Clinical Implications of Virological "Failure": Interview with Steven Deeks, M.D., San Francisco General Hospital
Recently physicians have noticed that some patients, being treated with a protease-inhibitor-containing regimen which failed to control the virus, still maintained a better CD4 (T-helper) count, and seemed to be doing better clinically, than would have been expected given their viral load. The most extensive report yet on this observation--from a cohort of 143 "virologic failure" patients followed for a median of 18 months at San Francisco General Hospital--was presented as a poster at the 5th Conference on Retroviruses and Opportunistic Infections in Chicago.1 AIDS Treatment News interviewed lead author Steven G. Deeks, M.D. on the findings of this study, and also on the current practices of physicians at San Francisco General Hospital in treating patients whose viral load could not be reduced to undetectable levels by antiretroviral therapy.Summary of the Poster Presented
at the Retroviruses Conference
The study by Deeks and others was a retrospective review of the medical records of a cohort of patients at San Francisco General who met a strict definition of "virologic failure." To be included in this cohort, patients had to (1) have had more than 24 weeks of continuous treatment with regimen containing a protease inhibitor (either indinavir, ritonavir, nelfinavir, or ritonavir plus saquinavir), (2) have had two consecutive viral load measurements of greater than 500 copies/ml (using the branched DNA test) at their last two clinic visits, and (3) have received ongoing care at San Francisco General Hospital. 143 patients met these criteria and were included in the review. Of the 108 who had a viral load test available before starting protease inhibitors, the median baseline viral load was 4.84 logs (about 70,000 copies), and the median baseline CD4 count was 82.
In the entire cohort of 143, followed for a median of 18 months, there were seven deaths: four were not AIDS related, two were from AIDS-related dementia which preceded the protease inhibitor treatment, and one was from invasive aspergillus. In the entire cohort, there were only eight new AIDS-defining events during the followup period: two cases of non-Hodgkin's lymphoma, two pneumocystis, one cryptococcal meningitis, one cutaneous Kaposi's sarcoma, one PML, and one invasive aspergillus. There were no new cases of CMV or MAC.
More analysis was possible on the 108 patients who had a baseline viral load. Before the analysis, the patients were divided into three mutually exclusive groups: (1) Non-responders, with no viral load reduction greater than one log below the baseline (meaning that the antiretroviral regimen never reduced the amount of virus in their blood to a tenth or less of its pre-treatment level); (2) Transient responders, who did have at least one viral load measurement one log below baseline, but whose viral load had returned to within one log of baseline at their last visit; and (3) Durable responders, who had a greater than one log reduction in viral load at their last visit (but who, like all the others, still met the strict "failure" definition of not being below 500 copies on either of their last two visits). By these measures of response, there were 22 non-responders, 45 transient responders, and 41 durable responders (total 108).
At 18 months, even the "non-responders" had a median CD4 count increase of approximately 50. The transient responders had a median CD4 increase of more than 100, and the durable responders had over 150. The median CD4 increase for all 108 of these "virologic failure" patients was approximately 100.
The following conclusions are quoted from the poster:
"Despite a median duration of 18 months since initiating a potent PI-containing [protease inhibitor containing] regimen, and greater than 12 months since the emergence of virologic failure, the median absolute CD4 T-cell level in this cohort remained approximately 100 cells/mm3 above the pre-PI baseline.
"A significant proportion of patients with an incomplete response to a PI-containing regimen had a durable 1 log or greater decrease in viral load. This durable suppression was associated with a sustained increase in the absolute C4 T-cell count.
"Patients who had a transient virologic response to a PI-containing regimen had also significant and sustained increase in the absolute CD4 T-cell count.
"New AIDS related events were uncommon in this cohort of 143 patients 'failing' a potent PI containing regimen (n=8). There were no new cases of CMV or MAC.
"These observations suggest that failure of a PI-containing regimen, defined virologically, may not be predictive of subsequent CD4 T-cell decline (over the following 12 months).
"These observations must be considered in light of the retrospective nature of this study, and the limited duration of follow-up. Long-term follow-up of this cohort is needed to further examine the clinical implications of PI failure."
