Durban World AIDS Conference
July 9-14, Durban, South Africa

REPORT 33

WHEN TO BEGIN THERAPY? Reported by Jules Levin, NATAP

The topic of "when to begin therapy" has been much discussed and and controversial in recent months. This is in part because eradication of HIV is highly unlikely, so starting therapy early in the hopes of eradicating HIV does not appear to be in general a preferable approach. In addition, side effects, difficulty in tolerating treatment regimens and in adherence add to the general feeling that therapy can be started at a later point in disease progression than was previously used to start therapy. It is difficult to maintain strict adherence over a long period of years.

Oftentimes doctors and researchers don't agree with nor follow the recommendations of the IAS and PHS Treatment Guidelines. But, the IAS and PHS Treatment Guidelines have been modified to address the concerns mentioned above. The IAS Guidelines recommend therapy when CD4s are <350 regardless of viral load, when viral load is >5000 even if CD4s are between 350-500, and when viral load is >30,000 if CD4s are >500. They recommend "considering therapy" if CD4 is 350-500 and viral load is <5000, and when CD4 is >500 and viral load is 5000-30,000. They recommend "deferring therapy" if viral load is <5000 and CD4 is >500. And, therapy is recommended for all patients with symptomatic established HIV infection. I think an important consideration in deciding when to begin therapy is if the patient is ready, willing and prepared to begin.

Several studies related to this question were reported on at Durban. Most of the studies are not randomized prospective studies but are retrospective observational studies. It has been suggested that a large randomized controlled study be started to try to answer the question "when to begin therapy". This has been controversial because questions have been raised about: is such a study feasible, would doctors and patients enroll in such a study, would we get the answers we want as conditions and treatments change over the years. Many researchers, doctors, community advocates, and NIH officials feel such a study needs to be conducted, while others feel such a study either is not necessary or is not feasible. Nonetheless, several pilot studies are in the planning stages.

So, below is a review of the studies presented at Durban on this topic.

BRIEF SUMMARY

One study from Tim Sterling at Johns Hopkins found that men with viral load of 40,000 copies/ml after seroconversion did not progress to AIDS. However, a person with 500 CD4s and 40,000 viral load vs,. a person with 250 CD4s and 40,000 viral load may present different situations. Still, these findings warrant further discussion about when is the best time to start therapy. In addition, he found similar results regarding women as other studies have found. Viral load after seroconversion in women was about half that of men (see study below). These studies report that the gender viral load differences seem to disappear 4-5 years later and that disease progression is the same for men & women. Still, further discussion is needed about-- if it's ok to delay therapy for women, the viral load criteria may need to be different for men & women regarding eligibility for clinical studies. As well, the full implications regarding gender viral load differences are not understood and need further research and discussion.

A second study at Durban suggested that CD4s, not viral load, may be a better way to decide when to begin therapy. Some researchers agree but this is controversial as some data suggests viral load is better to use. Many doctors consider both CD4 and viral load in making judgements, which I think is preferable. A CDC study suggests that 350 may a preferable time to start therapy. A study looking at a number of CPCRA studies from Dr El-Sadr at Harlem Hospital in NYC reports no significant difference  in the progression of HIV disease or death between women and men across these studies and the findings were consistent across the studies and over time.

Initial HIV-1 plasma RNA level after seroconversion does not predict progression to AIDS in women

Tim Sterling from Johns Hopkins  University reported at a late breaker oral session on this retrospective analysis on the relationship between a person's initial viral load after HIV-1 sero-conversion and it's use in predicting progression to AIDS.  Progression to AIDS was defined as an AIDS defining opportunistic infection (OI). Sterling said initial HIV-1 plasma RNA level (VL) after seroconversion  (SC) predicts progression to AIDS in men but this has not been well-studied  in women. Several recent studies have demonstrated lower VL in women  than men at the time of sero-conversion, and sex differences in HIV-1 viral  dynamics.

We measured plasma VL among all HIV-1 seroconverters  enrolled in a longitudinal cohort study of injection drug users  (IDU) and correlated the initial VL with the subsequent development  of AIDS. 3,380 IVDUs were enrolled and followed every 6 months between March 1989-December 1998. VL was subsequently quantified for each visit after HIV-1 SC. Inclusion criteria  for this study were HIV-1 SC within 12 months of last sero-negative visit and before 12/1/97, and > 2 plasma VL's after SC. VL  was determined by RT-PCR (Roche).

There were 295 sero-converters, of whom 202 met the inclusion  criteria; women were not disproportionately excluded. Although 29/156 (19%) male and 15/46 (33%) female sero-converters progressed  to AIDS, time to AIDS did not differ by sex (p = 0.19). Time  between SC and 1st VL did not differ by sex (median = 4 mos;  p = 0.8).

