Wednesday, September 10, 2014

DiaPep277 Development Canceled Due To Alleged Misconduct

This is the first paragraph from yesterday's press release from Hyperion:
Hyperion Therapeutics, Inc. (HPTX) today announced it is terminating development of DiaPep277 for newly diagnosed Type 1 diabetes. The company has uncovered evidence that certain employees of Andromeda Biotech, Ltd., which Hyperion acquired in June 2014, engaged in serious misconduct, including collusion with a third-party biostatistics firm in Israel to improperly receive un-blinded DIA-AID 1 trial data and to use such data in order to manipulate the analyses to obtain a favorable result. Additional evidence indicates that the biostatistics firm and certain Andromeda employees continued the improper practice of sharing and examining un-blinded data from the ongoing DIA-AID 2 trial. All of these acts were concealed from Hyperion and others. The Company has suspended the Andromeda employees known to be involved, is notifying relevant regulatory authorities, and continues to investigate in order to explore its legal options. Hyperion employees were not involved in any of the improper conduct.
A note on terminology:  Hyperion did not use the term "fraud" in describing what happened, although Globe News did, and another news service said "falsified data".  Instead, in their conference call and press release, Hyperion used terms like "serious misconduct and deceit", "collusion", "extensive measures to conceal their wrong doing", "actively and consistently lied", "dishonesty and deceit", "deception was extraordinarily serious", and so on.  Based on all that, I do think that "fraud" is the right term, but it is important to remember that I mean this word in the English dictionary meaning [d1], not the legal meaning.  No one has been convicted of any crime, not even charged, and I doubt anyone ever will be.

The sound track for this posting is here:!/s/Saturday+Night+s+Alright+for+Fighting/2UTbqy
(Note this is The Who's version, because I can hear the words better in it than in Elton John's.)

As usual [d] notes are at the bottom of the posting, and provide more details.

Background on DiaPep277

DiaPep277 is a peptide (a part of a protein).  It is a small part of a naturally occurring protein called "heat shock protein 60".  The hope was that it would cause the immune system to stop attacking beta cells.  Development was done by Andromeda (either as a separate company or a division within another company), and no other company is doing work in this area.  It is one of the potential cures for type-1 diabetes that I have followed from the very beginning of my research.  It had already finished phase-II trials when my daughter was diagnosed in 2003.  I have made more postings on DiaPep277 than any other potential cure, except for Diamyd.  You can read them here:

In 2008 I published a blog based on DiaPep277's earliest data from a phase-III trial and I felt the results were so small it was unlikely to be successful.  You can read that here:
However, in 2011 I published this slightly more upbeat blog:
I continued to follow it until 2013 when I "threw in the towel" stating that the results seen so far were so small that they could not lead to a cure (although I still held out hope they could lead to a new treatment).

What Happened?

As you read my description of what happened, it is important to remember that all my information comes from Hyperion (except for a tiny bit from Evotech), and none of it comes from Andromeda, or any of the specific employees who are alleged to have participated in the dishonesty.  If Andromeda or the people involved publicize their side of the story, I will likely need to update this, based on that new information.

It is normal practice to run two large clinical trials to get the data required by the FDA and the EMEA, and Andromeda had started two: DIA-AID-1 and DIA-AID-2.  They were designed to be twin studies and have 450 people each [d2].  Just last June, DiaPep277 was sold to Hyperion.  This sale included all rights to the new drug, and the transfer of some Andromeda employees who were working on it.  At that point DIA-AID-1 was complete and had been published, but DIA-AID-2 was not quite finished.  The completion date is early 2015.  So the Hyperion statistics team were evaluating the entire DIA-AID-1 data set, as a sort of practice run to get ready to analyse the DIA-AID-2 data, when it was ready.

You can read the DIA-AID-1 results in this paper:
Thanks very much to Diabetes Care, published by the ADA, for making the whole paper available on line.

When the Hyperion statisticians looked at the full data set for DIA-AID-1, they noticed something very odd.  If they analyzed the entire data set, then DiaPep277 did not have a statistically significant good effect in the primary outcome measurement.  The clinical trial had failed.  However, Andromeda had excluded 30+ patients from the analysis because they had violated the rules about who should be signed up [d3].  With those exclusions, the data showed a statistically significant good effect in the primary outcome [d4].  The study had succeeded. That's unusual, because the exclusions are supposed to be made "blind" (not knowing if the drug worked for those people, or even if they got the drug), and excluding people randomly from a trial, should not change the outcome.  These exclusions were done by an outside company which was involved in the clinical trial, and that company was not supposed to know who got the drug and who got the placebo.

Except Hyperion's investigation found (according to Hyperion) that some of the Andromeda employees passed data to the outside company, and people at that company used that data to selectively remove patients from the study to bias the results.  Also, Andromeda employees changed the primary end point of the study [d5], so that it would be successful. In both cases, the decision was made "unblinded" (i.e. knowing who received DiaPep277, and who received placebo), when the decision should have been made "blind". This completely undercuts the results of the clinical trial.  Hyperion said that this did happen in the DIA-AID-1 trials results, and was also in process of happening in the DIA-AID-2 trial [d6]

Although Hyperion was careful not to name the company that did the statistical analysis for Andromeda (they always referred to it as an Israeli Biostatistics company), nor did they name any of the people involved.   However, on page 1399 of the DIA-AID-1 paper, there is discussion of what companies did statistical analysis as well as describing what each author did in running the study and writing the paper.  (In the future, I'll be checking to see if this company or these researchers are involved in any research I report on.)


Hyperion has said that there is no way to move forward with regulatory approval for DiaPep277, and they will not attempt it.  So DiaPep277 is dead.  That's the short term impact.  I suppose they could try to sell it to someone else, but who would want to buy it now?

