This week Michael Savage questioned the legitimacy of most diagnoses of autism because they are based only on symptoms – not on any objective tests. I have also been engaged in a blog discussion with Dr. Jon Poling about the exact nature of his daughter’s diagnosis – does it represent autism or something else. Both questions are ultimately about the very nature of a neurological diagnosis (or medical diagnosis in general). So I thought I would dig deeper on this question for background.

As a clinician I often discuss with patients what I think about their diagnosis. In my experience most people begin with many misconceptions about how diagnoses are typically made. I also train student doctors and they too typically come to me needing a far more complete and complex understanding of how diagnoses are made and what they mean. So even outside of the context of autism controversies this is a very useful topic to cover.

Pathophysiological Disease

There are different types of diagnosis, or labels, that we can attach to patients. The gold standard is the pathophysiological disease. This refers to a discrete medical entity that has a known cause, specific physiological processes and effects, and a natural history. This leads to a recognizable set of signs (things we can see about a patient) and symptoms (things the patient experiences) as well as specific abnormalities on diagnostic testing.

Just to take one example – pernicious anemia. This results from decreased or absent secretion of intrinsic factor by the parietal cells in the inner lining of the stomach. Intrinsic factor is a glycoprotein that is necessary for vitamin B12 to be absorbed in the intestine. The lack of vitamin B12 leads to anemia, because B12 is a necessary cofactor in the production of red blood cells. The anemia results if pale skin, fatigue, and bleeding gums. Severe B12 deficiency can also lead to other diseases, such as neuropathy and dementia. In this disease there will be specific laboratory abnormalities upon which the diagnosis can be made to a high degree of sensitivity and specificity, and all signs and symptoms can be explained. It is, as we say, a discrete pathophysiological entity.

Clinical Syndrome

At the other end of the spectrum are what are referred to as a clinical diagnosis – a diagnosis made solely on the basis of a certain constellation of signs and symptoms. This is often how diseases are first recognized and only later their underlying pathophysiology sorted out.

Let’s look a migraine headaches as an example of a clinical diagnosis. Migraines have been recognized for a long time and they have a fairly specific clinical picture – no one doubts the existence of migraines as a real biological entity. But in practice the diagnosis is made purely on the basis of symptoms – intermittent headache which is usually pounding, ususally one sided but may occur on both, and associated with sensitivity to light and sound and sometimes nausea and vomiting. There may also be present what is called migraine aura, which can be a focal neurological deficit or something like flashing lights off to one side that seem to migrate across the visual field.

Few patients have all of these symptoms, most have many but not all of them (and there are other features as well I did not list). How many, exactly, do you need to qualify for the diagnosis? Are some symptoms more helpful than others? For example, there are many causes of headache, so not everyone with a headaches has migraines – but a migrating flashing light aura is very specific for migraines.

It is also true that, for everyday diagnostic purposes, there are no laboratory or imaging studies that are evidence for or against the diagnosis of migraine. It is a purely clinical diagnosis.

Diagnostic Categories

Sometimes a clinical diagnosis is probably a single disease that is just not yet understood. However, sometimes a clinical syndrome represents many underlying different diseases that all have similar ultimate effects, and therefore symptoms.

At times this is very conscious. For example, we might diagnoses someone with heart failure – knowing that this is just a description of a physiological condition not saying anything about the underlying cause, and understanding that there are many possible underlying causes.

Diagnoses are therefore sometimes hierarchical – there are diagnostic categories with subcategories and perhaps further subcategories, and finally discrete diseases. But we cannot always drill down to the specific disease, so we have to make due with the closest categorical description we can make.

Putting it together

Most diagnoses are somewhere in between a rock solid discrete pathophysiological disease and an ambiguous clinical syndrome without any idea of cause. Most diagnoses rely heavily upon the clinical picture, are supported by laboratory findings but not definitively diagnosed by a single abnormal result, and rely upon ruling out other possibilities that can look similar.

Diagnosis is often a matter of probability, not certainty. Also – sometimes the focus of clinical research and evidence is not centered around making a specific diagnosis but knowing what to do with certain diagnostic categories. In other words, it is not always helpful to obsess over finding a specific label for what a patient has. Sometimes it is more important just to know how they should be treated given their diagnostic category or their specific features.

