Intro: A recent article on the website i09 titled, ‘The Most popular Antidepressants are Based on an Outdated Theory” has again raised the issue of Chemical Imbalances. It is interesting that the author of the i09 piece cites Dr. Peter Kramer and states, “Some psychiatrists vehemently disagree with the way journalists and other psychiatrists have pushed back against the chemical imbalance theory….” Several years ago Dr. Kramer published two essays in support of the chemical imbalance theory – one in the New York Times, and one on his blog site. In both cases he cited what he considered the best evidence in support of the theory, but he did not discuss the research in any depth. Back in 2008, we took an in-depth look at the evidence that Dr. Kramer used to support the chemical imbalance theory. When one takes a closer look at that research we do not think it supports the theory. For this reason, we are reposting our 2008 essay about this.
The following essay was originally posted in 2008. Because of the recent round of media attention we are reposting it here.
A Reply To Peter Kramer: Do Serotonin Imbalances Cause Depression?
The idea that depression is caused by an imbalance of serotonin is often made in the mainstream media. In a recent paper in Society we reported on our efforts to determine what evidence reporters who make these claims were using as supporting documentation. We also had correspondence with a pharmaceutical company, several psychiatrists, and the National Institute of Mental Health. For the most part we received few citations of scientific papers, and those that were cited did not provide evidence for a causal connection between serotonin and depression. Our experience is that there are few scientists familiar with the data that will publicly defend the serotonin theory of depression.
We also believe that debate between researchers holding conflicting viewpoints is an essential part of scientific progress. We were therefore pleased to see that Peter Kramer, author of Listening to Prozac, recently posted a defense of the theory on his blog page. Dr. Kramer had previously made public comments responding to critiques of the serotonin theory in 2006, stating, “While it’s true that one could say that these drug companies are using a very oversimplified metaphor — and a metaphor for something that may not even exist at all — it’s also wrong to suggest that it has no relationship to contemporary theories of mood regulation.” Since we agreed with this statement, we were intrigued by Dr. Kramer’s recent defense of the serotonin theory. In his blog posting, Dr. Kramer cites evidence to support the serotonin theory, much of it from a recent review article in the New England Journal of Medicine. Below, we analyze the evidence presented. All of the studies Kramer discusses are interesting research projects, but for us, the important question is: Do these studies provide enough scientific evidence to support the ubiquitous statements in the media about depression? As just a few examples: “Mental illness are simply chemical imbalances,” or “Depression is thought to be caused by a chemical imbalance in the brain.”
We believe that an in-depth look at the scientific studies cited by Dr. Kramer is enlightening:
1. Alteration in 5-HT1B Receptor Function by p11 in Depression-Like States (2006) Science. 311:p77. In reference to this paper, Kramer states: “Other studies have found that serotonin receptors function less efficiently in depressed patients. A protein named p11 is implicated: depressed patients have less of it in the brain.” The paper in question is a four page paper that examines the interaction of the serotonin 1B receptor with p11 in rodents. Out of the four pages, the authors devote one sentence to p11 in humans. In what can only be called a preliminary report, as very few details are supplied, the authors report that p11 mRNA is decreased in the anterior cingulate cortex in patients who suffered from depression (p.79). The study used fifteen patients and fifteen controls. Seven of the patients committed suicide. Patients who commit suicide are very likely not a fair representation of the average patient diagnosed with depression, but this problem aside, the authors do not mention whether the patients had a history of medication exposure – an obvious question.An e-mail inquiry and subsequent reply from E. Fuller Torrey, the head of the Stanley Foundation, which supplied the brains for the study, confirmed that the majority of the patients had been prescribed medication – whether they filled these prescriptions and/or took the medications is unknown. Dr. Torrey did mention that one possibility is that the patients who committed suicide stopped taking their medication, implying that the lack of medication was a contributing factor to their suicide. Thus the lower level of p11 in the depressed patients could be due to long- term antidepressant use, to the withdrawal effect, or to a combination of the two. Therefore, the major piece of evidence that Kramer uses to defend the serotonin theory is a comparison of 15 patients with a history of medication use, which was only one minor component of a larger study. We do not think that our concern with prior medication use is excessive. In the words of Ross Baldessarini, esteemed professor of psychiatry: “Almost any psychotropic drug that’s given for more than a few weeks leads to changes in brain function such that when you stop, the brain has to reset its thermostat.”
