Resistance in the UK: new approach to epidemiological studies

Simon Collins, HIV i-Base

Incidence of resistance is frequently presented from an analysis of single time point data from specific populations such as new infections, primary infection and national or local cohorts. This is often used to interpret prevalence and trends in resistance to specific drugs and drug classes.

However, these approaches do not provide an accurate estimate of the percentage of the population at any time who are resistant to a given class, or the characteristics of patients who are developing individual patterns of multi-drug resistance.

Annual reports from resistance databases only provide information relating to current or very recent treatment, as historical and archived mutations will not necessarily appear on in these results. The denominator used for these analysis can also be inappropriate depending on whether the total number of tests or the total number of treated patients is used.

A clearer indication of the incidence of drug resistance in the UK was presented by Deenan Pillay from six large database collaborating in the UK Collaborative HIV Cohort (CHIC). [1]

The number of annual resistance tests in this cohort rose steadily from 500 per year in 1997 to 2500 in 2000 and has remained stable at that level. Data was presented up to end of 2002.

Previous analyses of single time point analyses from this dataset have shown fairly stable percentages of tests with resistance to 1 class (75%), 2 class (55%) and 3-class (15%) resistance from 1998 – 2002. They also showed percentage of samples with NNRTI resistance overtaking PIs resistance in 1999 and remaining higher (50% vs 30%) in line with changes in prescription practice.

When the data was analysed by cumulative number of resistant classes compromised it became clearer to see the increase in 1-, 2-, and 3-class resistance in the population as a whole over time. Prevalence figures were obtained by using the number of patients on treatment in the UK (from SOPHID) as the denominator.

Table 1: Cumulative resistance

≥1 class ≥2 classes ≥3 classes
Cumulative resististance in 2000 2250 1600 500
Cumulative resististance in 2002 3500 2500 800
Prevalence 2002 (% UK pt on Tx) 18% 12% 4%

Prevalence of resistance by class showed approximately 17% patients have resistance to any class, 16% to RTIs, 10.5% to NNRTIs and 7.5% to PIs. These figures represent a minimum lower level.Despite inclusion in treatment guidelines, resistance tests were not universally used in 2002.

The percentages in this study were from all resistance tests that were requested, and many of these patients would have been on failing treatment with suspected resistance. The definition of ‘class resistance’ was based on a single mutation in any one class.

A second poster from the same group, presented by Andrew Phillips, looked at results from resistance tests from routine care of patients from six clinics in London/Brighton who were using combinations of three of more drugs.

From 1996-2003, around 4450 patients started ART with >/=3 drugs. 56% started with an NNRTI, 41% with a protease inhibitor. The cumulative risk of virological failure (two viral load >1000 copies/mL after 24 weeks from start of ART, unless during interruption) was 24% by 2 years, 34% by 4 years and 42% by 6 years.

Almost 1000 patients (22%) had a resistance test result at some time after start of ART, 859 of which produced an evaluable result. Cumulative risk of resistance is shown in Table 2 below.

Table 2: Cumulative risk of resistance

2-year 4-year 6-year
Cum. risk of vir. failure 24% 34% 42%
Resistance to any class 10% 20% 27%
Resistance to 2 classes 6% 14% 18%
Resistance to 3 classes 1% 2.5% 3.5%

Risks by 6 years for class-specific mutations were: M184V/I 18%, >/=1 TAM 15%, NNRTI 17% (25% when restricted to those who started with NN), major protease mutation 8% (10% when restricted to those who started with a PI).

The study noted that these are lower limit estimates as test results were not available for many with virological failure, and resistance below sensitivity limits of assays will be missed. Factors presented in poster.

Significant factors associated with development of resistance in a multivariate analysis is included in Table 3.

Table 3: Risk factors associated with the development of resistance in multivariate analysis

Viral load <100k 0.64 (0.51-0.79) (vs10-99k) p < 0.001
Previous AIDS 1.33 (1.10  1.61) p = 0.003
No PI or NNRTI 1.83 (1.32  2.52) p<0.005
Using >3 drugs 0.71 (0.52  0.97) p=0.03

Although most patients achieve durable suppression using 3 or more drugs, this study shows that an appreciable number do not.


  1. Pillay D et al on behalf of UK CHIC Group. Estimating HIV drug resistance in the UK treated population. XIII Intl HIV Drug Resistance Workshop, 2004. Abstract 77. Antiviral Therapy 2004; 9:S88.
  2. Phillips AN et al on behalf of UK CHIC Group. Risk of development of drug resistance in patients starting antiretroviral therapy with three or more drugs in routine clinical practice. XIII Intl HIV Drug Resistance Workshop, 2004. Abstract 135. Antiviral Therapy 2004; 9:S151.

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