We asked Dr. Deeks about this study, and about treatment of such patients at San Francisco General Hospital, which is widely recognized as a leader in AIDS care. The interview took place on February 12, 1998.
AIDS Treatment News: In this study of clinical consequences of virologic failure of protease inhibitor therapy, why were the "durable responders" counted as treatment failures?
Dr. Deeks: They did not meet the strict criterion that has become the gold standard of therapy, which is to achieve sustained undetectable viral load. They had a clear response; the viral load went down and stayed down, but at least in the last two visits it was detectable (greater than 500 copies/ml using standard bDNA assay).
ATN: In this study, were there predictors of who would do poorly--predictors that might be helpful for clinicians in managing patients?
Dr. Deeks: Our presentation was only meant to report, "This is what we saw." To take this observation and conclude how clinicians should use viral load and CD4 data would be premature.
Our data simply state that the patients we identified as having "failed" protease inhibitor treatment still clearly had a sustained CD4 increase. That is a major point of this presentation. We are not saying that patients will not progress eventually. But as far as we can tell, over the 12 to 18 months after we recorded failure, for the vast majority of patients, CD4 decrease or clinical progression did not occur.
We did have a small group who developed opportunistic infections: only eight in this cohort of 143. What was constant with most of them is that even though they may have had a viral load response to their therapy, they never had a CD4 count response. This was consistent with what Dr. Judy Currier reported in her analysis of ACTG 320;2 that analysis showed that for patients who received AZT and 3TC, or AZT, 3TC and indinavir, just a small increase in CD4 count above their baseline (after eight weeks of treatment) was highly associated with a good clinical outcome. Our data were consistent with their observations.
One possible take-home message is that if patients do get even a small T-cell response, that may translate into prolonged clinical benefit. Patients who do not see a T-cell response to protease inhibitor therapy might benefit from more aggressive prophylaxis of opportunistic infections.
Yet our results cannot prove that if you fail protease inhibitor therapies virologically then you should stay on that particular regimen--or even that you should necessarily stay on therapy. To answer those questions would require a prospective clinical trial comparing the different options that patients have: stopping therapy, switching therapy, or staying on therapy.
ATN: What are the risks of staying on therapy despite virologic failure?
Dr. Deeks: By allowing viral replication in the face of a selective pressure, high level drug resistance may develop. This process may continue to occur long after the viral load starts to increase. It may limit future options with so-called "salvage therapy" down the line.
A second risk is continued side effects, which for some patients may outweigh any durable benefit from these therapies. And the cost of these drugs will be a factor for some.
In our clinic the vast majority of patients, once they failed the protease inhibitor virologically, either stayed on that therapy or switched to a second protease inhibitor containing regimen. And virologically, those groups basically did about equally well. We have not had many patients achieve durable viral suppression after switching to a second protease inhibitor containing regimen. Whether or not these people would have done as well had they stopped all their drugs, we do not know.
Of note, clinical investigators at Johns Hopkins University have reported reasonable success in patients who switch early to a second protease inhibitor containing regimen.3 The fact that some of their patients switched early may be critical. Delaying the switch for months, which was common in most of the patients we studied, may allow for the development of high-level resistance.
ATN: Is there any observation from those who did choose to stop everything?
Dr. Deeks: We only had about half a dozen patients who chose to stop antiretrovirals because of virological treatment failure. Among clinicians at San Francisco General Hospital, there is a strong sense that stopping therapy is not a good thing to do. So we do not have data to make a comparison.
One possible criticism of our data is that only those patients who did well stayed on therapy. We do not believe this was a significant issue in our analysis, since very few patients stopped therapy--and all patients, including those who stopped, remained in the analysis.
So again, we can only offer the observation that virological failure is not necessarily equal to clinical failure, over 12 to 18 months. But people should not take this and say, "I failed my protease inhibitor, but I should stay on it." We really do not know that, and certainly there are risks in staying on the drugs.
ATN: Have you seen increased viral resistance in these patients?
Dr. Deeks: We have not looked for it yet, and do not have much resistance data on these patients now. We need to do this analysis.