The medium-term impact has three questions:

1. Will the paper describing the results from the DIA-AID-1 trial be retracted?  It was published in Diabetes Care (a journal of the American Diabetes Association), so it will be interesting if the authors retract it, or if the editors/publishers retract it [d7].

2. Will there be a civil lawsuit?  Will anyone face a criminal charge?  Remember that Hyperion paid tens of millions of dollars for DiaPep277 based largely on results which were invalid.  Andromeda and the nameless Biostatistics company are Israeli, while Hyperion is American, and I'm sure that will complicate both civil and criminal legal matters.

3. In addition to the results paper, Andromeda employees also published a research paper comparing the primary end points (new and old) of their study.  If Hyperion's can show that this data was manipulated, then this study should be retracted as well.

The abstract of the paper is here:
and says specifically that the findings were "unexpected", which Hyperion claims is untrue.

The long term impact is less predictable, but could be much wider.  How many other studies used the same biostatistics company?  Some of the researchers involved in this study are very big names, and have published many other papers, and worked with other companies, including some companies doing clinical trials.

Some Personal Notes

I gave up on DiaPep277 long ago, so in that sense, it was dead to me even before this came up.  But it is still deeply shocking.  (A statistician that looked at this situation described it as "staggering".)  At the end of the day, new drug safety and effectiveness are supposed to be shown via scientific testing. There is a lot of good statistics used to show that results are meaningful, and not due to chance, accident or mistake.  But all those statistics assume that the people running the study are not liars or cheats.  Detecting people who are willing to commit serious misconduct in their scientific studies is not easy.  Implementing the procedures necessary to detect active deceit in all scientific studies would be horribly expensive.  Currently, the FDA uses statistical methods, to find "honest" mistakes, rather than do the sort of investigations and surveillance required to find premeditated fraud.  That's part of the reason why I think it's important to the scientific process to see what consequences the people and corporations face in this case.   Because if other people and other corporations see that they don't face huge consequences, then they will be more willing to risk the same kind of misconduct that is alleged here.

In general, this blog does not report on financial transactions.   That is specifically because of Andromeda and DiaPep277.  Whenever a company buys a new drug, there are always very positive press releases, and early on I thought about reporting those in the blog.  But I noticed that DiaPep277, in particular, was getting passed around to a lot of different companies, and for no good reason that I could see.  I don't remember now all the moves (this was years ago), but I'm pretty sure that Andromeda sold it, got it back, sold it to someone else, and got it back again.  All the companies were Israeli, and as I remember it, some of them were partial owners of each other [d8].  I did not understand it.  That's a lot of moving around in a small market.  It convinced me not to report on company-to-company movement in my blog; that it didn't mean anything, or at least I didn't know the meaning.  But now, in retrospect, I wonder if it was a sign of trouble.  

Finally, I want to personally urge Hyperion to make public the researchers involved, and the evidence about each one's involvement in this alleged misconduct.  There are 21 named authors of the DIA-AID-1 study, and 6 named authors of the change-of-primary-endpoint paper (with much overlap).   Right now, all of these researchers are "under a cloud"  but it is likely that many of them did nothing wrong.  Maybe none of them did anything wrong, and the misconduct was done by others, or never happened at all.  But in any case, the whole type-1 world deserves to know who did what, and the supporting evidence.

Extra Discussion

[d1] For example (from "deceit, trickery, sharp practice, or breach of confidence, perpetrated for profit or to gain some unfair or dishonest advantage."

[d2] This is a quirk of the American approval process.  It is law that there must be two studies to confirm the results.  Therefore, you can not run one 600 person study, you must run two 300 people studies, even if they are otherwise identical.

[d3] It is common for some patients enrolled in a clinical trial to be excluded from the results for a variety of reasons.  The most common is that someone drops out, and complete data for that person is not available.  However, people are sometimes enrolled by mistake in violation of the study's rules. For example, a trial might require first treatment within 100 days of diagnosis, but a review of paperwork, after the study completes might determine that the patient signed the paperwork within 100 days, but didn't actually get the drug until later.  Data for that patient would be (properly) dropped from the study results.

[d4] If you look at the patient flow diagram on page 1394 of the paper, you can see that a total of 34 patients were excluded from analysis.  It is the two boxes just below the "allocated" boxes, and these changes impacted both the "mITT" data and the "PP" data.  The claim Hyperion made was that these exclusions were made "unblinded" and were specifically tailored to bias the results.

[d5] The primary end point is the most important result of a clinical trial.  Usually, there is one primary end point, and several (less important) secondary end points.  If the primary end point shows a statistically significant good effect, then the trial is successful.  If this is not seen, then the trial is unsuccessful.  The FDA generally determines effectiveness of new drugs based on primary end points.  So therefore, changing the primary end point of a study, in the middle of the study is very unusual, and if it was done "unblinded" (ie. with knowledge that the current primary end point is failing, or the new one succeeding) that is scientific fraud (in my opinion).  Its the moral equivalent of moving the goal posts to the ball's location, rather than putting the ball through the (stationary) goal posts.

In the case of DIA-AID-1, both the original primary end point, and the new primary end point involved measuring C-peptide, but the two end points measured it in different ways.  The details are described in the comparison paper.  Abstract is here:

Remember that although Hyperion has said that this happened in the DIA-AID-1 study, I have not seen their supporting evidence, and so have no way of knowing if this is correct or not.

[d6] As part of this research the DIA-AID-2 data collected to date was unblinded, and Hyperion said that there is almost no chance that it would end up showing a good effect in the primary end point.  (Remember that most of the DIA-AID-2 data has already been collected, even as they are still waiting for the last few patients to get the last few data points.)

