Fiasp – going off-label. Being intentionally non-compliant!

Fiasp – going off-label. Attempting to handle the unwanted side effects and retain the rapid effectiveness
Fiasp – going off-label. Attempting to handle the unwanted side effects and retain the rapid effectiveness

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

As has been discussed in many places on Social Media and the internet, whilst the speed at which Fiasp takes effect has made pre-bolusing a thing of the past for many, a fair number of users have highlighted issues with it, many of which are detailed here. I have also discussed what has been reported.

Side effects of using Fiasp with a pump that have been reported in various forms of social media include:

  • Irritation at the infusion site
  • Pain on bolusing
  • Lumps and soreness on removal of the set that persist for a period of up to two weeks after the removal
  • Shorter set life
  • Steadily increasing basal insulin needs
  • Increased insulin needs with meals
  • Random high glucose events requiring significant additional insulin with no obvious cause

As the number of people struggling with this seems to have grown, there was chatter on social media of people starting to mix their old insulin with Fiasp, in the hope of coming up with a hybrid that retained the fast action of Fiasp but reduced down the side effects that people were experiencing.

After a week of suffering at my infusion sites, I decided that I too would attempt this approach. With every bolus feeling as though someone was sticking a hot needle into me, and small, painful, ball bearing sized lumps appearing at every set change, I didn’t feel that I could continue with Fiasp alone. One of the biggest issues in Type 1 is protecting the injection/infusion sites, so this appearance of what looked like damage didn’t seem all that welcome. I therefore chose to embark on this “WeAreNotWaiting” approach to Insulin.

Sometimes, quality of life outplays other things and leads us in new directions, and in the case of Fiasp, the underlying insulin is the same, it has just been mixed with something new to make it act faster. As we used to mix Short and Long acting insulins to create the equivalent of 70/30 in the past, it wasn’t an enormous step to try mixing Ultrafast and Fast.

In the 28 days prior to making this decision, my stats were tending away from where I wanted them. I was seeing greater variation and too regular lack of effect. The distribution shown below gives a time in range of 86% or so. While this is, by any point of view, a good result, the ups and downs around that were starting to get wearing.

And while the loop was patiently reeling me back in when I went high, or desperately trying to suspend insulin after I had “dragon chased” as the Fiasp seemed to have no effect, there was rather too much roller-coastering going on. And it all seemed to be related to eating. I’ve previously posted about how I was seeing significant variation in carb ratios, which lead me to dig further, and the below percentile graph indicates some of the concerns I was having, with the 2am lows being the result of trying to catch the 10pm highs. As the graph below shows, overnights with no food were really pretty good. As soon as I got into eating (a relatively higher carb diet) that’s where things started to go wrong. I also didn’t need to be eating a lot of carbs in a day to see issues, indeed one of the worst days was relatively lower carbs.

Running this alongside the site issues, I took the step of mixing Fiasp with Novorapid, mainly to see if I could retain the fast action of Fiasp whilst gaining some more predictability and less pain from the Novorapid.

My mix is 50% of each, which has become denoted on Social Media as F50N, i.e a mix that is 50% Fiasp with 50% Novorapid. F30H would be a mix of 30% Fiasp with 70% Humalog. I achieve this by simply extracting half a reservoir of Novorapid, topping it up with half a reservoir of Fiasp, then agitating the reservoir. Others are mixing in an empty phial.

At this point, I must reiterate that this is not recommended by anyone and goes directly against the insulin packaging, which states:

Filling the pump
► Fiasp® should never be diluted or mixed with any other insulin.

As it’s a package statement, there is no reason given and what we’re doing here goes directly against what the manufacturer and regulatory sign off state. This is very much “off label” usage.

However, after a week of running with this mix, I’ve been rather surprised at the outcomes.

The site issues I was experiencing have all but disappeared. There are no more “ball bearings” and bolus pain is seemingly gone. The redness and lasting after effects just don’t seem to be there. This is a welcome change, but does require further monitoring to see if anything comes back.

