A Blood Test For Schizophrenia—Not Now—Not Ever?

  Do not think about, write about or deal with  human behavior without determining the effects of incentives.

   Chicago, another city like Detroit, ruled by “progressives” for decades, is facing bankruptcy as the pension fund of city employees runs out of money. Too bad they can’t tax the dead who have voted for years in Chicago elections. Fortunately, it’s never to late for me to say, “Told ya so!”

     Don’t forget, All workers are created equal, government “workers” are just more equal. Some gravy trains eventually stop running.

   Wherein we see another key to understanding mental illness has been lost. They should keep these keys in a safe place, they’re always being misplaced.

A Blood Test for Schizophrenia with 83% Accuracy?

An NBC online News article dated October 15, 2010, carried the noteworthy title New blood test may help detect schizophrenia.  Thanks to Francesca for the link.

The article was written by Natasha Allen, a freelance medical journalist.  The gist of the article is that there is a new blood test called VeriPsych which “researchers say” is 83% accurate in discriminating people who are “schizophrenic” from people who are not.

One of the researchers – Michael Spain, MD, Chief Medical Officer of Rules-Based Medicine, is quoted as saying:

“There is a certain amount of denial when a child is diagnosed with schizophrenia. You wish that your child did not have that…It is a good test to convince parents or even the patient to stay on medication, as opposed to just subjective opinion.” [Emphasis added]

The article also points out that Rules-Based Medicine is the company that makes the test and funded the study.

The study itself is published in Biomarker Insights, a peer-reviewed journal that began publishing in 2006 and is owned by Libertas Academia.  The study appeared in the May 2010 issue and is called Validation of a Blood-Based Laboratory Test to Aid in the Confirmation of a Diagnosis of Schizophrenia. There are 24 authors.  The lead author is Emanuel Schwarz of the Institute of Biotechnology, University of Cambridge, UK.  The Institute of Biotechnology had reportedly served as consultants to Rules-Based Medicine.  Five other authors, including Sabine Bahn, MD, PhD, MRCPsych, have links to the Institute.

An additional four authors, including Dr. Spain, report links to Rules-Based Medicine Inc.

Another author reports links to Psynova Neurotech Ltd., Cambridge, UK (a subsidiary of Rules-Based Medicine).  And another reports links to the Stanley Medical Research Institute, Chevy Chase, Maryland.

Here’s the study’s abstract:

“We describe the validation of a serum-based test developed by Rules-Based Medicine which can be used to help confirm the diagnosis of schizophrenia. In preliminary studies using multiplex immunoassay profiling technology, we identified a disease signature comprised of 51 analytes which could distinguish schizophrenia (n = 250) from control (n = 230) subjects. In the next stage, these analytes were developed as a refined 51-plex immunoassay panel for validation using a large independent cohort of schizophrenia (n = 577) and control (n = 229) subjects. The resulting test yielded an overall sensitivity of 83% and specificity of 83% with a receiver operating characteristic area under the curve (ROC-AUC) of 89%. These 51 immunoassays and the associated decision rule delivered a sensitive and specific prediction for the presence of schizophrenia in patients compared to matched healthy controls.”

Which means that if you perform these 51 tests on a person’s blood and collate the results using VeriPsych’s algorithm, the result will predict schizophrenia with 83% accuracy.

In the conclusions section of the article it states:

“In this multicenter study, we discovered and validated a biomarker panel for schizophrenia based on biological and technical reproducibility of the molecular signature.”

“High classification performance demonstrated that the decision rule could identify schizophrenia patients with high accuracy irrespective of the disease duration or treatment state.”

“In summary, the present findings demonstrate the applicability of a rapid and non-invasive test to confirm the presence of schizophrenia.”

And their work is not finished!

“We anticipate that the 51-plex assay panel will result in the future development of a differential diagnostic test that can distinguish among various neuropsychiatric disorders such as schizophrenia, bipolar disorder and major depressive disorder.”

Under “Acknowledgement,” the authors tell us that the study was “instigated and supported” by:

    Rules-Based Medicine
    Psynova Neurotech Ltd
    Stanley Medical Research Institute

The authors express their thanks to various colleagues who assisted in the research.  And they singled out for special thanks “…Dr. Fuller Torrey for his support and suggestions.”

Promotion and Publicity

After the research article was published, the information about VeriPsych was picked up by the following media sources:

October 6 2010, a site called MPR put up a sort of ad that reads “VeriPsych schizophrenia diagnostic aid available,” and gives a number to call RBM for more info.

A site called Fast Company put up an article called Veripsych Says It Can Spot Depression, Schizophrenia In Blood on their site, no date given.

Singularity Hub put up an article Blood Tests to Diagnose Schizophrenia, Other Brain Disorders on the Horizon, by Jeremy Ford on Jan 18, 2011.

