Can We Predict Response to Antidepressants?

September 15, 2007

In this article, we use the example of major depressive disorder (MDD) to review research efforts to identify predictors of treatment response, both to antidepressant medications and to psychotherapy. We describe the promises and limitations of this research, with some emphasis on brain imaging studies, and then discuss how this work may be integrated into clinical practice in the future.

A long-standing problem facing practicing psychiatrists is our limited ability to select treatments for patients based on the likelihood that a specific treatment will be effective for a particular patient. In this article, we use the example of major depressive disorder (MDD) to review research efforts to identify predictors of treatment response, both to antidepressant medications and to psychotherapy. We describe the promises and limitations of this research, with some emphasis on brain imaging studies, and then discuss how this work may be integrated into clinical practice in the future.

THE PROBLEM

MDD is a significant cause of morbidity and mortality that affects 13.1 to 14.2 million American adults annually.1 Following an adequate trial of the SSRI citalopram in a recent multicenter study, only 28% of patients with MDD achieved remission.2 Remission rates with other classes of antidepressants are similar. Until clinical or biological subtypes of MDD that are more likely to respond to one treatment than to another are identified, we are left to consider other factors in selecting treatments for our patients. Common strategies include choosing medications with side-effect profiles thought to be most advantageous for a specific patient, or choosing treatments that have previously been effective for a patient's family members. While these approaches are quite rational, no empirical studies have been conducted that provide evidence to support them.

CLINICAL PREDICTORS

One clinical subtype of depression with good evidence of differential response to different treatments is atypical depression. Patients with atypical depression, a subtype defined in DSM-IV by reverse neurovegetative symptoms (increased sleep and appetite), rejection sensitivity, leaden paralysis, and mood reactivity,3 have higher response rates when treated with monoamine oxidase inhibitors (MAOIs) than with tricyclic antidepressants (TCAs),4 a finding replicated in at least 4 studies. While probably a valid distinction, it is of less clinical relevance today, since both these classes of medications are used less frequently because of the risks of hypertensive crisis with MAOIs and fatality in overdose with TCAs. Future studies will hopefully examine whether transdermal selegiline, an MAOI that does not carry the same dietary restrictions as orally administered MAOIs when given at dosages up to the 6 mg/24 h patch, is similarly effective in atypical depression.

An interesting body of literature has examined whether there are certain clinical variables that may predict response to treatment with psychotherapy. The NIMH Treatment of Depression Collaborative Research Program, a particularly informative study, analyzed factors predicting differential response to interpersonal psychotherapy (IPT) and to cognitive-behavioral therapy (CBT) for major depression in a 16-week, multicenter, randomized trial.5 Low levels of social dysfunction predicted a better response to IPT, and low levels of negatively biased thinking patterns as assessed by the Dysfunctional Attitudes Scale predicted a better response to CBT. IPT works to improve relationships and social functioning, and CBT works to challenge and reformulate patterns of negative thinking. This suggests that a certain level of functioning in the psychological domain that is the focus of a given psychotherapy must be present when beginning treatment in order to derive maximal benefit from that treatment. The idea that treatments that act on a certain area of functioning may work best when this area is not overly impaired has interesting analogies in the brain imaging literature.

BIOLOGICAL PREDICTORS Techniques

Positron emission tomography (PET) and single photon emission CT (SPECT) are related techniques that can be used to visualize and, when performed rigorously, quantify molecules of interest in the brain. PET can also be used to measure blood flow and glucose metabolism. Both PET and SPECT can involve the intravenous injection of a radioactively labeled molecule (a radio-ligand) that has been chemically designed to bind specifically to a protein of interest in the brain. The radioligand is then visualized using a PET or SPECT camera. These methods are very powerful in that they allow us to measure, with some assumptions, the number of available molecules of interest in vivo in the human brain. Some of the drawbacks to these approaches include cost, radiation exposure, and the occasional need for placement of a radial arterial catheter to accurately quantify proteins in the brain.

Functional MRI (fMRI) is a technique that can indirectly measure brain activity by taking advantage of the different magnetic properties of oxyhemoglobin and deoxyhemoglobin. It does not require radiation exposure and is less invasive than PET, but it yields relative, rather than absolute, information regarding brain activity and no information on location and quantity of proteins.

Pharmacogenetics identifies genetic variants, including single nucleotide polymorphisms (SNPs) that are associated with differential response to a specific treatment. These approaches are less invasive and less costly than PET or SPECT, but they are a less direct way of assessing the chemical milieu in the CNS, which arises not merely through genetics but also through gene-environment interactions.