[d7] To be blunt, the paper contains the following two quotes (pages 1393 and 1394), which Hyperion now claims are untrue:  "Participants, investigator site staff, persons performing the assessments, and data analysts were blinded to patient allocation from the time of randomization until database lock" and "The study protocol was amended, and the statistical analysis plan was planned and finalized before the study was unblinded, with the GST clearly defined as the primary end point."

[d8] Even now, I don't have a complete list, but here is a partial list of companies involved in DiaPep277 development: Andromeda, Clal Biotechnologies, Teva, Peptor, DeveloGen, and Evotek.

Joshua Levy 
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Thursday, September 4, 2014

ADA 2014: Other Topics, Pediatric Approval and Big Data

This blog posting covers two topics which I found interesting at ADA's Scientific Sessions 2014.  It contains much more of my personal opinions than a regular post.

Barriers to Pediatric Treatment Approval

This was an ADA session on getting drugs/treatments/devices approved for pediatric use.  It focused on one topic: why is it so hard to get pediatric approval for diabetes treatments (both in type-2 and type-1)?  The entire discussion panel took the position that getting pediatric approval for new treatments was too long/slow/difficult, and so there was no discussion of safety trade offs, or the that the extra work of pediatric approval led to commensurate increased safety.  Everyone just assumed that it didn't.

The answer to the question of why pediatric approvals were so hard was basically this: The FDA requires testing on adults before testing on children (part of the Helsinki protocol on medical testing).   So treatments get approved on adults.  But the FDA requires separate testing on children.  So approval for children comes years after approval for adults.  You don't want to start your pediatric testing before you know that the treatment is going to be approved for adults, so there will always be a gap of many years.  However, doctors can prescribe a treatment for anyone, once it is approved.  So if the treatment looks safe for kids, doctors will prescribe it, once it is approved for adults.  Many parents will encourage this, by clamoring for the newest treatments available.   Now the drug companies think to themselves, why bother testing on kids?  We're already getting profits from kid's prescriptions, and that's only going to increase.  So there is little motivation for them to do pediatric testing.  Of course insurance companies try to say "we won't pay for it for kids, because it hasn't been approved for kids", and sometimes they succeed and sometimes not.  At the end of the day, that just limits the treatment to more wealthy families, which creates another problem; newer pediatric treatments are limited to the wealthy, even for people who have insurance.

All of this leads to the worst possible outcome, from a patient's point of view: drug companies don't test in children, and doctors routinely prescribe for children.  We have use without testing.   Officially, the FDA can wag it's finger at both industry (for not testing) and doctors (for prescribing), but the bottom line is that it's the FDA's policies that promote this sad state of affairs.  (And the FDA can blame the Helsinki protocols for the problem, when the problem might really be their simple minded implementation of those protocols.)

In Europe, they have shifted towards requiring a plan for pediatric testing as part of the adult approval process, and the FDA is considering a similar change.  The idea, is that since some doctors will prescribe a treatment for children once it is approved for adults, the regulator agency should require some testing (or at least planning for testing) on children even as part of adult approval.

Another idea, promoted by researchers and industry, is to share control groups.  Right now, if you want to test a treatment, you need a treated group and a control group.  That often means you must recruit twice as many people than actually get the drug.  For example 150 people get the drug, and another 150 go through the study, but never get the drug.  To test a second drug you need to recruit another 300 people, and so on.  The idea here is to test several drugs at once, all sharing the same control group, so maybe recruiting 600 people to test 3 new drugs, instead of the 900 people it would take now.

Discussion (My Opinions)

I don't have a easy solution to this problem, but I think there are a few obvious things to consider:
  1. Although children are not identical to adults, we are all the same species (well, maybe not teenagers, but everyone else :-) and the current FDA policies that require testing on adults and then testing on children need to be relaxed to take into account both the similarities between adults and children, and the differences.
  2. Many devices are likely to work very similarly in children as in adults, and such devices should have extremely simplified child testing requirements.  In some cases having combined phase-III trials and approval.  (How can anyone claim that a BG meter is going to be different for children than for adults.  Even for a pump or CGM, the differences are likely to be tiny or nothing because the physiology that they are based on in the same.)
  3. On the other hand, drugs which impact growth or hormones are likely to have a different effect on children, and at the very least should follow the same adults first policy we have now.
  4. In between, there are lots of drugs which are likely to work very similarly for adults and children, and these could undergo overlapping (but not completely combined) testing, where some testing was done on adults first, but abbreviated testing can be done on children.
  5. In general, I think the EU's solution (more work up front) is not a good one, because it slows down approval by creating more hoops to jump through.  Also, if there was child-only problem with the drug, it might still block adult approval, and I don't think that is a good thing.
  6. Also, I think allowing doctors to prescribe "off label", but having many insurance companies not pay for "off label" use creates some bad effects as well.  But the question of insurance is much too big a "can of worms" to discuss here.
Two side-discussions on childhood type-2 diabetes:
1. In general, ADA 2014 sessions labeled "pediatric" were about half type-1 diabetes, and about half type-2 diabetes.  In itself, this shows how fast childhood type-2 diabetes is growing, since even 15 years ago, pediatric diabetes was almost a synonym for type-1 diabetes.
2.  I was really shocked by how bad the outcomes were for people diagnosed with type-2 diabetes in childhood.  In type-1 we are used to serious side effects that happen decades down the line, and can be minimized with good control during all that time.   But that's not the reality of type-2 diabetes in childhood.   Very serious side effects can happen during childhood.  They are hit with bad complications much earlier and much more commonly than people with type-1 diabetes.

Big Data

The term "big data" refers to using huge amounts of data to answer questions that were not even considered when the data was collected.  A "classic" data base task might be to record all the books a person buys, so that you can see what authors they like.  A "big data" task might be to record every book a person views while shopping, and every book they discuss on-line, and how quickly they read each page of each book they own, in order to answer questions about what they like, why they like it, and what they do based on their likes, etc.

