MARCH 4, 2016
The Holy Grail of Medicine
THE RIGHT DRUG for the right patient at the right time is the holy grail of medicine. Doctors come closest to this grail when they treat an infection: they identify the offending bacterium and then, based on its identity, prescribe a specific antibiotic. Still, the strategy is rather generic: it’s aimed at the affliction itself, not tailored to the particular patient with the affliction.
To be sure, this one-size-fits-all approach often works, but it can miss the mark, sometimes with tragic consequences. What if doctors could treat diseases based on the specific genes, enzymes, and biochemistry of a patient? This is in fact the goal of an ambitious new project called “precision medicine.” In the eyes of many scientists, precision medicine shimmers on the therapeutic horizon. For physicians and patients, it’s the stuff of dreams.
My friend Rick is living that medical dream. In the fall of 2014, his doctor discovered the cause of a persistent pain that had settled in his lower back. It was a fist-sized “osteochondrosarcoma,” a notoriously aggressive tumor of bone and cartilage. The tumor did not respond to chemotherapy, and so, as a last resort, researchers tested it for mutations.
Analysis revealed that the cancerous cells carried a mutated gene that produced an aberrant form of an enzyme essential for cellular metabolism. The deviant enzyme had unleashed a process that caused the cells to remain immature and replicate too quickly. Fortunately, researchers had just recently, in the knick of time for Rick, developed a “designer” medication (a single molecule, really) that inhibited the chaos-inducing action of the defective enzyme. The glorious result: Rick’s tumor has shrunk markedly, and he is doing well.
Though precision medicine is far from established therapy, a handful of lucky patients with cancer and other conditions are benefiting from tailored therapy. Take cystic fibrosis. Its prognosis is usually grim — suffocating levels of mucus clog the lungs and other organs; but it is one of the few diseases caused by a single dominant gene mutation. Last year, scientists developed a medication for a variant of cystic fibrosis in which there is a specific mutation in the “cystic fibrosis transmembrane conductance regulator” gene. That gene codes for a protein that regulates the migration of ions in and out of cells in the lungs. The medication corrects the protein so that the cell membrane allows chloride to pass, thereby averting dangerous mucus build-up.
In January 2015, President Obama allocated $215 million to the kind of therapy that saved Rick. Indeed, much of the president’s so-called Precision Medicine Initiative targets cancer — and understandably so. It is after all a disease of genetic mutations leading to unregulated cell division. Not only is cancer common and often lethal, but many of the standard treatments are ineffective, and side effects of chemotherapy and radiation are brutal. Scientists have made great strides in understanding cancer biology, and so the moment is certainly ripe for precision medicine. The daunting challenge, however, is that most cancers are not the product of a single renegade mutation but rather of a complicated lifetime’s worth of interplay among genetic influences, environmental exposure, and lifestyle choices such as diet and smoking or drug abuse.
The classic “precision” model thus has two key steps: locating a mutation that drives the pathology and developing a treatment that disarms the harmful product of that mutation. But what about precision medicine in other arenas, such as psychiatry? Can precision medicine address mental illness? Is this realistic, even possible? The National Institute of Mental Health, part of the National Institutes of Health, believes it is possible, though even the most avid boosters acknowledge that it is a formidable project. The foremost challenge is to envision how alterations in particular brain functions translate into defects of the mind.
Let’s set the stage for such a discovery by affirming that Cartesian dualism is dead. The material brain, in other words, produces all the contents of the ephemeral mind. Every subjective experience, from the ache of nostalgia to the frisson of a Christmas morning, corresponds to physical events in the brain. The mind — the realm of feelings, desires, ideas, memories, intentions, and subjective experience — is produced by the action of neurons and brain circuits. The fact that physical mechanisms generate mental experience is not a new discovery. Hippocrates famously noted that “it is specially the organ which enables us to think, see and hear, and to distinguish the ugly and the beautiful, the bad and the good, pleasant and unpleasant.”
