The New England Journal of Medicine in February published the results of a deeply flawed clinical trial in South Africa. The SAPIT (Starting Antiretroviral Therapy at Three Points in Tuberculosis Therapy) trial was designed to determine the most effective way to treat patients infected with HIV and tuberculosis. But it raises a number of disturbing questions about the oft-debated and vexing issue of appropriate standards of care in clinical trials undertaken in developing countries like South Africa. It also raises serious concerns about the quality of ethical review undertaken in those countries, and it highlights some surprising deficiencies in existing U.S. regulations regarding when ethical review should be undertaken by American IRBs.
In the SAPIT trial, 642 individuals co-infected with HIV and TB were randomized to receive one of three treatments: early integrated ART (started during the intensive phase of TB therapy), late integrated ART (started during the continuation phase of TB therapy), or sequential ART (started after completion of TB therapy).
Individuals randomized to the two integrated arms received ART within three months of study entry. Antiretroviral treatment of individuals randomized to the sequential treatment arm, however, was delayed on average for almost nine months. For several unlucky study participants, ART was delayed even longer; mean time to initiation of ART in this group was six to 11 months (260±71 days), regardless of CD4 cell count, an indicator of HIV disease progression, at study entry. This matters, as patients with CD4 cell counts lower than 200 cells per cubic millimeter face a significantly increased risk of death.
Forty-four months after the first SAPIT participants were enrolled, the study’s Data Safety Monitoring Board stopped the sequential treatment arm and required that all of its participants be started on ART immediately. In its interim analysis, the DSMB had found that participants in the sequential treatment arm had a much higher rate of death than participants in the integrated treatment arms.
The majority of these deaths occurred among patients with CD4 counts below 200 cells per cubic millimeter at study entry. Twenty-seven of 213 participants, or 12.7 percent, in the sequential therapy arm died during the abbreviated treatment and observation period, of whom 21 had CD4 counts of less than 200 per cubic millimeter at study entry, while only 25 of 429, or 5.8 percent, in the integrated therapy arms died over the course of the study. Integrated HIV and TB treatment thus reduced the likelihood of death among study participants by more than twofold.
The trial conclusively showed that integrated HIV and TB treatment is more effective than sequential treatment. The question is whether obtaining that data justified the 10 or more preventable deaths that occurred among trial participants. Even before the study began, years of observational data had shown that the risk of HIV-related death among patients not receiving ART therapy was highest during the first few weeks of TB treatment, calling into question the need to conduct a randomized, controlled trial in which ART was delayed by as long as 11 months for some study participants.
The side effects of concurrent HIV and TB treatment can often be managed by observant clinicians; given this, the WHO Global TB/HIV Working Group recommended as far back as 2003 that “any decision on treatment should be individualized, based on response to anti-TB therapy, side-effects, and readiness for ART.” More recently, based primarily on data from the SAPIT trial, WHO urged that ART be given within eight weeks of starting TB treatment in co-infected individuals.
Current treatment guidelines recommend that patients infected with both TB and HIV should start ART as soon as possible. Where CD4 testing is routinely available, these guidelines recommend starting ART when CD4 cell counts drop below 350 cells per cubic millimeter, and they require ART once cell counts drop below 200 cellsper cubic millimeter. Where routine CD4 testing is not available, these guidelines suggest that HIV-infected individuals with TB should start ART immediately.
In a recent article in The Lancet, several of the SAPIT investigators argued that “an estimated 10,000 deaths could be prevented every year by the initiation of ART in HIV/tuberculosis co-infected patients with CD4-cell counts below 500 cells per cubic millimeter.” However, the level of care provided to participants in the sequential treatment arm of the SAPIT trial fell considerably below this standard. The level of treatment provided to these volunteers even violated existing domestic standards of care for patients in South Africa, which recommend starting patients with CD4 cell counts of less than 200 per cubic millimeter on ART after just two months of TB treatment.
Although the SAPIT investigators noted that trial participants’ primary care physicians were free to start them on ART as necessary – in keeping with the 2003 recommendations of the WHO Global TB/HIV Working Group – this ignores the reality of public sector HIV care and treatment in South Africa. It is unlikely that most of the study participants even had a primary care doctor, let alone a physician who would initiate ART despite the restrictions imposed by the SAPIT protocol. Who then could have started these patients on life-preserving treatment if not the clinician-investigators on the study? Individualized treatment decisions based on medical need were not made in the sequential arm of the study, despite the ethical obligation of study clinicians to ensure the safety and well being of individual study participants.
Aggravating these problems, the trial design violated the golden rule of ethical study design and conduct: No clinical equipoise existed between the trial arms. Given decades of research showing that untreated patients with CD4 cell counts of less than 200 cells per cubic millimeter are at great risk of HIV-related complications and death, coupled with observational data showing that this risk is even higher during the first few weeks of TB treatment for co-infected individuals, the sequential treatment arm never should have been considered a viable treatment strategy. Patients randomized into the sequential arm were knowingly subjected to substandard clinical care.
The fundamental ethical (and legal) issue is this: the SAPIT study caused foreseeable harms and preventable deaths for a substantial number of impoverished and poorly educated South African trial participants. In doing so, the trial violated the Declaration of Helsinki’s requirements on standards of care of the then-applicable 2000 revision:
The benefits, risks, burdens and effectiveness of a new method should be tested against those of the best current prophylactic, diagnostic, and therapeutic methods. This does not exclude the use of placebo, or no treatment, in studies where no proven prophylactic, diagnostic or therapeutic method exists.For the South African researchers involved with the SAPIT trial, the Declaration of Helsinki is not only a moral exhortation but also a legally binding obligation. South Africa’s Good Clinical Practice Guidelines require that all investigators “follow fully the guidelines set out” in the Declaration of Helsinki, in addition to other national research ethics guidelines.
Organizations such as the Welcome Trust in the U.K. and the National Institutes of Health in the U.S. have directed millions of dollars to train members of African ethical review committees, and yet this trial was reviewed and approved by a committee in South Africa. The failure of this research ethics committee to recognize the clinical, ethical, and legal deficiencies in this study is shameful and, we hope, will be investigated by the relevant South African authorities.
Another issue is worth noting. As published in the New England Journal of Medicine, the study boasts a large number of authors. This is not an uncommon practice for large clinical trials like SAPIT. The authors of the article are located in South Africa and the U.S., and some authors list both South African and U.S. institutional affiliations. It is unfortunate, however, that U.S. research ethics guidelines (at least as interpreted by the American institutions involved) did not require these authors to submit the study protocol for ethical review by committees in the U.S.
If, as these institutions suggest, the study investigators were not acting on behalf of the U.S. institutions or were not engaged in what these universities define as human subjects research, it is worth asking why these institutions are listed on the published research papers. If they want to share in the academic glory associated with a study published in a prestigious journal like the New England Journal of Medicine, then they have the obligation to ensure that the trial meets accepted standards of ethical conduct and the responsibility to protect the rights and safety of all study participants. Further review might have made an appreciable difference to the ethical design and conduct of this study.
Sean Philpott is an assistant professor of bioethics at Union Graduate College and a former science and ethics officer for the Global Campaign for Microbicides. Udo Schüklenk holds the Ontario Research Chair in Bioethics at Queen's University, Canada, and is the joint editor-in-chief of BIOETHICS.