Little-Noticed Ban of Little-Used Drug Triggers Little-Known Investigation
Whither the Checks and Balances?
On April 9th, a New York City treating physician, Dr. Paul C. Bellman, submitted a treatise-length letter of protest to members of the Public Health Service (PHS) panel responsible for meting out the government’s Guidelines for the antiretroviral treatment of HIV-infected persons in the United States. What prompted Dr. Bellman’s action was his discovery of a decision on the panel’s part to ban the use of hydroxyurea (a.k.a. hydroxycarbamide) in the setting of HIV — a use for which Dr. Bellman had experienced considerable success in some of his trickier patients. One ridiculously poorly designed (and executed) ACTG study was cited as the sole supporting evidence for the decision (ACTG 5025), several authors of which serve on the Guidelines panel itself. The role of Bristol-Myers’ influence, particularly in the design and interpretation of the 5025 study, was also explored.
As these catalytic events often do, Dr. Bellman’s response to the hydroxyurea ban led him (and subsequently, one TAGline editor) to undertake a much broader look at the make-up and operations of the Guidelines panel itself — and then deeper still to the incestuous nature of the entire enterprise: from drug development to clinical trials to regulatory approval to the final prize of being designated preferred or unpreferred status. Regardless of one’s pre-existing or reflexive prejudices about this particular drug, might it be useful to examine the larger issues? TAGline thought it might.
As much of Paul’s letter that would fit is re-printed below. But first, a few disclosures: Dr. Bellman was this editor’s treating physician shortly after the Vancouver faux eradication hoo-hah — when it was discovered he was one of a very few AIDS docs in New York at the time to offer the ultra-sensitive PCR assay. Said newsletter editor occasionally seeks Bellman’s medical advice even today. They also published a letter together in a 1998 issue of Lancet describing patient’s use of hydroxyurea to wean him off HAART. An unedited version of Dr. Bellman’s letter, complete with references, a detailed analysis of the 5025 study as well as other relevant studies — and short asides on both immune activation and cellular resistance, is available for the asking.
April 9, 2004
To Whom It May Concern:
This letter is to bring to your attention my concern that the recent Guidelines issued by the Department of Health and Human Services (“Guidelines for the Use of Antiretroviral Agents in HIV-1 Infected Adults and Adolescents”) contain a serious error that, if followed, could potentially jeopardize the health of my patients as well as others and also potentially restrict physicians and patients from choosing appropriate and reasonable treatment options in a range of settings of HIV care. This guideline was brought to my attention by my patient who … mentioned [in his magazine] that he was being treated by me, his physician, with a combination of ddI and hydroxyurea and that the current Guidelines advise that hydroxyurea “should generally not be offered” to any HIV-infected a patient. In his letter, [this patient] states that I had suggested ddI and hydroxyurea as a possible alternative to a treatment interruption which he was keen on doing.
After reviewing the Guidelines and the supplement explaining why hydroxyurea should not be used, I feel that a serious error has been made in which the actual known and well-reported efficacy of this therapy has not been properly regarded. In addition, the toxicity of the therapy has been dramatically exaggerated. I believe that the efficacy and toxicity of hydroxyurea-based therapy was not properly regarded, and that the rest of the relevant medical literature was excluded from consideration.
Given the potential life-saving value of hydroxyurea for some very sick patients living with HIV infection, the Guidelines’ strong disapproval of this medication is of great concern and, I believe, raises important questions regarding potential biases regarding how these Guidelines have been developed. I feel these biases also need to be addressed and will discuss them in some detail in this letter.
The Guidelines, issued in July 2003, specifically state that hydroxyurea “should not be offered” by physicians to HIV patients “at any time.” The Guidelines state that they are evidence-based (DII) and weigh the strength of the recommendations based on evidence. I was surprised and disturbed that the committee felt they could distill all the data and research on hydroxyurea to this simple prohibition despite the fact that numerous clinical studies have been published showing the importance, efficacy and safety of this drug.