ATN: What research ought to be done next?
Dr. Deeks: The virologists and immunologists need to take a close look at this phenomenon, and try to determine if there is any underlying reason to think that virologic failure and resistance to a protease inhibitor confers loss of virologic virulence. This is a major question clinicians have been asking. There are some small studies suggesting that it may be true. A paper published a few months ago suggested that resistance could result in a switch from a syncytium inducing phenotype (SI) to a non-syncytium inducing phenotype (NSI)4; this could explain why resistant virus would be less destructive to T-cells. Other observations along those lines would also suggest that resistance to a protease inhibitor comes at some significant cost to the virus. These are the kinds of virologic studies that need to occur. I am told that it is very difficult, however, to test a virus for virulence in the laboratory. We are interested in working on this problem here at the Gladstone Institute [a privately funded nonprofit research facility associated with San Francisco General Hospital].
The other issue is what exactly is happening in the immune system--which goes back to the fundamental question of how HIV causes immune destruction. Is it a direct effect of the virus on the CD4 T-cell, or is it something much more fundamental earlier, a view many investigators are coming to support?
ATN: In other words, not the virus killing the CD4 cells, but doing something that prevents the cells from being produced in the first place?
Dr. Deeks: Yes. And it is possible that the transient virologic response to a protease inhibitor could lead to a sustained immunologic benefit. If the problem caused by the virus occurs early in the development of the mature immune system, one could envision that even a brief period of time when the virus was not around could let the immune system regenerate itself. Therefore immune function could be more sustained in the face of the virus coming back up, and that could lead to a sustained clinical benefit. The implication, however, is that eventually progression will occur. But when that may happen is unknown.
If you look closely at the data from the nucleoside analog trials, the CD4 count response is more sustained than the virologic response. So what we have reported may not be associated with protease inhibitor failure specifically. It may be a phenomenon of antiretroviral drug failure.
ATN: What would we need to answer the question of whether people should stay on treatment, or go off? Would a clinical trial be necessary?
Dr. Deeks: I doubt that people would support such a trial now--taking patients who "have no clear options to switch to" and randomly assigning them to go off all antiretroviral treatment, or to recycle drugs that had not worked for them. My sense has been that many clinicians and patients would not participate in such a study. So we will always have inconclusive data.
What needs to happen for these patients is not to answer the question of stop, or change, or recycle--but to develop new and better drugs.
Caution on Starting NNRTI Drugs
After Protease Inhibitor Failure
Dr. Deeks: At San Francisco General Hospital, for our patients who have failed more than one protease inhibitor--and they are very common here--one approach has generally been to keep them on whatever drug combination is best tolerated, and avoid adding or changing to any of the newer drugs, particularly the non-nucleoside reverse transcriptase inhibitors. The goal here is to save these drugs for when we know how best to use them in "salvage" regimens. Until then, the goal of therapy for patients who have failed their first-line therapies is to move from seeking viral suppression, toward seeking sustained clinical benefit. It can be difficult to distinguish between these two. But a strong feeling of physicians here is not to play the non-nucleoside RT inhibitor card for these patients until you have to--that is, until patients are progressing clinically. Perhaps over the next year we will have clinical evidence on how best to approach patients no longer responding virologically to protease inhibitor therapy. Presumably, the non-nucleoside reverse transcriptase inhibitors will be a critical component of that strategy.
ATN: Do you mean efavirenz in particular?
Dr. Deeks: Or nevirapine or delavirdine. These are very potent drugs--but they rarely do much, unless they can be given with two other drugs which will be very effective. And for these patients who have gone through the first line of antiviral therapies, there probably are not one or two other drugs that are going to be effective enough to lead to complete suppression.
Viral Load: How Low Is Low Enough?
ATN: Do you have any thoughts on the issue of whether a viral load below 400 copies is good enough, or do you really want to be below 50, 40, or even less?
Dr. Deeks: This is one of the questions that clinicians are going to struggle with over the next year. The data that Dale Kempf from Abbott Laboratories presented about a year ago clearly shows that the lower you get, the longer you stay down;5 that makes sense. So the goal of therapy will be to get the viral load as low as possible; the way to measure that would be with one of the new ultrasensitive assays.
