SAMSON Trial Makes Its Case to Cut Statin Side Effects

John M. Mandrola, MD; James P. Howard, MB, PhD


November 18, 2020

John M. Mandrola, MD: Hi, everyone. This is John Mandrola from | Medscape Cardiology. I'm very excited to be with Dr James Howard, who is a clinical instructor at Imperial College in London. James presented a very beautiful trial at the American Heart Association meeting called the SAMSON trial, which examined the nocebo effects of statins. It was a very cleverly designed N-of-1 trial. James is here to tell us about this trial and the results. James, welcome.

James P. Howard, MB, PhD: Thank you.


Mandrola: Can you tell us the topline results of SAMSON?

Howard: The take-home message from SAMSON is that patients really do suffer from statin side effects. But the thing is, they are not caused by the statin molecules; they are caused by the act of taking the pill. And the symptoms are almost as bad if they take a placebo.

Mandrola: That is remarkable. How did you design this and how did you work through this trial?

Howard: We took 60 patients who previously had taken statins and had to stop them because they had intolerable side effects. Our median patient had taken two different statins before — some more, some less. We brought them to Hammersmith Hospital in London and we gave them a case of 12 jars at the start of the year: Four of those jars contained atorvastatin 20 mg, four contained placebo, and four were empty. Every month they were told which jar they were going to start next in a completely random order. They had a smartphone — if they did not own a smartphone we gave them one — and they would say, on a scale of 0 to 100, how bad their symptoms were.

At the end of the year, we brought them back and sat down with them and showed them their results. We showed them whether their symptoms were worse taking statins or not and how their symptoms were with placebo. This allowed us to work out the average symptom score when people were taking nothing, when people were taking statins, and when people were taking placebos. Six months later, we went back to our patients and said, "After we went through the discussion with you about how bad your symptoms were, have you now been able to go back on to a statin?"

Mandrola: What did you find?

Howard: We found that when people took nothing, they actually had very minimal symptoms, but when people took statins, their symptom burden was much worse. But interestingly, when people took placebo, it was 90% as bad as when they took statins. During the trial, 71 times people had to stop taking tablets because their symptoms were so bad, but only 40 of those were in people taking statins and the other 30 were in people taking sugar tablets. We also found that by sitting down with our patients and talking about their symptoms and showing them their personalized results, 30 of our 60 patients were able to go back on to their statin.

Mandrola: That seemed remarkable to me; I would not have guessed that number would be so high. Were you surprised?

Howard: We had an inkling that something funny was going on because if you look at meta-analyses of statin trials where they did not do run-in periods, the number of patients that stopped their medicines in those trials is very similar between the placebo arms and the statin arms. That raised our suspicion that something was a little bit funny considering how many of our patients say they have to stop statins.

We had two possible explanations of why this might be the case. The first is this nocebo effect which we are talking about today: that statins are actually making people feel worse because they are expecting to (and sugar tablets are just as bad). Our other theory was that maybe the tablets were not doing anything but people are at an age where they are developing arthritis, they are developing aches and pains; and when they occurred, people just blamed them on the statins, which is a perfectly understandable thing to do. Because SAMSON had periods of statin, placebo, and no treatment, we could differentiate between those two phenomena and we showed that the nocebo effect really does exist.

Nocebo Effect

Mandrola: I think some clinicians are a little skeptical about the nocebo effect, I guess because it's in our heads. Why might someone have as many symptoms from a sugar pill as from the true chemical?

Howard: It's a very good question. It's quite difficult to say the "true chemical" because SAMSON shows that we don't really know what the true side effects from the statin molecule are. How something manages to have such a strong nocebo effect is very poorly understood. Some people hypothesized via various biochemical pathways. Some people have shown that if you give someone a sugar tablet and tell them it's going to give them pain, cholecystokinin levels rise. It's not really that well understood.

But one thing that is interesting is that there must be a relationship between how much people think they are going to feel worse and how bad they do feel. We know that if we tell the patient in clinic that we need to treat their blood pressure, it's actually quite rare that someone comes back saying they feel terrible after starting their ramipril. But a lot of people feel terrible after starting their atorvastatin, their rosuvastatin. And if it's not the molecule itself, and yet it's a very similar cohort of patients, it must be because of our perceived conceptions of what statins do to us. That is probably related to the media and word of mouth from friends and family.

Mandrola: I was going to ask you about the social learning aspect of the nocebo effect. It seems in some of my reading that that may be very important.

Howard: Exactly. And actually, our results support that, because we've shown that if you get people who previously have been unable to take statins — who have even tried multiple different ones — and you sit down with them, show them their results, and talk through it with them, they are able to go back on them. That just proves the fact that good communication and a good understanding are able to overcome this phenomenon.

Nocebo Ratio

Mandrola: As a clinician, I really love that. Can we spend a minute or two on the nuts and bolts of the data? I noticed in the letter that the first analysis showed kind of a crazy ratio of 2.2. It seemed like more people were symptomatic with a placebo, but then there was an adjustment. Give us the thumbnail on that.