The consensus committee, perhaps anticipating the criticism their recommendation might cause, did outline the reasons for their decision more specifically in a supplement, dated March 23, 2004. In my opinion, a careful reading of this supplement shows the consensus recommendation that hydroxyurea “should generally not be offered at any time” is based on a highly selective, even restrictive, reading of the body of medical literature on hydroxyurea. This reading excludes the substantial body of medical evidence regarding its efficacy and actual toxicity, and cites information from only one clinical trial run by the ACTG trials program, ACTG 5025, published in AIDS which I will discuss in detail. However, this trial was not designed to specifically assess the antiviral efficacy or toxicity of hydroxyurea, and its use by the panel to draw a conclusion regarding hydroxyurea is poorly considered.
It is my observation that the NIH/PHS Guidelines panel is rooted in a medical paradigm that unfortunately has not been properly examined, questioned and criticized internally by the panel members or externally by other interested parties. This clinical paradigm of HIV care, as represented by the consensus Guidelines, includes multiple features. It includes not only the synthesis of scientific research, clinical studies, and clinical experience that it presents itself as, but also the individual and collective biases that inform these Guidelines.
These Guidelines have been generated in an economic, cultural and political context that more specifically includes factors such as how doctors are educated, the influence of drug companies in setting the agenda for most clinical research, the way that health care is delivered in clinical trials, the economic incentives of all the participants in the health care system, the way that only some doctors are selected to be on the consensus panel, the sources of income for those doctors, the payments made by drug companies to doctors, and so on. In the case of hydroxyurea, a generic drug, I believe that drug company interests to maximize profit embedded in a paradigm in which there are not appropriate checks and balances have led to its unwitting exclusion.
My proposal of this underlying paradigm is not intended and should not be construed as a personal or moral attack on the panel members but rather a factual examination that, if considered properly, can help the paradigm evolve to more effectively meet patient care needs.
The panel’s supplement begins by stating that hydroxyurea is indicated for the treatment of certain malignancies and sickle cell anemia. The supplement also states that its potential safety and efficacy for treatment of HIV have not been established, and that clinicians should be aware of certain safety precautions regarding its use. The supplement then proceeds to describe hydroxyurea’s potential mechanisms of action. Up to this point, the information presented is correct and reasonable. In the next paragraph, however, it is claimed that hydroxyurea is both highly toxic and ineffective as an HIV treatment, based on a study referred to in detail. I have several points of contention with this statement, which are elaborated below.
First, ACTG trial 5025, the only controlled clinical trial cited in the supplement as well as the only clinical trial done by the ACTG regarding hydroxyurea, is used to draw significant and, in my opinion, unwarranted conclusions regarding hydroxyurea’s safety and efficacy. ACTG trial 5025 was a multi-center, partially blinded, prospective randomized trial of 202 HIV-infected subjects conducted to determine whether therapy intensification improves long term rates of virologic suppression. Essentially, the protocol took patients who were already doing well on AZT/3TC/indinavir, as determined by the maintenance of undetectable viral loads, and randomized them in the following way: one third continued their previous treatment, one third changed to d4T/ddI/indinavir and one third changed to d4T/ddI/indinavir and hydroxyurea. The purpose of this study was to see if patients who changed treatment had better outcomes in terms of maintaining an undetectable viral load. This would show that ddI/d4T might be a more potent nucleoside backbone than AZT/3TC with a protease inhibitor.
The supplement reports that the trial shows “that there were no differences in viral load suppression between patients receiving hydroxyurea 600 mg twice a day in combination with ddI/d4T/indinavir compared to those receiving the combination regimen without hydroxyurea.” This falsely suggests that the trial showed that hydroxyurea lacks antiviral potency. As a clinician who is familiar with the various studies on combination therapy and is concerned with maximizing antiviral potency, achieving and maintaining viral suppression and minimizing the development of resistance and toxicity from long-term treatment, the results of this trial mean nothing in terms of measuring hydroxyurea’s antiviral effects for the following reasons.
