8th Annual Retrovirus Conference
Late Breakers
Chicago, Feb 4-8 2001

When To Begin Studies Reported at Retrovirus Conference

Because it is recognized that eradication is no longer achievable with the current crop of drugs, and because of the toxicities from these drugs and the difficulties in adherence and the consequences of developing resistance if not adherent doctors have been deferring the start of therapy. The Public Health Service Guidelines recently recognized this by changing the document to recommend consideration of deferring therapy til CD4s are 350 or viral load 55,000 by PCR or 30,000 by DNA. The text of the guidelines says doctor & patient can consider starting earlier. A link to the Guidelines is on the front page of the NATAP web site and can be downloaded.

I'm not a fan of the Guidelines. There is no clear answer on when to begin therapy, and conducting one large study to answer this is very difficult, and I think would not yield the answers to our questions. Deciding when to begin therapy is a personal & individual decision. Every patient has an individual situation, which is different from everyone else. Lifestyles, eating habits, preferences about therapy (which side effects are tolerable to them, etc), CD4s, viral load are all different between individuals. Each individual has their own personal concerns, which will differ from other's concerns. A person should weigh in deciding when to begin therapy are the potential risks & benefits of starting or deferring therapy. Below is a discussion of many of the things to consider. This list of benefits & risks may not be exhaustive.

Why should a person start therapy early (when cd4s are 350-500)? After starting therapy people tend to develop certain adverse effects, side effects, & toxicities. But there are potential benefits to beginning therapy early, when CD4s are over 350 or at 500. The risks and benefits should be carefully weighed by the patient, with their doctor before deciding when to begin therapy. HIV-related CNS disorders may be more likely to occur as viral load gets higher. HIV and certain HIV medications can lead to peripheral neuropathy.

To be clear, without ever taking any HIV medications, just having HIV may lead to peripheral neuropathy and other HIV-related CNS disorders. We know that shortly after a person becomes HIV-iinfected HIV enters the Central Nervous System. Data has shown that viral load over 10,000 increases risk for peripheral neuropathy by 25% over a 10-year period. Lower CD4s can also increase this risk. HIV by itself can also lead to neuropathy. In general, when HAART is started the immune system stops the decline caused by HIV: CD4s increase, the cd4/cd8 ratio improves, etc. Deferring therapy & letting cd4 count decline permits continual decline in the immune system. We don't know what cd4 count is the cut-off for increased risks. Over the course of many years cancers may be more likely to develop if CD4s go to low before starting therapy. We don't have data proving this but this is a concern. It could take many more years before we see cancers starting to pop up. Allowing viral load to replicate unchecked permits a patient's virus to mutate & become genetically diverse. This may have the effect of reducing the response to therapy. Several studies show that individuals who start therapy when cd4s are 350-500 respond better to therapy which in theory (if person remains adherent) should lead to more durability in response.

Why should a person defer therapy & til when when cd4s are >350 or 200)? After starting therapy certain side effects & toxicities may occur. Some people may prefer to delay therapy to defer experiencing them. After starting therapy you risk developing lipodystrophy. A significant percentage (estimated to be 20-50%) of patients taking HAART develop what's generally called "lipodystrophy". This syndrome includes body changes: fat loss in the arms, face, buttocks, hip area, legs; fat accumulation in breasts for women & much less likely in men, stomach, fat pads developing on back of neck called "buffalo hump). Protease inhibitors can lead to elevated cholesterol, triglycerides and less likely elevated sugar & diabetes. Elevated liver enzymes can occur after starting therapy. For the person with hepatitis B or C, these elevations may be higher & people are concerned if elevations are high. We don't know how risky it is for people with hepatitis & HIV to experience high elevations in liver enzymes (ALT), but we are concerned about it. Elevated liver enzymes create inflammation in the liver. The question we don't have an answer to --does this inflammation leads to increased fibrosis & liver damage harmful to HCV progression. For some individuals, it may be preferable to start HCV therapy before HIV therapy. A number of studies show that increasing cd4s to 300 or more after starting HAART appears to protect people from getting the major opportunistic infections they were in risk of getting when their cd4s were <200 or <100. Some doctors feel that a patient has the best chance of achieving the highest cd4 increases when starting therapy earlier. They feel deferring therapy until cd4s are lower risks that cd4 increases may not be as large. However, a couple of studies suggest this may not be true. These studies suggest cd4 increases may be the same whether you start at 250 or 400 cd4s.