There was an entire session on "Big Data" at ADA 2014. Although I don't work in Big Data specifically, I am a software engineer, and I do understand the topic.  It was interesting to hear how medical researchers view big data, and also interesting that none of the papers in this session would be considered "big data" by software engineers.

Monitoring Data from Doctor Office Visits

Two of the talks focused on what I would call "more data" (but no where near "big data").  These guys were talking about integrating more medical data from more sources.  But the amounts of data they were talking about was so small that they would not qualify as "big data" for anyone in the industry. (Indeed, the data was so small, it would easily fit "in memory" for a mid-sized virtual machine at my work site.)

One talk focused on scraping information from medical records and aggregating it.  Basically, they installed a server at a 100 or so doctors' offices ("medical practices") that used electronic medical records.  Every night, the server software would look at the newly updated records, and pull useful information and send it to a central server for analysis.  No identifying information was sent, so all data was anonymous and there were no privacy issues.

This data could be used to get an early warning of a flu outbreak or a rare side effect in an approved drug or an unusual drug interaction.  I very much hope that this can be used as a "safety blanket" to encourage more streamlined drug approvals, followed by more rigorous real use surveillance.  I think this combination can lead to the win-win of faster approvals and safer drugs.

In a real world (although small) application on this idea, the researchers looked at all problems reported by type-2 diabetics.  They noticed that many people who took two specific drugs at the same time had complaints about high BG numbers.  Now each of these drugs were supposed to lower BG numbers. Both had been extensively tested and found safe and effective in lowering BG numbers.  But by looking through 1000s of medical records, they found over 100 people who happened to take both, and they often had complaints (also in the their medical records) of high BG numbers.  The two drugs had never been systematically tested together.  The researchers gave both drugs to mice, and saw the mice BG numbers go up.   So the statistical discovery was confirmed in animals.  (Since it was a bad side effect, confirmation in animals was enough, you don't need to test something like that in people.)

A factoid from another talk at ADA 2014: In the US, 76% of people over 60 are taking more than one prescription drug.  And I'm sure the number is much higher for people diagnosed with a chronic disease like diabetes.  Yet drugs are often not tested together; indeed, people taking other drugs (especially for the same disease) are often specifically excluded from clinical trials to avoid uncertainty as to which drug is having what effect.

Recruiting and Running Clinical Trials on a Social Network

Another talk focused on using members of type-1 diabetes on-line groups as recruitment pools for studies, and making study participation much more like social networking.  I would view this as "crowd sourcing" clinical trials.  Therefore, it would be more natural to people who grew up updating Facebook status and sharing pictures.  Presumably such people are comfortable sharing their A1c, which drugs they take, and complications they experience.  These researchers have published some studies based on data from members of the TuDiabetes on-line community.

The researchers were generally worried about such things as "informed consent" to participate in a research study (and ethics in general), and also the quality of the data (especially self selection of the participants).   They did notice that early adopters tended to have better A1c numbers than expected, which suggested to me that they were "skimming" the people with good control, rather than a representative sample.

Discussion  (My Opinions)

Personally, for the "trials via social networks" idea, I'm more worried about deliberate manipulation, and I asked the speaker about this issue.  Not in type-1 diabetes, but in some other diseases, there are organized patient groups that have very strong points of view about their disease, and have actively tried to manipulate scientific research to agree with their views. (See discussion below.)  The researcher I questioned hoped that by using reputable groups (in this case TuDiabetes) they could minimize manipulation, and that statistical analytics might detect or prevent it.

I think that is wishful thinking because people can organize on one forum and then move to another to implement their manipulation, and also because the kinds of statistics usually used are not designed to detect or prevent deliberate, organized manipulation.  At the very least, we would need new kinds of statistics and new kinds of surveillance to protect these studies.

Details about Manipulation

For example, already in vaccine, abortion, and chronic fatigue syndrome (CFS) research, I have seen organized attempts to bias research by selectively submitting reports of side effects, boycotting research that might show something they don't believe, influencing participants in the research, etc.

Currently, the most common form of manipulation is creating spurious VAERS reports.  VAERS is a reporting system for vaccine side effects.  However, any medical professional can submit anything they want into the system.  So anti-vaccine groups run organized campaigns to ask doctors and nurses they work with to submit "side effects" that they claim are caused by vaccines.  Anti-abortion groups do the same for those vaccines which were developed using cell lines from aborted tissue.  Conservative religious groups do it for vaccines targeting sexually transmitted diseases.

More advanced forms of manipulation are also already in use.  Some chronic fatigue syndrome patients have started campaigns to actively discourage fellow patients from signing up for studies that might disprove their pet theories, boycott all studies by researchers who have previously published results they don't like, and even sabotage studies by getting patients to drop out of studies they have already started if those studies might come to a conclusion they don't agree with.  Both CFS groups and anti-abortion groups have organized mass ethics complaints against researchers whose work they disapprove of.

Research done via social networks would be even more open to similar organized attacks.

Joshua Levy 
publicjoshualevy at gmail dot com 
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Wednesday, August 20, 2014

ADA 2014: Scientific Study Shows CarbDM Is Good For You!

As Dave Berry used to say "I'm not making this up!".  There was a study presented at ADA 2014 where the researchers offered college aged (18-25) people with type-1 diabetes a weekly "diabetes club" where they could interact with others who had type-1, ask questions, and hear 15 minute talks on topics of interest.   (Sound familiar?)  The people who attended these meetings lowered their A1c numbers by about 1.2, which was statistically significant compared to the people who did not attend.  As a general rule, the FDA considers drugs that lower A1c by 0.5 to be effective, so this result would be more than enough to get a drug approved, if this treatment had been a drug.

