The Puzzle of Lithium

Why does lithium work for some people with bipolar?

This one question has puzzled and continues to puzzle the scientific community.  It is known that in the brains of people with bipolar, schizophrenia, and other mood disorders, the volume of gray matter is decreased relative to healthy controls.  Lithium works very well for a large portion these people ergo the hypothesis that lithium works by increasing gray matter in the brain.

G.J. Moore, J.M. Bebchuk, K. Hasanat, G. Chen, N. Seraji-Bozorgzad, I.B. Wilds et al. Lithium increases N-acetyl-aspartate in the human brain: In vivo evidence in support of bcl-2’s neurotrophic effects? Biol Psychiatry, 48:1-8, 2000.

Indeed this seemed to be true from imaging studies that appeared to show an increase of gray matter in people who take lithium (for both people with and without bipolar).  However a recent study

Cousins et al. Lithium, Gray Matter, and Magnetic Resonance Imaging Signal. Biological Psychiatry. 73(7):652-657, 2013,

shows that the when the lithium ions are mixed with water, they show increased volume on the images. This finding contradicts the hypothesis that lithium works by increasing the volume of the brain.

We are still left with the puzzle of why and how lithium works.  Without a definitive scientific explanation, we are left to rely on phenomenology to determine whether lithium works for a given person.  It is up to the patient based on their first-person experience to determine whether lithium works for them.  This is why a respectful therapeutic relationship is needed, and the role of the patient's experience of the world must largely influence treatment.  Despite the wishful thinking of some doctors or psych hospital staff, doctors cannot determine on their own whether a particular course of treatment is going to work.

Mixed Episodes Redux

Earlier in a post titled Mixed Episodes we discussed the DSM definition of a mixed episode and we discussed some other ways of viewing mixed episodes.  Adding to that discussion is a recent article:

Swann et al. Bipolar Mixed States: An International Society for Bipolar Disorders Task Force Report of Symptom Structure, Course of Illness, and Diagnosis. Am J Psychiatry. 170:31-42, 2013.

This article is the result of discussions and several conferences involving a long list of people who appear on the author list of the paper.  The point of the article is best summarized by its conclusion:

Components of what are considered manic and depressive states can combine in bipolar disorder. Mixed features may be associated with illness course and treatment response characteristics distinct from more exclusively depressive or manic states. Clinical characteristics, including co-occurring conditions, suicidal behavior, anxiety, poor treatment outcome, and severely recurrent and complicated course, appear stable across definitions and criteria for mixed states. The importance of recognizing and monitoring mixed features during a hypomanic, manic, or depressed episode is highlighted by their relationship to recurrent course, treatment resistance, co-occurring substance use, and potential for suicidality.

The article does give a specific clinical algorithm for identifying mixed states, similar to the algorithms that the DSM gives.  The primary type of episode is identified (manic or depressive) with concomitant identification of an episode as mixed if it has sufficiently many non-overlapping symptoms from the 'opposite' type of episode.  Symptoms are said to be non-overlapping if they are unlikely to occur in both non-mixed depression and non-mixed mania.

There were several interesting sets of symptoms brought up by the article.  The authors segregate symptoms as to whether they are manic or depressive.  Then they segregate mixed episodes as to whether they are primarily manic or primarily depressive.

The authors say that manic symptoms of maniac mixed episodes are "greater mood lability and irritability and decreased grandiosity, euphoria, pressured speech, and need for sleep ," and depressive symptoms of manic mixed episodes are "dysphoric mood, anxiety, excessive guilt, and suicidality" are symptoms of mixed episodes.    In depressive episodes, manic symptoms include "irritable mood, distractibility, racing thoughts, and increased talking."

There is a collection of symptoms that the authors say characterize mixed episodes regardless of the primary categorization (into mania or depression) of the mixed episode.  These include "anxiety, agitation, or psychosis."  About anxiety the authors say

Anxiety has been shown to correlate with depressive symptoms in manic episodes, with manic symptoms in depressive episodes, and with the degree to which symptoms were mixed regardless of polarity. This pervasive role of anxiety is consistent with mixed states being driven by hyperarousal. 

