Five-year survival rates of 87% without routine CD4 or laboratory monitoring in DART study demonstrate an important model for ARV access programmes

Simon Collins, HIV i-Base

The most important study results presented at the 5th IAS conference were from the DART (Development of AntiRetrovial Treatment in Africa) trial. [1]

Sponsored by the UK’s Medical Research Council and University College London, DART randomised over 3300 treatment-naive patients in Uganda and Zimbabwe to be managed by either routine three-monthly CD4 count and laboratory monitoring (LCM group), or by clinically driven monitoring (CDM). Laboratory monitoring was also performed for this group, but results were only given to the treating doctor when a grade-4 toxicity was identified. Viral load was not monitored in either arm.

Criteria for switching patients to second-line therapy included a CD4 count less than 100 cells/mm3 in the LCM group or by symptoms/progression in the CDM group.

Patients were also randomised to one of three treatments, all with background AZT+3TC: abacavir (9%, n=300), nevirapine (7%, n=247) or tenofovir (74%, n=2469). d4T was not included as a first-line regimen, even though some patients switched to d4T during the study (for example to avoid anaemia). They were then followed for five years (median 4.9 years, IQR 4.5 – 5.3).

The Data and Safety Monitoring Board (DSMB) closed an earlier component of DART looking into treatment interruptions in March 2006.

The rationale behind DART was to determine whether ARVs could be used effectively without routine monitoring, in order to broaden access to treatment in settings where CD4 and laboratory monitoring are either no available or where they are difficult to access.

Enrollment criteria included being treatment-naive with a CD4 count <200 cells/mm3. Baseline median CD4 count and mean viral load were 86 cells/mm3 (IQR 31-139, range 0-199; one-third of patients had a CD4 count <50 cells/mm3) and 5.4 log (SD +0.7) copies/mL. WHO stage 2/3/4 was diagnosed in 20%, 56% and 23% patients respectively.

At the conference DART had a separate satellite symposium within the main conference programme and this is one of the sessions that has been web cast. The top line results were both impressive and challenged common assumptions. Both arms showed a remarkable and similar 5-year survival rate – 90% vs. 87% in the lab and clinical arms respectively – separated only by a small percentage difference that only occurred after the first two years on study. This compared to an historical 5-year survival rate prior to HAART of only 8%. Clinic attendence was >98% with high reported adherence and only 7% patients lost to follow-up over five years.

The event rates for a new WHO Stage 4 event or death were 6.94 versus 5.24 per 100 person-years in the CDM vs. LCM arm [n=459 (28%) vs. 356 (22%) HR 1.31 [1.14-1.51], log-rank p=0.0001]. Death rates/100PY were 2.94 in CDM versus 2.18 in LCM (p=0.004). Differences between strategies occurred from the third year on ART whereas lower rates of switching to second-line ART occurred in CDM from the second year. There were no differences between strategies in time to first serious adverse event, grade-4 toxicity or ART-modifying toxicity (see Table 1).

Table 1: 5-year event results from DART study

Event rate/100 PY HR [95%CI] p-value
New WHO Stage 4 event or death 6.94 n=459 (28%) 5.24 n=356 (22%) HR 1.31 [1.14-1.51] log-rank p=0.0001
Death 2.94 2.18 p=0.004
Time to first SAE HR 1.12 [0.94-1.31] p=0.20
Time to grade-4 toxicity HR 1.08 [0.97-1.20] p=0.18
Time to ART-modifying toxicity HR 1.01 [0.88-1.16] p=0.85

Around 60% of patients in each arm remained on their first line therapy after five years, with 20% modifying one or more drugs for tolerability and 20% of patients in each arm switching to second-line.

The higher mortality in the clinical monitoring arm was explained by patients being switched to second-line treatment at lower CD4 counts than the laboratory monitored group.

