Northern Lights? Canada and the New Tricouncil Draft

Since it's issuance in 1998, Canada's Tricouncil Policy Statement (Canada's policy on the ethics of human research) has had an influence on the practice of research ethics that has outsized Canada's population.  The three research councils– CIHR, NSERC, and SSHRC– are presently revising the Tricouncil, and a few days ago, a revised draft was presented on the CIHR web site.  

There is much to commend the newest version.  There are also a number of disappointments. I won't dwell on these here, however. Instead, I will focus on Tricouncil's revised language on phase 1 and gene transfer research.

The revised Tricouncil contains a definition of phase 1 that, in my view, is somewhat outmoded and not much of an improvement on the old version. Both emphasize the role of phase 1 in toxicity and dose determination, but do not encompass the many other purposes to which phase 1 trials are put (e.g. for deciding whether to pursue phase 2, for gathering evidence of biological effects, etc.).  On the other hand, the new Tricouncil requires prospective registration of all trials– including phase 1. And it contains a lengthy discussion of "therapeutic misconception," which it defines as "the tendency of trial participants to believe that the primary intention of research tests and interventions is to provide a therapeutic benefit to the patient-participant."  The document urges research ethics boards and researchers to "emphasize  which specific elements  of a clinical study are required for research purposes, as well as the differences between research and the standard clinical care they might otherwise receive."  Bravo.

The new Tricouncil also, for the most part, replaces the old language of "gene therapy" with the more neutral "gene transfer."  In a section on "Gene Transfer," the new draft warns about therapeutic misconception. It shrinks from any ethical statement on germline modification by deferrinng to Canada's Assisted Human Reproduction Act.  The remainder of the text notes the irreversibility of genetic alterations (not quite accurate), the potentially latent nature of gene transfer risks (a point I agree with), and states that research and ethical debate is evolving rapidly (a point I mostly agree with). Somewhat disappointingly, there is no mention of the need for centralized, transparent, or specialized review of such protocols.

On balance, these two sets of modifications are pretty good, though rather than anticipate the issues that are likely to arise in the next 5-10 years while this draft is in force, I worry a little that instead the draft represents a policy that I and others wish we might have had in the previous 5-10 years. Onwards and upwards! (photo credit: Studiolit 2006)

Sourgene

Heaven help those perseverant souls who pursue translational research on neurodegnerative disorders.  New interventions in this area have just about the highest failure rate of any area of medical research. And last week, yet another promising strategy was shown ineffective in a phase II study.

The trial in question was testing Ceregene's gene transfer strategy against Parkinson's Disease (PD).  Ceregene was second out of the gate testing gene transfer against PD (the first was Neurologix).  In April 2008, I wrote about their phase 1 study results, which on the one hand seemed to suggest safety, while on the other hand did not show signs of a dose-effect nor a change in dopamine metabolism by imaging.  For me, the most troubling aspect of this protocol was its aggressiveness: researchers delivered vector along eight needle tracks to deep brain structures. Based on surgical complication rates in the published literature, the risk of causing permanent neurological deficits from cerebral hemorrhage was on the order of 7% in this study- and that's not including the additional risk associated with the vector itself.

On November 26, Ceregene announced results of its sham controlled phase II study of CERE-120.  First the good news: the press release described CERE-120 as "safe and well tolerated."  The bad news is that the investigators saw no difference in outcomes between the sham and active arms. The press release does not say whether any adverse events were reported; nor does it say anything about surgical complications. In the next few weeks, expect to see data on why, precisely, the strategy might have failed. (photo credit: bebob717 2006).

In Brugge / No Compassion (Part II)

Further to the therapeutic outlook on first-in-human studies at the Brugge meeting was Adrian Thrasher's thoughtful presentation on his own X-SCID study at Great Ormand Street Hospital. Thrasher's study was able to restore immune function in nearly all volunteers. Recently, however, his team reported a lymphoproliferative disorder like those seen in a very similar Paris study.

