WHY AREN’T WE THERE YET? Valuable but incomplete measures of brain changes in babies with autism

In my opinion the recent research paper, Differences in White Matter Fiber Tract Development Present from 6 to 24 Months in Infants with Autism (Am J Psychiatry 2012 ; 169 :589-600), reports a very important finding and represents a huge amount of work, but the study is quite incomplete both in what was chosen to be measured and how the findings are interpreted by the investigators, Jason Wolff et al.  Moreover, even though this study has been touted in the news as a way of detecting autism early, it really isn’t ready for prime time as a clear indicator of the autism diagnosis. I think this finding is about things that occur downstream of other biological factors that are driving these changes.  — And because these changes are downstream, they are fuzzier in that they reflect a mixture of lots of other influences.   This makes it hard for them to clearly demarcate risk from lack of risk.  Measuring what’s upstream might bring us closer to a clinically useful brain measure of autism risk in young infants.

Because I am a brain researcher, I spend a lot of time thinking about different techniques to measure things in the brain.  Let me explain to you why people study things like this, what was measured in this study, and what it means. Then I’ll be able to explain to you why this study is both important and incomplete.

Why people study babies at risk for autism with brain imaging

Autism is a neurodevelopmental – and many would say also chronic – condition that is presently defined by behavioral criteria.   But these behavioral features don’t emerge in babies – they only become detectable when a baby becomes old enough to perform these behaviors.  The “holy grail” of autism research prominently includes being able to detect signs of autism, or autism risk, at a younger age than when current gold-standard tests can diagnose autism.

Finding brain changes in infants that predict or diagnose autism would allow for early intervention that might reduce the severity of autism and the difficulties associated with living with autism – or possibly even prevent an autism diagnosis.

However while the brain is clearly important in autism, we do not have a diagnostic brain test for autism even in people old enough to be reliably diagnosed.  This means that research into brain changes in infants at high risk – i.e., infants who have older siblings diagnosed with autism – has to hunt for things without exactly knowing for sure what it is looking for.  Investigators therefore start from best guesses based on existing data.  There’s the rub – there’s a lot of data, and which data you choose as a starting point reflects underlying assumptions – and there are a variety of underlying assumptions that lead to different choices.

Explanation of this paper for lay people

The study chose to look at the development of white matter in the brain.  White matter is made up of fibers that come from neurons and connect to other neurons. The paper reports that infants who get diagnosed with autism have different trajectories of development of “fractional anisotropy”, or FA for short – they have more “FA” at 6 months and less “FA” at 24 months than those infants who did not get diagnosed with autism.  So what does this mean?  Let me explain.

FA is particularly used to look at so-called “white matter integrity.”  This is a general term with hazards since how you think about this abstract concept reflects what you know — and also don’t know of don’t integrate into your model — about white matter, brain tissue and metabolism as well as so-called “neuro-biology”  (a term which leaves out the glial cells, including the oligodendroglial cells that make the “white matter” white).

White matter and spaghetti

Fractional anisotropy (or FA for short) is a measure of how much the flow of water is restricted in brain tissue, particularly around bundles of axons, or nerve fibers.

A metaphor for FA I like to use is cooking spaghetti and adding spaghetti sauce after it’s cooked.  Think of the strands of spaghetti as stand-ins for the axonal fibers.

  • If you don’t stir it while you boil it, it sticks together and the spaghetti sauce can’t get in between the bundles.  That tight wad of stuck together spaghetti strands will have high FA.
  • If you stir it a lot so the strands separate, the spaghetti sauce will get in between the strands and you will have lower FA.
  • One more possibility that actually has implications here is that if you way overboil it the spaghetti strands themselves will start to become too soft and mushy around the edges and the sauce will even start to penetrate the strands themselves.  As I will explain below, this could relate to another measure, “diffusivity,” that was also used in this paper.

The word “anisotropy”  means that the “isotropy” – the ability of the fluid to flow unrestricted in all directions (iso = equal and tropy = in all directions) — is restricted.  Restriction of “isotropy” is “ANisotropy.” The extent or “fraction” of the restriction is measurable, and is called “fractional anisotropy”,  and FA is its nickname.

White matter tracts are bundles of fibers, or axons. The axons are myelinated – that is, wrapped in myelin, which is full of lipids. The myelination is provided by the oligodendroglial cells, whose lipid membranes wrap around the cells like jelly rolls.   Lipids (fats) repel water.  This is different than the spaghetti metaphor, since spaghetti strands are not wrapped in fat.  But the metaphor still is useful visually.  If the axon bundles are thick and tightly packed, water will only flow easily up and down along the length of the axons.  If the fibers are looser or the myelin structure is less intact or for a variety of conceivable reasons the fluid flow will be less restricted. There may be more space in between the fibers, or the fiber structure s themselves may be compromised.

How this applies to the data in the paper

Going back to the paper on FA in infants, when we compare the infants who got diagnosed with autism when they were old enough for behavioral diagnosis with the “non-autistic” infants in the study –the ones at high risk who did not wind up getting diagnosed with autism–  what we want to know is this: What does it mean to see MORE FA early on — more restriction of fluid flow — in the 6-month olds who later get diagnosed with autism, and LESS FA later on — less restriction of water flow — in those same infants when they reach 24 months of age?