AZT/3TC/indinavir and ddI/d4T/indinavir are highly potent regimens that will achieve and maintain undetectable viral levels in virtually every naive patient who takes them. The limiting factors of these combinations which prevent patients from achieving undetectable viral levels are not their potency, but rather their associated adherence and toxicity issues. Switching from AZT/3TC to d4T/ddI created increased adherence challenges requiring at least five separate doses on an empty stomach each day. Also, d4T/ddI is a markedly more toxic nucleoside combination than AZT/3TC. In fact, the trial results reported that “dose-limited toxicities rather than virologic rebound accounted for the differences in treatment failure among the study arms.” The study authors drew no conclusions regarding hydroxyurea’s potency, but the consensus panel did.
Amazingly, this study was not even designed to compare the effects of adding or removing hydroxyurea in terms of achieving undetectability, but rather the effects of hydroxyurea in maintaining undetectability. As a clinician, the limiting factors of maintaining the efficacy of a regimen once undetectability has been achieved are once more toxicity and adherence issues. Once patients achieve undetectability, studies show that almost every patient who continues to tolerate and adhere to the combination therapy maintains undetectability or maximal viral suppression. So how could the addition of hydroxyurea to highly potent combination therapy of adherent patients who tolerated the regimen and were already undetectable be expected to provide a measure of its antiviral potency by improving their efficacy or viral load suppression?
The supplement states that the results of the study show no additional antiviral effect from hydroxyurea and purports that this shows that hydroxyurea is not potent. In fact, no drug, no matter how potent, would show antiviral potency when added to regimens of patients who are already undetectable. No drug company would ever test a drug in this clinical setting to assess its potency! Even if the tens of thousands of patients could have been accrued in this trial that could have statistically powered any clinically meaningless difference between the hydroxyurea and non-hydroxyurea group in terms of maintaining undetectability, the stated objective of the trial, toxicity and adherence issues would drive the results over time. The differences between the groups in this study were, in fact, due entirely to adherence and toxicity issues. In my opinion, although this was a clinically controlled trial run by the ACTG, the trial itself is invalid if its goal or intent was to show or not to show an antiviral effect of hydroxyurea.
In fact, multiple studies have confirmed hydroxyurea’s antiviral potency in such clinical settings. For example, Montaner and colleagues showed that significant antiviral potency is achieved when hydroxyurea is added to ddI — and it is lost as soon as hydroxyurea is stopped. Four weeks after adding hydroxyurea to ddI monotherapy in ddI-resistant patients a 0.63 log reduction of viral load was seen in patients receiving a 1,000 mg dose of hydroxyurea. Biron and colleagues showed marked drops in viral load with hydroxyurea and ddI which were much greater than those from ddI alone. After twelve weeks of therapy the median decrease in HIV RNA was 1.7 log copies/mL with six of twelve patients completing the study with undetectable viral loads of less than 200 copies/mL.
The first placebo-controlled trials of hydroxyurea were conducted by Dr. Franco Lori and colleagues. Fifty-seven patients were randomized to receive either ddI alone or ddI plus hydroxyurea. After twenty-four weeks, the reduction of HIV RNA levels in the dual combination group was significantly greater than in the ddI monotherapy group, with the former exhibiting a 93% decrease (-1.32 log) from baseline level and the latter a 53% decrease (-0.78 log). Viral load rebound was observed with some patients in the ddI group, but none in the combination group. Interestingly, ddI-resistant mutants were observed to emerge under combination therapy which retained sensitivity to ddI in combination with hydroxyurea.
I believe I have established unequivocally that the exclusive use of the 5025 study by the panel to deny the efficacy of hydroxyurea is wrong. In addition, I will provide a few points which I believe also negate the panel’s claim regarding toxicity. In ACTG 5025, three patients were reported to have died in the study — all in the hydroxyurea-containing arm. This resulted in the data and safety monitoring board (DSMB) appropriately interrupting the study and preventing its completion. Two of the three patients were reported to have died of pancreatitis and one of liver failure. This led to the implication that adding hydroxyurea to ddI/d4T potentially unacceptably increased the risk of fatal events, and some people drew the conclusion that hydroxyurea was to blame for this problem. However, a careful review of the associated clinical literature and experience with hydroxyurea suggests that this is not the case.
