Lyumjev. Ultrafast? It depends…

After a week with this new insulin, I thought it was worthwhile having a quick review of what I’ve learned and how well it works for me.

As readers will know, after 24 hours, I wasn’t seeing very much in terms of the so-called ultra-rapidness of the insulin that had been displayed in the various trial data that had been put forward. This was using it in a pump, with a hybrid closed loop.

Using as an injection

Taking a step back, I decided to run some tests, injecting it as I would if on MDI. In doing this, it consistently met the target of being ultra-rapid. Pretty much like clockwork, 30 minutes after an injection, glucose levels would start to drop. Really fast. An example is shown below.

This behaviour seemed to work every time, and suggests that when I use it for discreet injections, I see the same effect as those who have participated in the trials.

Using with the pump

With the pump, however, it was a different matter, for me. And I stress the “for me” as this is n=1 and doesn’t mean it will be the same for others. The infusion sets in use throughout were Medtronic MiO Advance Teflon sets.

As I mentioned, after the first 24 hours, I wasn’t seeing the alleged ultra-rapid effect. Instead I was seeing something more akin to massive insulin resistance.

In this first image, you can see a bolus undertaken 15 minutes before eating. If the effects were as when I injected, you’d expect that upwards trend to be interrupted about 30 minutes after bolusing, but nothing. And OpenAPS SMBs appear to be having no effect either.

In this second image, you can see a meal eaten as glucose levels were dropping and then insulin and food given as the levels inflect. Firstly with the amount given early in this image, I’d have expected levelling off earlier than it did. Secondly, a half hour after the second large bolus, I’d expect the levels to at least start coming down, as IOB was significantly in excess of the COB. Instead they remained almost flat as there seemed to be no effect from the insulin. This raises questions for me about the effectiveness of continuous infusion of Treprostinil for me.

Site issues

The pump study showed that site issues are a thing with Lyumjev. And the patient Information Leaflet also highlights this.

My experience was that my sites with Lyumjev didn’t hurt when I bolused. However, the site itself became quite tender and felt like there was a large bruise. This made it quite painful and sensitive to touch. This was highlighted when I was in bed and rolled into the site, and the pain woke me up.

In addition to this, the bruise sensation took 24 hours to disappear after I had changed sites. The slightly odd thing was that there really wasn’t very much to show for this. Maybe a slight redness, but nothing abnormal from a used site.

It’s worth bearing in mind that my infusion sets are Teflon and the injection was done with a steel needle, so I need to check whether I see the same effects when using a steel infusion set.

Okay, n=1 and all that. Any other data?

Feedback from other early adopters has been varied. All the users I’ve seen feedback from have been using either a pump or an open source closed loop. Bear in mind that this is still a very small sample, n<10.

They fall roughly into three groups:

  1. Those for whom it worked perfectly
  2. Those for whom it didn’t work
  3. Those for whom it worked but they suffered a site reaction and stopped.

Those in category 1 seem to make up about half. It’s worth noting that one of the success stories here stated they use steel cannulas. I don’t know whether this plays a part. Categories 2 & 3 make up the rest, with one person experiencing a major allergic reaction after 18 hours and stopping completely.

What I take away from this is that individuals will not all react the same way, and care should be taken if providing Lyumjev to users on a pump, as they will need to closely observe what’s happening.

Why this variety in outcomes?

It’s a good question. Lyumjev is accelerated by the addition of Treprostinil to Lispro insulin, which in theory causes vasodilation at the injection site. For me with Teflon cannulas, it would appear that it’s causing other issues. Could it be that the concentration of the additive when used for infusion is simply too high for some of us, and as a result there’s a build up at the site causing issues? Would lower concentration work better in pumps?

We probably need to see more observation on this as more people start to use it and we get a better view on how people get on. If you read the patient information leaflet, there’s a statement that says this product is subject to additional surveillance, so I’d encourage any early adopters that see issues to record them with their local regulatory body.

What next for you?

On its own with my current setup, this has clearly not met my expectations. As a result I’ll look at some of the other things that have been done in the WeAreNotWaiting world relating to using Fiasp. Suffice to say that I’ve stopped using it “neat” for now.

Firstly I’ll try the steel cannulas and see if that makes any difference. I found with Fiasp that they helped, so I assume that while they do more physical damage, there’s less bodily reaction to them, which may ameliorate that impact of Lyumjev.

Secondly, I’ll take a look at mixing the Lyumjev with Lispro and see if that makes any difference.

Finally, it’s worth having some in the fridge purely for stand by. Knowing how fast it does work for me when injected, there may be times where that will come in helpful.

But in total, it has turned out to be somewhat of a conundrum, and not quite what I was expecting. Let’s see what happens next!