Here is their summary:
The mean HbA1c level was reduced at 3 months after the start of the program (from 8.8% to 7.6%), but even more pleasing was that also the long-term follow-up, after one year, confirmed the sustainability of the results (with mean HbA1c level 7.4%), ... The frequency of severe and mild hypoglycemia was also reduced
Obviously, in spite of the headline, this is not CarbDM specific. It shows the importance of peer support groups for type-1 diabetes.  I don't think anyone on Bravebuddies would doubt this, but it's good to see it supported in the scientific literature.

Source: Abstract 2312-PO

Although I focused on this one abstract, the concept of peer support, and different kinds of support groups, was a very popular one at ADA 2014.  Several of these studies focused on using on-line support groups of various kinds.  My memory is that there were 10+ abstracts on related topics, including one or two from Stanford University Hospital.

Joshua Levy 
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Thursday, August 7, 2014

ADA 2014: Type-1 Research and Information

This posting covers things I learned at the 2014 ADA Scientific Sessions, about type-1 diabetes, that were not cure related.

Transplants for Type-1 Diabetes

There was a paper at the prestigious "President's Session" on the results of a large, recent phase-III trial of pancreatic transplantation.  These were "classic" transplants (these patients were on immune suppression for the rest of their lives) so this is not a cure by my definition.  However, the results were impressive:
  • 50% used no insulin 1 year after treatment.
  • The rest used very little insulin (average insulin usage at 1 year was less than 0.5 units / kg / day)
  • Even after one year, C-peptide production was improving, so it is likely that these results would continue for some time.  (These patients will be followed to confirm that.)
While I continue to think that long term, whole body immune suppression is too toxic to consider a cure, I know that some people consider this path, and I think if you are going to consider it, you should have the most up to date data.

Accuracy of Diabetes Alert Dogs (DADs)

I've always wondered how accurate diabetes alert dogs were.  Especially since they are trained by many different organizations all over the country, with very little standardization or regulation.  It turns out, I'm not the only person who wondered about it.  There was a research poster on this subject by Shepard (et al.) from Charlottesville, VA, USA.  They collected data from diaries kept by 18 diabetes dog owners.  All dogs came from the same (unnamed) organization.

Overall the hit rate (an accurate alert) was between 50% and 65% (combining hit rate for low BGs and high BGs).  There was a lot of variation.  For example, one dog had a 100% hit rate for detecting lows, but another dog had 33%.  The minimum-maximum for dog detection of highs ranged from about 30% to about 77%.

I remember there being more information on the poster (I'm writing this off the abstract of the poster), but we are not allowed to photograph the poster.  I hope these guys publish the whole report.  I suspect they have a lot more data to share.  Poster was 73-LB.

There was also a presentation (874-P) on Diabetic Alert Dogs in general.  Here were their main conclusions about the industry (quoted from the abstract).  I've added some of my own opinions in square brackets; kindly remember that I know nothing about training dogs.  The term SD means Standard Deviation, which is a measure of how "spread out" the data is; higher numbers are more spread out:
  • Roughly half of the groups surveyed were designated non-profit organizations while the remainder were for-profit. 
  • Length of time organizations had been training DADs ranged from 1 to 13 years, with a mean of 4.6 years (SD = 3.3). 
  • The number of trainers in groups varied, ranging from 1 to 8. Number of DADs currently in training ranged from 2 to 175 across groups. Number of DADs placed per year also varied greatly, ranging from 1 to more than 100. 
  • The cost for DADs ranged from $1,000 to $30,000 or more, with a mean of $12,429.00 (SD = $8,121.10). 
  • Some but not all non-profit organizations assist prospective DAD owners in fund-raising efforts to cover costs. [I think some of the for-profits do as well.]
  • Dog age at placement in homes ranged from 2 months to 2 years (mean age = 19.1 months, SD = 6.2).  [Isn't 2 months barely over weaning age?  How can a dog be trained so young?]
  • There appears to be consistency in DAD training procedures, with all but one group using saliva samples. [I don't think this represents much consistency, as they could be using saliva in completely different ways.  I don't think the exact body fluid used in training is particularly important compared to other variables in the training process.]
And I just saw this TV news article on problems with diabetic dogs:

Dulaglutide: Once a Week Insulin

Dulaglutide is a new insulin which is being tested in type-2 diabetics for once a week bolus use (as a replacement for daily doses of Lantus or Levemuir).  There were at least 9 abstracts on this.  I don't think any of them included type-1s, but several of them were phase-III trials (including hundreds of people), and at least one trial ran for a year.  That suggests to me that this insulin is not far away from approval.

Sources:  962-P, 964-P, 979-P, 980-P, 981-P, 962-P, 110-LB, 122-LB, 330-OR

There was another weekly insulin, called HM12470, which is being tested in animals, and was reported on in only one abstract: 89-LB.

Stem Cells

There were over 42 abstracts on stem cells.  They covered many different areas, including both animal and human research, and many different types of stem cells.  A lot of the abstracts reported on progress in creating stem cells.  I skimmed through the abstracts, but found nothing particularly interesting for people with type-1 diabetes.

If you had type-2 diabetes, the most interesting paper might have been 123-LB.  These researchers tested giving patients stem cells from their own bone marrow.  It was a phase-I pilot study, but included 61 people and was placebo controlled.  Results focused on safety, which was fine, but they did have some effectiveness data, and that was good as well: A1c dropped from 0.2 to 0.5 depending on time and dose.  That's a pretty small result, but not bad for a phase-I trial aimed more at safety than effectiveness.  My understanding is that the stem cell treatment they used was the oldest, best understood, safest, and easiest stem cell treatment around.  So if it does turn out to be effective, it will be relatively cheap and widely available.

Type-1 and Cancer

It is well known that people with type-2 have a higher chance of cancer than others, but what about people with type-1 diabetes?  If the higher cancer rate was caused by high BG numbers, then one would expect similar higher cancer rates in type-1s.  One poster reported on cancer rates on people with type-1 diabetes from four countries that have diabetes registries (Australia, Denmark, Finland, and Sweden).