2. Major Depressive Disorder, New England Journal of Medicine (NEJM), 358:55. In this review article, the authors discuss numerous avenues of research into depression. While the authors discuss serotonin they do not supply any direct evidence for the idea that low serotonin is the cause of depression. For instance, they discuss tryptophan depletion studies. Tryptophan is a rate limiting enzyme involved in serotonin production in the brain. It was originally hypothesized that tryptophan depletion would lead to low serotonin, which in turn would lead to a lowering of mood. Yet, as the authors point out, in healthy subjects this has not been the case.The authors also discuss an oft-cited study by Caspi et al. which investigated common polymorphisms of the promoter for the serotonin transporter gene. Yet, the NEJM’s piece also includes the oft-made characterization about the study, which leaves out some important details. According to the NEJM, “Caspi et al. found that 5-HTTTLPR predicted depression only in association with defined life stresses” (p.56). However, this needs to be qualified by pointing out two important points: 1) Just the presence of the short form of the gene did not predict depression – after all following major life stressors only 35% of the short form carriers developed depression (If 65% of the carriers do not develop depression, it seems hard to characterize it as predictive). One needs to qualify the statement in the NEJM by saying that carriers of the short form were more likely than carriers of the long form to develop depression after major life stressors (It is the comparison that is important). Yet, besides pointing out that it is the difference between long and short form that is important, even more qualifications are warranted. 2) In addition, “defined life stressors” must be qualified. Because the carriers of the short form were at a disadvantage only if they were exposed to three or more major life stressors. For those people who only suffered one or two life stressors, the findings were reversed and the patients with the short form actually were less likely to be depressed – in this case the short form was protective (See Figure 3, p. 389 of the Caspi study). And a subsequent study by Shelley Taylor and her colleagues at UCLA found that people with two copies of the gene’s short form were actually less likely to have depression symptoms if they had positive childhood experiences. Also of note with the Caspi study is that the marker in question is not a rare mutation found in only a small percentage of the population. Approximately 70% of the population carries this “susceptibility “gene. One possible conclusion is that they have not discovered something unique about the subset of people diagnosed with depression, but have discovered something about human nature in general. Of course there are mechanisms in the makeup of the human brain that respond to environmental stressors. But finding these mechanisms does not mean that we have found a marker for a disease. If 70% of us have a gene that plays a role in our response to stress should this be considered “normal” or “pathological?” Complicating interpretation of this data, while some studies have found similar results to Caspi, other groups have not (Gillespie, 2005).
Kramer also touches on the efficacy of the SSRIs when he says, “The NEJM notes that a third of depressed patients do not respond to antidepressants. In those who do respond, what the authors call ‘the monoamine-deficiency hypothesis extended’ remains the most powerful explanation of the drugs’ mechanism of action.” Leaving efficacy aside, it is important to point out that showing how a drug acts at the molecular level does not necessarily equate to understanding the etiology of a condition – in this case determining that a drug acts on the serotonin receptor or serotonin system does not prove that depression is caused by low serotonin. If we followed the logic that “mechanism equals disease” then we could assume that every drug with a behavioral effect on the nervous system does so because of a chemical imbalance – something which is certainly not true. In many cases a drug will relieve a symptom but not treat the root cause of the problem. Take a long distance runner with a pulled muscle. Pain medication will relieve the pain – a symptom – but we do not attribute the cause of the pulled muscle to a chemical imbalance. Many drugs – both legal and illegal – act on the monoamine system: SSRIs, amphetamines, Ritalin, mushrooms, coffee, cocaine, etc. The authors of the NEJM piece state that: “A strong point of the monoamine theory is its predictive power. Almost every compound that has been synthesized or discovered for the purpose of inhibiting norepinephrine or serotonin reuptake has been proved to be a clinically effective antidepressant.” We would go one step further and point out that based on the predictive power of the monoamine theory, even though the legal and illegal drugs have similar pharmacological properties, the chemical imbalance theory will only be applied to the use of drugs sold by a pharmaceutical company.