Adherence & Lifestyle Changes. Once a person begins therapy they need to adhere closely to a strict schedule of taking the medications. This requires taking pills at least twice daily, although there are a few once-a-day regimens. There are several twice daily regimens that have very minimal numbers of pills but you have to speak to your doctor about whether they are appropriate for your individual situation. Still, a patient will HAVE TO remember to take the pills on time every day. Certain pills must be taken with food and others don't have to be taken with food. But the patient is required to remember these details & to follow these requirements. This restricts your lifestyle. Since adherence is difficult and many people are not able to adhere, they can develop resistance to the HIV drugs. Some people think this limits future treatment options. But many doctors do not feel that way since there are many options available now and a number of additional treatments are in development.

The Studies Reported At Retrovirus

Data from Cozzi & Phillips presented at the Glasgow Conference in October 2000 show starting therapy when CD4 is 200-350 and when viral load is below 100,000 does not affect CD4 & viral load response to therapy. These data can be seen in the Glasgow Reports on the NATAP web site. One researcher told me at Retrovirus that if his patients are completely adherent then therapy when CD4s are <200 also shows same ability to respond to HAART as measured by CD4s & viral load responses. At Retrovirus, there are conflicting results being reported by different studies. The Sterling study finds HAART slows disease progression only when started after CD4s are below 200, but this study has a 2-year follow-up, compliance is not monitored, and it compares HAART to no-ART plus double nukes.

The Opravil study shows delaying therapy accelerates progression and highlights the difficulties for patients in adhering to regimens & in tolerability and adverse events. The CDC study below yields interesting data as it looks at survival and when to begin therapy. Their follow-up of 2 years is too short but the data suggests that starting at >350 may be less risky than starting at 200-350, and certainly starting at <200 is risky. All the studies have flaws & limitations. First, they all have very short follow-up time. Following people for 2-3 years may not yield reliable data. After 2 years on treatment you may not see a difference in clinical HIV progression between two individuals starting at 400 cd4s vs 250 cd4s. But if you follow them for 6 years you may see a difference. In the Opravil study, patients and/or doctors decided to defer therapy or start therapy earlier. This may have created a bias. These might be patients who may be less likely to be adherent, who may use illegal drugs or for some reason may think therapy will not help them.

So, in the end a patient needs an HIV treating physician with good judgement and experience in treating large numbers of patients, so he/she has a good basis for helping the patient make the decision on when to begin therapy. Of course, a well-informed patient can play a key role in making this and other treatment decisions. If a person is not ready to make the commitment necessary when starting therapy, it may be better to defer it. When to begin therapy differs depending on individual circumstances and so is different for everyone. You must weigh & consider the requirement for adherence to a regimen while on therapy (non-adherence leads to resistance to drugs), concerns about tolerability and developing adverse events, and the risk for disease progression if therapy is deferred.


In brief, Opravil found clinical progression occurred more quickly for individuals deferring therapy after CD4s declined past 350, so people who started therapy at >350 CD4s (actual CD4 count was about 500) did NOT progress as quickly, when measured by progression to CDC stage B/C (17% in the untreated group vs 4.5% in the treated group) progression to CDC stage C (4.5% vs 0.8%), progression to AIDS events (16 vs 3) and death (5.3% vs 1.4%). The occurrence of 16 AIDS events in the untreated group vs 3 in the treated group is concerning. BUT, 4.5% discontinued HAART due to virologic failure, 29% changed therapy due to adverse events (listed below), 41% had a treatment interruption of 1 or more, and 19% were not on ART anymore at the end of the follow-up period. He concluded that risk of clinical progression was reduced 7-8 fold by starting therapy when CD4s were above 350 when persons were asymptomatic. BUT, if untreated, there was a relatively low risk of severe clinical events (4.5% for AIDS, 5.3% for death). He said the risks & benefits must be weighed in making a decision on when to begin therapy. Patients in this study have been followed ONLY for a short time (1-2 years), as in all these studies.

