Read with caution!

This post was written during early stages of trying to understand a complex scientific problem, and we didn't get everything right. The original author no longer endorses the content of this post. It is being left online for historical reasons, but read at your own risk.

The goals of this post are to review the evidence for tetracyclines as prion therapeutics, document the nature and results of studies that led to doxycycline making it to the human clinical trial stage, and ultimately come to understand how promising these drugs are.  Important note: this blog does not provide opinions on the efficacy of any medical treatment and nothing said here should be construed as medical advice.

Gianluigi Forloni of the Mario Negri Institute in Milan gave a 20-minute talk on this subject at the 2011 CJD Family Foundation Conference, as we’ve mentioned previously.  Watching his talk, I went through and dug up several of the original papers he referred to, as a starting point for my review of the literature.

Interest in tetracyclines for prion diseases arose in 1995 when Gianni and Merlini each discovered that 4′-iodo-4′-deoxydoxorubicin (aka IDX or IDOX), an anthracycline then under study for anti-cancer properties, inhibited amyloid fibril formation.  By 1997, a team (including Forloni) had shown that PrPSc-infected brain homogenate was less infectious (i.e. prolonged incubation time) in Syrian hamsters if it was coincubated with IDX before injection into the hamsters [Tagliavini 1997].  As far as I can tell, IDX eventually underwent clinical trials for primary systemic amyloidosis but never made it to FDA approval.  Its parent drug, regular-old doxorubicin, is a widely-used cancer drug (read: “chemo”) and the side effects are awful (again: think “chemo”). Forloni alludes to this in the video, and states that because patients cannot tolerate doxorubicin for long, the above discoveries touched off a search for chemically similar compounds that might be more benign while still having antiprion properties.  This search led quickly to tetracyclines.  A note here: tetracyclines are a class of compounds (including tetracycline itself as well as doxycycline and many others) often used as antibiotics, including in livestock.

By 2002 Forloni had managed to repeat his 1997 Syrian hamster experiment but with tetracycline instead of IDX, with the same result: “a significant delay in the onset of clinical signs of disease and prolonged survival time” [Forloni 2002].  In a separate study, he and colleagues also used several in vitro models (synthetic PrP-like peptides; actual PrP from brains of human CJD patients; rat neuron and astrocyte cultures in the presence of PrP 106-126) to demonstrate several anti-prion properties of tetracycline: it inhibits amyloid fibril formation, reverses PrPSc‘s proteinase K resistance, prevents neuronal death and astrocyte proliferation, and can be shown by NMR spectroscopy to bind to hydrophobic residues of PrP between residue 106 and 126, a region “which is thought to undergo a profound conformational change in PrPC to PrPSc conversion” [Tagliavini 2000].  All of these findings were taken as signs of the drug’s potential therapeutic value.  Even though amyloid formation and proteinase K resistance now appear not to be necessary conditions for prion pathogenesis, these in vitro findings do tell a molecular story: in the presence of tetracycline, PrPSc acts more like regular PrPC.  Barret 2003 validated Tagliavini’s findings by confirming that tetracycline reduces PrPSc‘s proteinase K resistance and binds to PrP 106-126; moreover, he also found that tetracycline inhibits the amplification of PrPres (i.e. proteinase K resistant PrP) in vitro.  (In contrast, quinacrine, which was the focus of Barret’s study, had less or no effect in each of these experiments.)  Though Forloni doesn’t mention them, there have been a number of other in vivo and in vitro studies of tetracyclines.

A few pieces are missing for me: Forloni touches on the stronger antiprion and weaker antibiotic properties of other tetracyclines such as 4-epitetracycline and rolitetracycline, but says no more on the subject, and a Google search doesn’t reveal any articles about these in relation to PrP.  At 6:54 in the video he also shows preliminary results from a patient preference human trial of doxycycline for CJD (10 years old at the time, so it must have started in 2001), with 21 treated and 77 untreated patients.  I understand that these were symptomatic CJD patients and so what he’s discussing is survivability after onset.  He actually shows a chart with an earlier age of onset for the treatment group, but I infer that these are just descriptive statistics on who the (already symptomatic) patients were going into the study, because his message is clear: doxycycline improved survivability compared to the untreated group.  (I say “untreated” rather than “control” because these don’t quite count as controls since they didn’t get a placebo.)  It looks like the results are as yet unpublished, but they appear to be referenced by a collaborator of Forloni’s, Luigi 2008:

In the last five years, a small group of CJD patients has received compassionate treatment with daily doses of 100 mg/kg [sic] doxycycline, and retrospective analysis showed significantly longer survival than untreated patients (Tagliavini F, et al., manuscript in preparation). Once again whether this effect is related to anti-prion activity or to protection of patients from bacterial infection is yet to be established. A major limitation of this study is that the results are not the outcome of a formal clinical trial but are based on open observations. The data reported in this paper provided the experimental basis for an ongoing Italian phase II, multicenter, randomized, double-blind, placebo-controlled efficacy study of doxycycline in CJD patients funded by the Italian Drug Agency.

