Note: Prion2014 has no formal policy on social media, so I have been limiting my blogging to published work that speakers have discussed, except where speakers have given me permission to mention unpublished portions of their presentations.
Making last-minute changes to our speech, Sonia and I missed all but the last minute or so of this talk. However it was interesting to see the effect on scientists with whom we spoke afterward. A couple of people told us that they had never seen or met anyone personally affected by prion disease before, and one researcher said that the patient narratives she hears at these conferences are what keep her going in “those dark science moments.” I look forward to seeing this talk when NeuroPrion releases the videos from the conference.
Will addressed the risk of prion transmission through blood transfusions. He reviewed the four confirmed cases of vCJD transmission through blood transfusion [three of which were reported in Hewitt 2006]. He then reviewed the finding that ~1 in 2000 individuals in the UK harbor subclinical vCJD infections of the spleen. Finally, a subclinical 129MV vCJD carrier has been reported for the first time, and it has been shown that the spleen does harbor infectivity and that this infectivity can cross genotypic lines [Bishop 2013]. Together, this transmissibility and the surprisingly high prevalence of subclinical infection argue that there is a risk of blood transfusion transmission from subclinical individuals. These cases will be hard to figure out because we can only confirm transfusion transmissions when both the donor and recipient eventually become ill. Will has scoured the blood transfusion records of vCJD cases to look for evidence that the infection arose from blood, and suspects there may be a handful of additional tranfusion-transmitted cases, but none could be definitively confirmed [Davison 2014 in press, Vox Sang]. The donors in these cases mostly lived > 20 years after donating blood, and many have died of other causes but none have died of CJD. Overall, he reminded the audience that most vCJD cases have no history of receiving blood transfusion, which rules out the possibility that there are a large number of additional tranfusion-transmitted cases. He also introduced some unpublished findings from Public Health England in which they failed to find any evidence of PrPSc in an at-risk tranfusion recipient. Citing Jean-Philippe Deslys’s poster and Prion2013 talk regarding atypical myelopathy in macaques receiving vCJD-infected blood, Will looked for any evidence of atypical neurological phenotypes in individuals who received blood from vCJD-infected individuals, and did not find any evidence [Gillies 2009].
Finally, Will moved on to the question of transmission of sporadic CJD via blood transfusion. The assumption and evidence to date have supported that sCJD has not transmitted via blood and there is no infectivity in sCJD blood. However, a few groups have given contrary reports [Puopolo 2013, Drouet Prion2014 poster]. Will believes the best way to determine if sCJD is transmitted via blood is to look at the donors of blood received by people who later died of cCJD, and ask whether the donors are enriched for deaths soon after donation or for sCJD deaths; or alternately, to look at recipients of blood donated by individuals who later died of sCJD, and ask whether the recipients are enriched for deaths soon thereafter or for sCJD deaths in particular. Evidence from the US indicates there is no such enrichment [Dorsey 2009]. Will’s unpublished results from a similar study in the UK are in agreement.
You can’t prove the negative. Will emphasized that he thinks the most likely explanation for the negative results in the US and UK surveillance studies is that sCJD is not transmitted via blood, but acknowledged that the statistical power of the studies is not sufficient to rule it out entirely, and so further work is needed.
In the Q&A, Deslys agreed with the general conclusions of the talk but argued that the future is uncertain since it has now been shown that prions are replicating in the periphery in subclinical humans. Will agreed there is ample justification for continuing the surveillance work. Another audience member raised the concern of transmissibility of other “prion-like” diseases. He noted that animal transfusion transmission experiments are negative and so we think that there is probably not transfusion transmission in humans, but that this isn’t proof that there is no transmission, so he asked whether there is an epidemiological study design that could answer whether these diseases are transmitted via blood. Will agreed that there probably are epidemiological efforts that could address this, and also pointed to a recent relevant paper by Inga Zerr [Schmidt 2014].
