Monday, January 7, 2013

Why Some Just Shouldn't Write About Science

We've taken a hiatus from blogging for too long and for mundane reasons after being so productive.  But leave it to Dr. Bob Sears to provide irresistible blog fodder.  This time he tries to critique a novel test for autism which results in the usual clumsy and error-riddled claim when he attempts to write about a subject he is clearly out of his depth on...science.

This is actually a novel and useful test with a premise that has tremendous diversity in terms of autism aetiology and treatment research.  But that doesn't stop Dr. Bob from not only getting it wrong out of the gate but proffering an inept criticism of it.  Dr. Bob states:
Researchers at the Children’s Hospital of Boston have developed a genetic test to detect the most common gene mutations that occur in children with autism (1.)
No they didn't and in fact stipulate quite clearly that they have developed a test based upon perturbed gene expression signatures in ASD children as compared to the control groups.  This is vastly different than genetic tests based upon mutational nucleotide sequences.  Dr. Bob's one and only "source" is a CNN report yet there is a very good, plain English press release from Boston Children's Hospital that explains very precisely what the test is and isn't.  There is also the primary research study open-accessed published in PLOS ONE which he obviously didn't bother to read before commenting.  Dr. Bob also states:
The small initial study only looked at 66 kids with autism, and compared their genetics to 33 neurotypical kids.
No they didn't.  The researchers started with a pool of 489 gene expression candidates, drew blood samples from their P1 group which was on a sample cohort of 66 male ASD cases and 33 age-matched male controls.  They examined perturbed expression (either up or down regulated) of those genes in the ASD group as compared to controls.  They then narrowed that pool of genes down to 391 based upon the strength of signature (you want a consistently detectable difference between groups) and applied the test to a subsequent group (P2) consisting of 104 male and female ASD cases and 82 age/sex-matched controls (although females were over-represented in the control group).  From the original study:
First, the training set (P1) was utilized to determine a classification signature (i.e. a combination of gene expression measurements) that was used to classify ASD patients in P1 (compared to controls). We ranked the 489 differentially expressed genes according to their area under the receiver operating characteristic (ROC) curve (AUC). Next we excluded those genes with low expression, requiring the minimum expression level across all samples to be at least 150. A total of 391 differentially expressed genes were then utilized in building the prediction models, which were subsequently tested against the samples in our independent validation cohort (P2).
The researchers then found that 55 gene expression signatures provided the the best prediction model while minimising the number of gene expression signatures needed and tested them again on the P2 group. The resulting accuracy of the test is both interesting and impressive.
Analyzing the blood samples, Kong and colleagues flagged 489 genes as having distinct expression patterns in the ASD group, then narrowed this to a group of 55 genes that correctly identified or ruled out autism in 76 percent of samples. They validated their findings in a second group of 104 male and female patients with ASDs and 82 controls, achieving an overall classification accuracy of 68 percent (73 percent for males and 64 percent for females).
This is an exciting find with positive ramifications that I will get to in a moment.  But since it's a genetic test, Dr. Bob doesn't like it and invokes an appeal to authority:
“They are measuring gene expression and then assuming it has something to do with mutations in the genetic code, which is not necessarily the case. In fact, there is a lot of information that suggests that differences in gene expression in autism are due to differences in gene methylation and gene regulation, which would not change the genetic code.”
I guess Dr. Frye didn't bother to read the study either where the authors explicitly state:
The classification performance in this study is encouraging, particularly as the two groups were heterogeneous and profiled using two different array-types. The classification of 73% of cases by expression profiling contrasts with the small percentage of ASD cases characterized by genetic mutations or structural variations to date. It also compares favorably to the performance of CMA, which, while high confidence, accounts for only 7–10% of cases of ASD. Together, these results suggest that gene expression signatures, which comprise multiple perturbed pathways, may serve as signals of genetic change suggestive of ASD in most patients. In this regard, this work parallels studies in neuropsychiatry where investigators have demonstrated that blood expression signatures are significantly different in schizophrenia [53], Alzheimer's disease [54], and bipolar disorder [55].
And
“It’s clear that no single mutation or even a single pathway is responsible for all cases,” says Kohane. “By looking at this 55-gene signature, which can capture disruptions in multiple pathways at once, we can say with about 70 percent accuracy, ‘this child does not have autism,’ or ‘this child could be at risk,’ putting him at the head of the queue for early intervention and evaluation. And we can do it relatively inexpensively and quickly.”
Emphasis added.  They aren't assuming any such thing as they are examining the actual expression of a set of genes and NOT the nucleotide sequences.  I would have expected better from Dr. Frye given he has an academic research appointment but there is no accounting for intellectual honesty and accuracy when one decides to associate themselves with a biomeddler anti-vaxx group like TACA.