Update 2012-09-16: the 100mg/kg figure is almost certainly an error on Luigi’s part and probably is intended to state “100mg/day”.  See comment below.

Forloni also mentions the double-blind study that Luigi introduces above, which began in 2006.  The results also appear to be unpublished– the only reference I found online was under the heading of “Unpublished Studies” in a review of potential prion treatments [Stewart 2008].  Forloni also cites a German study of 51 patients whose results I did manage to find online in a review.  That study found “a delay in disease progression and a significant (p=0.005) prolongation in survival time from onset in patients treated with doxycycline  (median: 292d, range: 162d-635d, treated group) compared to historical data in untreated patients (median: 167d, range: 33d-1448d)” [Zerr 2009].  Another French study of doxycycline also began in 2009, again with no relevant hits on Google as of today.

As far as I can tell all of those human clinical results are on already-symptomatic patients.  Forloni introduces a new project, a trial of doxycycline on 11 asymptomatic FFI carriers in one pedigree, mentioned previously on this blog, set to run through 2022.

There have been a whole host of other studies on tetracyclines which Forloni doesn’t have time to mention in the talk. Gu & Singh 2004 found an apparently complete reversal of pathology in a cell culture model of CJD using doxycycline; Luigi 2008 showed that scrapie onset in Syrian hamsters can be delayed by injecting the hamsters themselves (as opposed to coinoculating the infectious brain homogenate) with doxycycline, tetracycline or minocycline.  Forloni mentions still other studies in a 2009 review.  Curiously, though, it appears that doxycycline has never been tested on an animal knock-in model of a genetic prion disease.  There have been purified peptide models, cell culture models and human trials on symptomatic patients, but all of the animal studies I’ve found were based on infection with scrapie rather than a prion disease genotype.

Tetracyclines don’t have much in the way of side effects and they can cross the blood-brain barrier, though apparently only if administered in the right way.  In a review of tetracyclines as prion drugs, Forloni wrote: ”Tetracyclines have limited toxicity and some derivatives, such as doxycycline and minocycline, cross the BBB efficiently when an appropriate route is used” (my italics; his citation is Goodman and Gilman’s The Pharmacological Basis of Therapeutics), and in reference to the drug’s apparently disappointing performance against Huntington’s disease in vivo, he wrote “The lack of in vivo effect of doxycycline has different explanations including the unfavorable pharmacokinetics when the drug is administered in the drinking water”  [Forloni 2009].

One critique I have heard of doxycycline as a prion therapeutic is touched on by Luigi 2008 and elaborated further by the UK’s MRC Prion Unit:

Early results of these trials show that there may be a modest survival benefit in treated patients, but this is far from conclusive, and any benefit may be related to the antibiotic preventing terminal chest infections.

But while this fact could plausibly explain away the extension of lifespan in symptomatic CJD patients in the trials mentioned above, it doesn’t explain away the results from animal studies.  In the initial Syrian hamster study [Forloni 2002], the hamsters were never treated with doxycycline or tetracycline, rather, the drugs were merely coincubated with the PrPSc brain homogenate before the animals were infected, with the total amount of drug used for each animal (doxycycline: 444g/mol*1mM*30μL = 13μg/hamster) far from being clinically significant.  The hamsters (in both control and treatment groups) lived for hundreds of days after inoculation, long after any antibiotic effect would have passed, and the authors reported a delay in disease onset commensurate to the extension of lifespan (though unfortunately they seem not to have published the data for that claim).  Moreover, MRI and PrP immunoblot data showed differences in thalamic activity and PrP accumulation that should not have been present if the effect was due to antibiotic activity.  And if you’re a Bayesian, the in vitro studies demonstrating antiprion properties of tetracyclines give us a stronger prior belief that doxycycline’s observed effects might be due to antiprion activity in vivo and not just an artifact of their antibiotic properties.

A comparison to quinacrine is also merited.  In 2005 Quinacrine became the first would-be prion therapeutic to make it to a clinical trial in the U.S., but ultimately it was not found effective [Collinge 2009].  Perhaps in the future I’ll devote a whole post to quinacrine, but for now, from the little bit of searching I have done on the web, it seems that quinacrine went straight from in vitro studies to humans without any successful tests in animal models.  In fact, even before it entered clinical trials, it was already shown to have only limited antiprion properties (as compared to tetracyclins) at the biochemical level [Barret 2003].  Flupirtine and pentosan polysulfate have been similarly panned.  By contrast, in all the searching I did while writing this post, I didn’t find a single study that tested tetracyclines– whether in vitro, in animals or in humans– and failed to find an effect.  That doesn’t mean it works, but it is certainly not to be dismissed.

update 2013-06-04: see latest update from Prion2013.  Human trials look negative after all.