On October 13, 2010, livescience put up Natasha Allen’s article under the title It’s in the Blood: New Hope for Detecting Schizophrenia.  Here Ms. Allen is listed as a MyHealthNewsDaily Contributor.

Mental Healthy (UK) site ran an article Blood Test to diagnose schizophrenia and depression by Catherine Walker (probably in 2011), after Dr. Sabine made a presentation at the 2011 APA annual conference.

Psychiatric Times ran the headline Blood Tests for Diagnosis of Schizophrenia and Depression? on August 10, 2011

But three months later, on November 8, 2011, they also put up an article called Blood Tests for Diagnosis of Schizophrenia and Depression: Not Ready for Prime Time.

On October 28, 2012, Oxbridge Biotech Roundtable put out an article An afternoon with Prof. Sabine Bahn: Bridging prognosis, diagnosis and treatment.

Bipolar burble (a blog) did a post on this on April 18, 2011.  But the author of the blog put this at the end:

“This is a money-grab taking advantage of desperate mentally ill people.

I actually find this ‘diagnostic aid’ blood test for schizophrenia to be bordering on unethical. VeriPsych can cover their ass with math and statistics and probabilities and legal-eze and I’m sure that makes it ‘OK,’ but if you ask me, they are a hair’s breath away from lying. It feels irresponsible to me to hand out these kinds of results about a very serious illness based on one study. One. And there is so much math needed to make this model work that I would fall down dead if there wasn’t a mistake in there somewhere. Nobody gets it right the first time.”

So we have a blood test for schizophrenia!  The Holy Grail – at last.  Schizophrenia, the darling “diagnosis” of psychiatry, can now take its rightful place in the halls of medicine, soon to be followed by bipolar disorder and major depressive disorder.  And disgruntled, misanthropic naysayers, such as myself, can slink cringingly into our narrow beds of shame and ignominy.

But Wait! There’s a Glitch!

On January 2, 2013, the following notice appeared on VeriPsych’s website.  (VeriPsych is the name of the blood test, but it is also the name of the company that marketed the test and is affiliated with Rules-Based Medicine, which apparently is now called Myriad Rules-Based Medicine.)

“Thank you for visiting the VeriPsych website and for your interest in VeriPsych, the first blood-based diagnostic aid for schizophrenia.

We have temporarily suspended offering the VeriPsych test in an effort to improve its utility. In 2010, we conducted a beta launch confirming that the test worked as intended; however, in close collaboration with our medical and scientific partners, we collectively determined that the product needed further refinement to better fit the needs of patients and healthcare providers. Accordingly, we have shifted our focus onto the development of new transformative molecular diagnostic tools that address bi-polar disorder and major depression, in addition to schizophrenia. We are extremely excited by the progress made in bringing these diagnostic products to physicians and, most importantly, to the patients that can benefit from them. Unfortunately, there is currently no timeline for the availability of these new products in the United States or any other markets.

If you would like to receive information about our next generation of tests and their availability, please enter your email below. We will use this email distribution to release updates and news about the development of these new tests.

Thank You.”

Note the phrases:

    in an effort to improve its utility
    the product needed further refinement
    there is currently no timeline for the availability of these new products

History of the Financial Aspects


Psynova Neurotech, established in 2005 by Sabine Bahn and Chris Lowe, PhD, Director of Cambridge University’s Institute of Biotechnology


AUSTIN, Texas–(BUSINESS WIRE)–Rules-Based Medicine, Inc. (RBM), the leading multiplexed biomarker testing laboratory, announced today that it is partnering with Psynova Neurotech to co-develop and commercialize a blood test for the diagnosis of schizophrenia. RBM and Psynova will focus on the unmet clinical need for an objective and reliable diagnostic test to accelerate and optimize the treatment of schizophrenia. Under the terms of the agreement, the companies will collaborate on the validation, regulatory approval and manufacture of a diagnostic blood test for schizophrenia that will be sold worldwide exclusively by RBM.


May, 2010 Schwarz et al article published in Biomarker Insights journal


Professor Sabine Bahn, MD, PhD, MRCPsych, Director of the Cambridge Centre for Neuropsychiatric Research (CCNR) and Director/Cofounder of Psynova Neurotech Ltd, presented her research at the APA annual conference in Hawaii (May 14-18)

April 27, 2011 Myriad Genetics announced that it was going to acquire RBM and set up a subsidiary known as Myriad RBM, running it from RBM’s existing premises in Austin

June 23, 2011 According to Business Weekly, Myriad Genetics completed the $80 million cash acquisition of the PsyNova + Rules-Based Medicine company.