"Clinical Benefit of Early Initiation of HAART in Patients with Asymptomatic HIV Infection and CD4s >350" (Late Breaker abstract LB6)
     In the oral Late Breaker Session, Milos Opravil (Division of Infectious Diseases, University Hospital, Zurich, Switzerland, for the Swiss HIV Cohort Study) reported clinical benefit from early initiation of HAART when CD4s are >350.

As background, Opravil mentioned the ongoing discussion about the value of HAART being initiated in patients with preserved CD4s. The pro being immune restoration & the con being long term toxicity, preserving future treatment options by avoiding non-adherence & resistance & the costs of therapy. He said that since it is difficult to conduct clinical endpoint studies and randomized large clinical when to begin trials, results from cohort studies such as this may approximate clinical trials.

The Swiss HIV Cohort Study is a prospective observational database. They collected clinical & treatment follow-up information every 6 months and all laboratory results were collected. There is a total of 11,200 patients, 6550 with follow-up after 1/96, 83% of whom have had antiretroviral therapy (ART), and reasons for stopping or changing therapy have been prospectively collected.

STUDY OBJECTIVE was to quantify the benefit and the problems brought about by HAART initiation in asymptomatic patients with >350 CD4.

The investigators used a case-control study design where results were compared between two matched groups: (1) the treated group were treatment naïve, asymptomatic & starting HAART with CD4s >350; (2) the controls were matched patients NOT starting ART for at least the next 12 months. The analysis was intent-to-treat.


The treated patients started HAART 1/96-12/99, were never before treated, were asymptomatic, had >350 CD4 and had at least one follow-up.

The controls matched for date as a case ± 1 year, no ART in the next 12 months, IVDU as risk for HIV, age ± 20%, HIV-RNA ±0.5 log, CD4 ±20%, and gender. So, it appears as if these were individuals who decided to delay therapy or their doctors decided to defer therapy. Therefore, it's possible that this was a pre-selection bias. This group could include individuals who are less likely to adhere or believe that therapy works. It might also include people who have a illegal drug problem.

There were 358 matched pairs. They performed a retrospective (look back) patient chart review. And all clinical events were verified and an update was kept for unclear reasons for treatment change/stop.

There were 358 treated & untreated patients in the 2 groups, respectively.

In treated group follow-up time was a median of 2.27 years and 1.27 years in the untreated group (p<0.001). Total person-years follow-up was 745 in treated & 548 in untreated.

Lost to Follow-up:
  51 (14.2%) in treated group vs
103 (28.8%) in untreated group

66 patients (18.4%) started ART during follow-up period (beyond 1 year) in the untreated group. The median delay was 1.58 years, and when starting median CD4 was 323 (n=63) & viral load 4.37 log (23,400 copies/ml), n=62.

Opravil showed graphs depicting CD4 changes over 4 years. My visual view of the graphs follow. In the treated group CD4s increased on average to about 800 at the 4 year point. In the untreated group it appeared as though CD4s on average declined to about 400 at the 4 year point. The percentage of patients with <200 CD4s increased over the 4-year period in the untreated group.

Over 4 years follow-up,

62 in the untreated (17.3%)
16 in the treated (4.5%) progressed (p<0.0001).

Over the course of 4 years,

16 untreated (4.5%)
3 treated (0.8%) progressed (p=0.0001).