Dysfunction in the serotonin system as a predictor of outcome

There is a large body of evidence supporting the idea that the serotonin neurotransmitter system is abnormal in patients with major depression.6 Serotonin (5-HT) acts on at least 15 different receptors and is cleared from synapses primarily by the serotonin transporter. One of these, the 5-HT1A receptor, is located on serotonin-producing neurons (in the dorsal raphe nucleus), where it provides negative feedback to communicate to these cells when to reduce the release of serotonin (Figure 1). In many regions of the brain, 5-HT1A receptors also receive the serotonin signal on postsynaptic neurons, where they mediate its effects.7 The antianxiety medication, buspirone, acts primarily on the 5-HT1A receptor. Lemonde and colleagues8 described a genetic variant of the 5-HT1A gene that is more frequently found in patients with MDD. This variant leads to elevated levels of the 5-HT1A receptor and has been associated with poor response to antidepressant treatment in many,9-11 but not all,12 studies.

In response to long-term treatment with SSRIs, the inhibitory 5-HT1A receptors on serotonin neurons are desensitized, leading to increased serotonin release, which has been proposed as the mechanism of action of SSRIs.13 The genetic variant of the 5-HT1A receptor identified by Lemonde may be incapable of down-regulation following SSRI treatment, which may limit the therapeutic effect of SSRIs in patients with this variant.

Our laboratory used PET to study the 5-HT1A receptor in patients with MDD. We found higher 5-HT1A binding potential (proportional to the total number of available receptors) in antidepressant-naive patients having a major depressive episode than in healthy controls.14 We also found a correlation between the genetic variant of 5-HT1A and the level of 5-HT1A receptor binding in the raphe nuclei, consistent with preclinical studies.

Next, we studied whether baseline 5-HT1A binding would predict outcome after 1 year of antidepressant treatment. In a sample of 22 patients who received naturalistic treatment for MDD, nonremitters had higher 5-HT1A receptor binding across all brain regions at baseline than remitters.15 Thus, both genetic and imaging assessments of the 5-HT1A receptor may be partially predictive of treatment outcome, although this imaging study is limited by nonstandardized treatment--an issue we are addressing in an NIMH-funded prospective study.

The serotonin transporter (SERT) regulates serotonin transmission by facilitating reuptake of serotonin into presynaptic neurons (Figure 1).16 We recently described lower SERT binding in the amygdala and midbrain in patients with MDD than in controls.17 A SPECT study of depressed patients found a correlation between lower SERT binding at baseline in the thalamus/hypothalamus and reduced response to SSRIs.18 This study showed a similar trend between brain stem SERT and treatment response. We also looked at SERT binding as a predictor of remission from MDD in a sample that overlaps that of the 5-HT1A receptor study. We found a trend toward a predictive effect of SERT binding on remission status at 1 year (Figure 2).

The genetics of SERT have been studied at length as a predictor of treatment response in MDD. Several groups have found that depressed patients with a certain genetic variant of SERT respond more favorably to SSRIs,19 although this finding has not been replicated in all studies.19, 20 This genetic variant may be independent of SERT binding,21 so that to the extent that each of these measures is predictive of outcome individually, they may have greater predictive power when taken together.

Therefore, it seems that patients who have elevated 5-HT1A receptor binding and decreased SERT binding are less likely to achieve remission with antidepressant treatment, whereas patients whose 5-HT1A receptor and SERT binding are more like those of healthy controls are more likely to achieve remission with treatment. This profile of elevated 5-HT1A binding and decreased SERT binding associated with worse outcomes could be at least partially explained by the genetic variant of the 5-HT1A receptor that we described previously. A genetic variant of SERT could contribute independently to the likelihood of remission. A model of how this profile could arise is presented in Figure 1.

Other genetic markers

Several other genes have been examined in pharmacogenetic studies as possible predictors of treatment outcome in MDD. These include polymorphisms in genes that affect drug metabolism, such as cytochrome P-450 2D6; downstream secondary messenger systems in neurons, such as G-proteins; the viability of neurons, such as brain-derived neuro-trophic factor; and neurotransmitter regulation, such as tryptophan hydroxylase 1.22 Whole-genome association studies may be used in the future to compare SNPs across the entire human genome between remitters and nonremitters in response to specific antidepressant treatments.23