Among male seroconverters, initial median VL was 77,822  copies/ml among those who progressed to AIDS compared to 40,634  copies/ml among those who did not (p = 0.009). Among female seroconverters,  initial median VL was 17,149 copies/ml among those who progressed  to AIDS compared to 12,043 copies/ml among those who did not  (p = 0.21).

Sterling concluded initial VL after SC did not predict progression  to AIDS in women but did in men, but this may be related to the low number of women in the analysis (n=46). Initial CD4 did not differ between progressors and  nonprogressors among both men and women. These results are consistent with a sex difference in  viral dynamics and suggests that these data should be incorporated  into treatment guidelines for women.

In addition, when looking at risk of progression to AIDS, by 1 log changes in the entire population (not in an individual) there was an increase risk progression to AIDS (AIDS defining OI) of 1.5 (hazard ratio) for women & men (not statistically significant either). Higher VL is associated with greater viral load. When you have a guideline uniformly applying same standard to men & women a question.

Comments: A number of studies including this Sterling study have found that although initial viral load may be 40% lower in women than men, the risk of progression to AIDS is about the same. The sex differences in viral load appear to even out after 4-5 years. But, if VL eligibility is the same for starting therapy as defined by the IAS and PHE Treatment Guidelines its possible women are starting therapy too early. At this point this suggestion, that women may be starting therapy early, is preliminary and needs to be discussed. This study found men did not progress when VL was 40,000 copies/ml after sero-conversion and this ought to be discussed because Treatment Guidelines currently considering recommend therapy at lower levels.

Some experts suggest CD4 levels may be a better standard for deciding when to begin therapy. In his presentation Sterling reported from the ALIVE Cohort where 20,000 copies/ml were used for the cut-off, and 74% of men and 37% of women would be eligible for Therapy. But if you looked at a CD4 cutoff of 500 there was no difference between men & women. Despite same risk of AIDS more men than women are eligible for treatment. Using CD4, there would be less sex differences in when to begin therapy.

To start or not to start? Diminished effectiveness of anti-retroviral therapy among patients initiating therapy with CD4+ cell counts below 200

Julio Montaner from the Britsh Columbia Centre for Excellence in HIV/AIDS in Vancouver, Canada reported on which CD4 and viral load cut-offs for starting therapy were predictive of death. So, his outcome measure was only death, not AIDS defining opportunistic infections. This is a key point because in deciding when to begin therapy we want to prevent getting sick not just progression to death. Also important to the results reported in this study was that the follow-up is 20 months. It's possible the results and conclusions will change with longer follow-up.

Participants started triple combination therapy between Aug. 96 and Oct. 99. Rates of progression from the start of therapy to death were stratified by CD4 and plasma viral load levels. Risk ratios of factors associated with mortality were estimated using Cox-proportional hazard models.

A total of 1219 men & women were eligible. As of  Jan. 2000 72 patients had died of AIDS related causes, for a "crude" mortality rate of 5.9%.

In univariate analysis, age, gender, and PI use were not significant in predicting progression to death, whereas AIDS at baseline, CD4 count, and plasma viral load were significantly associated with death. But, in multivariate analysis only CD4 count remained associated with death. In other words, whether a person's viral load was 20,000 or 150,000 when they started therapy, it did not change the person's risk for death. But, the CD4 count mattered.

Montaner divided CD4 and viral loads into 7 categories to examine if there were any natural breakpoints where survival probabilities significantly differed between any 2 levels. Montaner divided CD4 categories into: <25, 25-49, 50-99, 100-199, 200-349, 350-499, and >500. He divided viral load categories into <10,000, 10,000-49,000, 50,000-99,999, 100,000-199,000, 200,000-299,000, 300,000-399,000, and >400,000. Montaner reported that in a multivariate model of analysis of his study data, CD4 cell count at baseline remained the only statistically significant factor associated with shorter survival and an independent predictor of mortality.

By my reading the Figure 2 graph in the poster, the probability of survival when CD4 count was "high"  (350-499, n=257) was about 98% after 2 years, when the CD4 count was very high (>500, n=192) it was about 98%, when the CD4 count was moderately high (200-349, n=327) it was about 98%. When the CD4 count was moderately low (50-99, n=102) the probability of survival was about 93%, and about 90% at the intermediate level (100-199, n=199). This suggests a difference in rate to death between a CD4 below or above 200.

Early Initiation of combination anti-retroviral therapy: does it affect long-term outcome?

Johnathan Kaplan from the CDC reported on a study whose goal was to determine whether starting therapy when CD4s are >350 affects the clinical outcome. In this study the outcome used for evaluation were AIDS OI, death, CD4 <200, and CD4%<14. So, this study differed from Montaner's in that Montaner's study used death as the only parameter for outcome while Kaplan used the development getting sick or low CD4s (<200). This is a more realistic measure.