Overall there was no statistically significant rise in cancer in type-1s as compared to the general population for men, and a really tiny effect in women. (The increased risk was 5%; for this kind of study, that is very small.)  I think type-1 diabetics should take a lot of comfort in this.   However, there were five cancers with significantly higher risk ratios in type-1s: colorectal cancer, liver cancer, pancreatic cancer, kidney cancer, and thyroid cancer.  I assume that there were also cancers that type-1s had a lower incidence of (as compared to the general population), so that the overall cancer rate was about the same.  But the researchers did not specifically list those.

This was poster 174-LB.

New Encapsulation Technology

A group in Italy has developed a new encapsulation technology, which is constructed layer by layer. They reported on the physical parameters of the new coating and success in mice.  It looked interesting to me.  I asked specifically about comparing the physical properties of this coating to the physical properties of LCT's coating or the IsletSheet coating, but they had not done any comparisons.  (They seemed to have a research mindset, rather than a competitive product mindset, which makes sense, since they are university researchers.)

Eye Targeted Transplantation

The immune system does not have access to the human eye in the same way it has access to other organs of the body.  This is called "immune privilege" and it means that things in the eye are not attacked as aggressively by the immune system.  For several years, the Diabetes Research Institute (DRI) has been exploring the idea of doing islet cell transplantation into the eye, so that the new cells would not be attacked by the immune system nor would they be attacked by the autoimmunity that triggers type-1 diabetes.  The researchers also liked the eye as a location for transplant, because it is easier to monitor.  If you put the cells near the pancreas (for example) you can't see them.  But if you put them in the eye, then you can just look in and see them.  So you can see if something is going wrong.

This presentation discussed progress in this area.  Unfortunately, it turns out that the eye is not completely free of immune response.  Early on there was the hope that the immune system was completely blocked from the eye, and transplanted cells would be ignored.  The immune attack in the eye is much lower than in other parts of the body, but it does still exist.  Studies have been done in mice and baboons.

I asked the obvious question: does it affect vision?  The answer is probably not, but more testing is definitely needed.  Mice don't see well, anyway, and tend to rely on hearing and smell, so it's not obvious if their sight is affected.  Baboons are much harder to work with, and these researchers did not systematically measure eyesight in either mice or baboons.

This was paper 437-OP and/or 348-OP.

SGLT2 targeted treatments

There were a couple of talks and a couple of posters on SGLT2 targeted treatments (mostly in the context of treating type-2 diabetes).  However, this research might lead to improved treatments for type-1 diabetes (although I doubt a cure).  This drug class works by causing kidneys to pee out more sugar, so it tends to lower post meal BG spikes.  This sounds helpful to me.  In addition, there was one paper (in the President's Session) that implied that SGLT2 was also involved in Glucose generation, such that the same SGLT2 drugs might also minimize lows.  If true, that would be doubly helpful: the same drug could lower the highest spikes and raise the lowest troughs of BG.

There was one session on SGLT2 specifically, and it had about 13 presentations, plus there were about 12 presentations on SGLT2 in other sessions in the conference.  That's a lot of coverage.

Joshua Levy 
publicjoshualevy at gmail dot com 
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Monday, July 14, 2014

ADA 2014: Type-1 Diabetes Cure Research, Immunology

This posting discusses two treatments which have the potential to cure type-1 diabetes, which have been tested on people, and which were reported on at the ADA's 2014 Scientific Sessions.  They were both tested in honeymoon diabetics (as I define "honeymoon").

The soundtrack for this posting (in honor of the Ramones,  RIP):!/s/What+A+Wonderful+World/1XjvmQ?src=5

Expanded Polyclonal Tregs

There was a presentation of results from a clinical trial on Expanded Polyclonal Tregs, which I've blogged about before:
Remember that this line of research is being pursued by two teams: the UCSF team (led by Dr. Gitelman) which is reported on here, and a team at Medical University of GdaƄsk lead by Dr. Trzonkowski.

A quick summary of this treatment is as follows: remove one specific type of T regulator cell (called "CD4(+)CD25(+)CD127(lo)") from a person with type-1 diabetes.   Grow them out so you have about 500 times more, and then put them back in the body.  Since regulatory T cells naturally regulate the body's immune system, the hope is that they will prevent the autoimmune attack which causes type-1 diabetes.

The UCSF team ran a phase-I clinical trial with 14 people.  There was no placebo group and the patients had type-1 for between 3 and 24 months.  The basic results were that after two years, these patients continued to generate C-peptide at the same rate as when they started the trial.  There was no drop off in C-peptide production.  That means there was no drop off in insulin production.  Since all these people were already diagnosed with type-1 diabetes, they were not generating much insulin, however people with type-1 generally generate less and less insulin over time.  So these patients did better than would be expected of an untreated group.   (Although as a pilot study, there was not an untreated group here.)

Source: ADA 2014 Presentation 174-OR.


There was presentation of results from a clinical trial on Antithymocyte Globulin (also called Thymoglobulin or ATG) and Granulocyte Colony Stimulating Factor (GCSF).  I've blogged on this trial before:

The basic idea behind this research is that ATG modulates the body's immune system, while GCSG causes the body to generate more T-cells directly from it's own bone marrow.  So this therapy is a combo therapy with one treatment to stop the autoimmune attack, and another to restore beta cells.

This study had 17 patients who got the treatment, and a placebo group of 8 who did not.  People had type-1 diabetes for 4-24 months when they received the treatment.  Basically, the 8 untreated people lost C-peptide production (which means they lost insulin production), just as you would expect.  The 17 treated patients ended up, after one year, at about the same C-peptide level from where they started. So they did significantly better than the untreated group.