Brain spaghetti and spaghetti sauce again

Please bear with me for one more important but technical thing before I tie things together.  This study did “secondary” measures of the “radial” and “axial” diffusivity to figure out where the FA changes were coming from.   Roughly speaking, “axial” diffusivity measures water diffusion along the length of the fiber, while “radial” diffusivity measures water diffusion perpendicular to that, across the fiber.  In the spaghetti metaphor “axial diffusivity”  would be whether the fibers are more spread apart from each other along their lengths  because you stirred while cooking and got them separated from each other,  while “radial diffusivity” would indicate  whether or how much the fibers themselves have gotten mushy and exceedingly soft.  This could conceivably suggests that things may be going on both with the spaces between the fibers and in the fibers themselves —  [Though remember, spaghetti is just a  metaphor, not science,  and we need to please BE CAREFUL because these diffusivity and FA measures DON’T DIRECTLY TRANSLATE to clear interpretations about the underlying tissue – they are just interpretations of magnetic signals in an MRI scanner, not direct measures of fibers, and there is a lot controversy about how to interpret them in the scientific literature.]

Now, what does this all mean?  Why does it matter?

On my website www.autismWHYandNOW.org, I look at the ways that three key questions are addressed in autism:

1)      What is autism?

2)      How is autism caused?

3)      How can we help?

Here’s how the first  two of my three core questions apply to this study: Clearly, the big questions are,

1)      Regarding “What is autism?”–So what do these FA changes have to do with how brains of people with autism act differently?  And

2)      Regarding “How is autism caused?”– So what happened to the brains of these babies o make their FA change like this

A good start, an important contribution

The big contribution of the paper is to show at a younger age than before that there are dynamic brain changes happening early in infant development, before symptoms arise.  That’s a gigantic advance over the older belief system that the autism is indelibly stamped into the brain by genes from conception or certainly before birth, with nothing of significance happening after birth.  But the paper’s discussion of HOW and WHY these changes are occurring, and what impact they may be having, is kind of weak.

This weakness is manifested in what they propose for future research directions: They want to link these fiber changes to behaviors, and to genes.  But is that all there is?  No, not really.  And for 3) “How can we help? what do they propose? Early behavioral intervention.  A great idea and very important, but does that cover all we can potentially do? No, it doesn’t.

These investigators are not wrong.  But their viewpoint is incomplete.  We need a more comprehensive approach.  Here’s a big piece of how to get there.

Toward a more comprehensive whole body systems approach to the brain

As I explain in Chapter 5 of my new book The Autism Revolution: Whole Body Strategies for Making Life All It Can Be, brains are made of cells which have lots of vulnerabilities not only genetically but in their day-to-day brain health.

  • This includes how well their chemistry works, versus how messed up it is by nutritional shortfalls and challenges from toxics that glitch things up.
  • It includes how healthy their lipid membranes are – how much they are nice and fluid from an abundance of essential fatty acids versus how much they’re stiff and fragile from not enough omega-3s in the diet and from toxic and free radical injury.
  • It includes whether the cell’s energy-generating mitochondria are healthy and intact enough to work at full tilt — or whether they are impaired by membrane damage or lack of vital nutrients like B vitamins and CoQ10 that are critical for their energy production assembly line.
  • It includes whether the brain is able to be adequately supplied by oxygen and nutrients from the bloodstream, or whether the blood is sluggish or sticky (“viscous”) – even a little sticky –and whether the blood vessels are a bit compressed.
  • And it includes whether the brain can keep itself clean through efficient disposal versus garbage or whether the clean-up mechanisms are sidelined by things like immune activation so that left-over debris piles up and gums up the works.

There’s evidence on every single one of these points that there are issues that could compromise function in brains of people with autism.

Brain health, brain connectivity and brain metabolism in autism: why and how

Why is this specifically relevant to autism?  How well and energetically the brain cells work and how healthy the micro-environment around the cells is shapes how well and in what ways cells function and communicate with each other. In particular, it greatly affects how electrical and chemical signals get generated and organized in the brain.   People with autism have issues with brain coordination and they also have brain and body metabolic issues. Vast amounts of basic science support a relationship between brain chemistry/metabolism and electrical function. So it’s a plausible hypothesis to say these are related to each other.

All of these aspects of brain health and function are affected by whether the baby’s food is nutrient rich or depleted, whether the baby is exposed to or protected from toxins, whether the immune system is resilient enough to handle microbes and allergens, whether activity is rich enough to help develop brain-body coordination, and whether things are stressful or tolerable.  These themes of food, toxins, “bugs” and stress are critical in the way I explain things in my book The Autism Revolution. Most of this applies to the mother, too, before and during pregnancy. Since these all of these brain day-to-day issues are going on all throughout brain development, they can also affect how the brain develops – including how the fiber tracts develop.  They could certainly affect things that determine FA and diffusivity.