  1. Ok, I left a comment how well mine was doing after a week. However it went down hill this Sat. I changed sets and locations and still could not get it down. I could not get b.s. below 200. So I tried using insulin from pen in pump (tandem x2). Way way way more aggressive! Even more aggressive than when I had things leveled out using a vial in pump. I could not keep my b.s. up no matter what. Also burning sting was all but gone. So yes…so far (based on only a one off test) insulin does seem to work better from a pen. All that said, when I changed sets saturday I noticed some blood that looked diluted. But didn’t think much about it. However tonight mine started rapidly climbing again after an all day fighting to keep it up. I noticed a wet feeling under my sleeve. I looked in mirror and felt. There was clear liquid where line plugs into infusion set. And a lot of diluted blood on tape that was not there earlier. Also the line was now w very easy to “wiggle” where it plugs into the port as if the port sleeve seal had eroded allowing backwash. I am thinking there may be to issues POSSIBLY. One it irritates enough it can cause bleeding on some people. It is rare I ever bleed with my infusion sites. If I do it is a dark red not really diluted. And i have never had humalog leak insulin out with the blood into the tape much less two in a row. Also I have never had an infusion set back leak between line and port. And for sure I never have a tight line into port that becomes very very loose. It definitely seems to be breaking down material allowing a loose seal allowing the leak at least with tandem infusion sets. Again a very small amount of data. But from 0 diluted blood leaks to 2 out of 2 and from 0 leaks to 2 out of 2, it is leaning strong.

  2. One more addon to my post. I removed my infusion set this morning. I returned to Humalog. However, before bed I squeezed where the line plugs into the infusion set tightly closed and taped. It no longer leaked out. However in the morning there was a lost of insulin (clean without blood in it) loose inside the hollow cavity ov the infusion set above the skin. This is outside of the tube that takes insulin through the cath. Also there was more bleeding and the tape holding infusion set to skin was very wet with diluted bllod and areas of simply wet from clean insulin.More suspected thought that with my brand of infusion set the insulin may possibly be breaking down some material inside the infusion set port. Again this is a single person with only one example since I did not do as detailed with the last infusion set that had diluted blood.

  3. Hi Tim! Thanks for sharing your experience with Lyumjew. I wonder if those resistence issues with FIASP and Lyumjew arise due to the insulin itself or if it is due to the peak time and length of distribution and potential gaps in the insulin distribution? For example: users with a flat line basal rate or wonky basal rate might run more into resistance issues? I am using a circadian profile and my hb1ac is most of the time below 5,5% and (moderate carb eating). Also Autotune user might run into similar issues. I am keen to start using it myself and to see how its working for me. I tried FIASP but I couldn´t see advantages for myself, so I switched back to Humalog.


  4. Be aware that Lyumjev has a substantially different action profile. More of its action happens within the first ~2-3 hours compared to Humalog. This in itself is a great advantage, because many hypos happen due to that “Humalog tail” after ~4-5 hours. That “tail” is greatly reduced in Lyumjev.

    However, that tail also helps with covering gaps in basal profiles. Setting up a good basal profile is difficult to do, so gaps are likely to be present. Autotune produces results that are better than that of many endos, but that’s not because Autotune is great. Instead, many endos simply program in absolutely dreadful basal profiles, so the bar is rather low. Autotune’s basal profiles still aren’t ideal, and the included gaps have a potentially much bigger impact with Lyumjev than with Humalog. Other loopers I’ve talked to who use circadian basal profiles also observed this. Once their basal was well tuned, Lyumjev worked amazingly with the loop.

    So, to summarize: With Lyumjev, your basal profile must be set up really, really well, and Autotune won’t be enough. Time to try out circadian profiles.

    • Thanks for your feedback. As the post pointed out, this wasn’t an issue with the basal profile but with the insulin per se. Bolusing was having no effect, which is rather different to the assumption you’d made.

      As it happens, I already run a circadian profile, and have never had issues with Autotune working off that or with other insulins with it.

      • Hi, that’s interesting. I thought you are running autotune according to your blogs. But admittedly I haven’t read all of your posts. Have you tried fasting as well and do you run also there in stronger resistance? (I mean even a circadian profile there’s resistance just a lot less). For me, if there’s resistance then boluses aren’t working. Can be tedious to work it out…Have you tried time shifting your profile, like +1 hour. Could be worth trying, but it’s faster to realise on AAPS…

        • I do run Autotune, however it runs nightly on an OpenAPS rig. As I’ve adjusted the peak time to 45 mins and the dia, it shifts the whole curve forward quite a bit and is able to pick that up in Autotune.

          The variance between segments on my basal profile around mealtimes isn’t all that high hour to hour, and multiple SMBs + Basal increase should have had much more of an effect than I saw.

          Can’t say I’ve tried fasting with it.

          • I have recently tried to use Lyumjev in a Tandem pump and have similar reaction at site of blood and soreness, along with little to no response or reaction on my levels. I found myself changing sites for my pump thinking it was site or pump failure, to which I found out was not the case. Three different sites for three days and still no proper injection response with Lyumjev. Went back to Humalog and I’m back to being able to control my levels successfully without blood leaching or leaking of product.

1 Trackback / Pingback

  1. What are the clinical benefits of faster insulins? #EASD2020 | Diabettech – Diabetes Wisdom

Leave a Reply

Your email address will not be published.