This news article covers this research as well:
But note that this story has some phrases like this "there was an increase in the insulin-producing beta cells in the pancreas" which is overselling, in my opinion.  This treatment preserved beta cell levels; I don't see evidence that it increased them.

The most interesting quote in this story is the following forward-looking view from the researcher:
[Dr.] Haller said the eventual goal, years down the road, is developing a therapy that first uses an IV infusion of Thymoglobulin and then a Neulasta [trade name of GCSF] treatment once every two weeks for three months to greatly reduce or eliminate the need for some Type 1 diabetes patients to take insulin injections.
Another interesting point, is that both ATG and GCSF are already FDA approved for other uses.  This makes them easier to use in clinical trials, and means they could be used "off label" for type-1 diabetes, if prescribed by a physician.

Source ADA 2014 Presentation 173-OR.


First, these two studies highlight the lack of standardization in terminology used to describe "honeymoon" type-1 diabetes.  The first study enrolled people 3-24 months after diagnosis, and used the term "recent onset" (which I interpret as a more scientific way of saying "honeymoon").  The second study enrolled people 4-24 months after diagnosis, and used the term "established" (which I interpret to mean "non-honeymoon").

For my part, I'm considering both of these clinical trials to be "honeymoon" tests, because they included people who had been diagnosed for less than a year.  That's the dividing line I've used informally in the past, and I'm going to continue to use it, until I see a better definition.

Second, I view both of these results as honeymoon style results.  If they gave these treatments to people with long established type-1 diabetes, one would expect no improvement because the group would start out with no C-peptide production. On the other hand, these treatments have real benefits to recently diagnosed type-1s (who still have some insulin production), and could be even more beneficial to people who are losing insulin production, but have not yet been diagnosed. This is the hallmark of a honeymoon style result to me: it is highly dependant on how long a person has had type-1 diabetes.

Joshua Levy 
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Sunday, July 6, 2014

ADA 2014: Type-1 Diabetes Cure Research, Artificial Pancreas

At the 2014 ADA Scientific Sessions, there were several reports on progress on artificial pancreas (sometimes called "closed loop").  Unfortunately, all of them were reported on a day that I was not at the convention, so the information below is mostly from the printed materials at the convention, news reports, and convention "buzz".

The "Bionic" Pancreas:
Bihormonal, Closed Loop, Artificial Pancreas Progress

This was clearly the big news of the scientific meeting.  Here is my previous coverage on this (and it includes links to DiaTribe's more complete coverage):

Bihormal refers to supplying both insulin and glucagon (so it can raise or lower a person's blood glucose).  Closed loop artificial pancreas refers to automatic dosing as needed with data from a CGM to a pump without human intervention.  Bionic is a marketing name used by Dr. Damiano's group at Boston University.

There were two big publications on the Bionic AP.  The first was in a scientific journal, published the month before the show, and the second was a presentation at the show.

First, I'l discuss the study published just before the show.  The basic set up was that people wore the devices for one day of calibration, and then two days of data collection.   Data was collected for four groups: adults and adolescents, and people who signaled when they were going to eat a meal, and those that didn't.  No one counted carbs or dosed in response to meals.  The signaling group just told the AP that they were about to eat a breakfast, lunch or dinner; nothing about the content.

  Average BG  
  Estimated A1c  
  % in range  
Adults Before Treatment7.3
Adults who signaled meals1326.280
Adults without meal signaling1426.770
Adolescents Before Treatment7.9
Adolescents who signaled meals1627.368
Adolescents without meal signaling1757.760

What this means, is that for adults who did not signal when they were going to eat, they had an average BG level of 142, a likely A1c level of 6.7 (if they had done this for 3 months), and their BG levels were in range 70% of the time!  Now, that looks pretty good, but the news gets better.

Here are the results from the follow up study, done by the same researchers, and given as a scientific talk.  This study was "free range" adults who were free to roam over 3 square miles of Boston, staying in a hotel, working out at a gym, and eating mostly at restaurants, while the adolescents were attending camp.  For this study, no one signaled meals.  It included 20 adults and 32 adolescents, which makes it phase-II sized by my reckoning.

This study has only two data points that matter:

Average BG Number
  (for both adults and adolescents)  
  Estimated A1c  

There was slight complexity in the data.  That 138 number was the average over all five days of the test.  The researchers expected that the first day would be worse than the other four, because the unit was calibrating itself to the patient the most during that first day.  For adults, this worked out, the next four days average BG was 133 suggesting that long term use would result in an even lower number, and might even drop a few more points (over time, as the AP better learned how the person reacted to insulin, glucagon, and food).  But for adolescents, that's not what happened.  They averaged 147 over days 2-5.  Even if 147 (A1c of 6.7) is the long term number, that is still a complete success.   It is lower than the ADA standard of 7.5 for adolescents.  But it is a mystery to me why those days should average higher than the first day.

Summary of NEJM data:

Note: information for this section came from an ADA abstract, a JCEM paper, and a NEJM abstract.   You can read the whole NEJM article here:
JCEM abstract here:

Single Hormone vs Bihormonal Artificial Pancreas
A group from Canada gave a talk where they directly compared injected insulin, an insulin AP, and an insulin and glucagon AP.  For average BG numbers, they found that both types of APs were similar to each other (the dual pumps were only very slightly better), and that they were both significantly better than injections.  However, when they looked at low BG events, then the dual hormone APs had significantly fewer such events than single hormone APs.  This makes sense, since the dual hormone pumps can directly prevent lows by dosing glucagon.

So this Canadian trial suggests that a bihormonal AP might do a little better than a "classic" AP, but it should not do vastly better, if measured by average BG.  When I first saw that poster, I was a little dubious.  Two hormones seemed like much better technology than one.  But then I saw the results below.  One of the complexities, is how does one measure an AP?  Using average BG is easy and straightforward, but should we also measure low BG events and/or high BG events?  If you do (especially low BG events), then the dual hormone APs might look better in comparison.