Kaplan analyzed data from the CDC Adult & Adolescent Spectrum of HIV Disease Project (ASD), a medical review study of HIV-infected persons in 11 US cities. They considered only patients who enrolled in Jan. '94 or later, had a CD4 count >350, and had no history of AIDS OI at the time of enrollment. They were also treatment naÔve and were started on 0, 2, or 3 antiretroviral (ART) drugs while there CD4 count was >350: 2 drugs= 2 NRTIs, 3 drugs= 2 NRTIs plus either a PI or NNRTI.

They estimated 2 year incidence of AIDS (CD4<200, CD4%<14, or AIDS OI) using Kaplan-Meier procedure. They estimated hazard ratio (HR) for developing AIDS in proportional hazards regression model after adjusting for ART regimen, age, sex, HIV risk mode, site, baseline CD4 count and hemoglobin, PCP and MAC prophylaxis, history of pneumococcal vaccination, and average number of outpatient visits/year.

In case the table below is not readable here's the data:

Through Dec. '99, patients starting 2 or 3 vs, 0 ART drugs were followed for 953 patient years vs, 1701 patient years, respectively. There were 590 patients in the 2-3 drug arm and 1030 in the 0 drug arm. AIDS was diagnosed in 34 patients receiving 2-3 ART drugs and in 126 patients receiving 0 drugs. The hazard ratio for developing AIDS prescribed 2-3 drugs was 0.21 (.13, .34) and remained significant when controlling for prescription of ART after CD4s dropped below 350. (*)There were too few  HIV-related deaths (4 and 9, respectively) to allow analysis of death as an endpoint. The 2-year AIDS incidence in patients starting on 2 or 3 vs, o drugs was 7.7% (95% CI 4.7, 1.7) vs. 16.6% (95% CI 13.3, 20.0).

PATIENT-TIME, AIDS CASES, DEATHS

(#) 95% confidence interval

Comments: This study found that starting therapy when CD4s were >350 was preferable and safer than starting therapy when CD4s were <350, because in a 2 year follow-up the incidence of AIDS was lower when starting therapy at >350 CD4s (7.7% vs. 16.6%). However, the study did not compare starting therapy at 500 CD4s vs. starting therapy at 350 CD4s. The results of this study is to some degree at odds with the results of the Montaner study where he appears to find no difference in progression to death (2 year follow-up) whether you start therapy at 200 or above 200 CD4s. However, the key difference in the studies and their conclusions may be simply related to the outcomes measured--Montaner used only death and Kaplan looked at the development of AIDS (CD4 <200, CD4% <14, and AIDS OI).

Kaplan discussed limitations of his study in his poster. His study is observational, patients were not randomized. However, multivariate model included covariates related to stage of disease and health care: ART regimen, age, sex, HIV risk mode, site, baseline CD4 count and hemoglobin, PCP and MAC prophylaxis, history of pneumococcal vaccination, and average number of outpatient visits/year. They could not measure adherence to therapies. There were insufficient numbers of deaths to analyze death as an endpoint. And, other outcomes were not measured: toxicity, development of resistance, and quality of life.

Kaplan concluded that the initiation of ART at CD4s >350 is associated with decreased AIDS incidence in this observational cohort database. However, insufficient numbers of deaths preclude a definitive conclusion regarding the benefits or risks of early ART at this stage of the analysis.

The effect of gender on progression of HIV desease among patients on combination antiretroviral therapy

Wafaa El-Sadr from Harlem Hospital in New York City reported on this study. She assessed  risk of progression of HIV disease or death among patients enrolled in randomized CPCRA studies in the past decade. These studies were grouped based on the type of combination therapy prescribed.

Group 1 patients were randomly assigned dual nucleosides and Group 2 patients were randomly assigned triple ART. Hazard ratios (HR) and chi squares were calculated for predictors corresponding to gender, race/ethnicity, age, baseline CD4 and history of IVDU.

Demographics: Group 1 patients (n=730,(62 Women, 668 Men) and Group 2 patients (n=885), (136 W,  749 M). Blacks or Latinas were 86% and 79% of W in Groups 1 and 2,  respectively. Injection drug use (IVDU) was noted among 42% of  W in Group 1 and 20% in Group 2. Percent of W increased from 9% in Group 1  to 15% in Group 2. Median follow-up was 26 and 21 months in Group 1  and Group 2, respectively.

El-Sadr conclusions: Previous reports that suggested that women differed from men in rates  of HIV progression, involved studies in which patients were not randomly  assigned to treatments, did not have similar access to care and  did not have similar follow-up. In our population, where the  mix of drugs and the intensity of follow-up were the same for  the two genders, as a result of randomization, baseline CD4 cell  count and type of ART had a significant impact on the risk HIV  progression. History of IVDU was associated with higher risk of progression.  No significant difference was noted in the progression of HIV disease or mortality between women and men across these studies and the findings were consistent across the studies and over time.