I personally have a strong hunch that the FA findings in this new paper are related at least in part to the emergence of neuroinflammation in the babies that develop autism.  But this study alone can’t be used to prove that, because it doesn’t measure that directly – in fact at present we have no really good ways to measure neuroinflammation in the brain using brain imaging, so we have to piece together a set of indirect clues to come to that conclusion, and we don’t have all the right studies yet to do that.  When we do get that information, we may learn important and surprising things about what is going on and HOW this happens that may affect how we intervene and treat.  So we shouldn’t jump to conclusions too soon.

How else can we approach this to get more of a full picture?

It is possible to study the brain’s biology using other brain imaging techniques than the ones that measure FA, but it doesn’t occur to most brain imaging researchers to do this.  Hence we get this wonderfully interesting, important and useful study, that is also incomplete.

I would propose that

1)     Regarding “How autism is caused“, it is dysfunction in the health of these many cellular, chemical and electrical functions that is UPSTREAM of the problems with FA development in infants with autism, and that if we measure these upstream biological functions we might get more robust and clinically useful early brain indicators of autism and autism risk. 

2)     Regarding “what autism is“, we might also understand autism better in terms of what I think are the ground zero mechanisms that intimately interrelate brain cell health problems with brain cell communication problems — and that are disrupted by the extent to which the Total Load of exposures, insults, deficiencies and vulnerabilities overwhelms brain/body resilience, and

3)     Regarding “how we can help,” looking at brain cell health problems in a whole body strategy context gives us a lot we can do with what we know right now to improve the situation — reducing “Total Load” and improving Resilience. 

I would also propose that starting from this framework we can generate more powerful brain research, better markers, and research that is more immediately helpful to people with autism and many others with related challenges.

Home stretch: summing up

So to sum up for now, here are four important questions:

  1. Is this study’s measure robust enough for us to march our 6 month old babies down to the local MRI scanner and test them for “autism”?  No, it is not.  First of all, if you look at the graphs in the paper you see that there is huge overlap between babies with autism and controls at the start point so you can’t say much of anything at six months from a scan measuring FA.  Then, to really make the determination, you need the trajectory measure — that is, you need to connect the dots over time from 6 months to 24 months — because it’s the slope (rate and direction) of the change that tells you something about risk, not any individual measure.  So this is pointing toward something that might be an early measure, but this study has not arrived at the early measure itself– as the authors clearly acknowledge but some press reports did not.
  2. What’s missing from this study  The findings are described meticulously and displayed beautifully.  They’ve done a lot with what they’ve got.  What they don’t have is measures either of the underlying cellular chemistry and blood flow, or measures of the differences in brain function.  Granted, one  study can only talk about so much.  But they don’t advocate future research in these directions, either.
  3.  Why did these highly trained, dedicated investigators design a study that did not look at these upstream factors that probably are driving these FA changes?  Two reasons:
    Reason 1: Because they believe that structure causes function, and structure is caused by genes.  They don’t think about how functional differences can lead to structural changes, and how there is a two-way street between structure and function.  Their view tends toward being static and compartmentalized, rather than dynamic and systems oriented. Based on this belief system their future study directions include functional MRI — which will show what parts of the brain activate differently in these autistic infants — but it will not look at the electrical signaling differences (e.g. by measuring brain waves with EEG) which are arguably closer to the underlying physical properties of the brain, such as brain chemistry (which can also be measured non-invasively using magnetic resonance spectroscopy).  This will produce more beautiful papers, but on its own this kind of data may take years or forever to add up to a comprehensive practical approach to helping and heading off autism. And it won’t illuminate the moment-to-moment underlying mechanisms driving the behaviors we label as “autism.”
    Reason 2: Because they believe that autism is fundamentally a social disorder.  Certainly autism is defined by social criteria.  But from a whole body perspective, these social and behavioral features are emergent properties arising from underlying alterations in the biological substrate – the cells, the metabolism, etc.
    There is also a third reason which is that the methods to perform physiology measures in living human beings are somewhat harder to use at this point in time.  But that doesn’t explain why this possible future research direction is not discussed in the paper.
    In order to generate research designs that will illuminate the underlying biological levels as well as the social and structural manifestations, we need: 1) a shift from a bottom-up genetic determinism approach to a “middle-out” physiology-driven approach, and 2) a shift to a systems biology approach.
  4.  What can we do about this?  First of all, the concerned public needs to understand what this is about and put this paper into a broader perspective.  Second, the concerned public and scientists need to understand the assumptions they are making (like that autism is fundamentally social), and that other defensible assumptions (like that underlying biology produces social “symptoms”) may provide useful insights.  THEN we need to get the research done that will produce truly whole-body, whole systems data about the brain, the brain in the body and the brain-body system in the world.  That is my own research agenda, and the agenda of my research program,  the TRANSCEND Research Program.  My group and others with similar approaches need to network together so we can get this complex task done in a coordinated, efficient and timely fashion to more directly address the public health crisis that autism represents.

And last but not least -or rather or first and highest priority, while that is going on we need to use the knowledge we already have about whole body strategies for making life all it can be to create an Autism Revolution that helps large  numbers of people right now.

Blog by Martha R. Herbert, PhD, MD, Posted July 15, 2012