Geschwind spoke on his work on improving the diagnosis of CJD. He began by giving some descriptive statistics on his UCSF cohort:
- Total N = 2288 referred
- Peak age of onset in the 55-75 year old range
- Mean survival ~7 months
- Estimated incidence of 1-1.5 per million population per year
- Estimated lifetime risk of CJD ~1 in 10,000
Of all the patients referred, 10% are genetic prion disease patients, 3% are mutation-positive asymptomatic carriers, and 3% are at-risk individuals who turn out to be mutation-negative. Of the ~500 patients seen in person at UCSF, they are slightly enriched for genetic prion disease, probably due to referral bias. Overall they’ve seen 20 different mutations.
Geschwind gave an example of one patient’s disease course, a 61 year old male with a 3-month disease course. In the first month of symptoms, he had problems with confusion and reading. In month 2 he lost his speech and many of his daily routines. By the time he was seen at UCSF in month 3, he had problems with anger and frustration and needed 24 hour care. This individual had restricted diffusion in the cortex on dwMRI, and autopsy-confirmed MM1 sCJD.
He then gave an example of a more difficult case to diagnose. This individual had a 3.5-year disease course, and surprisingly good cognitive function – he kept driving up until a couple of weeks before his death. He was referred with a diagnosis of primary progressive aphasia, but had a distinctive sCJD signature on dwMRI. On autopsy it was proven to indeed be MV2 sCJD. This individual was case reported [Johnson 2012].
In the UCSF quinacrine clinical trial [Geschwind 2013], almost 20% of patients referred actually turned out not to have CJD. They then reviewed the most common misdiagnoses and time to diagnosis, which averaged 2/3 of the disease course [Paterson 2012].
Geschwind compared three sets of diagnostic criteria for CJD: the WHO criteria, the UCSF criteria, and the European criteria. He argued that our criteria need to be updated. Many times, the first reported symptom in CJD is not even in the list of symptoms considered to be typical of CJD [Rabinovici 2006]. In terms of biomarkers, he reported that EEG is only about 65% sensitive [Steinhoff 2004], and 14-3-3 is only ~50% sensitive and not very specific either [Geschwind 2003]. Geschwind now considers the 14-3-3 test to be a test for rapidly progressive dementia and not for prion disease in particular. CSF Tau has slightly better sensitivity and specificity. However, MRI is currently better than all the biomarkers. He hopes that RT-QuIC will prove to be yet better.
Still, even with MRI, diagnosis is challenging. He found that 64% of the time, radiologists reading an MRI missed the CJD diagnosis, and he emphasized the need for better training of radiologists on this issue.
In his current work, Geschwind is working to make the MRI diagnosis more quantitative and less subjective, with automated parcelling of MRI images to identify regions of reduced diffusion. Surprisingly, they find that restricted diffusion is less severe in the most advanced patients, as defined by the Barthel index of executive function impairment. Diffusion appears to be get progressively more restricted early in disease, and then to get less restricted later on. The nonlinearity means that time in disease course cannot be determined from diffusion data alone.
In the Q&A, Simon Mead argued that we should be cautious about changing diagnostic criteria, because consistency in criteria over time is useful for epidemiology. He argued that the biggest problem in CJD diagnosis is simply that neurologists are not motivated to get a diagnosis quickly because they think there is no benefit to having the diagnosis since the disease is untreatable. He said that we need to emphasize the value of diagnosis for end-of-life planning, and we also need to do more clinical trials. Geschwind responded that the most useful thing for epidemiology is autopsy data, and that the current diagnostic criteria are almost useless for clinical trials because patients do not meet the criteria until they are too advanced for any hypothetical treatment. Geschwind said he is encouraged by the demonstrated feasibility of recruiting patients for clinical trials, for instance in the recent doxycycline trial, but that we really need earlier diagnosis in order to get patients at a more potentially treatable disease stage.
Zerr said she agreed with Geschwind’s discussion of diagnostic criteria for prion disease in general, and that her talk would focus instead on the question of how to get a subtype-specific diagnosis so that, for instance, we can predict which patients will have a rapid vs. slow progression. By subtypes, she is referring to the types defined by [Parchi 1999]. Some of her work has focused on looking for differential diffusion restriction on dwMRI between different brain regions in MM1 vs. MV2 vs. VV2 sCJD [Meissner 2009].
