At the CJD Family Foundation Conference in Washington, DC earlier this month, Simon Mead of the MRC Prion Unit in London was present to introduce his group’s preparations for future clinical trials. He focused primarily on the development of a new functional rating scale for measuring treatment efficacy. He also spent some time discussing MRC Prion Unit’s vision for a possible clinical trial of PRN100, an antibody against PrP, as a treatment for sporadic Creutzfeldt-Jakob Disease.
MRC Prion Unit has been studying antibodies as potential prion therapeutics for over a decade now. They showed that intravenous injections with ~28.6 mg/kg/day either of two antibodies – ICSM-18 or ICSM-35 – could completely cure mice of peripheral prion infection as long as the treatment was started before the infection reached the brain [White 2003]. Unfortunately the antibodies had no measurable effect late in disease nor in mice infected intracerebrally. That’s at least partly because antibodies have a hard time crossing the blood-brain barrier.
It is thought that some tiny fraction of antibodies in the bloodstream do reach the brain [Bard 2000, Banks 2002] and so some researchers have administered enormous doses of antibody peripherally in an attempt to reach therapeutic concentrations in the brain. In the ongoing controversy over the role of PrP in Alzheimer’s disease, one study used large intravenous doses of a different anti-PrP antibody (6D11) to occlude PrP and reported a behavioral improvement in Alzheimer’s model mice [Chung 2010].
Two other studies have attempted to use large doses of anti-PrP antibodies to treat prion disease in mice, with mixed results. A dose of about 1.2 mg/kg/day infused directly into the brain for 2 weeks was reported to be effective even when begun fairly late (timepoint 0.8) in the disease course, although the effect size (~8% delay in terminal illness) was small and it is not clear whether the results are statistically significant after multiple testing correction [Song 2008 (ft)]. Intravenous doses of ~3.6 mg/kg/day, on the other hand, did not produce a significant therapeutic effect overall, although it was argued that the mice with larger amounts of antibody detectable in the brain appeared to live longer than those with less [Ohsawa 2013]. It has been speculated that the possible blood brain barrier disruption in prion disease [Chung 1995, Brandner 1998] could actually be an asset for peripheral treatments with antibodies.
Antibodies may occlude PrPC and block its binding to PrPSc, but it is thought that they work primarily by stabilizing the native conformation of PrPC, making it impossible for it to convert to a misfolded form [Antonyuk 2009].
One study raised concerns that the cross-linking of PrPC by antibodies could trigger neuronal apoptosis [Solforosi 2004]. That study found that two antibodies (D13 and P) against the central region of PrP (amino acids 95-105) caused extensive neuronal loss, while a third antibody (D18) against alpha helix 1 (amino acids 133-157) did not. MRC Prion Unit was unable to replicate this result using ISCM-18 / PRN100 (which bind helix 1) nor ISCM-35 (which binds the central region) [Klohn 2012].
For at least several months, perhaps over a year, MRC Prion Unit’s treatments page has had a notice that antibodies against PrP are “currently being manufactured for clinical trials in the UK by the MRC Prion Unit”. Dr. Mead’s presentation at the CJD Family Conference was the first public announcement we’ve seen, though we’re told a few details have been discussed at other conferences, and the proposed clinical trial recently made it into the news:
Dr. Mead stressed that the project is still pending “financial and regulatory hurdles” – read, neither approved nor funded as of today – and that details are not finalized. Still, the presentation offered a preliminary look at what shape a clinical trial might take.
The antibody is named PRN100 and represents the humanized version of the mouse antibody ICSM-18. It binds to alpha helix 1 of PrP, at approximately codons 146-153, and is thought to stablize PrPC in its native conformation. The first clinical trial would aim to assess the safety and efficacy of PRN100 administered intravenously to sCJD patients. Dr. Mead said that MRC Prion Unit had not ruled out the possibility of intrathecal or intraventricular infusion of the antibody down the road, but that intravenous administration would be the first target. He observed that we don’t yet know what fraction of PRN100 may prove to cross the blood-brain barrier, and that it’s not known to what degree the blood-brain barrier may already be disrupted in sCJD patients, which would make IV administration of antibodies more effective.
The first stage of the clinical trial would recruit multiple cohorts of 4 – 8 sCJD patients each for a dose escalation study to establish the antibody’s safety for human use. These patients would be eligible to re-enroll at each higher dose and to subsequently re-enroll in the trial’s second stage aimed at determining the antibody’s efficacy against prion disease.
