Researchers at the University of Miami are exploring whether a combination of two well-known supplements—omega-3 fatty acids and vitamin D—might help preserve insulin production in people with type 1 diabetes. The study, called POSEIDON, is testing whether high doses of these supplements, taken together, can slow the progression of the disease by protecting the remaining beta cells in the pancreas. These substances are classified as nutritional supplements rather than traditional pharmaceutical drugs. In the USA this means that there is no approval process required to show they are safe or effective.
The results for this trial were uploaded to the US FDA's clinical trial registration web site. However, I cannot find them published in a journal. Since the trial was unsuccessful, they might not have been published. An "after the trial" analysis was published, which I discuss below.
Results
The primary outcome results focused on the change in C-peptide levels, after eating a meal, over one year. C-peptide levels declined in both groups over the course of the year, which is the typical progression of the disease. C-peptide levels dropped about the same amount in treated people as in untreated people (there was no statistically significant difference), so the trial was unsuccessful.
This trial was randomized, open-label format, meaning both the researchers and the participants knew which treatment they were receiving. The study enrolled 27 participants who were divided into groups based on their age and how long they had lived with the disease. This included both children (ages 6 to 17) and adults (ages 18 to 65), as well as individuals with a new diagnosis (within six months) and those with established T1D (up to 10 years).
Discussion
Post-Hoc (After The Study) Analysis
The phrase "post-hoc" refers to studies that use the data collected from one study, but are designed knowing what that data is. They are not blinded, but quite the opposite, publish results that only come apparent with analysis done knowing the data.
These studies are generally not used for drug or device approvals, because it is easy to design a study that shows good things, by carefully crafting it to avoid the bad things that the researcher knows are in the data. However, they can show interesting directions for future work, or results that were not even under consideration when the study was originally designed.
These researchers published a post-hoc analysis which suggested that at the time of diagnosis, people with lower Vitamin D levels also had lower fasting C-peptide levels, but no difference in post meal C-peptide levels. I found these results uninteresting for a couple of reasons. The biggest is that there have been much larger studies looking at Vitamin D and T1D diagnosis. The second is that in other studies that I've seen have fasting and post-meal C-peptides tightly related (going up and down for the same reasons). So I think seeing an effect in one, but not the other is likely a random effect.
Words vs. Reality
This is how AI chatbots describe the two treatments tested here:
Omega-3 fatty acids, found in fish oil, have been studied for their potential to reduce inflammation and support heart health. Vitamin D, plays a role in immune function and may help regulate autoimmune responses. When taken together, these supplements might offer a dual benefit for people with T1D by reducing inflammation and supporting beta cell survival.
That sounds positive, but shouldn't. Notice the use of weasel words: "studied for their potential", "support", "help regulate", "might offer", and "might benefit". In fact, neither Omega-3 nor Vitamin D has been found to help inflammation or immune function in clinical trials such that they could be approved for use in the US. It is important to remember that these words are widely used in blogs, news reports, and sales literature to trick people into thinking these treatments help in some way. (Their widespread use on the web causes them to seep into AI chat bot results.) The unsuccessful results from this clinical trial, and several other clinical trials, show that they do not work, or at the very least, we do not have evidence that they work, and their promoters are speculating on future results which may never come.
I want to stress that this is not a simple minded failure of AI. It is a simple minded failure of the world wide web. AI is accurately repeating the mistake made by the web and people on the web, which is to treat opinions (and especially opinions designed to attract viewers) on an equal footing to data from scientific studies. To put it bluntly: 1000 bloggers saying Vitamin D is a wonder cure is less substantive than one well done study showing it is not. In a sense, general AI is implementing the thought process of a random person who can hear volumes and count sources, but can not evaluate anything.
More Information
FDA Clinical Trials Registry: https://clinicaltrials.gov/study/NCT03406897
Trial Description: https://diabetesresearch.org/poseidon/
Trial Details: https://www.cellr4.org/wp-content/uploads/sites/2/2018/04/2018-03-POSEIDON-Protocol-Consent-Assent-2.pdf
The research publication related to the study's rationale: https://pubmed.ncbi.nlm.nih.gov/27467009
Post-Hoc Analysis: https://academic.oup.com/jes/article/9/6/bvaf061/8110156
VITAL Results: https://www.vitalstudy.org/findings.html
Joshua Levy
http://cureresearch4type1diabetes.blogspot.com
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official BreakthroughT1D or JDCA news, views, policies or opinions. I sometimes use generative AI ("chatbots") to generate draft blogs, parts of blogs, or drafter alternate wordings for these blogs. I always review every part of every published blog to ensure that it is saying what I want, in the tone that I want, truthfully, and accurately. My kid has type-1 diabetes and has participated in clinical trials, which might be discussed here. I am obese and right on the border of T2D and therefore may be taking drugs for those conditions. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog!