The study authors identified two subsets of autism based upon gene expression signatures.
The biological pathways implicated by the differentially expressed genes identified in this study are of interest because some of the gene sets link to synaptic activity-dependent processes (i.e., long-term potentiation and neurotrophin signaling in Table 4), for which several ASD mutations have been found [40], [41]. Immune/inflammatory pathways were also identified in this analysis (e.g. chemokine signaling and Fc gamma R-mediated phagocytosis), which have been implicated in several studies of children with ASD compared to controls through CNS cytopathology [61], serum and CSF proteomics [59], as well as in cadaveric expression studies of the CNS [51].
This result improves upon what has already been elucidated with regards to differential gene expression in autism and other neurological disorders as well as identifying a substantial portion of children at risk for developing autism who could then be followed prospectively.  Familial histories, in utero environmental and neonatal exposures could be recorded to help identify potential inducers of perturbed gene expression and identify additional subsets of autism.  The more that is known, the more can be done to reduce the risk of autism and develop evidence-based treatment modalities.  This is a good thing but obviously eludes the likes of Dr.s Bob and Frye.
In my opinion the most important of all is to never take a “wait and see approach” if early signs develop. If we witness any regression or developmental delays we need to recommend and begin developmental and biomedical therapies as soon as possible to achieve the best possible outcomes. The decision to seek out these therapies will be based on an infant’s development, not on a genetic test which needs much more research and verification before we get excited.
Dr. Bob critiques this test based upon his own deficits in science but thinks he can ameliorate autistic behaviours with "biomedical therapies" which consist of throwing everything at the wall and see what sticks and of course, withholding vaccines.  Whereas identifying gene expression signatures before a child develops developmental delays is nothing to get excited about.  The only thing he manages to get right is that this test needs wider validation before becoming commercially-available.  This is acknowledged by the study authors:
The application of these predictors to a prospective cohort would allow us to further assess their validity as a diagnostic and prognostic tool. Finally, our groups with ASD were compared to developmentally normal controls and not to individuals with other neurodevelopmental disorders. Nevertheless, the accuracy we have obtained in this study is a necessary first step towards a trial validating a set of predictive biomarkers.

In conclusion, this study of children with ASD describes a gene expression signature that shows promising accuracy in classifying children with ASD from controls. The ability of the ASD55 predictor to correctly classify ASD samples compares favorably to the DNA-based tests currently proposed for ASD diagnosis. The results presented here raise further questions that bear investigation but are outside this study's scope: At what age does this ASD55 signature manifest? Is it present at birth? Finally, we expect that larger studies can be used to determine whether particular characteristics of ASD can be classified or predicted from a gene expression signature (e.g. seizures and language delay) and thereby improve individualized treatment in the near future.
One of Dr. Bob's faithful asks the question on his Facebook page: 
 Is this MTHFR or another mutation?
To which he responds:
I am not sure if this test includes MTHFR.
Read the damn study Bob; the answer is no, MTHFR expression is not included.

Another asks:
I'm interested. Would it detect asbergers [sic]
The answer (not from Dr. Bob) is yes, the test was able to distinguish autism, Asperger's and PDD-NOS.

All in all this is a test to get excited about, not just for its potential commercial use as an early detection method for autism risk but its current utility for research that appears to be an expedient avenue to elucidating autism aetiology, developing preventative and targeted treatment strategies.  If there is one thing I have learned being involved with vaccine discussions is that biomeddlers/anti-vaxxers hate genetics research because the evidence keeps mounting against vaccine causation and thus "recovery" via DAN! nonsense.  Of course this makes "biomed" pedallers like Dr. Bob even more obsolete as they are getting left in the dust by actual technology.

No comments:

Post a Comment