Last Wednesday, diabetes researchers and advocates from around the country convened in Washington, DC, for an FDA hearing on the Artificial Pancreas with the aim of making this pipe dream a reality. The point of the hearing was to allow JDRF and other stakeholders — researchers, industry reps, and patients — a chance to provide input on the Artificial Pancreas Project, in particular “recommendations to ensure the safe and effective testing of artificial pancreas technology in real-life situations.”
To bring you up to speed: various artificial pancreas-like devices have been tested in controlled hospital settings around the country, including clinics at Yale University and Stanford University. But in order to gain FDA approval, the artificial pancreas must be tested safely in “real world” conditions, i.e. in field studies where patients wear the device on a daily basis in their own home environments. In preparation for this hearing, JDRF formed a panel of international D-diabetes experts to make specific recommendations on how these studies can and should be accomplished.
A summary of their preliminary recommendations can be read in PDF format here.
You’ll note that the panel gives advice on how to transition from inpatient to outpatient settings for studies — things like establishing “a transition step where each patient demonstrates the ability to operate the system and be responsible for calibration without intervention from medical personnel (who are in close proximity) is advisable before moving to an outpatient setting.”
Next, the expert panel talks about which groups of patients should be allowed to participate in such studies: “Otherwise healthy T1D patients experienced with using insulin pumps and continuous glucose monitors, since such patients have the highest likelihood of using the closed-loop system safely. This would then be followed by the study of more clinically relevant populations that may be more likely to benefit.”
But at the end of the day, it’s all about safety.
What constitutes a safe artificial pancreas system? you may ask. Well, Folks, that is the $64 million question — and the one at the heart of this hearing.
All the players are understandably concerned about the risks of a system that could deliver insulin, or stop delivering it, automatically. Because of the potential hazards and the complexity of these systems, the FDA has yet to establish “clear and reasonable” regulatory guidelines for the artificial pancreas, or specific rules for any “automatic” features that might be added to current insulin pump systems. Without guidelines, as you can imagine, it’s difficult for any company to present a product to the FDA for evaluation, or even set up a clinical trial that they can be confident would be recognized to that end.
In Europe, Medtronic Minimed has already introduced an insulin pump that includes a “Low Glucose Suspend” feature (its Veo system), which enables the pump to shut off automatically for 2 hours if the attached CGM alarms that blood sugar has reached the patient’s “low threshold.” But this pump has not yet been approved by the FDA for US distribution, presumably because they see the automatic shut-off as potentially harmful to patients’ health.
Another main concern for the FDA is the fact that current CGMs (continous glucose monitoring systems) aren’t 100% accurate, so how can they be trusted to determine how much or how little insulin a patient needs? Sounds like a path to a tidal wave of hypoglycemic events, no?
Well, the data from “proof of concept” studies (studies conducted in a very small number of patients to prove whether a product is viable for larger scale studies) is telling another story, according to Dr. Fran Kaufman, Chief Medical Officer of Medtronic Diabetes. Data from 27 patients using the company’s Minimed Veo combined with a CGM showed a significant drop in A1C after 6 months, and notably: “There’s been no evidence of an increase in hypoglycemia,” she says.
But Dr. Patricia Beaston, an endocrinologist and member of the FDA’s Artificial Pancreas team, explained her hesitation: “The CGM and meter are part of the system and they have inherent errors. The fact they agree with themselves makes it hard to make any determination. If you look within individual patients, the accuracy of the sensor for the 3-6 days varies across those patients. A lot of it is dependent on how well that patient does the calibrations, how often, and the quality of the glucose meter they use to make those calibrations.”
Bruce Buckingham, an endocrinologist at California’s Stanford University and Packard Children’s Hospital and an investigator into the artificial pancreas, countered those fears, saying, “There’s not a very big risk if you turn it off for a couple of hours.” He states that in early studies, when the CGM indicated a low when there was none, leading to up to 2 hours of no insulin, the chances of ketones was fairly rare. (Not exactly the most comforting words when you’ve been on a CGM that seems to alarm for no reason, but for folks with hypoglycemia unawareness or who consistently do sleep through lows, the auto-shut-off system could be a life-saver!)