An important detail which still has not been finalized is whether the efficacy trial would be open label, meaning patients could choose to enroll and receive the drug, or double-blind, meaning that patients choose to enroll for a 50/50 chance of receiving drug or placebo. Dr. Mead said that a hypothetical double-blind trial would seek to recruit a total of 60 patients in order to have 30 treated and 30 placebo-controlled. This number, he stated, would give 80% power to detect a 20% slowdown of disease progression at p = .05.
However, he also presented a great deal of analysis of the natural history of prion disease, analysis which could help to make an open label trial informative despite the lack of placebo controls.
Citing the lengthy survival but lack of quality of life improvement in a subset of U.K. patients treated with pentosan polysulfate, Mead invoked a need to develop measures of disease progression that are more relevant to patients and their families than plain survival. For this reason his group has developed a functional rating scale of neurological impairment [Thompson 2013] known as the MRC Scale.
The MRC Scale measures ability to carry out daily tasks of living, and can conducted by telephone interviews with patients’ caregivers. It attempts to group cognitive, psychiatric and motor impairments into a single score reflecting that different patients may decline in different ways but that comparisons are still needed. It’s most similar to the Barthel scale or the Clinical Dementia Rating Scale Sum of Boxes [O'Bryant 2008] in Alzheimer’s – and does not involve clinical tests, in contrast to the UHDRS used in Huntington’s Disease.
According to the MRC Scale, sCJD patients show a wide variability in the rate of decline. Surprisingly, neither age nor sex nor PrPSc biochemical type (Type 1 or Type 2) turned out to significantly stratify the rates of decline. Instead, codon 129 genotype turned out to be the only statistically significant modifier of the rate of decline which would need to be included as a covariate in order to assess therapeutic success in a clinical trial.
Whether MRC Prion Unit ultimately decides on a double-blind placebo-controlled study or an open label study, the result will probably prove at least somewhat controversial. There are good arguments on both sides.
A double-blind model is the only way to ensure that the groups of treated and control patients being compared are statistically the same groups. In an open label trial, the risk is that comparatively healthier patients, earlier in their disease course or with slower progression, will be especially likely to enroll, while patients nearer to death and already heavily disabled may not enroll. In such a scenario, one can see a significant difference in survival (or functional decline) between treated and control groups even if the drug does nothing. This is a risk in the open-label trials of doxycycline in Germany presented at Prion2013 and on which Inga Zerr presented updates at CJD2013 as well.
Of course, this potential for bias is widely recognized and there are ways to control for it. PRION-1, the clinical trial of quinacrine in prion disease, was open label partly by necessity: because quinacrine is an approved drug, patients could refuse to enroll in a placebo-controlled trial and get the drug from another doctor instead. That clinical trial utilized functional rating scales to assess how impaired patients were at the time of enrollment and how their disease progressed. The treated patients actually did live longer than untreated patients, but not after controlling for how advanced their disease was when they enrolled [Collinge 2009].
The MRC Scale, it is hoped, will do even better than just controlling for degree of impairment at the time of enrollment in the study. Dr. Mead implied that the slope of decline is actually independent of the intercept – in other words, measuring how quickly patients decline is on the MRC Scale is independent of how advanced their disease is when they enroll. That means patients can be compared directly, stratifying only by codon 129 genotype.
Detractors will argue that without a randomized placebo group, you can never guarantee that the treated patients are statistically the same set of patients as the untreated patients to which you compare them. Even if you control for impairment at time of enrollment, there still may be some subtle selection bias that is hard to control for.
It occurs to me that one should be able to model the threshold of effect size needed in order for an effect to be unattributable to selection bias. For instance, if 100 patients are offered the opportunity to enroll and 50 enroll, then the treated cohort represents at worst the top 50% of patients, so one could compare the top half of the historic distribution to the full historic distribution to see the maximum artifactual effect size obtainable by selection bias alone (I’ll explore the math here further in a future blog post). Meanwhile, an advantage of an open label trial is that more treated patients can be enrolled more quickly (since you don’t lose half of them to placebo), which increases power.
Of course, if it can be done without compromising the drug’s ability to become approved, the other advantage of an open label trial is that patients can be offered a real potential treatment and not just a 50% shot at one.