Howard: To work out the nocebo ratio, you have work out the incremental symptoms above no treatment for both placebos and statins. Once you have that increment, you then divide the placebo proportion by the statin proportion that says how much of it is placebo. What we were not expecting, and it was our oversight, was that some people have such tiny excess symptoms when taking statins and quite a bit more excess symptoms while taking placebo, that you ended up dividing a very large number or a moderately large number by a tiny number. And that gave you astronomically high ratios.

The problem is that if you then go and process that using a parametric test, the mean is very skewed by outliers. (A median isn't, but a mean is.) This became a problem because our initial plan was to work out a ratio for each patient and then process those afterwards. What we had to do instead was bring all of the data for all of the patients together in a multilevel model, where we're controlling for days within months, within treatments, within patients, and then work out the ratio. That is all we did the second time around and that is why the number is much more sensible looking.

Symptoms Due to Nocebo

Mandrola: I noticed that the mean score was like 15 for nocebo [Editor's note: Dr Mandrola misspoke; the mean symptom score of 15.4 was on placebo.] and 16 for statins, and the P value was not significant. But yet, one of the messages is that 90% of the symptoms are due to the nocebo effect. How should I think about that?

Howard: We're not trying to say that statin molecule side effects do not exist. Every drug, whether it is acetaminophen or simvastatin, in some patients will cause side effects. So we should not be too surprised that it isn't 100. But obviously, there is a confidence interval around that 0.90 (we don't know it). So we can't say for sure whether it is exactly 90%, but it is reasonable to expect that some patients will have true statin side effects, and there may be some of those patients in SAMSON. We had to exclude patients who had previous myalgia symptoms with creatine kinase levels greater than five times the upper limit of normal, because some patients do clinically develop a biochemical myositis. Of course, in the real world, it won't be 100%, but what SAMSON wanted to show was that the majority, if not all, of the symptoms that our patients seemed to suffer is from this nocebo effect.

Limitations of SAMSON

Mandrola: What is the biggest limitation of SAMSON? I see online that a number of people are criticizing the small numbers, or that these were not de novo patients. What do you say about all that?

Howard: I'll talk about the de novo patients first. If they want us to do a trial of people who have never had statin side effects before and show that they don't have statin side effects, my gosh, are you lining yourself up to be criticized by the public, because they will just say we selected the kind of patients who don't get statin side effects. We have set ourselves up to fail by picking the people who suffered the most and demonstrated that even among these patients there isn't a huge amount of evidence that there is a direct statin effect.

Now, in regard to the patients not being representative, this is a common problem in clinical trials, and one could say that not every patient will be happy to put themselves through possible pain and suffering associated with going back on a tablet that they have tried several times before, and then use a smartphone every day to tell us how they are. We owe our patients a lot, and they were really invested in this process. And a lot of our patients pursued us asking to join the study. Maybe that is not typical of a lot of the patients in our clinics, but this was a very pragmatic trial and we did not exclude patients because they had atypical symptoms, for example. Sixty percent of our patients had myalgia, but not 100%. We let people with any symptoms in. We tried to be as inclusive as possible. But yes, when you run a trial like this, there will always be criticisms, and some of those are fair.

Using SAMSON in Clinical Practice

Mandrola: What is the take-home message for clinicians? How should we use SAMSON at the bedside?

Howard: What SAMSON shows you is that it actually makes a difference if you spend the time to sit down with the patient and show them actual data, ideally data relating to them (which is obviously going to be hard in the clinical setting). It makes their symptoms better and it reduces the nocebo effect. This shows that sitting down and spending time discussing this will probably reduce the symptom burden your patients suffer from.

One other thing that would be nice going forward is if we could continue to do these N-of-1 studies at the patient level outside of clinical trials but in clinical practice. Now, there are some barriers to that: You have to get identical-looking placebos to whichever statin you are going to give. But after that, our patients were very happy to do this experiment, and they really wanted to know whether they would be able to take these lifesaving drugs. Maybe in the future, clinicians will be able to run these trials with a little bit of help from funders or placebo manufacturers.

Mandrola: It's almost like we need a SAMSON(ite) suitcase with a smartphone and the appropriate placebo. I guess it's very difficult to make placebos because it is expensive.

Howard: It is. It's one of these things where it's very expensive to make only a few. If you decided that you wanted to make many of them, like 50,000 patients' worth, then it becomes much cheaper. But setting up an industrial process to mimic off-the-shelf generic atorvastatin, which is what we did, has large upfront costs. I think it would be great if we could get suitcases of these medications to try.

We should call them "wood boxes" after Frances Wood, because although I'm presenting it, she was the joint first author and she was the research nurse who really kept this trial together.

Mandrola: Excellent. James, I really appreciate the time you spent with us. I think that SAMSON is so important. Clearly, statins are beneficial drugs, and beneficial drugs are only beneficial if people take them. I think it transcends statins to how we speak with patients and how we introduce therapies. I'm just so excited about SAMSON and I'm so excited to talk to you. Thank you for being with us.

Howard: You are very welcome. Thanks for having me on.

Follow Medscape on Facebook, Twitter, Instagram, and YouTube


Comments on Medscape are moderated and should be professional in tone and on topic. You must declare any conflicts of interest related to your comments and responses. Please see our Commenting Guide for further information. We reserve the right to remove posts at our sole discretion.