Thrasher stated clearly "The purpose [of X-SCID protocol] is therapeutic effect; it is not a safety study."  Fair enough: the study was in a pediatric population (standard research ethics requires clear therapeutic warrant for such risky studies), and Thrasher's protocol did not range doses the way typical first-in-human studies do. And I should add, there is some grounds for thinking of the study as having therapeutic warrant, not the least because it was supported several unsuccessful X-SCID human studies and a successful one in Paris). Still, putting the therapy before the learning- this made me somewhat uncomfortable.  Therapy might have been his (and his hospital's) intent, but to describe the study as ontologically "therapeutic" and not "research"?  Intent only gets us so far...

Thrasher revealed some unusual properties about the molecular events leading to this leukemia (see? told you it could be construed as a safety study).  And now, here's the compassion part. Thrasher was circumspect about this particular leukemia, because the patient who developed the leukemia had originally been ineligible for the protocol because he had a matched unrelated bone marrow donor. The regulatory agency made a "one-time" exception to waive the normal risk-benefit balance.

Of course, one should be very careful generalizing from this one case where "compassion" seems to have led authorities astray. And presumably, the boy's parents were thoroughly informed about the risks going in to the protocol. Still, the example is somehow instructive. (photo credit: missinguigga 2008).

No Compassion

"They say compassion is a virtue...." so sang David Byrne of the Talking Heads.  But what about "Compassionate Use?" 

This refers to the practice of giving terminally ill patients who are otherwise ineligible for early phase clinical trials access to investigational agents. At the Brugge ESGCT meeting, Finnish researcher Akseli Hemminki described providing 125 patients compassionate use access to a novel oncolytic gene transfer vector– Ad5/3-Cox2L-D24. Hemminki was not reassuring when asked by an audience member whether a concurrent clinical trial was actually testing the approach, and rumors swirled that he had charged patients for the agent.

Compassionate Use is highly controversial.  Patient advocates view it as a lifeline, and over the years, drug regulators like FDA and EMEA have eased restrictions on patient access to untested agents.  In late 2006, for example, the FDA proposed new rules that would make it easier to provide untested agents to groups of patients (rather than individuals); the rules would allow companies to recover manufacturing costs from patients seeking access. A related set of proposed new rules would allow companies to charge patients for entering early phase clinical trials. 

Compassionate Use raises troubling questions for ethics and policy. With respect to the former, is it really an act of compassion to offer terminal patients a completely untested composition of matter? If the answer is "yes," well, that makes any future clinical trial that will randomize some patients to standard of care (which, for terminal patients, is nothing) diabolical. That leads me to the policy concerns: if you can get access to a drug outside a study, why enter the trial at all? Compassionate use potentially complicates the collection of rigorous data about the safety and efficacy of new interventions, and thus has public health implications. As for the idea of actually charging patients for access– whether on or off a trial: that's plain wrong.

According to the American Cancer Society, no records are available on the number of agents provided through compassionate use, the volume of patients who receive drugs through compassionate use, or their outcomes. Regardless of where one stands, the practice would seem a policy black box.  (photo credit: fgm878, 2007).

In Brugge: The Cure

One of the most striking themes at the European Society of Gene and Cell Therapy was the extent to which continental European researchers conceptualize first-in-human gene transfer experiments as therapeutic interventions rather than research protocols.

Perhaps the most extreme and explicit expression of this was view was presented by Bonn internest Thomas Heinemann (he also studied philosophy and serves on several ethics committees in Germany). Heinemann advanced the notion of the "controlled individual therapeutic attempt," for which the primary objective is therapeutic gain; the scientific dimensions of such studies (e.g. collecting safety data) are necessarily secondary. As he put it, research is only justified "ex post facto."

I found this argument intriguing for several reasons. First, Heinemann justified this claim largely on grounds of autonomy and instrumentalization of desperately ill patients. In contrast, North American bioethicists typically use autonomy and instrumentalization to argue the opposite: that research is primarily intended to serve the ends of others, hence the paramount importance of obtaining consent from volunteers and their guardians, hence the need to be extremely cautious going into a desperately ill population, where autonomy might be compromised.