Finally, John Knight, a professor of computer science at the University of Virginia, offered a metaphor for thinking about the idea of “safety” in inherently “unsafe” machines: When we get into a car or fly in an airplane, we know there is a certain chance that they could stop working and crash, or that something externally could go wrong. But we do it anyway. Why? Because, as Dr. Knight explained, “Safety is defined as an ‘acceptable’ level of risk.”
In his testimony, Dr. Aaron Kowalski, Assistant Vice President of Treatment Therapies at JDRF, seemed to be hinting at the same thing. “In the hospital setting, these systems display fantastic results. The biggest need is to move these studies to the real world. We need to show these studies, and I’m completely convinced that if we use these systems in the real world, they will show an improvement in efficacy.” JDRF has also released a statement on their current thinking about the Artificial Pancreas Project.
By the end of the hearing, four main points were laid out that the FDA and industry need to work out, as presented by Dr. David Klonoff, head of the Diabetes Technology Society*:
The patients who testified — and those of us at home — know that while safety is a priority, there are inherent risks in living on insulin anyway. Some talked about the dangers and fears of nighttime low blood sugars… So is it prudent to wait for the technology to be “perfect”? If it ever will be?
The FDA will be reviewing feedback from the various speakers, and as always, we’ll be here to bring you the latest as things develop. Until then, what are your immediate thoughts on the Artificial Pancreas Project and the Low Glucose Suspend? Are you chomping at the bit to get started, or are you in the “slow and steady” camp?
{* Thanks to Crystal for the screen grab!}
[...] This post was mentioned on Twitter by Kerri / Diabetes, DiabetesMine, Bernard Farrell, Ellen Hoenig Carlson, JCampbell and others. JCampbell said: DiabetesMine: FDA Artificial Pancreas Hearing: Beating a Path to the Real World http://bit.ly/8ZCJiB #diabetes [...]
With meters and CGM’s in their present form I can’t see how the FDA could approve an artificial pancreas. Meters right now only have to be 80% accurate and this is what we use to calibrate the CGM? To me that is not acceptable. I have been using a CGM for about 2 years now and I’m a very active person. I have found that the CGM is fairly accurate when I’m not doing much but when I’m active the accuracy goes out the window. I am a avid cyclist and there has been many times when the CGM says I’m 80 when I’m 180 according to my meter. I also had times when the opposite is true CGM says 180 meter says 80. I think the FDA is going to say get the meters and CGM’s close to 100% accurate then we’ll talk.
I’m the mom of an 11 year old with type 1 diabetes. I’m certainly not a scientist, but I gotta believe we can solve the low/high management problem and improve the quality of life for people with diabetes. I guess this is my way of saying I’m chomping at the bit to get this thing started. The current system is far far from perfect; maybe by taking the leap of faith with an artificial pancreas type system we can move in that direction of making that reasonably safe.
I could rant for hours…..frustrations run high.
I listened to some of the webcast, and I wonder if I was being overly sensitive in one aspect. I felt like the speakers were implying that “compliance = good control” and “high A1C number = non-compliance”. (Whatever compliance is). Anyway, I know with my son that we can do everything “right” and end up with a high blood sugar, and other times realize we’ve counted carbs totally wrong and end up with a blood sugar of 100! Was I being too sensitive?
Love your blog.
Everyday diabetics are getting themselves in trouble and going to the ER with the tools we have. Either because of bad decisions or equipment failure or just bad luck. If you view that as your control group then whats the harm of doing some careful testing on willing participants. Diabetics are already making big mistakes.
If we wait for the FDA – Sensors will NEVER be accurate enough, insulin will NEVER work fast enough and algorithms will never be smart enough if we dont get some real world testing started.
The FDA and the JDRF live in a bubble – they dont realize how often the average type 1 takes a wild guess at insulin adjustments. The world is NOT perfect and never will be. We need to make incremental improvements.
I’m willing to help anyway I can. I meet the criteria listed. I understand the risks.
rant:
One thing that definitely benefits from this is JDRF fund raising – its cool to be associated with this kind of technology. Medtronic loves it and the cash is flowing to JDRF.