Franco Lori, a pioneer in the use of hydroxyurea, published a study in AIDS in response to the issue of pancreatic risk. His study showed that the risk of pancreatitis is a function of high peak drug levels of ddI. The Havlir study used the old FDA-approved formulation of 200 mg buffered tablets twice a day and instead gave two tablets, or 400 mg, once a day which led to much higher peak serum levels of ddI — which correlate with risk of pancreatitis. The new ddI 400 mg EC formulation, like the old 200 mg twice a day tablet, reaches much lower and safer peak levels than giving two 200 mg tablets once a day.
In addition, he published a paper in AIDS entitled “The Direct Mitochondrial Toxicity of Antiretroviral Drugs” in which he assayed the mitochondrial toxicity of nucleoside analogues with or without hydroxyurea. Based on in vitro work, which correlates with known and reported clinical experience, he found that hydroxyurea with ddI increases pancreatic cell toxicity in the presence of high concentrations of ddI. Hydroxyurea itself does not cause pancreatic toxicity. He did not study the potential synergistic effect that co-administered ddI and d4T might have had on pancreatic toxicity.
Dr. Havlir acknowledges that the degree of pancreatitis seen in ACTG study 5025 is not consistent with the published literature on hydroxyurea and ddI. Lori reports that in four ddI/hydroxyurea trials consisting of about five hundred patients, mostly treatment naïve, no pancreatitis was seen. Lori explains that hydroxyurea can increase intracellular levels of ddI and that the higher doses of hydroxyurea may also have played a role in these patients’ deaths. In his multi-year studies of hydroxyurea, Lori mentions that he has not seen any deaths from pancreatitis and has seen no more pancreatitis than from ddI alone. In early studies of ddI, a dose-related correlation to the development of pancreatitis was established with doses of 500 mg or more being correlated with a much higher risk of pancreatitis. Could it be that BMS sought to deflect attention away from the ddI/d4T duo (both double “d” arms showed an increase in serious side effects) and finger hydroxyurea as the fall guy? At that time, ddI/d4T was still considered a potentially (lucrative) alternative to combo nuke market leader AZT/3TC. We now know, of course, not to use these two “d” drugs in combination.
In a correspondence to the journal AIDS, Barreiro and colleagues recently reported a “low risk of pancreatitis” in HIV-infected patients on hydroxyurea and ddI. Their report was based on a group of 114 patients who began ddI and hydroxyurea after maintaining an HIV RNA level of less than 50 copies/mL with triple combination therapy for one year. Data was collected after a full year of treatment — much longer than the Havlir study. The results placed the incidence of pancreatitis at 0.87% per person-years, which falls within the range expected for ddI monotherapy. Interestingly, triglyceridemia dropped from 40% on HAART to 18% on ddI and hydroxyurea, possibly lowering the risk of pancreatitis. The authors concluded that the association of ddI and hydroxyurea seems to be safe, as determined from individuals with good immunologic status.
In summary, the consensus panel misused the results of the ACTG 5025 study to falsely claim that hydroxyurea failed to show antiviral potency and exaggerate its toxicity. The toxicity concerns reflect the wrong use of the old formulation of ddI with a highly toxic partner of d4T and an unusually high dose of hydroxyurea. In addition, some patients were included and died who should not have been allowed in the trial because of pre-existing high triglyceride levels. Had the panel concluded that the combination in question, ddI/d4T/hydroxyurea together, should not be used, I would support their decision as long as options were left open for rare clinical exceptions.
It upsets me that there are so many useful controlled trials that could have been done on hydroxyurea through the ACTG to extend or challenge the results of smaller, positive clinical studies. As a clinician who is trying to offer my patients the best therapy, which requires balancing short- and long-term efficacy with toxicity concerns, I believe there are many clinical situations where hydroxyurea could be part of an effective regimen. I would welcome controlled clinical trials that actually address the clinical situation in which a knowledgeable and experienced clinician might choose hydroxyurea. Why the ACTG 5025 trial was conducted and then used so uncritically by the consensus panel — and even more importantly why larger controlled trials of hydroxyurea designed to test its efficacy and toxicity in more relevant clinical settings have not been done — needs to be examined.