August 10, 2011 Michael Spain, MD, Chief Medical Officer of Myriad RBP is quoted in Psychiatric Times as saying that “…psychiatrists in a variety of treatment settings have ordered the test for hundreds of psychiatric patients,” at a cost of $2,500 per test, and that “…numerous insurance carriers…” were paying for it.

January 2, 2013

VeriPsych suspends the test.

So what happened?  What went wrong?  Within psychiatric circles, a valid blood test for the condition known as schizophrenia would be the media equivalent of aliens landing in Times Square.


   In other words, if this were true, it would be well-known and widely used. It’s not so it isn’t.

Unfortunately, everybody concerned, including Fuller Torrey and the Stanley Medical Research Institute are keeping their heads down.  The timing of the sale of PsyNova for $80 million seems significant.  That is an awful lot of money for something that “needs further refinement” and for which no marketing date is available.

Crunching Numbers

I have no inside information as to what went wrong on this study, but here’s a little analogy that might go some way to explaining the matter.

Suppose my wife and I decide to rent out the upstairs of our house, but we particularly don’t want tenants who play loud music.  So we decide to develop a test that will enable us to distinguish these individuals from people who play their music quietly.  We live in a small village, and we decide to conduct our research here.

There are 100 houses in our village, and for simplicity’s sake let’s say that there’s one person living in each house and we have 10 items of information on each person.

So we walk around the entire village every day for, say, a month.  We carry a decibel meter and we take noise measurements at each house.  At the end of a month, we average the daily readings from each premises and then we begin bumping this data against the ten items that we know about each occupant.

Let’s say that one of these items is age, and we find a correlation between age and noise.  The older the occupant is, the lower the decibel number.  But the correlation isn’t all that good.  There are some noisy old folks, and some quiet young people.  So we arbitrarily decide that we’re going to use , say, 20% of a person’s age, as a negative factor on the noise predictor scale.

Another item of information that we have is whether each individual is right or left-handed, and we notice that the left-handed people are far noisier than those who are right-handed.  So we give being left-handed a 60% weighting.  People who are left-handed will get 0.6 (60% of 1) added to their noise-prediction score.  Right-handed people get 0.  And so on for the rest of the information.

Then we add up each person’s scores on the noise prediction scale, and if we’re lucky, all the high scorers will be noisemakers and all the low scorers will be quiet people, and voila, we have a simple way to predict if a prospective tenant will be noisy or quiet.  The string of fractions (60% left-handed – 20% age + … etc) is called an algorithm.

In actual research, however, results are seldom that clear cut.  It is more likely that we will have a scale that discriminates with less than perfect accuracy – say 60%.  So we start to “tweak” the weightings that we assigned to each item of information.  Instead of 60% for left-handedness, perhaps it would be better it if were 65%.  Or perhaps the age score should be weighted at 27%, and so on.  Doing this by hand would be tedious and time-consuming, but with computers one can bump any combination of weightings against the criterion measure with little difficulty.  And in this way, we find the combination of fractions that gives us the most predictive algorithm.

The authors of the study started with 181 blood tests, from which they identified 51 tests that had some correlation with the “schizophrenia” group.  They then ran these 51 tests on 806 participants (577 schizophrenia; 229 controls), and from this data they developed an algorithm that separated the schizophrenia participants from the controls with an 83% accuracy rate.  They report that “…all elements of the data set were used to train the algorithm.”

“Training” the algorithm is what I’ve called tweaking in the example above.  But there’s a problem.  By tweaking the data so thoroughly, what I’ve actually produced is a noise algorithm that may work reasonably well for our village at this point in time.  It might not work in the next village or even in our village next year.  (It may be, for instance, that the high correlation that we found between noisiness and left-handedness is a complete fluke that has no validity outside our village.)  This is particularly pertinent in that noisiness is not some kind of inherent trait like left-handedness or tallness.  Rather it’s a behavior, and behaviors are acquired (or not acquired) through a complex and highly individualized process of interaction between a person and his environment.  Two people who are inherently very different might both score high on a measure of noise, while identical twins might score at opposite ends of the scale.

The general point is that if one is working with a discrete set of data and a fairly large number of variables, it’s usually possible to construct an algorithm that will separate the individuals with a reasonable degree of accuracy along a given criterion.  In other words, if one can tweak the algorithm more or less indefinitely, and add, or drop, variables at will, a pattern will eventually emerge.  The pattern is not necessarily spurious.  It may be a real pattern, but it only applies to the individuals concerned.  I don’t know if this is what happened in the Schwarz et al study, but it might be something along those lines.  All concerned are staying fairly quiet about it.  So perhaps we’ll never know.