Agitation is a combination of hand-ringing and uninhibited impulsive action.  Psychosis, as we know, is typified by thought disorder, including paranoia, delusions, and hallucination.  Psychosis is thought by the authors to be present primarily in either manic or mixed episodes.

While this article represents a step forward in terms of diagnostic tools for bipolar, I have reservations about it.  This article makes heavy use of the binary description of bipolar while simultaneously breaking the binary thinking and expanding it to be trinary.  The authors want to have their cake (the dichotomy) and eat it (make it trinary), too.  Also, they assume that one can always identify a primary episode type.  I suspect this last assumption is based on circular logic.  I think there are examples of episodes not having a clearly distinguishable primary type, and yet these episodes should still be classified as being bipolar episodes.

I am an example of a person having episodes that fall between the diagnostic cracks.  This has previously been discussed in the post titled My Diagnosis or Lack of One.  One comforting thing about the article mentioned above is that I finally find a description that matches my experiences.  This is comforting, because I know that I am not the only one who experiences the combination of symptoms that I experience.  In particular, my episodes have been marked by anxiety, agitation, and psychosis.  According to the authors it would seem that I have been having mixed bipolar episodes.  However, I find this categorization of my episodes to be unsatisfactory.  The popular conception of bipolar is wedded to the idea of a dichotomy between mania and depression.  My experience has nothing to do with such a dichotomy.  I experience no highs, no extreme lows.  On the other hand, the drugs that help me also help people with the traditional bipolar.  So, it seems to me that we are missing some important biological connection between patients exhibiting these disparate sets of symptoms.

The analysis performed by the authors is on individuals who are incontrovertibly diagnosed as bipolar, according to the binary definition of bipolar.  Then the authors take these people and group their symptoms into the binary categories of mania and depression. The authors then use that grouping of categories to decide whether an episode is primarily manic or primarily depressive.  Does this sound like circular logic?  They first assumed there were categories for mania and depression, then used patients diagnosed according to this dichotomy to establish whether symptoms and subsequently episodes were primarily manic or depressive. The correct statement of their argument goes something like this: if there exist two poles for the bipolar disease, then there is a grouping of symptoms into the two poles and a subsequent categorization of episodes into primary types of mania and depression.  But the classification of episodes cannot provide evidence for episodes having a primary manic or depressive categorization, unless we first assume that there is a dichotomy matching the labels mania and depression, making the logic circular.

Another problem with the article is that most reasonable experiments involving phenotype data would yield the result that there is a mixed trait.  This is because such a study would be based on the false binary assumption that there are two clusters of phenotypes.  I can take any cases (individuals with a disease) having any binary disease along with a set of multiple symptoms that have some noise (meaning that the symptoms do not exactly predict the presence of the disease).  The binary analysis described above can be performed whereby the cases are grouped into two poles based on some of the symptoms.  Most certainly, since there is noise, there will be symptoms that do not segregate exactly with the two poles as they are defined by a few of the symptoms. As a result it will seem as if there are cases with 'mixed' symptoms.  And the 'mixed' symptoms would be an artifact of the incorrect assumption that there is a dichotomy describing the disease.

It would seem that the simplest explanation for the presence of mixed episodes is that we have a collection of noisy symptoms that do not describe a disease having binary poles.  This would be an application of Occam's razor where we reject the introduction of a more complicated mixed bipolar model involving manic, depressive, mixed manic, and mixed depressive episodes, and we instead entertain simpler models which not based on a binary description of the disease symptoms.  For example, if we are committed to having phenomenological diagnostic tools, we could diagnose bipolar and schizophrenia from a single bucket of symptoms.  The diagnostic algorithm would be something like the following: if a person has at least 6 of the symptoms displayed in an episodic fashion then they might benefit from lifestyle changes, therapy, and anti-pychotic and mood-stabilizing drugs.  This approach would better reflect what we know about these diseases than current approaches based on false dichotomies.