Neurological Localization vs Etiology

In neurology particularly we need to separate out thinking about diseases between those based upon localization (what part of the nervous system is not working) and those based upon etiology (the underlying pathology). By and large, the symptoms of a neurological condition are determined solely by localization. Of someone suddenly loses the ability to speak, with certain features on detailed exam, I may be able to say that they have a lesion in Wernicke’s area of the cortex – a very specific localization. But I can say nothing about what is causing the lesion. It may be a stroke, a bleed, a tumor, or even a seizure.

I treat ALS so I also often give this as an example as well. ALS (amyotrophic lateral sclerosis, also called motor neuron disease or Lou Gehrig’s disease) is characterized by progressive loss of upper and lower motor neurons, causing weakness, atrophy, and hyperreflexia (increased deep tendon reflexes). It is a clinical diagnosis supported by certain laboratory findings and also dependent upon other entities – like spinal cord disease -being ruled out. But it is not a pathophysiological disease. It is, rather, a category of disease and likely contains many underlying causes that all have in common the ultimate death of motor neurons.

It is the specific cells within the nervous system – the motor neurons – and the pattern and time course of their death that determines the diagnosis of ALS, not the underlying cause. When we eventually learn more about the various causes of ALS it will likely break up into many specific separate disease diagnoses. This process has already started. For example, we recognize familiar ALS due to a mutation in the SOD1 gene as one specific cause.

Bringing It Back to Autism

With all this in mind, what kind of diagnosis is autism – or ASD, autism spectrum disorder? It is a clinical diagnosis made by the presence of certain neuropsychological features. The primary feature is a relative loss of social ability, attention, or focus. Like ALS, it is a neurological category diagnosis. This means that it is not one discrete pathophysiological entity. ASD likely represents many underlying entities that share in common a similar effect on brain function.

Another way to look at this is that there are probably many possible pathways to the same destination – that destination being features of autism. We are beginning to separate out subtypes by clinical features. For example there appears to be a regressive form of autism where children develop normally for 2-3 years and then lose some of that progress.

We are also learning a great deal about some of the causes of autism, with the evidence so far pointing strongly to various genetic factors. But everyone recognizes that there is no one cause for autism, because autism is not one thing but a category including many entities.

One of the challenges for diagnostic categories like ASD is to know where the edges are, and what symptoms can overlap with other entities. For example, diseases processes that we do not usually associate with autism may impair brain function in such a way as to cause some of the same symptoms seen in ASD.

As I mentioned with ALS, damage to the spinal cord can cause most of the same signs and symptoms as ALS. Likewise if someone sustains brain trauma they will likely have signs and symptoms similar to Alzheimer’s disease – loss of memory, poor concentation, psychomotor slowing, etc. Therefore we need to separate out those features of a diagnosis which just tell us which nervous tissue is involved and those that likely tell us something about the underlying cause.

Regarding the Hannah Poling case, Dr. Poling wrote in his letter to me that his daughter has the DSM-IV diagnosis of autism. This is based upon the fact that she has certain features of autism, such as decreased language and gaze avoidance. But she also has many features that are not specific to autism, such as seizures, growth decline, and mild hypotonia. She seems to have a neurological disorder that is broader than just ASD. It is known that she also has a mitochondrial mutation, and so it is possible that this causes some of these additional features. It is also possible that whatever happened to Hannah caused a more diffuse neurological injury. Those parts of the brain that are involved in ASD were affected, but only because many parts of the brain were affected.

At this point it is not possible to say what implications, if any, Hannah Poling’s case has for the more typical cases of ASD. She may be no more relevant than head trauma is to Alzheimer’s disease.

Within the realm of clinical diagnoses, like autism, it is very difficult to know how to interpret cases that are atypical – like Hannah Poling. Since the diagnosis is clinically based, having unusual clinical features may mean that the underlying cause is completely different. There are no objective laboratory tests to tell us that the underlying pathophysiology is the same.

Conclusion

The nature of neurological diagnosis is a complex topic often mired in imprecise language. In my reading of popular discussions of the autism-vaccine issue, from Michael Savage to David Kirby, there is a great deal of confusion and ambiguity regarding the very nature of diagnosis itself. Hopefully this background will facilitate future discussion.