Validity and Reliability of the Schizophrenia Label

Another critical issue in this matter is the nature of the criterion variable and the accuracy with which it can be measured.  In my hypothetical noise study, I have a fairly objective measure (decibels).  But it’s not perfect, because, firstly, I’m taking only one measurement per day, and secondly, I’m taking measurements from the street, and, for this reason, houses that are built closer to the street will, other things being equal, score higher than houses that are set further back.  In the case of the condition known as schizophrenia, the situation is hopelessly confounded because all the DSM items that define the condition are vaguely-defined behaviors, the assessment of which is inevitably subjective.  In other words, if you choose, say, 1,000 people at random and ask 20 psychiatrists to examine all of them and identify and list those who “have schizophrenia,” you will get 20 different lists.  (There will be some overlap, of course, but you will not get perfect concordance.)  So an algorithm that’s been trained on the basis of one of these lists may not work very well on another.


   In other words, if schizophrenia cannot be reliably defined, how can a test be 83% accurate?

In this regard, there are two interesting quotes from Emily Deans, MD, a Psychiatry Department instructor at Harvard, in the second Psychiatric Times article mentioned earlier.

“Since the diagnoses are based on a recipe list of symptoms from DSM-IV and not known brain pathology, new biologic markers and tests are re-searched and validated against the formal diagnostic criteria.  These criteria are designed to be assessed by mere observation and questioning of the patient.  Thus, biomarkers only end up as valid as the original criteria, or less so, depending on the validation of the scale used in research…” [Emphasis added]


“…biomarkers based on DSM-IV will never be as useful as ground up research to link known brain, gene, and MRS findings to the patient’s symptoms.”

What Dr. Deans is saying here, in effect, is that schizophrenia, as defined by DSM (which is the only way it can be defined) will never be linked reliably to specific neural pathology.  This is something that we “mental illness deniers” have been saying for decades.  Psychiatrists, on the other hand, have been saying, with a level of confidence bordering on recklessness, that schizophrenia (as defined by DSM) is a brain illness, and they even claimed to have identified the neural deficit involved (the now discredited dopamine theory of schizophrenia.).

But it should not be concluded that Dr. Deans or psychiatrists generally are retreating from the bio-psychiatric perspective.  Rather, psychiatry’s position is shifting from the “schizophrenia-is-a-brain-illness” stance of former years to “schizophrenia-is-many-brain-illnesses” which is becoming the rallying cry of the present.  And they’re going to identify each one through ground-up molecular research any decade now.  Meanwhile, by some extraordinary coincidence, neuroleptic drugs are the appropriate “medication” to correct all of these illnesses.  What a stroke of luck!

Stop Press

I was about to publish this post last week, when I ran one last Internet check on VeriPsych to see if there were any updates.  To my surprise, I found a promotional video which was published on YouTube on February 11, 2014.  The video is titled New Blood Test for Schizophrenia.  Here’s a quote from the narrator:

“Researchers…at the University of Cambridge have been working on a blood test for schizophrenia for many years.  A first test was launched in 2010, but later withdrawn from the market, as the price tag of around 2000 Euros was too expensive for wide usage.”

So VeriPsych was withdrawn because it was too expensive.  The earlier announcement on the VeriPsych website (that is still there at the time of this writing) said that it needed “further refinement” in order to “improve its utility.”  So does it need refinement or a price cut?

Another quote:

“They have now developed a new version, which they claim is cheaper, and provides more detailed information for the diagnosis.  The test looks at certain proteins in blood samples of patients to distinguish between different kinds of mental illnesses.  Researchers say that the new test is able to diagnose schizophrenia with a certainty of 83%, and depression with a certainty of about 90%.  Although the test could never stand on its own, it provides doctors and patients with valuable backup information.”

But the test isn’t quite ready for market yet.

“Sabine Bahn and her colleagues want to launch the new test within this year.”

The video is professionally produced and will probably catch some attention, but there are too many unanswered questions.  Firstly, why was the test pulled if, as reported, it had been selling well and its cost was being reimbursed by insurance companies?  Secondly, if it just needed a price cut, couldn’t this have been done more or less instantly rather than being off the market for 13 months?  Thirdly, is this the same test with the 51 “disease signature” analytes that was described in the original study?  If so, then where did the additional information concerning depression screening with 90% accuracy come from?  If not, has the new test been written up in a peer-reviewed journal?  Fourthly, – and most importantly – what prompted the January 2013 statement that the test “needed further refinement,” if the 2010 beta test had confirmed that “the test worked as intended”?

To me, it just seems like we have too many questions.

Philip Hickey, PhD.


    Another breakthrough broken.

    To order my novel, “Days of Songs and Mirrors: A Jacobite in the ‘45”, click here.

Government Job or Respect–Which’ll It Be?
Cheerio and ttfn,
Grant Coulson
Cui Bono–Cherchez les Contingencies


Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: