Monitoring treatment in resource limited settings: results from PHPT-3 and Stratall ANRS12110/ESTER trials

Polly Clayden, HIV i-Base

In resource-limited settings, optimal monitoring and switching criteria from first-line to second-line therapy is unclear. Results from two trials were shown as oral presentations that suggest that monitoring viral load is not essential for switch to second line. [1, 2]

Marc Lallemant showed data from PHPT-3, which was conducted in Thailand. This was a randomised double-blind (until first switch) non-inferiority trial. Participants were randomised to CD4 or viral load monitoring, which was conducted every three months.

Dr Lallemant explained that the trial was designed for a setting with only two lines of treatment and where second line is far more expensive than first line. The investigators wanted to test whether monitoring and switching people without viral load compromised their health or their future options.

PHPT-3 enrolled HIV-positive adults (CD4 count 50 to 250 cells/mm3, not hepatitis B or C co-infected), starting NNRTI-based HAART.

In the CD4 monitoring arm, patients switched to second-line protease inhibitor (PI) -based therapy when they had confirmed CD4 decline of 30% or more from peak, and in viral load monitoring they switched when they had confirmed viral load >400 copies/mL.

The primary endpoint was death, new AIDS-defining event or clinical failure – defined as CD4 <50 cells/mm3 – at 3 years. Secondary endpoints included proportions switching to second line, time to switch, resistance mutations at failure and future treatment options.

The trial enrolled 716 patients of which 60% were women. Their median CD4 count at baseline was 144 cells/mm3 (range 90 to 200 cells/mm3).

Regimens were 65% efavirenz-based regimen and 66% of participants received tenofovir/FTC. Other study drugs were nevirapine and AZT/3TC. At 3 years of follow up 93.3% of patients were evaluable. Ten percent stopped treatment for toxicity across both groups.

There were 58 clinical failures overall, 28 and 30 in the CD4 and viral load groups respectively. The respective rates of clinical failure per patient years were 2.3 vs 2.5 and of death 1.1 vs 1.4.

In multivariate analysis, anaemia, adjusted HR 2.7 (95% CI 1.5-4.8), p=0.001; CD4 <150 cells/mm3, AHR 2.3 (95% CI 1.2-4.2), p=0.009 and viral load >5 log, AHR 1.8 (95% CI 1.0-3.0), p=0.04, were predictive of clinical failure at 3 years.

The probability of switch to second-line (excluding toxicity/intolerance) was 5.2% (95% CI 3.2-8.4%) vs  7.5% (95% CI 5.0 -11.1%) in the CD4 and viral load groups respectively, p= 0.10.

The respective median times to switch were 11.7 months (95% CI 7.7-19.4) vs 24.7 (15.9-35.0), p=0.001. And the median duration of viraemia >400 copies/mL was 7.2 months (IQR 5.8 to 8.0) vs 15.8 months (8.5 to 20.4), p= 0.002. But the median CD4 counts were 426 cells/mm3 vs 420 cells/mm3, respectively.

Dr Lallemant noted that 15/31 patients in the CD4 monitoring arm who switched to second-line had viral load <50 copies/mL at the time of switching.

Viral load was <50 copies/mL in 99% of patients at 3 years follow-up and patients with CD4 monitoring did not have fewer future treatment options, with the exception of one patient with multiple thymidine analogue mutations (D67N/M41L/L210W/T215Y).

Dr Lallemant concluded that, after 3 years, the rate of clinical failure was very low and did not differ between the two strategies. Most mutations had been selected at the time of virological failure. The additional time spent on failing treatment in the CD4 arm did not result in reduced future treatment options.

He noted that the conclusions from PHPT-3 are similar to those from DART and HBAC in adults and PENPACT-1 in children. He added that the need for viral load monitoring may be less important than close and regular safety, tolerability, adherence, and immunological monitoring. He remarked that the nurse/patient team with expert assistance from doctors, biologists and patient networks “maximizes efficacy and durability.”

This was followed by a related presentation of data from the Stratall ANRS12110/ESTER trial.

Charles Kouanfack showed findings from a trial designed to compare clinical monitoring alone with laboratory and clinical monitoring. This trial was conducted in 9 rural district hospitals in Yaounde, Cameroon.

Dr Kouanfack explained, in Cameroon, the national programme followed WHO guidance for a public health approach based on decentralised, integrated HIV care delivery in facilities where laboratory monitoring is generally unavailable. He noted that the 2010 guidelines also state that using viral load monitoring to detect treatment failure and switch is recommended but has “low quality evidence”.

Stratall ANRS12110/ESTER was a randomised non-inferiority trial enrolling HAART-na, HIV-positive adults with a WHO stage 3-4 disease or stage 2 and total lymphocyte count <1200 cells/mm3, who were followed for 2 years. Management was by the health workers in charge of routine activities.

The primary endpoint was mean increase in CD4. The increase in the clinical monitoring arm was judged to be non-inferior to that in the laboratory monitoring arm if the difference was less than or equal to 25%.

Secondary endpoints included: viral suppression, death, new stage 3 or 4 events, resistance, loss to follow up, adherence, treatment changes and toxicity.

Participants were monitored clinically 3 monthly in both arms and those in the laboratory monitoring arm also had CD4 and viral load measured every 6 months.

Switching to second line was indicated by grade 3 or 4 events in the clinical monitoring arm and persistent viral load >5000 copies/mL in the laboratory monitoring arm.

Of a total of 493 patients, 256 were assigned to clinical and 237 to laboratory monitoring. Of these, 93% were followed and included in the analysis. Patients were similar at baseline with CD4 counts of 179 cells/mm3 and 182 cells/mm3 in the clinical and laboratory monitored arms respectively. Both arms had high baseline viral loads of 5.6 log10 copies/mL. Overall 70% were women. About 65% started treatment with d4T + 3TC + NVP.

The trial failed to demonstrate non-inferiority of clinical monitoring:  the mean increase in the CD4 count was 175 cells/mm3 (95%CI 151-200) vs 206 cells/mm3 (95% CI 181-231) in the clinical and laboratory arms respectively. This gave a difference –31 (–63 to +2), the non-inferiority margin was –52 (–58 to –45). The analysis was last observation carried forward.

The analysis also revealed that 13 (6%) laboratory-monitored participants switched to second-line regimens because of treatment failure, compared to none of the clinically monitored participants, p<0.001. But, viral suppression (49 vs 52%), resistance (both 10%), mortality (18 vs 14%), disease progression (36 vs 29%), adherence (both 64%), loss to follow-up (9 vs 8%), and toxicity (19 vs 25%) were similar between the two groups.

Dr Kouanfack concluded that failure to demonstrate non-inferiority of immunological recovery and the need to switch to second line in this trial supports the WHO recommendation of laboratory monitoring of HAART where possible.

He also concluded that the difference between the two strategies suggest that clinical monitoring alone can be used for at least the first two years of treatment in order to expand scale up and to take into account financial and infrastructural constraints in resource limited settings.


  1. Jourdain G et al. PHPT-3: A randomised clinical trial comparing CD4 vs viral load ART monitoring/switching strategies in Thailand. 18th CROI, 27 February–3 March 2011, Boston. Oral abstract 44.
  2. Kouanfack C et al. HIV viral load, CD4 cell count, and clinical monitoring vs clinical monitoring alone for ART in rural hospitals in Cameroon: Stratall ANRS 12110/ESTHER trial, a randomised non-inferiority trial. 18th CROI, 27 February–3 March 2011, Boston. Oral abstract 45LB.

Both webcasts: Research on Delivery of Care in Developing Countries. Monday 28 February 11 am.

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