But what about the pharmacodynamic effects? Well these also look to be showing up something interesting. So far, there doesn’t seem to have been a dramatic change in the time to insulin action (indeed on my OpenAPS rigs I am still running with the Ultrarapid curve designed for Fiasp without it causing any issues). The other thing I have observed is that the dramatic climbs where the insulin seems to be having no effect seem to have stopped, at least in the week in which I have been testing this.

Statistically, the results look like this:

The majority of low readings can be attributed to a period of sleeping on the sensor and are clearly visible in the day by day stats while the highs have dramatically reduced. Either way, the distribution looks a lot better. In addition the variation indicators show a lot less variation. The key thing to note is that eating hasn’t changed in these two comparisons.

The percentile graph backs this up, showing a lot tighter 10/90 spread, and much faster reversion after postprandial highs. What’s notable is that a week’s data isn’t really enough to properly make a comparison though so longer term data is required.

But why would mixing Fiasp with Novorapid appear to make things better without losing the speed of action? Well we need to look at the discussions on Social Media to try and understand some of this.

Why do we think this might be making a difference…

Smarter folks than me, who have a much better grasp of the biochemistry involved in the process have read through the patent and speculated as to what is going on, and why, by reducing the amount of Fiasp, we aren’t seeing a significant reduction in speed of action. Thanks to Dr. Joel Milner* (DPhil Biochemistry, University of Oxford) for articulating why this may be happening, based on interpreting the ingredients list and the patent. What is below is a theory as to how the insulin Aspart is bonded with the Nicotinamide.

Fiasp appears to be manufactured by mixing solutions of insulin and nicotinamide, achieving first-order kinetic binding of the nicotinamide to the insulin B-chain, thereby increasing its solubility.

According to the US patent, Novo Nordisk looked at solutions that contained between 133 and 333 times as much Nicotinamide as Insulin Aspart. This is presumably because the binding is quite weak and this was needed to drive the process to completion.

So what we take from this is that there’s a high concentration of Vitamin B3 in the form of Nicotinamide in the solution we inject. What we also speculate is that due to bonding of Nicotinamide with Aspart being weak, a lot of B3 is required to get a “best bond”.

But why, when we mix with other insulins does it appear that the fast effects are retained? After all, at a 50% Fiasp/50% Novorapid mix, there’s obviously less of the faster acting stuff. Going back to Joel:

To understand what happens when you mix the two insulins, you have to look at the theoretical ligand binding curve (shown in the picture below). This models the interaction between a ligand (for example B3) and its binding partner (for example, insulin). The shape of the curve shows that in order to achieve a high level of binding (on the vertical axis) you require a significant concentration of ligand (B3 on the horizontal axis).

If the interaction between B3 and insulin in Fiasp follows these well established characteristics and if the B3 can bind equally to any fast acting analog added to the mix (and accelerate its action) , a 50% reduction in B3 concentration will result in only a small reduction in the proportion of total insulin bound to B3.  The result will be almost as much acceleration of action but with fewer side effects from the B3. This is shown above using the red dot and arrow for Fiasp and the green for the mix.

This seems like a reasonable explanation as to why the mixes of non-reformulated insulin are still providing effective speed of action, and also why the site specific issues seem to have been reduced. It would also be an explanation for why the random jumps in glucose levels seem to have abated, however, in looking at these explanations, we have to bear in mind that this is all postulation and  we also need to take care that we are not applying Occam’s Razor.

Alongside this, we also need to bear in mind what a former pharmaceutical industry employee mentioned, and that regards the regulatory approved formulation of the insulin.

  1. The released formulation is going to take into account a number of factors and will not only solve for efficacy, but also stability, ease of manufacture, temperature sensitivity and other things.
  2. Bearing in mind that this is a refactoring of an existing product and not the introduction of a new active ingredient (the insulin protein remaining the same), the clinical trial and sign off process it may be required to undertake for regulatory sign off could be less intense. Note that this is not saying it was, but that the processes may differ.

We know from the submission for the EMA sign off that the CSII test period was only 6 weeks, and that’s probably not long enough for all the issues that have arisen to surface.

