Genomics And Metabolomics In Psychiatry As A Combined Tool - Predicting SSRI Response

From Reference 1

The last two posts here were an introduction to some recent work that combine aspects of genomics and metabolomics.  The paper that I will briefly discuss here is done by researchers from the same groups focused on studies in genomics and metabolomics.  They present work that is quite exciting because it illustrates the amount of information necessary to analyze the biological complexity present in brain science and they produce results that may prove very useful from a clinical standpoint.  The study involves serotonin metabolism in depression. I have discussed serotonin in many places on this blog as a significant neurotransmitter that refuses to go away despite the various critics.  This work will illustrate why that may be.  The first paper (1) looks at the relationship between baseline serotonin levels and levels after treatment with selective serotonin reuptake inhibitors and what those phenotypes may be associated with at a genetic level.  A second paper (2) that I will discuss in a subsequent post looks at kynurenine metabolism in depression and the association of that metabolite with symptoms severity and the genetic correlates.  Both papers offer a new look at serotonin metabolism in depression and its genetic basis.

In paper 1, the sample included 366 patients in the Mayo Clinic Pharmacogenomics Research Network Antidepressant Medication Pharmacogenomics Study (PGRN-AMPS). 918 samples from these patients after 4 and 8 weeks of SSRI therapy were used for the metabolomics part of the study.  Baseline serotonin and changes in these levels were measured.  Patients with higher baseline levels and/or greatest drops in serotonin levels were determined to have the best response to treatment with SSRI medications.

From Reference 1

When a GWAS study was performed looking at the baseline serotonin concentrations as the phenotype and at 4 and 8 weeks of treatment a significant single nucleotide polymorphism (SNP) cluster was noted at the Tetraspanin 5 (TSPAN5) gene on chromosome 4 and a number of SNPs were noted at the Glutamate-rich 3 (ERICH3) gene on chromosome 1.  Both are highlighted in the Manhattan plot at the top of this post. Both of these genes were noted to be novel genes involved in serotonin metabolism and plasma serotonin concentrations.

The SNPs 5' of the TSPAN5 gene were cis regulatory elements for that gene (referred to in the paper as cis-expression quantitative trait loci or eQTLs).  In the ERICH3 cluster SNPs, two were variants that were associated with proteosome mediated degradation of ERICH3.  Changes in the expression of this gene were correlated with plasma serotonin concentrations but the serotonin pathway expression was unaltered.  One of the SNPs was also associated with clinical response in the STAR*D study.

The study is interesting because of the plasma serotonin concentration phenotypes, positive treatment response, and identification of associated genes and SNPs. What has not been determined is the specific mechanism of the drop in serotonin levels and the specific genetic mechanisms.  In the experimental section of their paper they show how the serotonin concentrations in the periphery could also affect levels in the CNS by looking at TSPAN5 and ERICH3 expression in neuroblastoma cells.  The authors were able to demonstrate that both mRNA and protein fold changes of enzymes involved in the synthesis, degradation, and transport of serotonin were affected by knockdown (KD) - (a technique of gene silencing by the introduction of doublestranded interfering RNAs (siRNAs)) or overexpression (OE) (a technique leading to enhanced gene transcription by the introduction of regulatory elements for transcription.) genes.     

The paper is an excellent example of molecular psychiatry and a possible application in the field of precision medicine.  The ultimate goal is to determine the treatment in psychiatry that has a high probability of working as soon as possible and eliminate the long trials that many people have to endure before they find a medication that is effect for (in this case) depression.  Along the way it will be evident that just as clinical psychiatrists have know for some time -  the general categories of psychiatric disorders - just like all polygenic illnesses are really a collection of diverse disorders at the omic levels.

The rapid identification of these many subtypes will not only lead to more rapid and efficient treatment - but also prevent unnecessary exposure to medications that can be both intolerable and ineffective.