Such approaches could facilitate investigations into the biology of these diseases, perhaps recognizing that there are changes in the brains of people with bipolar and schizophrenia.  Maybe our current biological studies are inhibited by the ascertainment bias that comes from our insistence on a separation between mania and depression.  Perhaps we would discover biological diagnostic tools faster if we were to discard the notion that there is a polar distinction between mania and depression.  In a similar vain, perhaps the distinction between schizophrenia and bipolar is something that needs to be reconsidered as well.  Perhaps there is a common biological basis for both schizophrenic episodes and mixed episodes.

I see that the biggest weakness of this article is that it clings to the idea that there are two poles of bipolar.  While the article seems at first glance to break the binary by introducing a definition of mixed episodes, a closer look reveals that the authors diagnosis of mixed episodes only reinforce a possibly false dichotomy between mania and depression.  It is certainly not the case that there are two main types of episodes; this thinking seems to me to be an artifact of our attempt to understand this disease by categorization.  It is time to move beyond the binary or trinary thinking and look for a biological basis from which to diagnose this disease.

Antidepressants and Pharmacogenomics

Tansey et al. Contribution of Common Genetic Variants to Antidepressant Response. Biological Psychiatry. 73(7):679-682, 2013.

PsychCentral posted a news article about this study.

The above mentioned study is a pharmacogenomic study that performed a genome-wide association study of to detect genetic variants that are associated with antidepressent drug response.  Since the previous sentence uses a lot of jargon, I will try to break it down.

Pharmacogenomics is the study of how the genetics of individuals influences their biochemical response to drugs.  Namely, we are interested in how well a particular drug treats the disease that it is designed to treat.  So, these studies are usually an investigation of the main-effects of a drug, rather than an investigation of the side-effects.  (Although the pharmacogenomic approach can also be used to study side-effects, it is hard to get sample sizes that are large enough to find anything with such a study.)  The reason we are interested in pharmacogenomic studies is because drug response is partially genetic.  For many diseases we know that whether or not a drug works for your parent(s) can well predict whether the drug works for you.  So, not only are diseases genetic, but the effective treatments can also be genetic.

To explain how traits can be genetic, first we need to understand what genetic variation is.  Humans have two copies of every chromosome, and we have a distinct sequence of nucleotides (the molecular units that make up DNA) on each chromosome.  There are sites or positions in the genomic sequence where if we look across many individuals, we find that there are multiple nucleotides that appear at that site.  These sites of variation are called single nucleotide polymorphisms (SNPs).  If a SNP is responsible for a disease trait then there would be a single mathematical function that relates for each person the nucleotides that they have at the SNP site to disease trait.  The simplest examples involve single SNP sites that are Mendelian diseases (either recessive or dominant) and have binary disease traits, or traits that either exist or do not exist.  We call diseases complex (or polygenic, meaning that it involves many sites) when the trait is a function of the nucleotides at several SNPs.  Now we have described the generative model, i.e. given the genetic data and a mathematical function, decide what the trait value is for each person.  The aspect that makes this novel research is that the mathematical function linking the genetics to the trait is unknown.  This is why governments and foundations are spending millions of dollars on studies that try to discover these mathematical functions that link genetics to the trait.

The most common type of study that attempts to do this is called the genome-wide association study (GWAS).  These studies are used not only for in pharmacogenomics, but also for studying the genetics of heritable disease.  A GWAS study that looks for an association between some trait, in this case drug response by depressed individuals when given anti-depressant drugs, and the nucleotides at a particular SNP site.  An association (usually meaning a significant correlation) is a specific mathematical function, like the ones that we discussed above, that relates, with few errors, the nucleotides of the individuals at a SNP site to the value of the trait.  (The functions we described above were completely general whereas most GWAS methods rely on correlations which are a restricted class of functions.)

The study that was sited at the top of this post, had two main results.  The first result is a negative result in terms of being able to use the genome of a patient to identify which drugs will most help them.  The second result is a positive result saying that if a patient's close relative responds well to a particular drug then the patient likely will as well.

First, they found many SNPs that are associated with antidepressant drug response.  This means that drug response is polygenic and therefore is a complex trait involving multiple SNP sites.  Knowing that antidepressant drug response is polygenic helps us know what to look for in the genome.  We should be looking for interacting sites that work together to produce the trait.  This is a negative result, because it means that there is not just one genetic site that we can use to test for what drug each person will respond to.