Over 4 years, 3 in the treated group and 16 in the untreated group (1 after starting HAART) experienced events: treated with HAART-- 1 esophogeal candidiasis (ESO at 800 CD4s, 1 pulmonary/extrapulmonary tuberculosis (TBC/TEX), and 1 Non-Hodgkins Lymphoma (NHL) at about 700 CD4s; in untreated group-- 4 cases of Kaposi's Sarcoma at 700, 600, 250, and 400 CD4s, respectively, 2 cases of dementia at 400 and 500 CD4s, 2 cases of TBC at 500 and 250 CD4s, 1 case of TEX at 200 CD4s, 1 case of salmonella sepsis at 400 CD4s, 2 cases of NHL at 400 CD4s, 2 cases of bacterial pneumonia at 200 and 150 CD4s (after initiating HAART), 1 case of cryptosporidiosis (after starting HAART), 1 case of ESO at 50 CD4s.

Over 4 years 19 (5.3%) in the untreated vs 5 (1.4%) in the treated group (p=0.0006) died from all causes. After excluding suicide & drug overdose, 12 untreated (3.3%) and 4 (1.1%) died over 4 years. There was a higher probability of dying after excluding these causes (p=0.018).

Reasons for death, Treated vs Untreated
HIV related (4 lymphoma, 1 PML, 1 salmonella)  
0 vs 6
Other ID (endocarditis, meningitis, septicema)  
0 vs 3
1 vs 1
Cardiac reasons  
1 vs 2
Narcotic overdose  
1 vs 4
0 vs 3
2 vs 0
   5 (1.4%) treated vs
19 (5.3%) untreated

In multivariate analysis, the hazard ratio for CDC B/C event was 0.15 and 0.12 for death (p<0.001) for starting therapy at >350 & being symptomatic vs deferring therapy. Viral load at baseline was predictive for progression to CDC B/C event: a log difference in viral load had a hazard ratio of 2.10 (p<0.001). So, for every 1 log higher viral load (10,000 vs 100,000 is 1 log difference) the chances for progression to CDC B/C was increased. But viral load differences was not predictive in this study for CDC C event & death. Being an IVDU was predictive for progression to CDC B/C and C event and to death. NOT PREDICTIVE: differences in CD4s of 100, age (per 10 years), gender, hemoglobin at baseline.

Opravil said that the untreated group had higher drop-out rate and thus shorter median follow-up time but artificial prolongation of follow-up duration for all non-progressing controls until Dec 2000 does not impact on the differences between the groups.

TREATMENT ISSUES (n=333, 25 excluded in STI)

- 60% (200) changed >1 drug
- 15 (4.5%) for virologic failure
- 97 (29%) for intolerance & adverse event
GI 19%
Nephrotoxicity 3%
Nervous system 2.1%
Lipodystrophy 1.2%
Hyperlipidemia 0.9%
- 13.5% for "patient wish"
- 9% for other causes
- 3.9% for reason not documented

41% had >1 treatment interruption
19.5% were not on ART at the end of follow-up

Author's Conclusions

"CD4 Lymphocyte Count Level is Better than HIV-1 Plasma Viral Load in Determining When to Initiate HAART" (abstract 519)
     Tim Sterling (Johns Hopkins) et al evaluated the impact of HAART (>90 days treatment with PI, NNRTI, or triple NRTI regimen with abacavir) on new opportunistic infections or death assessed by CD4 & viral load before starting HAART. Patients in this study were enrolled in care after 7/96 and study period was thru 6/2000. Again, follow-up has been for ONLY 2 years.

1014 patients treated at the Hopkins HIV Clinic were either in group 1 (n=530) and received HAART for at least >90 days, or were in group 2 and received no ART or received dual nucleosides. If I heard Sterling correctly in that group 2 could have received dual nukes, this makes a difference to me in the analysis. Comparing HAART to dual NRTIs you may not see different results. In the Opravil study, its my understanding, patients received either HAART or no therapy. For this study disease progression was defined as developing a new AIDS defining illness (1993 CDC AIDS-defining illness, excluding CD4 count <200) and death. Viral load was quantified by HIV-RNA by RT-PCR.