In closing, some will question the utility of reading papers that contain a lot of terminology from what we used to call molecular biology.  I like to read these papers because it continues to consolidate what I learned in medical school and add to that knowledge. In my biochemistry seminars back then - there was still a lot of emphasis on enzymatic pathways and protein function. We had to know the basics of nucleic acid structure, function, and analysis - but nothing like the details presented in this paper.  As an example, we knew the synthetic pathways and enzymes for serotonin biosynthesis discussed in this paper - but the idea of analyzing human DNA with a chip encompassing 7.5 million SNPs across hundreds of research subjects would have been mind blowing and in many ways it still is.  Reading papers like this one also assures that you are not stuck in serotonin metabolism and receptor theory from the 1980s.  If that is all you know these days - it is not enough!

I realize this is not for everybody - but for some of us it is very exciting stuff.

 
George Dawson, MD, DFAPA

Graphics Credit:

All of the above figures are from reference 1, per a Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License.  The graphics are unaltered and are taken from the paper as they are printed in reference 1 below.  This is a non-commercial and not-for-profit blog.




References:

1:  Gupta M, Neavin D, Liu D, Biernacka J, Hall-Flavin D, Bobo WV, Frye MA, Skime M, Jenkins GD, Batzler A, Kalari K, Matson W, Bhasin SS, Zhu H, Mushiroda T, Nakamura Y, Kubo M, Wang L, Kaddurah-Daouk R, Weinshilboum RM. TSPAN5, ERICH3 and selective serotonin reuptake inhibitors in major depressive disorder: pharmacometabolomics-informed pharmacogenomics. Mol Psychiatry. 2016 Dec;21(12):1717-1725. doi: 10.1038/mp.2016.6. Epub 2016 Feb 23. PubMed PMID: 26903268

2: Liu D, Ray B, Neavin DR, Zhang J, Athreya AP, Biernacka JM, Bobo WV,Hall-Flavin DK, Skime MK, Zhu H, Jenkins GD, Batzler A, Kalari KR, Boakye-Agyeman F, Matson WR, Bhasin SS, Mushiroda T, Nakamura Y, Kubo M, Iyer RK, Wang L, Frye MA, Kaddurah-Daouk R, Weinshilboum RM. Beta-defensin 1, aryl hydrocarbon receptor and plasma kynurenine in major depressive disorder: metabolomics-informed genomics. Transl Psychiatry. 2018 Jan 10;8(1):10. doi: 10.1038/s41398-017-0056-8. PubMed PMID: 29317604.


3: Ji Y, Biernacka JM, Hebbring S, Chai Y, Jenkins GD, Batzler A, Snyder KA, Drews MS, Desta Z, Flockhart D, Mushiroda T, Kubo M, Nakamura Y, Kamatani N, Schaid D, Weinshilboum RM, Mrazek DA. Pharmacogenomics of selective serotonin reuptake inhibitor treatment for major depressive disorder: genome-wide associations and functional genomics. Pharmacogenomics J. 2013 Oct;13(5):456-63. doi: 10.1038/tpj.2012.32. Epub 2012 Aug 21. PubMed PMID: 22907730.

4: Mrazek DA, Biernacka JM, McAlpine DE, Benitez J, Karpyak VM, Williams MD, Hall-Flavin DK, Netzel PJ, Passov V, Rohland BM, Shinozaki G, Hoberg AA, Snyder KA, Drews MS, Skime MK, Sagen JA, Schaid DJ, Weinshilboum R, Katzelnick DJ. Treatment outcomes of depression: the pharmacogenomic research network antidepressant medication pharmacogenomic study. J Clin Psychopharmacol. 2014 Jun;34(3):313-7. doi: 10.1097/JCP.0000000000000099. Erratum in: J Clin Psychopharmacol. 2014 Oct;34(5):558. PubMed PMID: 24743713.

5: Ji Y, Schaid DJ, Desta Z, Kubo M, Batzler AJ, Snyder K, Mushiroda T, Kamatani N, Ogburn E, Hall-Flavin D, Flockhart D, Nakamura Y, Mrazek DA, Weinshilboum RM. Citalopram and escitalopram plasma drug and metabolite concentrations: genome-wide associations. Br J Clin Pharmacol. 2014 Aug;78(2):373-83. doi: 10.1111/bcp.12348. PubMed PMID: 24528284; PubMed Central PMCID: PMC4137829.