Second, the paper gives results about the heritability of drug response.  Heritability is the fraction of trait variability that is due to genetics.  When we say that the heritability is 42%, we do not mean that there is a 42% chance that a child of a person with the trait will also have the trait.  Rather, we mean that heritability is the ratio of the variance of the trait that is due to genetics and the total variance of a trait.  Heritability is not about the risk of having a trait but rather it is about whether the trait is genetic or not.  A trait having a heritability of 42% means that 42% of the variance of the trait value is due to genetic causes.  This main result of the study is best expressed in this sentence

We find that common genetic variants explain 42% (SE = .180, p = .009) of individual differences in antidepressant response.

What can we as potential users of antidepressants take away from all this?  First, it will be a while until the science advances to the point that it can predict which drug will work best for us.  Second, the best predication about which antidepressants would work well for us is the list of antidepressants that work well for our relatives.

In the wider discussion of which psychotropic drugs may be helpful to a person, it might be quite reasonable to assume that there is heritability of drug response.  For example imagine two siblings, one who has anxiety and one who has schizophrenia.  (This example is quite plausible because we know that families having bipolar or schizophrenia also have a lot of anxiety and depression.)  At any rate, in our imagining, the sibling with the anxiety was diagnosed first and found Zyprexa to be a very useful drug for pro re nata treatment of anxiety.  The second sibling, when diagnosed with schizophrenia, recognizes that Zyprexa has worked for their sibling and tries that drug to good effect.  As another example, suppose antidepressants cause a family member with bipolar to become manic.  Then it might be good for other family members to avoid that class of drugs regardless of which diagnoses they may have.

Although there is not much research yet that elucidates the heritability of psychotropic drug response, the experience of patients indicates that this is a critical aspect of successful treatment.  I fervently hope that the research will move in the direction of attempting to elucidate the heritability of drug response, because it would lead to successful treatment.

Academic Interviews

Interviewing is difficult for everyone.  Imagine interviewing while managing a mental illness and dealing with low grade psychotic symptoms.  Indeed, some people would hypothesize that the stress of interviewing would make the symptoms worse.

For those of you who are not familiar with academic interviews, the process requires some explaining.  (This description is specific to North America.)  Early in the fall each school wishing to hire someone will convene a search committee.  The search committees job is to

• locate candidates for hiring by advertizing the open position, 
• invite the best ones for interviewing, 
• interview the selected candidates in one- or two-day interview, and 
• recommend the best candidate to the hiring authority (usually the dean or the department chair/head).

From the applicant's perspective, they put in their applications to as many schools as they feel will consider them and wait to hear if they have an interview.  The application usually consists of a cover letter, a research statement that describes the applicants research aspirations, a teaching statement that describes the applicants teaching aspirations, some samples of the applicants academic writing such as journal articles, a CV, and letters of recommendation.  Getting all of this material collected together can take time, and submitting an application is considered the easiest part of the whole process.  Once an applicant is selected for interviewing, they need to prepare a 45-minute job talk (sometimes a short teaching lecture or chalk-talk is also required).  Again, preparing these talks is considered to be part of the due diligence of accepting an interview invitation.

The difficult part for the candidate and for the search committee is actually having the interview.  The search committee spends a significant amount of time preparing a one- or two-day schedule for the candidate.  This usually involves arranging numerous 30- to 45-minute meetings with various busy faculty members who each have their own tight schedule.  Typically all these faculty members are from the department that is considering hiring the candidate, but in interdisciplinary situations, meetings might be scheduled with faculty from diverse departments.  To fully understand the level of planning involved in this scheduling, it is necessary to explain that that everything is choreographed from the time the candidate arrives at the local airport to the time that they leave.  It is not uncommon for the candidates every meal to be scheduled and for faculty members to shuttle the candidate from appointment to appointment.  It is considered polite and hospitable for the interviewers to manage the candidate's schedule like this.

