Is Hba1C a poor yardstick for determining access to new T1D treatments in the real world

Is Hba1C a poor yardstick for determining access to new T1D treatments in the real world
Is Hba1C a poor yardstick for determining access to new T1D treatments in the real world

If you’ve ever looked at a clinical trials document for a type 1 treatment, you’ll nearly always spot two things.

  1. Reduction in Hba1C is a clinical outcome
  2. The people they want on the trial are those for whom a reduction could be measured

Setting aside what’s wrong with the latter for another day (and there’s quite a lot wrong with it), let’s look at the first, as it’s a real bugbear of mine.

This post comes as the result of a Twitter discussion I observed and then responded to, and the data I was looking at this morning, as it relates to the likes of Fiasp and new treatments.

Where are we starting?

A paper published in the Journal of Diabetes Science and Technology in September 2017 identified that pre-bolusing for meals resulted in a lower Hba1C than post bolusing, or completely forgetting to bolus. Wow. Big surprise there. If you don’t have insulin on board for a meal that contains carbs (or importantly carb equivalents) then your blood glucose level will end up higher than if you do. Sometimes I do wonder why research dollars are paid out for research to prove a truism. But that’s an altogether different issue.

The key point here being that the lower Hba1Cs are associated with better matching insulin action to food.

There’s nothing wrong with that statement alone, however, we then have to take a look at new treatments for Type 1 Diabetes, and a really good example here is Fiasp. Why Fiasp? Well it’s’ a recent example of a new formulation of a diabetes drug that reported changes in Hba1C as part of its clinical trials. Firstly, it showed non-inferiority in its trials versus Aspart.  You’d not really expect it to show inferiority given that it is a reformulation of Aspart. What it showed was a slight* reduction in Hba1C in the Onset 1 type 1 trials.

*Slight means between 1.7% and 4.2% reduction.

Firstly, it’s worth bearing in mind that under trial conditions, those using Novorapid pre-prandially saw a greater reduction in Hba1c than those using Fiasp post-prandially. The average reduction for Fiasp users bolusing pre-meal was 0.32 and Novorapid was 0.17. If we crudely account for the trial generating a 0.17 reduction using the Novorapid data, then we find that pre-prandial Fiasp reduced levels by 0.15, which is a little under 2% of the average. It’s crude, but it makes a point.

In addition, this graph was part of the data.

The result of this data? That many Healthcare Professionals that looked at the activity graph and trial data questioned the point of bothering with Fiasp as the reduction in Hba1C appeared to be tiny.

What they didn’t consider was that, in reality, the tiny difference in a graph translated to something much more in use, as has been widely documented, as have the longer term “side effects” for a subset of users. You could achieve those tiny Hba1C reductions by bolusing just before you ate.

This seems like a minor point, but for someone living with T1D, it takes away a major issue. Having to pre-bolus to time the insulin action with the food to avoid a post-prandial spike. Something that a non-T1 takes for granted. Fiasp, for the first time, freed people up from having to plan 30-45 mins ahead. It allowed freedom, as I’ve discussed before.

And no-one, least of all a specialist in Diabetes, should be ignoring that. But many did, or took their own point of view as regard to efficiency. As a result, the amount of time that Fiasp took to get on to the formularies around the UK as a result has been a bit of a joke. Well, I say “A bit of a joke” and by that, I mean a disgrace. Why? It costs exactly the same as Novorapid. Even now, a year since its release, we have only had 1,390 items prescribed in the UK in December 2017. There are 400,000 people with type 1 in the UK.

Even I, who ended up stopping its use in the pump, still use it as an injectable when I need that bump up.

And let’s be honest, if we want to bring down Hba1C to the UK National Target of 6.5%, there’s a lot more that needs to change, not least of which is not recommending that people go to bed at 8mmol/l then stay there all night, and perhaps better advice around pre-bed short acting insulin and basal testing to confirm that overnight basal rates are calibrated as best they can be.

After all, sleeping for eight hours at an average of 8mmol/l means that for the next 16 hours, you need to average no more than 7.7mmol/l. That’s a challenge at the best of times given how food affects glucose levels, and especially given the usual advice on when to dose the current fast acting insulins.

This is, of course, also associated with confidence around overnight hypos, and the only way to alleviate this is really to provide some mechanism for managing, or alerting to, hypos, which brings us back to either closed loop (which is DIY only in Europe at time of writing) or CGM, about which we already know the cost issues.

Leaving that aside then, when new drugs are introduced and some people want to try them, how can we enable early adopters to use them and provide real-world feedback? Well the first and easiest way, with the like of Fiasp, is to have the pharma companies list them by generic and get a prescription for a generic. Then no-one is any the wiser. That’s perhaps not in keeping with transparency though.

The second is that we need a way to make prescribing these types of treatment easier. Especially when they have potential to make life easier and don’t come at an additional cost. Whilst additional cost is always a tricky question, for early adopters looking to try something new, that generally means they are asking. They aren’t expecting involvement from the healthcare team to understand how something might work and what changes they might need to make. I’d call them “sophisticated users” that are experimenting with themselves. There is therefore no healthcare system cost.

All it takes then, is for the agreement between the healthcare system and the patient to require that there is feedback into the system to educate the prescribers to a greater level and not only are these drugs available to those who ask, but the system gains something in response. Feedback and experience. And then at least, for those people, waiting on the formulary would be a thing of the past.

Who knows? By implementing a system like this, might we get cost neutral solutions to patients with education driven from other patients much more quickly? Could it base the whole cycle more in the real world of patient involvement than in the world of randomised control trials? As my earlier comments on the Onset 1 trial mentioned, in a trial, even the controls improved. Is that really reflective of use in the real world?

1 Comment

  1. Hi Tim, good article and I hope you are well?
    I’ve done a few courses recently and have come across a few things that made me think about how valid HbA1C is. The first is that haemoglobin cells last approx 120 days (with quite a bit of variability-to be discussed later!). Theoretically, therefore, you could have your HbA1C tested when you have a whole tranch of new haemoglobin which would be glycated very little. This would lead to a lower score than otherwise. Conversely, your cells could last longer which would mean they were glycated a little more than ordinary. Now, I get that it’s an average, and a guide, but when you take into account that haemoglobin lasts longer the more stable your blood glucose levels are and you can see the potential for incorrect scores.
    The second point is that it’s possible to have 2 people with the same HbA1C with one person with fairly stable BG and the other with lots of variability, high and low. With glucose variability having been shown to be a risk factor for lots of health conditions, again, it doesn’t really reflect how well the condition is being managed. A test I’ve come across, which actually measures glucose variability, is Glycomark. Have you seen this? It basically measures variability over a 2 week period and so could be much more useful in terms of testing, making adjustments and testing again multiple times before you would have had another HbA1C.

    Anyway, you’ve probably seen all this stuff but just incase!

    On another note, Benji didn’t get on with dexcom so we went back to libre’s and are now running a Blucon Nightrider with Xdrip/Nightscout. Only had it a week but so far so good!

    Speak soon-enjoyed the videos from last weeks events.


Leave a Reply

Your email address will not be published.