In my opinion, the use of this trial by the consensus panel is driven by the current paradigm of HIV treatment. I believe that the paradigm contains a bias towards treatment with brand name drugs that are highly lucrative for Big Pharma. The Big Pharma-driven paradigm is against the use of off-label generic drugs such as hydroxyurea, is unwilling to pay for the research of such treatment, and certainly does not support studies investigating treatment interruption. The constantly promoted paradigm of HIV as a chronically manageable, but not curable, disease justifies a model of never-ending treatment, which of course can always be improved or tweaked, and consigns patients to a lifetime of antiviral therapy with brand-name drugs.
The concept of treatment interruption, especially in combination with immune-based therapies that would further reduce the need for treatment, is a threat to this paradigm. Importantly, the Guidelines advise against treatment interruption for all HIV clinical treatment settings — in opposition, I might add — to similar British (BHIVA) Guidelines and again, in my opinion, based on a restrictively biased reading of the medical literature.
I believe there is a critical lack of questioning of the Big Pharma approach by physicians, which therefore tacitly supports this bias and directly shapes the recommendations of the consensus panel as well as patient care. I believe that patients, physicians and all other interested parties need to know more about this bias. The individuals chosen by the Department of Health and Human Services are all highly qualified in HIV medicine. However, they also have significant direct and indirect ties to Big Pharma in terms of consulting arrangements, speaking fees, “med ed” and other public relations involvement, as well as participation in the Big Pharma-dominated ACTG system.
For example, Dr. Diane Havlir, the author of the ACTG 5025 study, is a paid consultant to Bristol-Myers Squibb, the maker of d4T and ddI. Martin Hirsch, a co-author of the study for ACTG 5025, is also on the consensus Guidelines panel. (Sometimes academic conflicts of interest can prove more compromising than financial ones.) Overall, I do not believe that the physicians on the consensus panel are sufficiently aware and critical of Big Pharma’s influence.
Perhaps as importantly, the role and selection of academic physicians who help design and conduct clinical trials through the ACTG system with drug company support and who also sit on FDA and consensus panels need to be examined. From my point of view, these physicians, while usually of the highest caliber, often do not have large HIV practices in which they have one-on-one doctor-patient relationships over many years. Academic physicians often have much of their time devoted to clinical research, teaching and administrative responsibilities and may be less involved in direct patient care. As a result, they do not directly take the pulse of the evolving nature of HIV treatment in terms of patients’ problems and treatments because the time frame of clinical trials is too short to capture this pulse.
The taking of this pulse is a vital part of the care of the HIV patient given the complexity and variability of HIV and HIV treatment in which evolving clinical knowledge needs to be strategically translated into individualized treatment recommendations. An academic physician may not experience the anguish of a patient and the concerns of his or her physician regarding such issues as progressing facial atrophy, abnormal bone density leading to hip fractures, hyperlipidemia, glucose intolerance, abnormal liver functions, heart attacks and diabetes, and consequently may not weigh toxicity concerns regarding therapy with potency or resistance concerns.
These sometimes life-threatening side effects can occur even though the viral load is undetectable and the T cells are great. In the third, fourth or fifth year of HIV treatment, the academic physician may not clearly hear the patients plea, “Do I have to keep taking these meds?” The academic physician may see things in black and white terms without seeing the necessary striving of the patient and doctor to find a middle ground between the efficacy and toxicity of medication, particularly in the absence of data from long-term clinical studies.
Kaletra (lopinavir/r), for example, is a great and perhaps even breakthrough drug with regard to potency. But should it be awarded the status as the treatment of choice — as it is in the Guidelines — given the toxicity concerns? It would be much more helpful if the Guidelines acknowledged the large gaps in our knowledge and then permitted physicians and patients to strive for that middle ground in treatment — rather than imposing rules and prohibitions that are not evidence-based and which, in addition, obscure the critical issues of balancing the benefits and risks of treatment in the context of our doctor and patient shared treatment goals of long-term survival, optimal quality of life, and the hope for a cure.
Dr. Paul Curtis Bellman