After interviewing several candidates like this, the search committee makes a recommendation of the best fitting candidate to the hiring authority.  The candidate, knowing little of the machinations of the search committee, is simply left to wait and wonder unless they are the lucky candidate that receives the telephone call letting them know that the school will make them an offer.  For the interviewed candidates that do not receive the telephone call, there typically is no indication that they have not been selected except silence.

For the candidate receiving the lucky telephone call, the process is not over.  They have received what is called the verbal offer.  Next they are required to negotiate with the hiring authority over salary, start-up, and other terms of the contract.  It is not uncommon for the negotiations to be completed before the paper offer is printed and sent to the candidate.  Until receiving the paper offer, the candidate must be careful, because the verbal offer could be rescinded at any time.  For example, offers can be rescinded,  because funding for the position is not granted.  Therefore, the candidate must be careful not to count their chickens before they hatch and continue to make a good impression.  Once the paper offer has been given, the candidate typically has a deadline for accepting the offer and for negotiating any remaining items that have not been agreed upon prior to the paper offer being extended.

Now, having described the process of getting an academic job, I will discuss interviewing from the perspective of the candidate.  The academic interviewing process is almost ideally designed to foster extreme emotions.  The amount of ambiguity, the lack of clear communication (i.e. deafening silence), and tricky negotiations can easily inspire feelings of uncertainty, insignificance, and nervousness.  On the flip side, being invited for interviews, perhaps many of them, can inspire feelings of grandeur.  In short, academic interviewing is not for the faint of heart, because nobody feels normal during the interview process.

For myself, I felt anxious over double scheduling and missed flights.  I worried about whether my taxi drivers were driving me directly to my destinations.  I felt indicted by the people who fell asleep during my seminars.  I worried over every little mistake that I made.  I double scheduled interviews by mistake.  I hated all the travel and the many time-zone changes.  And, I felt the rush of exhilaration over being invited to interview at good places.

There are many challenges one might face during such an interview process.  I will list each along with potential strategies for overcoming the challenges.

1. The anticipation before an interview can cause insomnia.  I find it most useful just to take something that encourages sleep.  Benadryl can be perfect for this.
2. Worries about travelling, getting stuck in snowstorms, missing flights, double scheduling and other hiccups in the schedule.  I find it most useful to pad the travel schedule with a bit of extra time, so as to avoid anxiety over being late.  Beyond that a Zen approach to uncontrollable events works well.  In the worst case, medication can be adjusted slightly and anti-anxiety medication can help ease the obsessive worries.
3. The demanding schedules can cause stress due to the constant demand of trying to impress people.  I found it useful to insist on bathroom breaks frequently.  This gave me a chance to catch my breath.  Some interviews scheduled some free time to prepare for the talk.  I did my best to be alone during that time.  The most demanding schedule that I had was a 3 day interview at one university, and I got a break by asking for the dinner on the second day to not be scheduled.
4. Good news like interview invitations and offers can be stressful.  I made sure to stay immersed in work when I wasn't interviewing, so that I had something besides the job hunt to focus on.
5. Flying around a lot can be very stressful due to noisy plane, busy airports, late or cancelled flights and time-zone changes.  I dealt with this by making myself as comfortable as possible on flights.  I got a small blanket, eye cover, and ear plugs.  I made sure that my carry-on contained all the essentials, such as medications, emergency food, my computer, and power cables.  I travelled light, so that most of the time I had everything I needed in my two carry-on bags.
6. Decision making can be hard.  The offers are when you begin to see your hard work paying off, but there is still decisions to be made.  I tried to deal with this by making them as quickly as I could.  Certainly, one might be stuck waiting for more information before making a decision, but it is best to let the hiring authority know as soon as you can what your decision is.  No's are important because they might allow the search committees to pursue other candidates.

Despite difficulties of the academic job search it is likely worth the effort.  The stresses can be managed with a little care, foresight, and fortitude.

Pills on the Go

Taking pills on the go can be challenging.  If you are like me, you are more likely to miss a dose when you are away from home than when you are at home.  Of course the main challenge is that the medication is so important that doses should not be missed.  The other challenge is to prevent questions from other people about your medical condition.