In an analysis of the two strategies, 3-monthly routine monitoring was determined to not be cost effective (based on the cost of treatment used in DART and relative to the WHO target for Incremental Cost Effectiveness Ratio (ICER) of 3 x GDP per capita). [2]

Lab unit costs were CD4 ($8.80), haematology ($5.30) and biochemistry ($29.50), with biochemistry carrying the highest cost with the least efficacy benefit. This was used to support the DART main conclusion that treatment shold not be witheld while waiting for monitoring and that resources for treatment access programmes should prioritise treatment over monitoring.

Several other presentations at the IAS conference presented a wealth of other aspects of the study including:

  • Impact of different WHO 3/4 events on ART on subsequent survival (Abstract MOPEB003)
  • Impact of cotrimoxazole in patients on ART: showing a 50% reduction in mortality during the first 72 weeks independent of CD4 count (Abstract MOPEB020)
  • 5 year follow-up of participants initiating ART with Combivir plus nevirapine or abacavir (randomised): showing >90% survival and >80% alive and event-free, with clear virological and CD4 advantages for the nevirapine arm (Abstract MOPEB057)
  • Assigning clinical endpoints in clinical trials in resource limited settings (Abstract TUPEB098)
  • Impact of ART on incidence of malaria in Uganda (Abstract TUPDB104)
  • 5 year follow-up of creatinine and estimated GFR in patients receiving and not receiving TDF first-line: showing overall low incidence of renal impairment on all regimens (2.9% GFR ever <30, 5.9% confirmed <60) with no differences between the LCM and CDM arms (Abstract TUPEB184)
  • Pregnancy outcomes in women in DART: showing 378 pregnancies (57% live births, 6% still births and 36% termination/miscarriage), similar rate of congenital abnormalities (~3%) as the ARV pregnancy register, and no HIV-infected babies to date (Abstract WEPEB261)

These and other presentations are now posted on the MRC website. [3]

See the next issue of HTB for detailed coverage of some of these studies.


These results strongly support expanding access to treatment to wider populations independent of access to routine laboratory monitoring and that delaying treatment access while waiting for laboratory infrastructure to be developed is resulting in extensive mortality and morbidity.

While the cost effectiveness analysis was limited to the strategy determined at the study outset, a sensitivity analysis looking at absolute real world costs based on using generic drugs (rather than subsided Western brands) and using an annual CD4 count from year two would be likely to bring this within the WHO target, and may reverse the slightly higher mortality linked to late CD4 switching seen in the clinical monitoring arm.

For example, in DART the annual drug costs for the three first line treatment arms were $432 (tenofovir), $444 (nevirapine) and $698 (abacavir) and $954 for second-line (lopinvir/r + ddI).

The modelling for the DART analysis concluded that CD4 tests would have to drop to $3.80 or less to be cost-effective, but this was based on providing 4 tests per year. Simply adding an annual test after two years would raise the minimum target cost per CD4 test to $15.20, which is well over the current costs. This implies that an annual CD4 test is actually cost effective and would detect those patients with CD4 counts <100 cells/mm3 who are at highest risk of serious events.

It is important to note that DART did not recommend excluding laboratory monitoring: the strategy was for treating without routine three-monthly monitoring, and the difference is critical to the study interpretation.

These results should also not be used to reduce the research and investment into the development of low cost diagnostics.


  1. Mugyenyi P et al. Impact of routine laboratory monitoring over 5 years after antiretroviral therapy (ART) initiation on clinical disease progression of HIV-infected African adults: the DART Trial final results. 5th IAS Conference on HIV Pathogenesis, Treatment and Prevention, 19-22 July 2009, Cape Town. Oral abstract TUSS102.
  2. Gilks C et al. Cost effectiveness analysis of routine laboratory or clinically driven strategies – DART trial. 5th IAS Conference on HIV Pathogenesis, Treatment and Prevention, 19-22 July 2009, Cape Town. Oral abstract TUSS103.
  3. DART study home page, MRC.

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