Is LADA the same as Stage 2 Type 1 Diabetes?

This subject has been bouncing around in my head for a while and came about from an interaction with one of the giants of diabetes research: Professor Chantal Mathieu.

As a textbook LADA, I was constantly frustrated, prior to becoming insulin dependent, that I fell through the gaps on most research. Either the research wanted people recently diagnosed as type 1 or people who were insulin-dependent. Having a 5 year honeymoon insulin-independent honeymoon excluded me from both. In discussions with researchers, a practical limitation of researching LADAs is we progress so slowly meaning to bubble up statistically significant results takes timespans of years, instead of months, which is expensive from a grant perspective.

Perhaps the difficulty is our definitions. If we accept LADA is a form of stage 2 type 1 diabetes, it could be included in research studying the transition from stage 2 to stage 3 which can occupy a time period similar to the LADA honeymoon.

Type 1 Diabetes Stages

A bit of Googling brought up Milk and Honey Nutrition‘s excellent summary of the stages of type 1 diabetes.

In theory, LADA could fit into stage 1 but, usually, it is high blood sugars which leads to a diagnosis of LADA as proactive auto-antibody screening is still rare in most countries.

I fell somewhere between stages 2 and 3 at diagnosis. I tested positive for three auto-antibodies and showed symptoms typical of stage 3 but, with dietary control, I could keep reasonably normal blood sugars and a healthy HbA1c. After five years, my HbA1c began to rise and I moved to insulin before the typical symptoms of DKA showed themselves.

So Why Is LADA Distinct From Stage 2?

According to Wikipedia, LADA as a term was introduced in 1993 and a formal model of the stages came around 2015. So, there was plenty of time for LADA to establish itself as a conceptual sub-type before the stages became part of the academic diabetes zeitgeist.

There have been dissenting voices as to the need for a distinct LADA classification. This particular link makes a statistical argument, rather than one based on observational evidence and, given I tested positive for three auto-antibodies on four separate occasions and yet still had a honeymoon of years, not months, in my opinion, debunks this particular paper. I mention it simply to show there is continued debate on the subject.

Professor Mathieu’s Twitter response at the start of this article, also confirms the distinction between LADA and ‘classic’ type 1 is far from having consensus.

Where To From Here?

To settle the debate needs a few elements. The first would be a large cohort of people with stage 2 type 1 and LADA. Reviewing their characteristics and defining distinct sub-groups based on measurable parameters could tease apart the groups. As mentioned, systematic screening is still not widespread so studies of stage 2 are still relatively limited and studies of LADA are even more so. Until that changes, the distinction between LADA and stage 2 will be conjecture, more than fact.

It is accepted that type 1 is a heterogeneous disease i.e. there is no ‘true’ type 1 but, rather, many, many diseases which exhibit similar broad behaviours so LADA could well be a distinct classification. At this stage we come back to the response I got at ATTD 2023, we simply do not have the data to know and must consider it in a way which suits our purposes.

A Free AGP Reader, Driven By AI

One of the common reports generated by people with diabetes who use Continuous Glucose Monitors is the Ambulatory Glucose Profile (AGP) graph.

If you are unfamiliar with this graph, it is a “heat map” of your blood glucose level (BGL) sampled from multiple days and put over a 24 hour period. So, for example, in the above which is one of my AGPs from last year, the data were over three months (03/07/22 – 03/10/22) and there are clear spikes around 14:30 and 21:30. These, of course, related to meals (lunch and dinner). Why no breakfast spike? Because, generally, I do not eat breakfast.

What I set out to do (and achieved) is create a bot which can be fed an image like the one above and reply with observations to discuss with your health care team. It works via email and generates replies like this.

If you are interested in trying it out, send an email with a screenshot of an AGP (in JPG, PNG, or BMP format) to AGPReader@hotmail.com and you should get back a response in a minute or two. Do not worry if your AGP does not look exactly like mine, I have trained the bot to review graphs from a variety of software sources.

If you are interested in seeing how I put it together, check out my post on my technical blog here.

Next Steps

While reasonably reliable, I am keen to refine and improve it. To this extent, I am talking to medical academics and professionals to get their input on both the training and the analysis the bot generates. I expect, with their assistance, the bot will continue to improve.