3. “Elevated Monoamine Oxidase A Levels in the Brain: An Explanation for the Monoamine Imbalance of Major Depression,” Archives of General Psychiatry. This paper is one of the more interesting papers on this topic. MAO-A is an enzyme that breaks down the neurotransmitters serotonin, norepinephrine, and dopamine. The authors’ hypothesis is that MAO-A levels are higher in the brain during untreated depression, which they believe could explain the monoamine theory of major depression. They used positron emission tomography (PET) scanning on 17 patients diagnosed with depression and 17 healthy patients, and measured their levels of MAO-DVs, an indicator of MAO-A. The group of depressed patients had higher mean (average) levels of MAO-A. Thus, one interpretation is that these higher levels of MAO-A enzymes cause major depression. After critically appraising this paper, we have several questions:.
Does this paper establish that MAO-A is the cause of major depression? Or, alternatively, does it simply document an association? We wonder if higher mean MAO-A among a group of depressives might be a sign of depression, rather than the cause. Causality requires that the higher MAO-A directly caused the depression; usually, this would be established by documenting that (a) higher levels of MAO-A were found in the brain before onset, or at the time of onset, and (b) by ruling out all other potential causes. This experiment accomplished neither.Consider that many people get nervous before public speaking, and, frequently, their heart rate quickens and their blood pressure goes up. But most people would not say that high blood pressure is the biochemical explanation for public speaking anxiety. Similarly, how do we know that lower MAO-A in depressed individuals is not simply a sign of depression, rather than the cause? Certainly, further experiments are required to establish a causal role for MAO-A in depression.
Have the results been replicated? This is a study of 17 depressives and 17 healthy patients in Canada. Time will tell if independent research teams are able to replicate this finding with other patients. It is possible that the finding has been replicated and we are unaware of it. Certainly, important findings should be replicated, repeatedly, by as many different research teams as possible.
Are the results specific to depression? Keep in mind that MAO does not just break down serotonin, but also norepinephrine and dopamine. These three neurotransmitters are implicated in almost every DSM-IV condition. In future experiments, rather than healthy volunteers, it would be interesting to see patients diagnosed with major depression compared to patients diagnosed with other non-affective mental disorders, such as ADHD, anxiety disorders, etc. Until such experiments are performed, we think many will question whether these findings are specific to depression. In the published paper Meyer et al mention that 6 of the 17 patients did have a prior history of medication use. They also report that based on secondary post hoc analysis of various clinical characteristics, one of which was prior medication use, none of the correlations reached the trend level. However in the paper they do not report which patients were medication naïve. In correspondence with the lead author of the paper, Dr. Jeffrey Meyer, he graciously shared with us which patients had a history of medication use. When one looks at this data the previously medicated patients tend to lie at the extremes for MAO-A levels. For instance in the prefrontal cortex and the caudate, the patients with both the highest and lowest levels for MAO-A are medicated patients, and as a result, these patients do not influence the overall group mean. However, this pattern is of interest, as the early studies by Seeman and Lee of altered dopamine receptors in the brains of schizophrenics were tempered by the fact that the studies also included medicated patients. Subsequent studies by other investigators were unable to replicate the findings in non-medicated patients.
Are the author’s conclusions congruent with the existing clinical trial data? Recent research has challenged the efficacy of SSRIs. For instance, a recent systematic review concluded, “Among adults with moderate to severe major depression in the clinical trials we reviewed, paroxetine was not superior to placebo in terms of overall treatment effectiveness and acceptability” (Barbui, Toshiaki, & Cirpiani, 2008, p. 296). We are confused as to how such findings lines up with the proposal that serotonin imbalance causes depression. Michael Thase, an academic psychiatrist and pharmaceutical company-funded clinical trialist, has proposed that one or two out of ten patients prescribed an SSRI have a clinically significant response. Would proponents of the chemical imbalance theory interpret this to mean that 80-90% of patients prescribed SSRIs do not have a serotonin imbalance?