The problem is, what do you do when the viral load does not go so low? This is a fundamentally different question. If you get down to very low levels, less than 400 copies, but the virus is still detectable, and it stays at this low level, what do you do then? Do you add drugs? Do you change the entire regimen? Do you continue to follow patients and watch what happens? The problem is that these new tests provide you with powerful information, but we do not know how to use it. So I myself do not see a dramatic need right now for access to the ultrasensitive assays. They will cause more confusion when they become more widely available.
The data from Keith Henry, from his clinic at the University of Minnesota, would suggest that a large proportion of patients who are "undetectable" with the standard assays will be detectable with the ultrasensitive assays.6 So it will be a real concern.
Some other posters at the Retroviruses meeting are relevant here. One looked at five vs. four vs. three drugs(7); this was a relatively small study done in Europe. Researchers looked at viral decay rates over the first two weeks [how fast the virus decreases when combination antiretroviral treatment is first started], and observed that as you increase the number of drugs the patients begin with, the decay increased substantially. To me this suggested that the three-drug combinations now used, for example the classic AZT plus 3TC plus indinavir, might not be enough to fully suppress viral replication--since if you go to four or five drugs, you get more rapid decline. This would suggest that current three-drug regimens may be suboptimal for some patients, which may be why we see drug failures even in patients who are completely adherent to these regimens.
How important is it to get the viral load to extremely low levels? Obviously the gold standard test for answering this question would be lack of clinical progression. But in people who get to these very low levels, we see very little clinical progression; it would take years to distinguish between those under 400, for example, vs. those under 40. In practice we will never know the answer for sure.
So it comes back to the theory. Theory drives much of what we do, and probably it should. Here theory would suggest that the less viral replication you have, the less resistance you will develop, and the longer these drugs will work. So getting the viral load as low as possible should be a central goal of therapy.
References
(1) Deeks SG, Beatty G, Cohen PT, Grant R, and Volberding P. Viral load and CD4 + T Cell Changes in Patients Failing Potent Protease Inhibitor Therapy. 5th Conference on Retroviruses and Opportunistic Infections, Chicago, February 1-5, 1998 [abstract #419].
(2) Currier JS, Williams PL, Grimes JM, Squires KS, Fischl MA, and Hammer SM, for the ACTG 320 Study Team. Incidence rates and risk factors for opportunistic infections in a phase III trial comparing indinavir + ZVD + 3TC to ZVD + 3TC. 5th Conference on Retroviruses and Opportunistic Infections, Chicago, February 1-5, 1998 [abstract #257].
(3) Gallant JE, Hall C, Barnett S, Raines C. Ritonavir/Saquinavir (RTV/SQV) as Salvage Therapy after Failure of Initial Protease Inhibitor (PI) Regimen. 5th Conference on Retroviruses and Opportunistic Infections, Chicago, February 1-5, 1998 [abstract 427].
(4) Ercoli L and others. HIV phenotype switching during antiretroviral therapy: emergence of saquinavir-resistant strains with less cytopathogenicity. AIDS. August 1997; volume 11, number 10, pages 1211-1217.
(5) Kempf D, Molla A, Sun E, Danner S, Boucher C, and Leonard J. The duration of viral suppression is predicted by viral load during protease inhibitor therapy. 4th Conference on Retroviruses and Opportunistic Infections, Washington, January 22-26, 1997 [abstract #603].
(6) Henry K and others. Levels of plasma HIV RNA measured by the Roche Ultrasensitive assay in HIV infected individuals with RNA levels below the bDNA detection limit. 5th Conference on Retroviruses and Opportunistic Infections, Chicago, February 1-5, 1998 [abstract #529].
(7) De Wolf F, Lukashov VV, Danner SA, Goudsmit J, and Lange JMA. Clearance of HIV-1 following treatment with three, four, and five anti-HIV drugs. 5th Conference on Retroviruses and Opportunistic Infections, Chicago, February 1-5, 1998 [abstract #384].
