A Breakthrough In Management Of Type 1 Diabetes?

I want to call your attention to an article published the other day in the  New York Times (February 5, 2010).  The article was entitled “Insulin Dose Automated In a Study” and written by Natasha Singer.  The article summarized a study just published online in the journal Lancet (5 February 2010) written by Hovorka and colleagues and entitled “Manual closed-loop insulin delivery in children and adolescents with type 1 diabetes: a phase 2 randomized crossover trial.”  There was an accompanying editorial comment entitled “Closed-loop insulin delivery: is the holy grail near?” written by Eric Renard.

If you have access to the full article and the editorial comment, I urge you to read them (it will cost you a bundle unless you have access to a medical school library and have privileges to obtain the articles).  In my opinion, the newspaper article did not really convey the gist of the study in a way that was easy to understand.  In reading the newspaper article I got the sense that some major breakthrough had occurred.  When I read the actual scientific report, I was much less impressed that we were on the brink of “the holy grail” as the title of the editorial comment suggested we might be.

What is a closed-loop insulin delivery system?

In people who do not have diabetes, their insulin delivery is within a closed-loop system.  This means that insulin delivery is automatically regulated based on need.  The cells in the pancreas that secrete insulin (the beta cells) are little computers that read the blood glucose level continuously and secrete insulin as needed to maintain  blood glucose levels in the normal range.  It’s the same idea as with a thermostat on the wall.  It reads the ambient temperature continuously and depending on the particular system, can heat or cool the air as needed to maintain the temperature within a very narrow range.  These are closed-loop systems.  In contrast, an open-loop system doesn’t do things automatically to maintain the status quo.  Someone needs to manually “close the loop.”  With respect to room temperature, it would be necessary to check the temperature and manually adjust the temperature up or down as needed. With respect to diabetes, the standard approach uses an open-loop system approach to adjust insulin injection doses or rates of infusion for people who use insulin pumps, based on blood glucose readings and/or anticipated food intake or activity.  Are you with me so far?

How was the study designed?

The investigators studied 19 patients with type 1 diabetes age 5-18 years.  The patients were treated with either standard continuous subcutaneous insulin infusin (OLCSII) or a closed-loop system (CLCSII).  Remember, the OLCSII is an open-loop system in which the patient adjusts the insulin infusion rate and/or the doses with meals based on fingerstick blood glucose testing.  Comparisons were made between OLCSII and CLCSII overnight, after rapidly and slowly absorbed meals, and after exercise.  Patients were masked to both blood glucose and interstitial glucose levels during all 3 protocols; the sensor glucose levels were interstitial glucose readings obtained almost continuously.  Investigators were masked to blood glucose levels but not sensor readings.  During the overnight studies, the sensor glucose readings were fed every 15 minutes to a computer which calculated insulin infusion rates based on an algorithm.  A nurse then “closed the loop” by manually adjusting the insulin infusion rates.  During nights when the patients used OLCSII, the insulin infusion rate was not adjusted based on either fingerstick blood glucose or interstitial glucose readings.  The primary outcome was time during which plasma glucose levels were in the range 70-144 mg/dL or below 71 mg/dL.

What happened?

The results showed no statistically significant differences between OLCSII (21 nights in 17 patients) and CLCSII (33 nights in 17 patients).  Likewise, there were no statisticallysignificant differences between the treatment groups with the meal and exercise protocols.  A secondary analysis of the data in whch the results from the 3 protocols were pooled showed increased time in the target range (60% vs. 40%) for CLCSII and reduced time with glucose levels below the target range (2.1% vs. 4.1%).  These differences were statistically significant.  The investigators reported that “no events” occurred in the CLCSII group when plasma glucose was below the target range, meaning no hypoglycemic symptoms; in the OLCSII group, 9 events were recorded.  The investigators concluded that CLCSII “could reduce the risk of nocturnal hypoglycemia in children and adolescents with diabetes.”

What does it all mean?

First, it was rather surprising to me that this study made such a media “splash” given that the data showed no statistically significant differences between OLCSII and CLCSII for the primary outcomes data analyses.  It was only with the secondary analyses (after the fact so to speak) that statistically significant differences were shown.  Basically, it was only by “data mining” that the investigators showed statistically significant differences between the two arms of the study that favored CLCSII.  What I am trying to say in polite terms is that the data manipulation in the study was not a very good way to show that CLCSII might be better than OLCSII in preventing nocturnal hypoglycemia in children and adolescents using CSII even though CLCSII might really be quite a bit better.  I thought the investigators actually interpreted their data rather conservatively and hinted that CLCSII might just decrease risks of nocturnal hypoglycemia but that further studies were needed.  Mostly, I agree with their conclusions and found the data interesting.  At the same time, I found the New York Times article a bit over the top with its summary of the study and the possible clinical implications.  It’s not for me to say but I wouldn’t have thought a small study with no statistically significant different outcomes for the primary data analyses between “standard care” and and a new approach would have merited a report on the first page of the Business section of the New York Times.  Maybe I have misinterpreted things?  Since the report was placed in the Business section and not the Science section, maybe it’s all about the commerical possibilities of  CLCSII?

What do the data really mean?

I do not want to be viewed as cynical in my criticism of the New York Times report.  Hypoglycemia is a serious matter and anything we can do to decrease diabetic patients’ risks for hypoglycemia should be embraced.  I would agree that the study was a sort of “proof-of-concept” for the use of a continuous glucose monitoring system and a computer to adjust insulin infusion rates to decrease patients’ risks for nocturnal hypoglycemia.  On the other hand, there was no discussion in the medical article or the newspaper article that previous long-term studies with continuous glucose monitoring using interstitial glucose monitoring as in present study has not shown less hypoglycemia (nocturnal or otherwise) in either children or adults with type 1 diabetes and only minimal overall improvement in blood glucose control in adults and none in children (see New Engl J Med 2008;359:1464-76).  In addition, most hypoglycemia in patients with type 1 diabetes, whether they are being treated with CSII or other types of insulin regimens can be prevented by appropriate adjustments in insulin doses based on fingerstick blood glucose testing.  It is just that such adjustments require patients to be highly knowledgeable about their diabetes and compulsively attentive to their self-care.  That’s not so easy.  Maybe while we’re waiting for a cure, it’s a good idea to work hard on temporary fixes such as closed-loop insulin delivery systems?  At the same time we must be very careful to separate hype generated by those commercial interests dreaming of big bucks and scientific truth.  Fear of hypoglycemia must not push us into expensive treatment approaches that are theoretically useful but cannot be shown to be clinically helpful without resorting to secondary data analyses.

Leave a Reply

Your email address will not be published. Required fields are marked *