She is interested in whether biological variables affect the sensitivity of biomarker tests. For instance she finds that the 14-3-3 test is less sensitive in 129MV individuals, and in individuals with longer disease duration (many of whom are MV).
Recently she has been performing proteomics on CSF samples to try to identify proteins that could be subtype-specific biomarkers [Gawinecka 2013].
Zerr announced that she is currently using RT-QuIC as a diagnostic tool and gave some preliminary descriptive statistics. She sees strongest seeding activity (earliest ThT fluorescence) in genetic CJD patients, somewhat less in FFI, and the least in sCJD. Seeding activity is also strongest in individuals with short disease duration.
In the Q&A, I asked whether Zerr is currently using RT-QuIC in a clinical setting, i.e. to obtain diagnoses to return to patients and their families, or if it is still considered to be at a research stage. She responded that only in the past few weeks she has begun using it as a confirmatory clinical diagnostic tool. In a subset of cases, RT-QuIC is performed and is considered in the overall diagnosis in consultation with neurologists. Pierluigi Gambetti, co-moderator of the session, added that since January 2014 the NPDPSC in Cleveland has been routinely returning to clinicians RT-QuIC results on the patient CSF samples they receive. Larry Schoenberger clarified that patients must still meet other diagnostic criteria in order for RT-QuIC to be performed.
After the talk, I spoke with Dr. Gambetti and he offered more detail on the U.S. rollout of RT-QuIC in his center in Cleveland. Since January 2014, RT-QuIC is being performed on all referred CSF samples which meet at least a very low (sensitive) detection threshold for either Tau or 14-3-3 biomarkers. These thresholds are estimated to be about 95% sensitive, so most true CJD samples are undergoing RT-QuIC. The RT-QuIC is performed on site in Cleveland. It is fairly labor intensive and expensive; a nominal portion of this cost is billed to patients’ insurance but most of the cost is still being covered by NPDPSC’s surveillance grants, hence the need for pre-screening based on Tau and 14-3-3 rather than testing all samples.
Mead began by reviewing previous clinical trials in prion disease [Otto 2004, Collinge 2009, Geschwind 2013, Haik & Marcon 2014]. Though no therapeutics have emerged from these trials, the PRION-1 trial in the UK was a valuable opportunity to learn about how useful our rating scales are. His analysis of the PRION-1 trial suggests that some functional metrics are more useful than others [Mead 2011]. Based on this he developed a new functional rating scale, the MRC prion disease rating scale [Thompson 2013].
When he then plotted time vs. rating on the new scale for his patients, it was immediately clear that the rate of decline was very different for some genetic prion diseases than for sporadic prion disease. He suspected that further stratification of the patient populations might reveal that the rate of decline is much more homogeneous within subtypes than it is overall, which would mean that stratification could increase statistical power in clinical trials. When he ran a linear mixed model to try to explain patient’s rating scale score as a function of time plus various covariates, he found that codon 129 genotype explained a large chunk of variance in the slope of decline, and no other variables were significant. PrPSc type (1 vs. 2) is a significant variable on its own, but it covaries heavily with 129 genotype; 129 genotype is the better predictor on its own, and when 129 genotype is included in the model, PrPSc type offers no additional explanatory power.Therefore he supports stratifying by codon 129 genotype in future clinical trials. He offered to make MRC Prion Unit’s natural history data available to any investigators who want to do an open label study comparing against historic data.
In Mead’s simulations, a one-armed randomized clinical trial of about 20-40 patients with stratification and using 50% reduction in slope of decline in rating scale at p < .01 as the endpoint would have about as much power as a 100-patient trial based on survival time alone.
In the final few minutes, Mead offered an update on his efforts to find genetic modifiers of prion disease. He reviewed the general evidence for the possible existence of genetic modifiers, PrP interacting partners and different neurotoxic pathways, and he argued that GWAS is a highly objective way to determine which other proteins are most important in prion disease. In his last GWAS, he had 1250 sCJD cases and 750 cases of other types of prion disease.
