Hello everyone! I’ve been wanting to talk about n of 1 clinical trials for a while, and here I am, finally writing about it. (As a caveat, I am not a person who actually knows much about n of 1 trials, or clinical trials in general; but I’ve heard about them enough that I started wondering — why aren’t these done for phages? Or are they?)
Are n of 1 trials real? Are they the same as compassionate use?
When I think back to my earliest phage conference memories, I distinctly remember the booming voice of FDA representatives saying ‘compassionate use is not a drug development pathway’ and ‘don’t use this as a way to get around doing clinical trials’.
I remember reading about the Helsinki Declaration around this time too (I even wrote a blog post about my understanding of that — do I dare dig it up? Ok here you go). My takeaway was that compassionate use is for patients, and clinical trials are for science. You don’t get to cherry pick your favourite aspects from each.
But then one day I heard about n of 1 trials. I remember thinking, oh, do people just mean compassionate use? And someone told me, no, it’s actually a separate thing. It’s an actual type of clinical trial, where a patient acts as their own control.
This was intriguing, given the phage field has been stressed about how hard it is to find enough patients for phage trials. I began to ponder, what are these n of 1 trials? Are people using them for phage therapy? If not, why not? (If there’s a way to do a ‘real’ clinical trial with one patient, wouldn’t we do it?)
So I finally delved into this (and yes, I actually DO use the word delve in my regular life). Here’s a glimpse into what I found and how I’m thinking about all this now.
Compassionate use and n of 1 trials are not the same thing… but I understand why I was confused
Compassionate use is when the patient gets an unapproved treatment because there’s nothing approved that can help them. It is not meant to be about data collection. (I’ve heard people say you’re not actually allowed to collect data systematically across compassionate cases without registering a clinical trial; though I don’t see how that can be true, given how many case report series outcomes get published these days).
N of 1 trials are when a trial is done to test drugs/interventions within one patient (often to see which will help the most). Like any clinical trial, the trial is pre-designed, stats are employed, etc.
Both n of 1 trials and compassionate cases are done for the patient (and not with a goal of getting generalizeable results about the intervention). Compassionate use is to help the patient, and n of 1 trials are to try to find the best treatment for the patient. But compassionate use gets driven by what the doctors think the patient needs; it isn’t generally registered as a trial with pre-specified dosing schedules. Whereas in an n of 1 trial you might see two drugs alternated, or one drug given for a while, then stopped, all at defined intervals.
N of 1 trials are a real thing (in other fields, mainly)
N of 1 trials are used in education, to test pain meds, and in palliative care, among others. (This article gives a great overview; thanks David Harper for sharing!) But the thing about them is, because a patient is acting as their own control, any drug or treatment that affects the underlying disease may confound the results too much to be worthwhile.
So a pain med would fit well, because once you take away the pain med, the person goes back to being in pain, and you thus sort of return to baseline before you switch to the next type of pain med. Contrast that to phages: if they killed the bacteria, and then say antibiotics were tested next, it would be hard to say that that was a fair trial of the antibiotic (since the bug might have been killed by the phage, leaving nothing for the antibiotic to do).
Once I learned this, I became less excited about n of 1 trials as a possibly good idea for phage treatments. That said, maybe for chronic infections and/or anywhere it can be demonstrated that the bacteria return to baseline when treatment has stopped, it could be viable?
At least one n of 1 trial of phages has been done (but I don’t have details!)
I did see recently in the news that Marisa Azad’s group in at The Ottawa Hospital in Ottawa, Canada did an n of 1 phage therapy trial. Dr. Azad, if you are reading this (or this makes its way to you), I would love to talk to you about this, how you set it up (were antibiotics involved too?), and why you did it — perhaps that can be the subject of a ‘part 2’ of this article!
For now all I know is that it went well (the patient seemed to have a good result), the Canadian government did approve it, and the phages were made by Cytophage.
(P.S. If anyone reading this has any leads on other n of 1 trials for phages — not compassionate use cases — please spam my inbox!)
Can n of 1 trials be combined?
One thing that’s been happening a lot lately in the phage space is performing a series of compassionate use cases and reporting on them in aggregate (for example see QAMH’s 100-patient series, TAILOR’s 12-patient series, Israeli Phage Therapy Center’s multi-patient series: this 16-patient series with the same phage, and this one about 5 that didn’t work).
Combining compassionate use cases is certainly useful, and I think it’s incredibly powerful that these reports are being compiled. But as with any experiment where everything is done differently, it’s hard to make conclusions.
