On April 29, 2013 Dr. Thomas Insel, the Director of the National Institute of Mental Health (NIMH), wrote on the NIMH Director’s Blog that the Institute will be re-orienting its research away from currently used diagnostic categories (the Diagnostic and Statistical Manual or “DSM”) and recommending the use of new criteria, the “Research Domain Criteria.” Hopefully, this will yield research outcomes more consistent with the actual biology of the brain. While this change is primarily aimed at researchers, it may prove to be the first in a series of steps that will radically alter the way we conceptualize psychiatric disorders. Given that research informs our clinical understanding, researchers and clinicians will need to eventually speak the same language. Autism will likely be one of the diagnoses that need to be reconceptualized.
In the current era of unlimited information, medical terminology has become a part of our general information. We all hear about medical research findings on a daily basis. Despite our familiarity with medical diagnostic terminology, few realize that many medical diagnoses are recent constructs – and often controversial – even among those who study those diseases. This is most certainly the case for psychiatric diagnoses and its catalogue, the DSM.
First, let’s review a little history. Many of the modern psychiatric diagnoses were first described in the late 1800s. Most prominently, Emil Kraepelin compiled the symptoms he observed into syndromes and they were named as diagnoses modeling this after the methods used in other branches of medicine. Originally, this was considered a radical innovation as it was at odds with the then prevalent psychoanalytic approach emphasizing the person’s lifetime narrative and premised on the understanding that mental illness is a reaction to difficult life events. The change from description of symptoms into diagnoses was harmonious with those who advocated a more biologic approach. Through the 1970s, a psychiatric diagnosis was still, for the most part, a matter of the physician’s judgment. This state of affairs changed dramatically with the advent of the third revision of the DSM in 1980. The DSM-3 brought some huge advantages to the field of psychiatry. For the first time, diagnoses were standardized such that any competent psychiatrist would arrive at the same diagnosis for a given patient. Research could be performed with standard diagnoses thus facilitating the psychopharmacologic revolution that was already taking place. Psychiatry took a huge leap forward to become a recognized medical specialty with similar vocabularies and methods as their medical peers.
Despite the benefits, there were some significant drawbacks to the DSM-3 that have largely been forgotten, including by many in the health professions. In order to obtain reliability, only observable symptoms were used as part of the diagnosis. The DSM-3 was “agnostic” as far as the etiology (i.e. cause) of the disorder. An etiologically based diagnosis (for example, strep throat) has predictable symptoms but there are various reasons why a person would get a sore throat. In order to make the diagnosis of strep throat, therefore, it is not enough to describe symptoms. A physician must get a throat culture and grow out the strep bacteria to prove it is indeed “strep.” With infectious diseases we can grow out the bacteria in a dish and then test the medicine in that dish for its efficacy. In contrast, without knowing the cause of a psychiatric illness, devising a rational, definitive treatment strategy is difficult. The creators of DSM knew this. Alan Frances, the leader of the DSM-4 effort, stated that the DSM diagnoses were meant to be a “way station” en route to more valid approaches and he cautioned investigators and clinicians not to “reify” (that is, to make real something that is not) the diagnoses. Despite that, the DSM diagnoses have been nearly universally reified. The evidence is now clear that most of the diagnoses are not good mirrors of nature, do not have zones of rarity (that is, natural boundaries between disease and normalcy), are not good predictors of prognosis, and are not generally predictive of treatment outcome. Little progress has been made in our understanding of schizophrenia, depression or, for that matter, autism over the past half century. This is a huge disappointment given that almost all of the prototypes of our current psychiatric medications were discovered prior to 1980. In this era of modern genetics and molecular biology, there have very few triumphs in the realm of the psychiatric disorders. One emerging understanding is that the symptoms that we are using (which are complex behaviors) are far removed from the basic molecules, genes and cells that can be studied. Too many factors go into influencing complex behaviors. Dr. Insel and many others are calling for the study of symptoms more directly related to neural circuitry.
Autism research also suffers from the dilemmas described above. Dr. Leo Kanner first described autism in 1943. Since then, Kanner’s descriptions have been tweaked by the various versions of the DSM but, essentially, we are working from the same model. Despite the belief that autism is highly heritable, researchers have found no genes responsible for more than a few cases of autism while over 100 genes have been described as having some association with autism. In terms of being a brain disorder, we have found many brain areas affected. As things go awry in development, it leads to abnormalities in other regions of the brain not initially affected, analogous to a house having a poorly poured foundation resulting in all sorts of structural problems. Despite the nearly universal understanding that autism is, in reality, a collection of many different disorders, the reified use of the diagnosis has hampered quality research. Much research on “autism” utilizes a large group of diverse disorders in the study. As an analogy, if we were to look for the treatment for “cough” we might not find one as cough is a symptom of not a single but many diseases.
Autism does not have natural boundaries. This is evidenced by the fact that even small changes in diagnostic criteria cause very large changes in prevalence numbers (an observation noted since the 1950s) and by research which shows that autism traits are found in a smooth distribution in the population. Therefore, the borders of autism must be arbitrary. The diagnosis is not good at predicting outcome (prognosis). Children diagnosed with autism might become very severely impaired or have a good outcome, and have cognitive functioning from the severely retarded to genius level. Research has demonstrated that the most powerful predictors of outcome are the child’s I.Q., the mother’s educational level, and having a non-minority mother. In two-year-olds diagnosed with autism, the most powerful outcome predictor was the quality of their motor (i.e. neurologic) functioning. All of these predictive factors are not part of the diagnosis. Treatment is also not necessarily guided by diagnosis. Behaviorists, starting from Ivar Lovaas, have said that the diagnosis is not necessary as behavioral treatment focuses on behaviors rather than the diagnosis. Medications are also generally targeted to specific symptoms so that the medications currently used by psychiatrists are given to patients based on their symptomology and across diagnostic boundaries.
The diagnosis only describes a small part of what seems to be present in individuals with autism. Close to 100% of individuals with autism have neurologic symptoms other than those described in the diagnosis. There is reason to believe that these symptoms might be central to the development of the diagnostic symptoms. Nonetheless, not being a part of the current diagnostic criteria, they are frequently ignored. Other psychiatric problems, such as anxiety and emotional dysregulation, are ubiquitous in the autism population and are often the target of treatment, yet play no part in diagnosis. So many medical problems have been associated with autism that one study, found that all diseases were found more frequently in autism with the exception of blood diseases, neoplasms and genitourinary disorders. Again, none of the physical disorders are involved in the diagnosis.
Could we do better with a different concept of autism? The answer is not clear; however, I believe we should try. One concept would be to reconsider all of the developmental brain disorders together. It is likely that there is tremendous overlap in the biological and developmental pathways that lead to the outcome. Our understanding of what produces the symptoms might emerge from the study of the symptoms that transcend the individual developmental diagnoses. For example, motor stereotopies are likely the same in a child with autism and in a child without autism but with an intellectual deficit. This is the goal that Dr. Insel is calling for in the research realm. A good question is, would this approach also benefit us in the clinical realm? I believe it would. By demystifying the diagnoses, it would lead to more confidence on the part of primary care physicians to treat the cases. Educators and other treatment professionals need to become more sophisticated at assessment of the child’s needs. Relying on a diagnostic name might make us feel like we are being scientific – even if we are not.
Stay tuned for developments as these changes may help lead to those much sought after research breakthroughs.