Lyumjev. What next? Another n=1 experiment…

Warning: What is discussed in this post is not recommended by the manufacturers and not aligned with the European Medicines Agency sign off for Lyumjev or Fiasp.

If you choose to do anything discussed in this article, you do so at your own risk with the full understanding of any side effects that are incurred.

After my initial issues with using Lyumjev in my pump (which have certainly been seen by others, along with plenty who have no issues), I figured I still wanted to have a faster insulin operating.  What you’re about to see here is off-label, and is not a recommendation for use of medication.

In the past, in order to find reduce the issues that a number of us saw with Fiasp, a number of people mixed it with either Lispro or Aspart, with reasonable results.

The whole basis for mixing insulins previously was to reduce the concentration of the additives with the hope that this would allow them to still work, but without causing the site reactions (or other reactions) that a number of us saw. So why not try this with Lyumjev?

And if you’re going to go off-label and mix insulin, why not do it with the two fastest ones on the market? This is an n=1 experiment, and as stated at the top of this article, is completely off label and against the manufacturer’s recommendations.

Before undertaking this step, I did a quick internet search (which I encourage anyone reading this article to do) to find out whether there were any contra-indications for Nicotinamide and Treprostinil in combination. Whilst there wasn’t anything specific to the two of these, there was a mention of potential concerns relating to high dose Niacin along with other vasodilatory medications, but nothing specific to Treprostinil. Given the doses of the two involved in this process, neither could be considered “High”.

With the above out of the way, I decided to mix Fiasp and Lyumjev 50:50, on the assumption that I was halving the amount of each additive per dose, and so in theory at least, reducing the site impact. I configured OpenAPS with a DIA of 6 hours and a custom peak time of 45 minutes.

I also told my endocrinologist that I was undertaking this experiment, and they raised valid concerns about the effects of all these additives on my subcutaneous layers, and whether they would affect future use of the sites. 

The results

And the results?

Much better than I could have expected. After 1.5 weeks of use, the site pain I was experiencing with Lyumjev on its own hasn’t occurred. There is also none of the burn experienced with Fiasp. When I remove the cannula, there is also none of the raised bump effect I saw with both a neat Lyumjev use or with a Fiasp/Humalog mix, although this maybe a “honeymoon period” whilst my body hasn’t decided it doesn’t like this yet.

And perhaps even more exciting have been the results. I’m using OpenAPS with SMB, eating soon not currently in use due to being on vacation and having inconsistent eating times and not being more active than usual. The seven day plot looks like this:

This is with an average daily carb consumption of 161g and a peak of 278g. The dosing strategy has been to announce carbs and only to bolus with food if there was high GI food involved (which for me included meringue, candy, bread). The bolusing with food has usually been around 50% of what was likely to be needed. Both the time in range and the standard deviation have been remarkable, and it feels like this is now operating like the presented data for Lyumjev at the various conferences.

Taking a look at the Ambulatory Glucose Profile for the week, we can see that not only has the interquartile range remained within the 3.9-9mmol/l for almost the entire time, so has the 90th percentile, which is quite remarkable. This is reflected in the average “high” level being only 10.8mmol/l.

Additional Observations

Additionally, what I’ve also noted is that the tail of the mix does seem to be substantially lower, with descending glucose levels regularly flattening out before getting into low territory, in a way that I haven’t seen with Fiasp/Humalog mixes. 

Another observation is that as the reservoir end is reached, the efficiency of the mix doesn’t appear to work so well, and on reaching the last 20u, I need to change the reservoir, which I hadn’t seen in other mixes. 

It’s also worth adding that moving off Lyumjev onto this mix saw a rapid change in sensitivity and carb ratio that required a high temporary target and careful observation and alarm setting. Although this was helped by Autotune’s overnight run, it’s something that I was wary off and needed to be observed carefully.

So what next?

For now I’ll continue to use this mix as it is working extremely well. It’s possible that in doing so, I may also be acclimatising myself to the Treprostinil in the Lyumjev. Why is that important? When Treprostinil is used in continuous subcutaneous infusion (CSI) for Pulmonary Hypertension, according to a couple of people I’ve spoken with, it takes some time to acclimatise, but then sites can be used for 20-30 days.

Whilst this is perhaps less of an issue with CSI for that purpose, when combined with insulin and the issues that poor insulin absorption can cause, it may be that a different approach is needed.

So far I’ve used this mix for 1.5 weeks, and it has worked well. Perhaps after this, moving to neat Lyumjev will cause fewer issues, but that remains to be tested. 

The ease of management I’m seeing with this mix far outweighs the issues with Lyumjev on its own, so I’m wary of trying it out for half a reservoir again, understanding that I might go back to using “water”, but lets wait and see.

Whilst I’m in agreement with the “If it ain’t broke don’t fix it” point of view, I also agree with “Nothing ventured, nothing gained” and there may still be gains to get…


  1. Hey Tim, great write up.

    Curious why you mixed Fiasp in vs Humalog given the previous issues you had with Fiasp? I know previous you had found the diluted effect of Fiasp long term to be reduced when mixed with Humalog if I remember correctly.

  2. Hey Tim, great write up.

    Curious why you mixed Fiasp in vs Humalog given the previous issues you had with Fiasp? I know previous you had found the diluted effect of Fiasp long term to be reduced when mixed with Humalog if I remember correctly.

    • I went back to Fiasp/Humalog mixed 50:50 around December time, which was about 9 months after my last dally with Fiasp on its own, primarily because I wanted something a bit faster over Christmas for the various treats involved.

      This time around it worked nicely with the mix (can’t tell you why) so I’d been using that for around 7 months effectively.

      • I am considering mixing Lyumjev with Humalog as I am one bottle through at this point with Lyumjev and I am getting site irritation, not crazy, but definite redness at the sites using metal cannula. I did try straight injecting Lyumjev and like you experienced, the action was much faster than via pump. If doing MDI wasn’t such a hassle, I would consider it, as the action is pretty great when not by pump.

  3. Hi Tim! Thanks so much for sharing your insights with us!
    May I ask how and how long you mix the insulins to ensure they are evenly distributed?

    • You can never really ensure this, but I generally put 3ml of each in an old insulin bottle using the pump reservoir, then agitate the contents before extraction. That gives me two reservoirs worth of insulin at a time.

Leave a Reply

Your email address will not be published.