Thanks for always keeping us up to date, Amy!
I agree with Doug above…if we wait till FDA approves everything we won’t even have Dexcom linked with Omnipod…what is up with that? I am truly disappointed with recent FDA decisions regarding diabetes drugs and device approvals. Is there an agenda here? Seeing as these more justified drugs/devices were put on hold again I see no hope for a closed loop system anytime soon.
The extent to which I would like an Artificial Pancreas to intervene would be if blood sugars hit the high 50s, a suspension so the child/adult does not drop further (which can be overridden if treated) and automatic correction for high blood sugars over 200. For safety’s sake, I would like the correction of the high by the AP to bring her down to a target of 130 or whatever target number is determined safe by the parent or adult. I do not trust the sensor 100 percent, I trust it about 75 to 80 percent. Personally, I would want some wiggle room and would not want the AP to bring her back to her actual target. If the AP can help keep overnight numbers somewhat stable — and I have every confidence that it can — we can do the rest. I would like to be able to either turn the AP feature on at the time I go to bed, however late that may be, or if she is in school and away from home sometimes. I absolutely would not want to turn over all management to the AP. I would like an AP FEATURE on the pump that can be utilized as necessary, still allowing users to control the pump when they want to. Need help at certain times of the night, do not want to turn over management to an AP 24/7.
The FDA should approve the Veo for use in the United States as quickly as possible. It’s use in alleviating nightime lows would be wonderful.
You asked “So is it prudent to wait for the technology to be “perfect”? If it ever will be?” It has been said quite often that perfection should never be the enemy of the good. Voltaire must have been thinking of the Veo.
We want to join the rest of the world in using the Veo. The FDA needs to fast-track this approval NOW.
The automated insulin pump is a very interesting read http://wps.pearsoned.co.uk/ema_uk_he_sommervill_softeng_8/48/12401/3174834.cw/content/index.html
I am extremely sensitive to IOB when exercising to the extent that I eat virtually NO CARBS until AFTER boot camp. Seriously, I do best blood sugar wise under intense exercise if I haven’t eaten any carbs for SIX HOURS prior. So I can’t fathom how an artificial pancreas could ‘plan’ to exercise the same way I do unless it also could deliver glucose.
It may work for largely sedentary folks who don’t have rapid changes in energy expenditure, but alas, I would never give it to an athlete without a 100% manual mode.
Susan
Exercise is only one challenge
Another is the difference between the speed that BG increases when you eat and the time it takes insulin to work. Even if insulin that worked in 5 min was available the CGMS is still 10 min behind so, in the fairy land that is “artificial pancreas” the patient will still have to count carbs and tell it what was eaten and due to the variability of high fat meals have on absorption combo boluses and the other complexities will be around for decades to come…
Unless they can figure out how to stop the auto immune response to the Beta cells without killing the rest of your immune response…
The good news is that through all this cash will still be flowing in truck loads to Minimed and JDRF ..
There is always a risk of technical malfunction and I suppose as with all devices- there will be an increasing role for individuals to monitor, calibrate and maintain their equipment. As with DCCT studies and others, increased involvement and attention to self-management can vastly improve control, A1c’s, etc. Add to that the use of CGM and pumps with proper education and the likelihood is even better control.
My concerns are more with the variable absorption issues with subcutaneous insulin delivery as well as glucose monitoring. I’ve heard many lectures on the “artificial pancreas” and many begin with a historic perspective of the biostator- a system that could monitor glucose levels and deliver insulin to maintain a stable, target blood sugar level. It worked beautifully. The difference was there were two IV lines used to sense glucose and deliver insulin.-And it was the size of a refrigerator.
Subcutaneous measurement of glucose and insulin delivery in response to those levels is fairly crude. If patients should not rely on CGMS readings for adjustments, why is there such a push to push this through FDA? It seems there are inherent conflicts of interest and the families living with diabetes are desperate for solutions.
[...] has had for the adult audience; it is something all diabetics can take advantage of and, if the FDA would get off of its butt this may happen [...]
[...] has had for the adult audience; it is something all diabetics can take advantage of and, if the FDA would get off of its butt this may happen [...]