How do I deal with these challenges?  Travel is required for my job, and some times a lot of travel is required.  Pills are required as well.  My strategy for dealing with this situation is:

1. I take the pills around the same time every day.  This is particularly important for people with bipolar (or perhaps bipolar-schizophrenia) which seems to be influenced by the circadian rhythm.  I use an electronic device that has an alarm set to ring every day at pill time which for me is 6pm.  
2. I do not take my pills in front of people that I do not want to have asking questions.  If I am with other people when the alarm goes off, I simply excuse myself and take my pills.  My alarm unfortunately sounds like a phone call, so I try to be polite.  Sometimes I have to pretend not to be taking a call, and later I will pretend to use the restroom as an opportunity to take my pills. 
3. I always carry my pills on me.  In order to take the pills, I must have them.  If I fly, they are always in my carry-on.  That way, if my luggage is lost or delayed, then I can still take my pills.
4. I always have extra pills.  This is in case a flight is delayed or cancelled.  The last thing you want to do when you are stuck in Chicago in a snowstorm is to worry about whether or not you have enough pills.
5. When travelling internationally, I always check the immigration and airport security rules to see if I must have with me my prescription or perhaps the original pill bottle.

These strategies will help anyone who has to consistently take pills.  In particular, people who have medical issues they do not want to discuss or have a particular timing regimen to their pill taking.

If anyone has any other strategies to share, please post them.

Bipolar-Schizophrenia as One Disease

Are bipolar and schizophrenia actually separate illnesses?  For technical accuracy, I should also include schizo-affective disorder which is the DSM's way of acknowledging that there is not a clear distinction between bipolar and schizophrenia.

Before I get into this, I should frame my participation in this discussion by explaining my controversial belief that bipolar and schizophrenia are not different diseases.  I believe that they are two faces to the same genetically-caused illness.  I think that there are clusters of symptoms whereby the people typically labelled with having schizophrenia may appear to have a different presentation than people typically labelled with having bipolar.  I do not deny that there may be some biological difference between the two presentations of symptoms, but I believe these differences are few enough as to consider bipolar and schizophrenia to be two subtypes of the same disease.

A primary driver of my belief is the growing evidence that bipolar and schizophrenia probably share common causes.  This is born out with both genetic and neurological evidence.  I will briefly describe the type of evidence along with providing references to the literature.

The evidence says that bipolar and schizophrenia have associations with some of the same regions of the genome.  What does it mean for a disease to be correlated with a region of the genome?  This means there is a statistical correlation between the presence/absence of the disease(s) and the nucleotides at a polymorphic position in the genome.  Typically researchers are testing Single Nucleotide Polymorphisms (SNPs), but some studies have examined Copy Number Variations (CNVs).  One can think of the association between a SNP and the disease as trying to place the disease in the genome at some position that explains the inheritance of the disease.

[Lichtenstein et al., The Lancet, 2009] This study looked at approximately 75,000 individuals with either schizophrenia or bipolar.  They examined relative risk, that is the risk of disease for a relative of a diseased person as compared with the risk for a relative of a diseased person.  They showed that relatives of people with bipolar had elevated risk of schizophrenia and vice verse.  They measured heritability of schizophrenia as 64% and of bipolar as 59%.  Heritability is the percentage of the disease variance that is accounted for by genetic factors.  (It does not mean that the child of a schizophrenic has that percentage as risk of getting the disease.)  The authors concluded that:

Similar to molecular genetic studies, we showed evidence that schizophrenia and bipolar disorder partly share a common genetic cause. These results challenge the current nosological dichotomy between schizophrenia and bipolar disorder, and are consistent with a reappraisal of these disorders as distinct diagnostic entities.

[The International Schizophrenia Consortium, Nature, 2009] This study did a genome-wide association study (GWAS) of roughly 3,000 individuals with schizophrenia.  They demonstrated the contribution of thousands of SNPs to the risk of schizophrenia.  And, they showed that many of those SNPs also appeared in a GWAS study of bipolar.