Is the statistical reporting adequate? We find it interesting that the Meyer et al. 2006 article, published in the highly-respected Archives of General Psychiatry, does not include basic information that is of interest to readers. For instance, in social science research, descriptive statistics are commonly presented. In this article, simple tables presenting the means and standard deviations for each brain area would have been helpful. The authors do not report the difference in mean MAO-DV between depressed and healthy patients; they only report the difference as “34%, or an effect size of 2.” We cannot determine how his effect size was calculated, and since there are several potential methods, this information would be useful. The authors address the uncertainty inherent in all quantitative studies by reporting p-values. Since the statistical techniques utilized require random sampling, we are skeptical of the value of p-values in a study like this. However, if one is going to quantify sampling error, confidence intervals are the commonly accepted technique (to put it mildly, only reporting p-values is the least conservative statistical approach). A confidence interval would express a range of uncertainty, for instance, “We are 95% confident that the difference in MAO-DVs between depressed and healthy patients is between 2.1 and 5.9 points” By eschewing confidence intervals and just presenting a single statistic to support their conclusions (“34%, effect size of 2” ), the authors imply an unwarranted level of precision. We do not know the reasons why this information was not presented, but, overall, we are surprised to find this type of statistical reporting in such a technologically sophisticated study.
Thus, while the Meyer study is an interesting piece of research, and is an important contribution to the field (and we certainly look forward to seeing future reports of MAO) we do question using it, as Kramer does, as the main piece of evidence in support of the idea that depression is caused by low serotonin. Keeping the historical context and scientific limitations of the study in mind, consider what Kramer is up against: 1) For almost 40 years, there have been unsuccessful efforts to find evidence for the serotonin theory of depression, and even Meyer et al. state, “…no convincing mechanism of monoamine loss has ever been found,” 2) after decades of failed research along comes a 2006 study that has not been replicated (if this is the first study to find evidence for the theory, presumably those media reports prior to 2006 which stated the theory had been proven were false), 3) the study only included seventeen depressed patients, 4) six of the depressed patients had a previous history of medication use, and we don’t know which medications the patients took, 5) there is overlap of MAO DV levels between patients and controls, 6) the medicated? patients lie at the extremes of the MAO DV levels, 7) within psychiatric research there is a long history of initial findings with failed replication attempts, and 8) we do not know if this finding is unique to patients diagnosed with depression or is found in other patients diagnosed with other conditions that list serotonin, norepinephrine, and dopamine as possible culprits. (As an addendum, as of 2015 this work has not been replicated)
We are heartened that Dr. Kramer has put in writing what he thinks is the best evidence for the serotonin theory, however we continue to wonder if this evidence justifies the massive marketing campaigns stating that chemical imbalances are the cause of depression. In the matter of informed consent we question whether patients who are told about the chemical imbalance theory really understand how little direct evidence there is for the theory. When scientists discuss the biological basis of mood with other scientists, it is almost as if there is a set of ground rules that everyone agrees with, such as: (a) There are extremely subtle nuances to many of these studies; (b) it is important to refrain from drawing too many conclusions from the latest findings; (c) that finding a biological difference between people diagnosed with depression and “normals” does not automatically warrant the jump to “causation; and (d) that environmental triggers play an extremely important role in why someone’s mood might be lowered. Yet, in the popular press all these qualifications disappear and instead the public is inundated with declarations about “chemical imbalances causing mental illness.” In a sense there are two entirely different discussions going on about the theory: In the media and the advertisements it is presented as simple and straightforward, but in scientific circles it is treated as tenuous, at best. In scientific circles the debate seems to be not so much about the strength of the theory but about the appropriateness of using it with patients – some prominent scientists have stated that the theory is a metaphor which they will not use with their patients, while others acknowledge it is scientifically problematic but defend using it with their patients.
We encourage reporters to dig deeper and to question whether the evidence in support of the theory justifies simplistic claims about the theory, and to question in more detail those psychiatrists and other experts who makes claims about the theory. In his essay Kramer points out that the miserably depressed Zoloft Ovoid creature is smarter than the critics have made him or her out to be. We agree with Dr. Kramer that the Ovoid is intelligent, however, given its ability to convince the media that there is substantial evidence in support of the chemical imbalance theory, it appears that the Ovoid’s talents seem to lie in the marketing, and not the scientific, realm.