Sterling found:

HAART (n=530) No-HAART (n=484)
Age (yrs)   38 38
Sex (male)   75% 66%*
Race (black)   69% 85%*
Men Sex w/Men   33% 16%*
IVDU only   14% 27%*
Hetero only   25% 24%*

In the no-HAART group there were significantly more IVDUs suggesting adherence may have been an issue.

HAART (n=530)
No-HAART (n=484)
 Baseline CD4
 Baseline Viral Load



Sterling showed Kaplan-Meier curves showing the risk for disease progression (new OI or death. For patients with <200 CD4s, there was no statistical difference in progression between individuals with viral load <20,000, >100,000 and 20,000-100,000.

Among patients with CD4s 201-350, there was no difference in disease progression whether viral load was <20,000, 20,000-100,000 or >100,000. (p=0.15).

Among patients with >350 CD4s at baseline again there was no difference in progression between the 3 viral load groups (p=0.78).

Using a multivariate proportional hazards model of and after adjusting for age, sex, race, IVDU, prior OI and CD4, viral load before starting HAART of above or below 100,000 copies/ml was not associated with disease progression.

Baseline Viral Load


Relative Hazard  95% CI  p-value 
<20,000    0.74 0.49 , 1.13 0.16
20,001-100,000    0.75 0.51, 1.10 0.14
>100,000   1.0    - -

Disease Progression (New OI or death) by Baseline CD4 Count Stratified by HAART Use

In the group with <200 CD4s, there was a statistically significane difference in disease progression between those receiving HAART & those not receiving HAART. Persons receiving HAART progressed more slowly (p=0.005). But in both groups disease progression did occur.

Among persons with baseline CD4s 201-350, there was no difference in progression whether or not a person received HAART (p=0.78). Disease progression was slower than in both groups in the <200 CD4 categories above.

Among persons with CD4 >350, persons progressed equally whether or not they received HAART, so treating with HAART (as in the 201-350) group did slow progression (p=0.65). Disease progression appeared more slow in these individuals than in those in the 201-350 groups.

In a cox multivariate proportional hazards model of disease progression among persons on HAART:

Baseline CD4


Relative Hazard  95% CI   p-value 
<200    4.23  2.38, 7.53  <0.001 
201-350    1.60 0.81, 3.15 0.17
>350    1.0    - -

The relative hazard for disease progression was slightly higher when starting therapy if CD4s are 201-350 compared to >350, but this is not statistically significant. These data suggest treating with HAART when CD4s are above 200 do not slow disease progression compared to treating with HAART when CD4s are >350. But that treating when CD4s are <200 this slows disease progression. But I think there may be some flaws in the study, when considering that adherence is not monitored, that this study compares HAART vs no ART or only dual nuke use (combined in group 2), and the short follow-up of only 2 years.

"Late Initiation of Antiretroviral Therapy (at CD4 count <200) is Associated with Increased Risk of Death" (abstract 520)
John Kaplan and colleagues from the CDC addressed this question by assessing the risk of HIV related death among 5,110 persons starting 2 or 3 drug antiretroviral therapy (ART) in 1994 or later in CDC's Adult and Adolescent Spectrum of Disease Project, a medical record review surveillance project conducted in 11 major cities in the US & Puerto Rico. Included are individuals >13 years of age in over 100 clinics and hospitals in the 11 cities. Over 51,000 persons have been observed from 1990 thru June 2000. The objective of this study was to determine the risk of death as a function of the CD4 count at which time ART was started. In other words, is there a CD4 threshold level for starting therapy at which patients don't do as well in terms of survival. This study has flaws, as well. The follow-up is short (2 years). Monitoring of lab tests were not done as often as would be optimal.