Mead now has an additional ~1400 sCJD samples. In the new analysis, he has a genomic inflation factor of 1.06 when controlling for the first four principal components of population variation. As expected, PRNP is the strongest linkage peak, even in an allelic model, which is just picking up the slight difference in 129 MM vs. VV susceptibility, since allelic models cannot pick up heterozygote advantage. After stratifying by codon 129 genotype, there is no signal at PRNP. Excitingly, Mead also sees 7 other linkage peaks which appear genome-wide significant at p < 5e-8. Odds ratios for all of the peaks are modest, as expected. Mead does not want to publish any of these peaks until he can do a replication study in a separate cohort, so he is seeking investigators willing to share sCJD patient DNA samples.
In the Q&A, Neil Cashman asked whether the effect of codon 129 on the slope of MRC rating scale is as strong for genetic prion disease as for sporadic. Mead replied that he didn’t have statistical power to answer that, though he offered some hunches. Cashman further praised the GWAS worked and asked, in a hypothetical future, once we’ve found all the genetic influences on CJD susceptibility, whether there will still be a purely stochastic component to CJD risk that cannot be explained by any genetic or environmental variable. Mead shared his opinion that yes, CJD has a large stochastic component that we will never explain.
Jean-Philippe Deslys asked what Mead thinks of the mixed type 1 / type 2 sCJD patients and how they fit into his conception of subtypes. Mead responded that he doesn’t think we have enough data to explain what is happening in these patients, but he believes that codon 129 genotype will still provide the most power for stratification in clinical trials.
Caughey spoke about two projects he is currently working on in different domains: the diagnosis of CJD using RT-QuIC, and modeling the structure of PrPSc.
He began by reviewing the many proposed models for PrPSc structure, most of which retain most of the native alpha helices of PrPC. This is at odds with the evidence that there is no alpha helical content, at least in amyloid forms of PrPSc. This is supported by [Safar 1993], deuterium exchange and updated FTIR data [Smirnovas 2011, Baron 2011]. Caughey acknowledged his own error many years ago in mis-interpreting FTIR data to support alpha helical content. He next reviewed data supporting the view that recombinant PrP fibrils derived from prion-seeded RT-QuIC adopt a parallel in-register intermolecular beta sheet (PIRIBS) conformation in the domain originally comprising alpha helices 2 and 3. “In register” means that the corresponding amino acids of each PrP molecule are in the same location. “Parallel” means these PrP molecules are lined up with one another. Each molecule is one latitudinal rung, perpendicular to the fibril axis. The PIRIBS structure work is not yet published.
For the final portion of Caughey’s talk, he gave a quick review of the basics of RT-QuIC as an introduction to Gianluigi Zanusso’s following talk. He noted the now large body of literature demonstrating that RT-QuIC can detect a wide range of prion strains in a wide range of tissue samples in a wide range of species. In general, the sensitivity is seen to be 80-90% and specificity is 99-100%.
In an effort to improve on sensitivity, he and Zanusso tried using nasal brushings rather than CSF from patients, and they found 97-100% sensitivity and 100% specificity. Though unpublished, the level of detail I’ve provided here is also in Zanusso’s Abstract O.31 and Caughey’s Poster P.196.
Because of Zanusso’s earlier work showing the presence of PrPSc in the olfactory epithelium [Zanusso 2003], he and Caughey set out to determine whether CJD could be diagnosed by performing RT-QuIC on nasal brushings. Zanusso and Caughey now have a paper in press at New England Journal of Medicine demonstrating that this technique has very high sensitivity and specificity for diagnosing CJD. Zanusso therefore provided detailed instructions on how to perform these nasal brushings for RT-QuIC. He narrated over a video of a dramatic re-enactment shot in HD.
He then summarized the findings of his new paper with Byron Caughey.
Looking forward, Zanusso wants to determine whether these techniques could be adapted for the diagnosis of Alzheimer’s or Parkinson’s.
In the Q&A, Jason Bartz praised the work and asked about the potential application to diagnosis of animals such as deer or elk: he asked whether the damage from brushing to the olfactory epithelium could actually heighten the susceptibility of the animals to exogenous prion infection.