For n of 1 trials, people do sometimes do meta-analyses, where multiple n of 1 trials are combined for the purpose of drawing general conclusions about efficacy of the intervention. To do this work, special stats magicians have developed purpose-built statistical tools that can actually make them interpretable. Things like accounting for crossover/spillover effects between one treatment and another, figuring out how much time needs to pass between treatments, etc. (Caveat that I am not a stats expert and won’t be touching that with a ten foot pole today — I only barely know enough to have laughed at this meme yesterday).
In any case, my takeaway is that it’s not out of the question that you could use a series of n of 1 trials to make conclusions that apply more broadly, beyond just the patients who were in the trial.
A thought on phages and antibiotics
If you read phage case reports/papers, you’ll see that antibiotics are often used with phages (they are often started and stopped during phage courses, etc). This makes sense if the physician is making the call in the best interest of that specific patient, rather than in the best interest of ‘science’.
This also makes me remember this discussion about BiomX’s recent CF trial (thanks Phage Club on Clubhouse!), where antibiotics were cycled along with phages, which made it hard to interpret the results.
Now that I’ve read more about n of 1 trials, it seems that cycling between different drugs can be a very useful trial design strategy, but it has to be pre-planned, and stats have to be done to determine the space between the cycles. And for phages specifically, it seems that we would need to make ourselves confident that the bug returns between treatments, to be fair to the next treatment.
Would it be worth designing an n of 1 trial for phages and antibiotics and doing these studies? Maybe if we chose the right clinical indication and bacterial target?
Bringing it back to the point: why do this?
Of course, now that I start thinking about trial design, I go back to ‘oh yeah, why not just design regular controlled trials instead of trying to do these single patient studies’. Well, one of the reasons phage therapy trials haven’t gone smoothly is because it’s so hard to find enough patients that need the same phage. So an n of 1 trial could alleviate pressure on finding a lot of patients.
From what I gather, this is not usually the driving reason other fields use n of 1 trials. Doctors appear to turn to these trials mainly when it’s unknown for a specific patient which drug is right for them (it’s apparently called ‘clinical equipoise’, which I found out has nothing to do with horses or ballet).
That said, I also read about how n of 1 trials are used in palliative care. Palliative care drug trials are hard to do, because the patients are hard to recruit (they are by definition near the end of their lives). This starts to sound a lot like compassionate phage therapy, since phages are given to patients that have no other options.
So perhaps there is a place for turning our compassionate case series(es) into n of 1 trials, and leveraging n of 1 stats magic to help us design these trials? Could this be a better mechanism for testing how best to do phage/antibiotic combinations?
On the other hand, why? Are we better off continuing with our compassionate use cases, compiling the data, not trying to generalize off of it, and just separately designing/running the traditional trials that involve lots of patients that all need the exact same phage prep?
Do n of 1 trials give us something we don’t have?
One reason to do compassionate use cases is to help patients. Another is to see your phage work in a human (and use that to get grant funding to keep studying it?). Another (related) reason is to use successful cases as a way to attract public attention to phage therapy (again to make it easier to raise grant/investment money), and to boost morale of phage scientists and students who get to see their work mean something. We can’t use our rapidly growing collective body of compassionate use cases in place of clinical trials, but they still have all these benefits.
So do n of 1 trials actually give us something we don’t already have?
Presumably, in the short-term, practical sense, some governments might more readily approve an n of 1 trial (I think this may have been true for the Canadian case mentioned above?) than a compassionate use case. And maybe it would be helpful to borrow the tools and language used to describe n of 1 trials to help us design clinical trials that fold phages and antibiotics in together in an intentional way.
At best, n of 1 may represent a nice way of teasing apart phage dosing and antibiotic synergy in humans, without requiring one-size-fits-many cocktails and large sample sizes. One could imagine designing trials that swap phage and antibiotic order, and testing combinations, etc, rather than (in addition to?) doing that work largely in animals and extrapolating to humans. That said, it may be harder to design these trials (especially to confirm that the patient returns to baseline between treatments — does that ever happen with infectious disease treatments? If it does, can we really prove it?) than to just do things the traditional way…
Thoughts?
I would love your thoughts. (And again I repeat my caveat, this is a brainstorm from someone thinking about this for the first time — I want to spark conversation with this piece, not give you the answers!)
Resources
- David Harper (Evolution Biotechnologies CEO / led the only successful phase 2 phage trial to date back in 2009 / now runs a phage consultancy) wrote about n of 1 trials on LinkedIn the other day — worth a read, and worth reaching out ([email protected]) if you want to know more!
- This paper on n of 1 trials
- This paper on combining n of 1 trials
- Another phage consultant for you to bug about this if you’re wanting more is Ben Burrowes - he apparently knows a bunch about n of 1 trials and has thought about their use with phages (Ben, if you’re reading this, and have anything to add, let me know!)