What do we take away from this?

I think I’d describe it as unexpected results. Those of us who have been mixing the Fiasp in differing concentrations with our previous rapid insulin have reported significant benefits:

  • on the site irritation and infusion pain front. It is almost like night and day.
  • The small number of users doing this have also maintained that the insulin performance doesn’t seem to have been detrimented.
  • In addition, as a DIY APS user, I have maintained the ultra-rapid insulin curve and seen no negative impact of late tail lows, which I’d expect if the Novorapid alone was playing a greater part. If anything, hypos have been reduced as greater predictability has resulted in less user intervention.

Personally I think this raises a number of questions.

  • Firstly, what is the requirement for testing a new formulation via each delivery channel for a medication? Given the lack of obvious public data about the body’s interaction with infused Nicotinamide, should additional long term testing have been required?
  • Secondly, what are the reasons for the formulation that Fiasp currently has? Could the Nicotinamide levels be reduced?
  • Finally, when we come to seeing the Lilly and Adocia enhancements make it through to phase III trials, will they include a six month CSII trial? I’m not sure whether anyone has infused a vaso-dilator (in however small amounts) subcutaneously for six months and observed the outcomes. Given this is what is in the mix for the faster Lispro that Lilly is working on, it might not be a bad idea to have that tested.

Don’t get me wrong, I love the speed of the Fiasp and hope that the additional, promising market entrants will have similar or better pharmacodynamics, but I’d rather not go through the oddities that have arisen with Fiasp, and I think many others would agree. With an eye out on what’s coming, can we make sure we get it right before release the next time please?

*Dr Joel Milner’s CV

Education:  BA in Biochemistry, DPhil in Biochemistry (both Oxford University)

Current Employment:  Reader in Plant Pathology, School of Life Sciences, College of Medical, Veterinary & Life Sciences, University of Glasgow.

  • Joel Milner has research interests in the molecular biology of plant-pathogen interactions and innate immunity in plants.  He has published well over 50 papers in refereed journals.  He iseditor or reviewer on a many scientific publications and a grant reviewer for BBSRC. He is currently a co-holder of grant funding to the value of approximately £1.4 million.

  • Joel was diagnosed with T1D in 1977 (aged 28), has been an insulin pump user since 2008 and has been self-funding CGM since 2010. He is currently open-looping using AndroidAPS  

    Joe l is also secretary and co-founder of IPAG (Insulin Pump Awareness Group Scotland).  He is active in the campaign for better understanding of and access to diabetes technology in Scotland


  1. Interesting piece if a little difficult to understand. Mixing the insulins seems to really help with the variability experienced on pumps. How could MDI users go about trying this? Would injecting the two doses of ultra fast and fast in the same area work? Or would the insulins need to be mixed before they enter your system?

    • I wouldn’t expect injecting two separate insulins in the same area to have much effect. I think pre-mixing is likely to be more effective.

  2. Very interesting article. I’m going to be in the UK in 2 weeks and will pick up some Fiasp and try this approach. Many years ago when I used to take Humalog, I had similar side effects to you, burning sensation, irritations, and lumpy scar tissue. None of that since Novolog, hopefully, this problem is a non-issue but I’m glad that you have found a solution.

  3. Is there an update for this? Fiasp seems to work very well for my body. My blood sugars respond so quickly and it makes corrections a breeze. However, I have had all the issues mentioned above. Did this enter testing or did you continue using this method? Any update?

  4. This article is fantastic! I, too, have experienced the dramatic benefits and the unfortunate side effects of pumping with Fiasp. I wasn’t sure if I would ruin the insulin completely if I tried a 50/50 mix, so I’d like to thank you for taking this risk and publishing your results. I’m going to give this a try!

  5. Do you know if Fiasp can be mixed with Admelog? My daughter switched from Looping with Admelog to Fiasp two months ago. The Fiasp worked great at first, but now we are having a lot of issues you described. We still have Admelog on hand, so I’m wondering if it’s possible to safely to mix the two.

Leave a Reply

Your email address will not be published.