Patients had no history of taking ART, they started ART during a period in 1994 or later by taking 2 or 3 drug regimen, and they had a CD4 count within 12 months before starting ART.

Kaplan compared the risk of death as a function of CD4 level at which ART was started (lowest CD4 count within 12 months prior to starting HAART). Two methods were used: they estimated 2-year survival using a Kaplan-Meier procedure, and they estimated the hazard ratio for death in Cox proportional hazards model controlling for age, sex, race, HIV exposure mode, history of AIDS illness, and 2 vs 3 drug therapy. They used an intent-to-treat analysis. That is, once therapy was started they disregarded changes.

5,110 persons were included in the analysis for a total of 8428 person-years of follow-up (median 17 months follow-up).

There were 902 deaths included in the analysis.

Number of Deaths & Person-Years
 CD4 @ ART   P-Years (#deaths)   Est. 2yr%   Survival* HR 
    0-  49  2918 (603)      64.8%    5.5
   50-  99   873 (114)      78.1       3.6
100-149  734   (57)      86.1       2.7
150-199 584   (24)      89.9       2.3
200-249 636   (27)      95.7       1.9
250-299 541   (21)      93.7       1.9
300-349 538   (21)      92.8       1.8
350-399 489   (11)      96.3       1.1
400-449 372   (11)      Ne 1.5
450-499 259     (2)      Ne 0.4
500+       564   (11)      96.5       referent

*uses Kaplan-Meier analysis. Ne means not estimable because I think Kaplan said there were no deaths. He said you can notice approaching 100% survival as you get around the 500 CD4 count level. Kaplan said that 300 CD4s appears to be a level where survival slightly changes and much more so at 200 CD4s. Two-year survival was only significantly reduced when starting HAART at 150 CD4 or lower.

Kaplan said that the results of the Cox proportional hazards model appears to parallel that of the Kaplan-Meier. The 500 CD4 level was used as the referent. The hazard ratios for death for people starting therapy at 200-350 compared to starting at 500 are slightly elevated. For persons starting therapy at <200 the hazard ratios are elevated and statistically significant. When grouping the 200-350 levels the hazard ratio is 1.8 and nearly statistically significant. Initiating therapy with a 2 vs 3 drug therapy was associated with a hazard ratio of 1.5, and history of AIDS opportunistic infections is associated with hazard ratio of 2.6, and both are statistically significant. 56% (2854) of the 5110 patients in the analysis started 3 drug regimens, and 37% (330) of the 902 deaths occurred in this group.

Cox proportional hazards ratio model for patients who started with 3 drug regimen (HAART):

 CD4 @ HAART   Hazard Ratio 
    0-  49        8.3
   50-  99        5.1 
100-149        3.0
150-199        3.0
200-249        3.8
250-299        3.0
300-349        3.1
350-399        1.7
400-449        1.5
450-499        1.0
500+       referent

Statistical significance occurs only for starting therapy when CD4s are below 100. Perhaps as time passes more clinical endpoints will occur and a more informative analysis will result identifying more clearly CD4 cutoffs on when to begin therapy.

Kaplan admitted there were many limitations of this analysis: patients were not randomized but this was an observational analysis; they don't know for sure whether patients were ART naïve before starting therapy; adherence was not followed; other outcomes like toxicity, development of resistance and quality of life were not followed; and including patients on 2 drug regimens (to increase sample size) dilutes the analysis but the 3-drug analysis helps.

Kaplan concluded by saying that initiation of therapy at <200 has an increased risk of death and therapy should not be deferred to this point. The data from the Kaplan-Meier curve suggests that starting therapy at 200-350 may also entail risk suggesting that starting therapy >350 may be beneficial. But he was unwilling to stretch that far until there is more data due to the limitations of the study and other considerations such as adherence & adverse events. So, if this study continues to collect, analyze & report data over longer follow-up periods on persons in the 3-drug regimen group, the study may yield important data on when to begin therapy.

 <  www.natap.org


All other articles >