Second, I was impressed by the speaker's conviction that first-in-human trials have therapeutic warrant. After almost twenty years of painstaking and at times discouraging research, we seem to have learned two things:  first, that first-in-human trials rarely go as expected, and second, that such studies often yield important insights about new interventions. I might have expected a more cautious and seasoned view about the therapeutic merits of first-in-human attempts: does it really enhance the autonomy of volunteers to offer so little by means of therapy, but to foreseeably get so much in terms of social good? (photo credit: virtualais //www.77click.it, Brugge, 2008)

Just the FACS: Reprise on Insertional Mutagenesis

I've just returned from the annual European Society of Gene and Cell Therapy meeting in Belgium.  Lots of great material for upcoming posts. For now, I want to follow on the last posting on the leukemias in the X-SCID study.  A warning: those lacking a stomach for science geek-talk might want to skip this posting.

In the previous posting, I stated that a recent paper provided evidence that retroviral integration in the genome ("insertional mutagensis") had triggered leukemias in the X-SCID study rather than over-expression of the corrective gene ("transgene"), the gamma c-chain (hereafter, "gc").  This was on the basis of data in the graphic above, which used cell sorting to show that levels of gc on the surface of T-cells was within a normal range.  In Belgium, Adrian Thrasher presented similar data for the fifth leukemia.

When I first encountered this figure, it bothered me: why did the authors measure gc expression by cell surface markers (a technique called "FACS") rather than Western or Northern blotting, or quantitative PCR, or something along these lines?  It seemed a very indirect way of seeing whether gc expression levels are in fact normal. Here are two possibilities that this figure fails to rule out:  1- gc is expressed at very high levels, but not packaged and presented on the surface of T-cells, perhaps because of insufficiency of other receptor components; 2- some gc transgene is aberrantly spliced, such that surface levels are normal, but intracellular concentrations of the alternate splicing product are abnormal.

A few years back, one team of researchers presented data indicating that gc transgene overexpression contributes to T-cell transformation. Another team claimed it was unable to reproduce this. The jury seems to still be out on whether the gc product contributed to the X-SCID leukemias, and I'm not yet convinced that the latest round of data fully exonerates the gc chain. (Graphic: figure from Salima Hacein-Bey-Abina et al, J Clinical Investigation 2008; 108: 3132-42).

Burst Bubbles

Among the greatest traumas for gene transfer was the development of leukemias in several children participating in trials using retroviral vectors against X-linked Severe Combined Immune Deficiency (X-SCID-- also known as "bubble boy syndrome").  About 20 or so children have had their immune systems fully restored by this gene transfer strategy.  Tragically, however, five children in two X-SCID studies (one in Paris, the other, London) developed T-cell leukemias that were causally linked to the gene transfer approach.

In the September 2008 issue of Journal of Clinical Investigation, Salima Hacein-Bey Abina and 28 other authors characterize the molecular nature of four of the adverse events, and report the outcome.  Before this article, it was known that one of the children died. This article now reports that the other three children have "sustained remission" after chemotherapy. 

The authors report that the vector inserted itself at the same genetic locus (LMO2) in 3 of the 4 cases.  In one case, vector inserted at a different genetic locus (CCND2); in another, vector inserted itself at LMO2 as well as a second locus, BMI1.  This suggests that LMO2 disruption is not the only path to causing cancer for this vector.  One other finding stood out.  Since the first leukemia was detected, many have speculated that the cancer was partly caused by the gene (rather than just the vector). However, the authors present evidence that the gene was expressed at normal levels in the children who developed leukemia. This lends support to the theory that the leukemias were not caused by the gene, but rather by some combination of the vector, the cell types used, and perhaps some characteristic of the underlying disease. (photo credit: concretecandy, boy in the bubbles, 2006)