Theoretically, mutations in the genes of neurons can contribute to mental illness. After all, many psychiatric conditions are highly heritable in that genetic influence, rather than environmental inputs, accounts for a high percentage of risk. But the cancer model — find a defective gene, devise a corresponding treatment — is not remotely realistic for mental illness. This is because various mental conditions are linked to dozens, even hundreds, of genes.
In schizophrenia, for example, 108 genes have thus far been implicated, each contributing only a small amount of vulnerability. Some of the genes are involved in dopamine receptor function and glutamine neurotransmission. Others are located in “histocompatibility regions,” which suggests immune, autoimmune, or infectious processes have a role in the development of schizophrenia. Several weeks ago, scientists made the discovery that a gene known as C4A within the histocompatibility region strongly contributes to the pruning of synapses (nerve cell connections) in the frontal lobes. Pruning occurs normally in adolescence, but it happens much more aggressively in people with schizophrenia, due to a mutation in C4A. As a result, the tissue in the frontal lobes, the region involved with planning complex cognitive behavior, personality expression, decision-making, and moderating social behavior is winnowed too much.
The unraveling of the genetic basis of pruning is a very exciting finding, but it only accounts for perhaps one-quarter of the risk of developing schizophrenia. Other factors need to be present too. Given the overall risk of schizophrenia is one percent, the presence of a C4A mutation increases the risk of developing the condition to 1.25 percent. Eventually, what we call “schizophrenia” will probably turn out to have hundreds of different causes and so will require a number of different treatments. It’s thus hardly any wonder that it is so difficult to develop a clear picture of who is really at risk for what.
Evidence regarding genetic risk is collected through “genome-wide association studies,” an approach that involves rapidly scanning markers across the complete sets of DNA, or genomes, of many people to find genetic variations associated with a particular disease. The NIMH has been examining the genetic basis of psychiatric conditions through the Psychiatric Genome-Wide Association Study consortium since 2005, but the prevailing theme in the results thus far is that tracing psychopathology to one or a handful of errant genes is a fool’s chase. Researchers have therefore turned their attention to a more macro level of explanation: brain systems (neural circuits) and their interconnections.
As researchers begin to undertake the gargantuan task of exploring the neuroscience of psychiatry, they face a big hurdle: namely, the fact that psychiatric diagnostic categories (e.g., schizophrenia, bipolar disorder, posttraumatic stress disorder, and so on) were not “built” with the underlying biology of these syndromes in mind. Rather, the diagnoses contained in the fifth edition of the Diagnostic and Statistical Manual (DSM), published by the American Psychiatric Association, are merely descriptive, codified around “symptoms” (subjective self-reports, such as feeling suicidal, hearing voices) and “signs” (overt manifestations, such as agitation, bizarre behavior, excessive drug use).
The psychiatric diagnoses that are by now household terms — major depressive disorder, obsessive compulsive disorder, anorexia nervosa, autism spectrum disorder — are famously “atheoretical.” This means that the diagnoses themselves hold no clue as to their etiology. Compare this to “streptococcal pneumonia.” The name itself tells us that the cause of the lung infection is the streptococcus bacterium. Analogous formulations for psychiatry do not exist even though the earliest figures in neurology and psychiatry, including Freud, longed for brain-based explanations.
For over a century, psychiatrists have believed that biological revelations were around the corner. In the early 20th century, Emil Kraepelin, the famous German psychiatrist who laid the foundation for the DSM, expected that the brain basis of psychosis would become manifest in his lifetime. In the 1970s, brain research and especially receptor biology were opening new avenues for psychopharmacology. By 1980, however, when the modernized DSM III (earlier editions had been organized around Freudian concepts) was published, its authors were resigned to the fact that their neatly circumscribed illness categories had failed to “carve nature at its joints,” as Plato might have put it. Still, they were largely confident that biological causes of mental illness would be uncovered by the turn of the 21st century. At the very least, they thought, we would soon have blood or spinal fluid tests, brain scan patterns, EEG information, hormonal indicators — surely, something — that would tell us what the patient “had.” Yet, three years ago, when the latest DSM edition appeared, it still contained no new indicators of pathology, only signs and symptoms as before.
