggests that treatment-naïve patients (potentially undiagnosed) could be the route for most TDR and that baseline genotype tests should be continued in chronically infected patients. Also, due to the high variability in TDR mutations the detection of one or more mutations may signal that undetected viral mutants may have been archived in latent cells. Systematic testing will also provide more detail on the existence of TDR in the population.


The finding that certain mutations are stable and not replaced by wild-type virus suggests that most cases of TDR may come from treatment naïve patients rather than from poorly adherent people on treatment, especially given the high rates of viral suppression once HIV is treated.

This is the first time that data have supported this explanation and these results deserve further investigation.

This finding is also important as the UK HIV MSM epidemic may be largely driven by undiagnosed people – in either acute or chronic infection – and this resistance analysis included a high proportion of MSM (70%). [3]

The high variability in the time to loss for many mutations limits the use of this data for estimating the time of infection in individual cases but in detection of M184V or Y181C appears supportive of relatively recent HIV infection.


  1. Castro H et al. Persistence of transmitted HIV-1 drug resistance mutations J. Infec. Dis (2013). Advance access online: 31 July 2013. doi: 10.1093/infdis/jit345. Free full text access.
  2. UK HIV Drug Resistance Database.
  3. Phillips A et al. Increased HIV incidence in men who have sex with men despite high levels of ART-induced viral suppression: analysis of an extensively documented epidemic. PLoS One, 1 February 2013, Volume 8, Issue 2. e55312. doi:10.1371/journal.pone.0055312.

Links to other websites are current at date of posting but not maintained.