Sonia and I were busy getting ready for our own presentation and missed the talks by Fabio Moda and Luisa Gregori.
Sonia and I took the stage to share our personal journey as patients-turned-scientists, a bit of our work in human genetics, and our perspective on the quest for therapeutics. Here is our talk:
Stohr spoke about Aβ strains. He created synthetic Aβ prion strains from Aβ40 and Aβ42 and showed that they maintained distinct strain phenotypes when serially propagated in mice. To demonstrate that this was truly a difference in “strain” i.e. conformation, and not just in the peptide, Stohr showed that when the Aβ prions were denatured in SDS prior to inoculation, they converged to a single strain. This work is in press at PNAS and will be out soon.
Amin spoke about stimulation of hippocampal neurons with PrPC. As background she cited work showing that recombinant PrP can stimulate neuronal differentiation and synapse growth [Kanaani 2005]. She used a laser tweezer to bring a vesicle full of PrP right next to a hippocampal neuron and then burst it, releasing the PrP. She showed some amazing electron microscopy videos showing rapid changes in the growth cone in the 20 minutes after exposure to PrP. This effect was not seen in PrP knockout neurons, so PrPC itself must be the receptor mediating this effect.
Andreoletti asked for the audience to bear with him, as he knew his topic was an unpopular one: the possibility of scrapie causing sCJD in humans. He reviewed the diversity of scrapie strains, citing [Birkett 2001] and another paper I didn’t catch. He introduced transgenic mouse lines expressing HuPrP 129M or 129V as tools for his work, and then a huge amount of data on transmission experiments – there is a very brief overview in Invited Abstract 24 to the effect that the scrapie strains propagate in these mice and become indistinguishable from sCJD. This is all unpublished.
In conclusion, Andreoletti said these results do not convince him that scrapie is actually a cause of sCJD in humans. He said his interpretation is he would not eat the brain of a sheep that had died with scrapie, but beyond that it is hard to speculate too much.
In the Q&A, Jean Manson apologized for the predictable question, and then proposed that the study had not adequately ruled out the role of PrP overexpression in strain convergence and she would be happy to provide HuPrP knock-in mice to help solve this. Andreoletti replied that he had indeed expected the question.
Gambetti spoke on the transmissibility of VPSPr and other atypical human prion diseases. Introducing VPSPr generally, he noted that VPSPr is apparently sporadic and that its genotypic pattern is almost opposite of sCJD: the majority of cases are 129VV, many are MV, and only a handful are MM. On Western blots, the PrPSc in VPSPr is exclusively monoglycosylated. He presented the results of attempts to transmit 12 human VPSPr cases to a total of over 100 transgenic HuPrP mice. All of this data is in Invited Abstract #25. They obtained zero transmission on first passage, though a large fraction of mice did have evidence of PrPSc or histopathological changes in their brains after > 600 days of observation. Based on this lead, they attempted second passage, but on second passage, there was no evidence of any transmission at all – no pathology or abnormal PrP even after 800 days. As an explanation, Gambetti hypothesized that VPSPr is indeed transmissible and is an authentic prion disease but that the transmissibility cannot be sustained over serial passages due to some limitation of, perhaps, HuPrP as a conversion substrate, or of the mouse brain. He showed preliminary results from further work done in a collaboration to figure out what exactly this limitation is.
Based on these preliminary results, they created a new mouse model that expresses human PrP but with mutations to remove glycosylation sites, dubbed Tg(HuPrP129MGlycKO). In this mouse model, they saw highly efficient transmission with severe pathology by 380 dpi.
Gambetti concluded by questioning altogether the distinction between transmissible and non-transmissible human prion diseases, speculating that all are transmissible if given the right substrate.
In the Q&A, Jean Manson noted that prion researchers are constantly being asked whether prion diseases represent a threat to public health, and asked for Gambetti’s opinion on whether transmissibility of, say, VPSPSr represents a threat. Gambetti opined that experimental transmissibility absolutely does not signify that the disease is contagious and exposes a risk to public health.