Nonetheless, the DSM system remains indispensable for an important reason that social scientists call “reliability.” A diagnostic scheme is reliable when two or more clinicians can examine a patient according to that scheme and agree (with high likelihood) on the category into which that patient falls. Clinicians use the DSM as a common language to discuss patients, and researchers use it as a shared blueprint to guide systematic inquiry. Yet, reliability of a diagnosis is not the same as “validity,” or accuracy, of a diagnosis. In medicine, accuracy can be determined in several ways, most prominently by whether a diagnosis links to a unique, identifiable biological origin. Other indicators of validity include whether a diagnosis predicts the course of an illness and what kinds of treatments the illness will respond to.
Current psychiatric diagnoses are limited because they have negligible validity. Abundant evidence attests to this. It is not uncommon, for example, for a psychiatric patient to meet qualifications for several diagnoses at once. This is because the sharply delineated diagnoses in the manual do not correspond to independent, discoverable entities in nature. Indeed, in real life, boundaries between diagnoses are fuzzy. As a result, many patients do not fit snugly into any of the procrustean categories.
It also explains why doctors typically use trial and error to find the most effective medications, dosages, and combinations for various symptoms. A large study of major depression found, for example, that recovery with the first selected selective serotonin reuptake inhibitor (e.g., Prozac) occurred only about one-third of the time. In addition, because diagnoses do not represent discrete underlying psychobiological mechanisms, psychiatrists typically, though not always, find themselves targeting symptoms rather than disorders. So, while psychiatrists will almost surely prescribe an antidepressant for a patient with major depressive disorder, the very same antidepressant drug can be helpful in obsessive-compulsive disorder, eating disorders, and panic attacks. Conversely, a single condition may require one medication for each symptom. The manic phase of bipolar disorder, for example, often requires a mood stabilizer to control the excursions of mood and a sedating antipsychotic to combat accompanying paranoia, and, perhaps, a benzodiazepine, such as Valium, to quell agitation.
Finally, the short-term prospects for new pharmacotherapies are not particularly encouraging. Industry has reduced investment because companies haven’t made progress in finding new chemical compounds, which in turn is probably due to the heterogeneity and crudeness of the treatment targets.
Early Inroads into Precision Psychiatry
Investigating the biological dimensions of mental illness means that researchers must be freed from the bounds of conventional diagnostic categories. In 2008, the National Institute of Mental Health embarked on a strategy rooted in “domains” of brain function. The strategy is called “RDoC,” or Research Domain Criteria project. Each of the five domains in the RDoC schema represents a capacity that is essential to human survival and, thus, has been evolutionarily conserved. These domains are as follows: cognitive systems; negative valence systems; positive valence systems; systems for social processes; and arousal/regulatory systems.
So, for example, the capacity for paying attention would fall under the domain of arousal and regulation. Humans need to focus their attention lest they be crippled by distraction. A problem with an attentional system that fails to filter extraneous inputs (as we find in attention deficit and hyperactivity disorder, or ADHD, and schizophrenia) will manifest as some degree of impairment. Within each domain there is a set of related constructs. For example, within the Cognition domain, we find Auditory Perception Construct and, within that, further breakdown into types of auditory hallucination. Within the Negative Valence System, we find Acute Threat Construct. Phobias, anxiety, and excessive fear responses represent types of acute threats.
Research under a strict RDoC paradigm thus dispenses completely with formal diagnoses and instead focuses on the neural circuitry tied to psychobiological systems. At this time, only a modest body of research has been performed under such conditions. The NIMH presents a recent study as an example of an RDoC-inspired investigation, but in fact it uses conventional diagnoses as a starting point. Psychiatrist Carol Tamminga at the University of Texas Southwestern Medical Center and her team examined about 800 patients diagnosed with one of three conditions in which psychosis is a prominent feature: schizophrenia, schizoaffective disorder, and bipolar disorder.
The researchers asked patients to complete a battery of neurocognitive and perceptual tasks to assess planning and memory capabilities, eye-tracking, inhibition, and brainwave responses to auditory stimuli. They identified sets of biomarkers that differentiated subgroups of patients. In addition, 1,000 of the patients’ family members and 250 healthy controls performed the same tasks. The researchers were able to identify three homogeneous “biotypes” underlying psychotic symptoms.