The Cambridge Artificial Pancreas

With all the excitement about the bihormonal AP, it is important to remember that there are also several "classic" AP projects out there.  For example, the results from the Cambridge AP, a "classic" insulin-only AP, were almost as good as the bihormonal results.  There were something like 7 presentations on various aspects of this project, so it was very well represented.

The "24 hours a day" trial included 17 people, and ran for 16 days (8 days with AP and 8 days with regular treatment).  They also reported on a nighttime only trial, which ran for 90 days!  Again, half with AP and half with regular treatment.

It's big results that matter, from the 24 hour and day trial, are:

Average BG Number
  (for both adults and adolescents)  
  Estimated A1c  

MD-Logic Artificial Pancreas Project

What's better than two closed loop, artificial pancreas projects?  Three!  The MD-Logic project uses a "fuzzy logic theory algorithm" to predict insulin dosing.  The research group presented a poster, which showed that using the MD-Logic AP at night, improved BG numbers the next day.  This clinical trial included 24 people and lasted for 3 months (6 weeks using the AP, 6 weeks not, for comparison).

People who used the AP woke up about 15 points lower (on average) than people who did not use it. Looking at all the BG numbers the next day, people who used the AP the night before had an improvement of about 11 points on average.  People who did not use the AP were in range about 66% of the time, while those not using the AP were in range about 62% of the time.  (Range was 70-180).

Source is poster 949-P.

The Virginia Artificial Pancreas

This is another ongoing research project into a "classic" artificial pancreas.  In the trial reported on at ADA 2014, 13 people were tested for 42 hours: 14 hours "open loop" treatment, and 28 hours of "closed loop" treatment.  People in the trial could move about a hotel.  This same research group is planning a 2 month trial of the same AP.

Source is poster 954-P and 104-LB.

Direct Comparison

Average BG
Estimated A1c
AP Use
Boston University138
Yes5 days24 Hours/Day
8 days24 Hours/Day
90 days
Night Only
2 days
24 Hours/Day

When you look at that, you might say the two hormones are better than one.  But I would not read too much into that difference.  It's not huge (8 BG points and 0.3 A1c), and remember that the single hormone solution is simpler all the way around: only one hormone to buy and load into the pump, less moving parts on the device, and so on.  (Not to mention the fact that Glucagon hasn't yet been approved for this application, although that is expected.)  Of course, the comparison is based on average BG, so might miss extra low BG events in the single hormone APs.

None of this competition bothers me in the least.  I love the idea of having four closed loop systems getting to market at about the same time with slightly different feature sets.  Having a bihormonal AP with slightly better control competing against a single hormone AP which is slightly simpler, sounds like just the sort of competitive situation that feeds progress in a capitalist economy.

Other Bits and Pieces

Poster 75-LB compared CGM data from actual BG data (measured using laboratory grade equipment) from blood pulled directly from a vein. They found that CGM data was very similar to the actual BG data, and that even when different, the differences were small. The researchers conclude that existing CGM technology is not the "weakest link" of AP technology.

Poster 747-P asked people who were testing a closed loop AP, what they thought of it. They liked it. They liked it because it provided better BG control, reassured them that nothing bad would happen while they slept, and improved BG control the next day.  Poster 110-LB contained similar information, but focused on the remote monitoring of an AP in a family situation (ie. parents remotely monitoring children).  A major conclusion of this research was that families wanted the AP/remote monitoring combo being tested; it did not need any improvements at all, it just needed to be made available.

Poster 948-P tested a closed loop system using diluted insulin compared to regular insulin, for small children (aged 4-7).  They found that diluted insulin worked a little better.  Average BG levels were the same, but time spent in range was 8% higher when diluted insulin was used.

Poster 951-P tested a closed loop system which (in addition to BG data) also used energy expenditure and galvanic skin response data.  These are two measures of energy use.  The hope was that by using energy expenditure data, they could make a better AP.  However, the data showed very little difference between using this data and not, and even this little difference was only when BG was above 250.


My summary of closed loop, artificial pancreas research is this:  We are seeing cure level control in phase-II clinical trials and for several different AP systems.  This is great news, for several reasons. First, it means they "only" need to get through phase-III trials (and marketing approval) for these APs to be sold in the US. They don't need to do better than the results they already have, just produce the same results in larger trials. Second, it means that if one falls apart, there are others which can still get marketed. Third, it means that the technology is ready. When one AP is successful, that team might just be ahead of the rest, but if four groups can do it, that means the technology is here for all.

Joshua Levy 
publicjoshualevy at gmail dot com
All the views expressed here are those of Joshua Levy, and nothing here is official JDRF, JDCA, or Tidepool news, views, policies or opinions. My daughter has type-1 diabetes and participates in clinical trials, which might be discussed here. My blog contains a more complete non-conflict of interest statement. Thanks to everyone who helps with the blog.

Saturday, July 5, 2014

More About ADA 2014

The JDCA recently published a "Flash Report"  based on try trip to ADA 2014.  You can read it here:

It covers some of the same topics that I already posted here, but has some different information in it as well.

Also, as part of attending ADA 2014, I have on line copies of almost all the abstracts presented there, and many of the posters, and a few of the presentations.  So, I'm going to experiment with "doing requests".  If you have a topic (a few keywords, or a sentence) and you would like to know if it was discussed at ADA 2014.  Send it to me, and (If I have time) I will summarize the ADA material that pertains to your topic.  Please note that I don't expect to start this for about a month, so send in now, but expect results in 4-8 weeks.  (Not very internet, I know!)  If you're interested in a specific researcher, I can tell you if they published anything at this meeting.

Joshua Levy