[Cross-Disorder Group of the Psychiatric Genomics Consortium, The Lancet, 2013] This study is actually of five psychiatric disorders: autism spectrum disorder, attention deficit hyperactivity disorder, bipolar disorder, major depressive disorder, and schizophrenia.  They found four SNPs that were significantly associated with all these diseases.  They also separated the people with the disease into two categories  child and adult onset of disease, which further refined the SNPs that they found.  The authors say

 Doubt remains about the boundaries between the syndromes and the degree to which they signify entirely distinct entities  disorders that have overlapping foundations or different variants of one underlying disease.

The pathogenic mechanisms of psychiatric disorders are largely unknown, so diagnostic boundaries are difficult to define.  Genetic risk factors are important in the causation of all major psychiatric disorders and genetic strategies are widely used to asses potential overlaps.

These results provide evidence relevant to the goal of moving beyond descriptive syndromes in psychiatry, and towards a nosology informed by disease cause.

On the neurology front, brain imaging provides evidence both of the similarities and differences between schizophrenia and bipolar.  Generally speaking, brain imaging can involve functional magnetic resonance imaging (fMRI) which images the blood flow in the brain and voxel-based morphometry studies which examine the gray matter.

[Ongur et al., Psychiatry Research: Neuroimaging, 2010] In this study the researchers selected 17 patients with bipolar and 14 patients with schizophrenia and did fMRI scans.  They used independent component analysis to analyze the data and identify synchronous activity patterns.  The data suggest abnormal functional organizations of the brain in both disorders when compared with controls.  The people with schizophrenia have abnormalities in the frontopolar cortex/basal ganglia while the people with bipolar showed abnormalities in the parietal cortex.  It is not clear how the abnormalities are related to the symptoms of the disorders. Indeed it is possible that eye movement during the fMRI explains some of the differences in the scans of bipolar and schizophrenic people.  This paper leads us to think that there may be a reliable way to differentially diagnose bipolar and schizophrenia, however the sample sizes are small and eye movement may have confounded the results.

[Ellison-Wright et al., Schizophrenia Research, 2010] This paper did a meta-analysis of around 2000 patents with schizophrenia and around 350 patients with bipolar.  The goal was to find whether there was similar brain imaging for the people with bipolar and people with schizophrenia.  This was done in light of the genetic studies suggesting a common cause for bipolar and schizophrenia.  The studies looked at gray matter using voxel-based morphometry studies. They found that the areas of gray matter reduction in bipolar overlapped the gray matter reductions found in schizophrenia with the exception of a region of anterior cingulate where the gray matter reduction was only found in bipolar.  They also found that gray matter reductions in schizophrenia were more extensive.

[Chai et al. Neuropsychopharmacology, 2011] This paper looked at 14 patients with bipolar and 16 patients with schizophrenia.  They remark that "a significant body of genetic, imaging, and neurophysiology research has established that schizophrenia and bipolar share substantial overlap in clinical features, as well as in contributing genetic factors."  Their goal was to find if there are common neural substrates for bipolar and schizophrenia.  This work also used fMRIs and independent component analysis (similar to the Ongur paper).  To paraphrase the authors, they found that there is a decoupling of the dorsal lateral prefontal cortex from the medial prefrontal cortex in both bipolar and schizophrenia, and that this decoupling is consistent with the impaired executive functioning seen in these disorders.  In addition to this commonality between bipolar and schizophrenia, they found a distinguishing feature.  The authors say:

Functional connectivity between [medial prefrontal cortex] and insula/[ventral lateral prefrontal cortex] distinguished bipolar disorder from schizophrenia, and may reflect differences in the affective disturbances typical of each illness.

Due to the weight of genetic and neurological evidence that is establishing commonalities between bipolar and schizophrenia, it is reasonable to hypothesize that they share common causes.  In light of this it might be better to revise the diagnostic categories so that bipolar, schizoaffective, and schizophrenia are subtypes under the same disorder.  As The Lancet, 2013 paper suggests, we may also need to significantly revise our categorizations of other diseases, as well.  We may be better able to diagnose and treat these diseases if we classify them according to common causes.