Biotype 1 were most impaired in exerting control over attention and information-processing, and were the most socially withdrawn. Biotype 2 cases showed intermediate levels of impaired cognitive control but had a normal to accentuated ability to detect and process visual and auditory stimuli. Those classified as Biotype 3 processed information normally but had some problems with sensorimotor reactivity. They performed the best socially and had the lowest levels of hallucinations and delusions.
These findings, pending replication, are important pieces of a daunting tapestry. The treatment implications, however, remain a matter of speculation. Perhaps, as Dr. Tamminga and her team think, treatments for Biotype 1 should target cognitive control and enhance brain mechanisms for discerning the relevance of environmental stimuli. Cases with Biotypes 1 and 2 might be good candidates for medications that correct neuronal activity levels through effects on cellular potassium or calcium channels. Biotype 3 cases, the team hypothesizes, might benefit from preventive efforts because environmental contributions, such as heavy marijuana use, seem to figure importantly in psychosis risk for that group. Anti-psychotic medication at a low dose would likely be needed as well.
Meaning, Context, and Biology
Precision psychiatry is a very long game. Many think it is also a very long shot. The odds of finding a single mutation that wields a large and straightforward influence on any given mental disorder are virtually nil. Thus far, said one of my colleagues, a schizophrenia researcher, “NIMH’s moonshot decoding of the genome has produced so little which is clinically useful.” With respect to RDoCs, Canadian psychiatrists Joel Paris and Laurence J. Kirmayer represent skeptics in the field when they write that it “seems to be based on the unrealistic assessment of how much neuroscience can explain without [paying] attention to patients’ experience and life world contexts […] [and thus] it downplays the role of psychosocial factors in psychopathology and treatment.”
The designers of the RDoC criteria say they are very much aware of the pitfalls of explanatory reductionism: the all-too-common temptation among some neuroscientists to believe that the mind can be explained solely by knowing about the brain. Indeed, we cannot use the physical rules from the action of molecules or cells to completely predict psychological or behavioral activity. This is because the brain and the mind represent different levels of explanation of the same phenomenon. When we talk about the brain, we talk about mechanisms underlying perception, emotion, and cognition; and when we talk about mind, we talk about awareness and meaning — and biological descriptions cannot capture that.
Now layer in the effect of the environment, both physical and social. The biology of our brains is shaped by the intrauterine milieu, toxic exposures, early life experiences, and cultural imperatives. Feedback loops abound. For example, a child whose innate disposition is calm and politely curious is going to elicit benign responses from adults, whereas a demanding child may provoke feelings of resentment. Those elicited responses further mold the child. Growing up in a home populated with hostile or indifferent adults has its own impact on a developing psyche, which, in turn, elicits responses. And so it reverberates.
Refracting treatment through the lens of biology can have unintended consequences, too. Consider a child whose innately high level of hyperactivity causes his parents to lash out; an unhealthy parent-child dynamic develops and the family seeks help. It would be unfortunate if the clinician reasoned that, because the etiology of the child’s problem is ultimately biological, the therapy must always be medical in nature. Strategies aimed at modifying the home environment and parental style may yield considerable improvement.
Psychiatry has made undeniable clinical and research progress over the past half-century, but within the last few decades the field has not scaled major therapeutic mountains. Unlike the case for cancer (where genomic analysis is proving fruitful), the future of precision psychiatry will not be found in genes. The kind of precision we can realistically hope for in psychiatry, most scientists believe, will likely be the identification of pathological subtypes according to defective brain systems. How well this will translate into the development of new therapies or the refined application of existing ones remains to be seen.
The brain is the most intricate system in the known universe. Frustrating proof of this complexity is the fact that the biological basis of mental disorders has eluded us for centuries and largely continues to do so. The current state of neuroscientific knowledge is simply not mature enough to devise therapies based upon brain mechanisms. Nonetheless, there is little question that researchers are on a great odyssey of discovery about how the brain works.