6: Athreya A, Iyer R, Neavin D, Wang L, Weinshilboum R, Kaddurah-Daouk R, Rush J, Frye M, Bobo W. Augmentation of Physician Assessments with Multi-Omics Enhances Predictability of Drug Response: A Case Study of Major Depressive Disorder. IEEE Comput Intell Mag. 2018 Aug;13(3):20-31. doi: 10.1109/MCI.2018.2840660. Epub 2018 Jul 20. PubMed PMID: 30467458; PubMed Central PMCID: PMC6241311.

Does Identifying Bipolar I Disorder Come Down To 6 Proteins?

The Mayo Clinic puts on very good conferences in Psychiatry.  Two of the last three that I have attended had a strong biological and genetic emphasis.  I was interested when I saw a reference in my Facebook feed to a study of potential biological markers of Bipolar I Disorder.  It was even better that the article was published in the open access journal Translational Psychiatry.  In the current article the authors looks at the results from the analysis of serum protein levels in controls and adults seeking treatment for depression and bipolar disorder.  The sera of their subjects and controls was analyzed by Myriad RBM in a quantitative immunoassay designed to search for biomarkers through large numbers of proteins.  The actual product and the proteins analyzed are described on the company's web page.  All of the six proteins identified as possibly being discriminating as listed in the above graphic including growth differentiation factor 15 (GDF-15),  hemopexin (HPX), hepsin (HPN), matrix metalloproteinase-7 (MMP-7), retinol-binding protein 4 (RBP-4), and transthyretin (TTR) can be located on this page and additional information is provided about the specific proteins

The authors emphasize in several places that this is a pilot or exploratory study but also point out that sufficient power to detect odds ratios for pairwise comparisons between mood disorders versus controls, bipolar disorder versus controls, and bipolar I versus controls.  They looked at 272 proteins from 288 samples (141 controls, 52 Unipolar depression, 49 Bipolar II, and 46 Bipolar I).  It was a one time cross sectional sample and no longitudinal sampling was done.  Rigorous patient selection was used to reduce the risk of substance abuse disorders and inflammatory conditions. In a table describing patient characteristics, the cases had significantly greater BMI, greater lifetime illicit drug use,  greater BMI, greater percentage of smokers, and fewer years of education.  Existing symptoms were rated with the following scales IDS-C (depression), PHQ-9 (depression), GAD-7 (anxiety), YMRS (mania), and AUDIT (alcohol use).  The cases were also being actively treated with antipsychotics, AED mood stabilizers, lithium, antidepressants, sedative/hypnotics, and thyroxine supplement.

The graphic from the article labelled figure 2 above shows the differences in protein concentrations for the six proteins that were significantly different after Bonferroni correction by diagnosis.  As can be seen from the figure all six proteins were at the highest levels in Bipolar I disorder.  ROC curves and the ROC-AUC was used to determine which proteins were better predictors of Bipolar I Disorder.  The text contains theoretical and speculative discussions of these particular proteins, what they have been associated with so far, and what importance that has for the issue of why their concentrations may vary in bipolar disorder.

There are a number of relevant considerations when looking at this type of proteomic analysis.  The most obvious is the assumption that the underlying dynamics of the biological substrate can be measured in meaningful ways by knowing the protein signature of those systems.  Although most of us are used to looking at cartoon depictions of neuron and synapses but the reality is much more complex.  Recent work in Science shows that there are 62 proteins associated with synaptic bouton (2) and vesicle trafficking and that the copy number of these proteins varies greatly.  The authors of that paper speculate that the production and number of those proteins may vary because some physical locations within the neuron may allow for an enrichment effect.  One of the implicit assumptions in the Frye, et al paper is that psychiatric disorders may have a unique configuration in terms of synaptic architecture and that it will be reflected in the proteins responsible for that architecture.  A further assumption is these CNS protein changes are all going to be reflected in the periphery and detectable in blood samples.  

Although it is premature to draw many conclusions about the data in this study, the implications may be far reaching.  It will be an interesting day in psychiatry if and when proteins will be used as biomarkers.  It will be an interesting day even if variants can be found and reliably detected.  Until then students of neuroscience and psychiatry will be able to appreciate that information flow in these systems is significant and we are just on the cusp of being able to understand it.  We are just at the stage of moving from cartoon versions of neurons - to the real thing.


George Dawson, MD, DFAPA

References:

1: Frye MA, Nassan M, Jenkins GD, Kung S, Veldic M, Palmer BA, Feeder SE, Tye SJ, Choi DS, Biernacka JM. Feasibility of investigating differential proteomic expression in depression: implications for biomarker development in mood disorders. Transl Psychiatry. 2015 Dec 8;5:e689. doi: 10.1038/tp.2015.185. PubMed PMID: 26645624.

2: Wilhelm BG, Mandad S, Truckenbrodt S, Kröhnert K, Schäfer C, Rammner B, Koo SJ, Claßen GA, Krauss M, Haucke V, Urlaub H, Rizzoli SO. Composition of isolated synaptic boutons reveals the amounts of vesicle trafficking proteins. Science. 2014 May 30;344(6187):1023-8. doi: 10.1126/science.1252884. PubMed PMID: 24876496.

Attribution:

The figure at the top of this post is from the above reference 1 and is used per the conditions of a Creative Commons Attribution 4.0 International License.

Mayo Clinic Counterpoint to FDA on Citalopram

The Mayo Clinic came out with their recommendations on what to do about the FDA's warning about citalopram.  By their own description they are more liberal with regard to their citalopram recommendations and more conservative regarding escitalopram.  I have previously reviewed the problem here and concluded that there is really a lack of data available on the likelihood of electrocardiogram abnormalities during normal clinical use and if citalopram is as cardiotoxic as the FDA is describing it - we should treat it more like a standard antiarrhythmic drug and used flecanide as an example.

For all practical purposes that would include baseline ECGs, ECGs at the max dose and taking it up one more level from either the Mayo Clinic or the FDA - a stress test looking for QTc prolongation at higher heart rates.  The other elements in the Mayo recommendations based on history and physical examination and expecting some physician knowledge of drug metabolism are fairly standard.  I thought it was interesting that they did not mention checking plasma levels of the drug especially in complex cases (eg. a patient with cirrhosis) who only responds to higher than recommended doses of the drug.  Regarding the statements:  "Selective serotonin reuptake inhibitors cannot simply be substituted for one another, not even escitalopram for citalopram."  That is generally true and where are these guys in the battle against PBMs saying that these drugs are all equivalent?  I have not found any patient that responded selectively to citalopram and not escitalopram.  I have generally been able to convert patients to an equivalent amount of escitalopram the next day.

Both the Mayo Clinic and the FDA are silent on molecular approaches to solving this problem and screening patient for potential risk before they are started on either drug.  The Mayo Clinic offers testing for cytochrome P450 genotypes.  The genetic basis for hereditary prolonged QTc intervals has been a hot topic of research over the past decade.  It is probably time to expand the search for additional genotypes that place people at risk during specific drug therapies.  Until then we have only very approximate methods of determining the at - risk population and keeping them safe and the Mayo recommendations are more reality based than the FDA.

I think it would also be possible to estimate the risk associated with taking citalopram across the entire population.  In fact, at this point the FDA seems to have the data to estimate the risk of any QTc effect at all to the risk of torsade de pointes - the most significant arrhythmia.  I think it is very important for patients making the decision to have this number and if I can provide numbers on rare but serious antidepressant complications like serotonin syndrome, a federal agency with more perfect information and no patient care responsibility can do better.

George Dawson, MD, DFAPA

Sheeler RD, Ackerman MJ, Richelson E, Nelson TK, Staab JP, Tangalos EG, Dieser LM, Cunningham JL. Considerations on safety concerns about citalopram prescribing. Mayo Clin Proc. 2012 Nov;87(11):1042-5.

FDA Drug Safety Communication: Revised recommendations for Celexa (citalopram hydrobromide) related to a potential risk of abnormal heart rhythms with high doses.