PET STUDIES OF BRAIN METABOLISM

PET can be used to measure the relative rate of glucose metabolism, an indirect measure of resting brain activity, with the use of the glucose radioligand [18F]fluorodeoxyglucose (FDG). Regional glucose metabolism has been studied as a predictor of outcome with antidepressant treatment.24 Depressed patients underwent FDG-PET, followed by 6 weeks of medication-based treatment. Glucose metabolism was compared between responders and nonresponders. Those who responded had higher resting glucose metabolism in the rostral anterior cingulate cortex than patients who did not respond. The authors described connections between the rostral anterior cingulate cortex and several other regions in the brain that could be involved in a poorly functioning circuit related to depression. This finding has been replicated in subsequent PET studies,25 indicating the possible importance of the rostral anterior cingulate cortex in the response to antidepressant medications. Another FDG-PET study found the opposite relationship between metabolism and treatment response in the left ventral anterior cingulate cortex, an adjacent brain region.26

fMRI STUDIES OF BRAIN ACTIVITY

A series of fMRI studies have looked at brain activations in response to specific probes as predictors of response to treatment in patients with MDD. In one of the first of these studies, patients viewed pictures of emotionally neutral or charged scenes, and their patterns of brain activation were used to predict subsequent response to the antidepressant venlafaxine.27 The investigators found that larger responses in the anterior cingulate cortex to presentation of emotionally negative pictures were associated with subsequent improvement with venlafaxine treatment, which is consistent with the FDG-PET studies previously described. This finding was recently replicated using responses to emotional faces to predict outcome with fluoxetine treatment.28 Interestingly, that study also analyzed structural MRI scans on the same patients, and found increased gray matter in the anterior cingulate cortex and other regions among fluoxetine responders, suggesting that the changes in brain activity that predict treatment response may reflect underlying differences in brain structure.

A more recent fMRI study looked at brain activation in response to the presentation of emotionally charged words as a predictor of recovery from MDD following CBT.29 It found a specific pattern of brain activity during the viewing of emotionally negative words that was associated with subsequent improvement with CBT: low activation of the subgenual cingulate cortex and high activation of the amygdala. The amygdala is involved in processing emotional stimuli, and its responses to emotional stimuli may be controlled by connections from the subgenual cingulate cortex.30 This finding involving the subgenual cingulate cortex contradicts the 2 fMRI studies involving medication treatment described above. Whether this is because of methodological differences between the studies or, more promisingly, caused by different patterns of cingulate activity predicting differential response to medication versus CBT remains an important question.

THE FUTURE

From this brief review, we hope to convey a sense of the breadth of approaches that show promise in predicting response to treatments of depression, as well as some of their limitations, including nonstandardized methodology for conducting functional brain imaging scans and conflicting results between studies. The measures we have described in this review are only partially predictive of treatment outcome when used individually.

To address this, we are currently developing statistical models that combine relevant clinical, brain imaging, and genetic data to predict which treatments will be the most effective for a particular patient. Whereas brain imaging studies to date have examined activity or chemistry in relatively large regions of the brain as predictors of treatment outcome, we are using measures of brain activity or chemistry in areas as small as a few cubic millimeters in these models, allowing us to rely less on preconceived notions about the anatomy and organization of the brain.31 In addition to predicting treatment outcome, these models may be used for predicting diagnosis (such as unipolar vs bipolar depression), as well as risk of suicidal behavior. Finally, studies are needed in which multiple biological and clinical measures are obtained before randomization to different standardized treatment conditions in order to bring these techniques closer to being ready for clinical use.

References:

References


1.

Kessler RC, McGonagle KA, Zhao S, et al. Lifetime and 12-month prevalence of

DSM-III-R

psychiatric disorders in the United States: results from the National Comorbidity Survey.

Arch Gen Psychiatry.

1994;51:8-19.

2.

Trivedi MH, Rush AJ, Wisniewski SR, et al. Evaluation of outcomes with citalopram for depression using mea-surement-based care in STAR*D: implications for clinical practice.

Am J Psychiatry.

2006;163:28-40.

3.

American Psychiatric Association.

Diagnostic and Statistical Manual of Mental Disorders

. 4th ed. Washington, DC: American Psychiatric Association; 1994.

4.

Liebowitz MR, Quitkin FM, Stewart JW, et al. Antidepressant specificity in atypical depression.

Arch Gen Psychiatry.

1988;45:129-137.

5.

Sotsky SM, Glass DR, Shea MT, et al. Patient predictors of response to psychotherapy and pharmacotherapy: findings in the NIMH Treatment of Depression Collaborative Research Program.

Am J Psychiatry.

1991; 148:997-1008.

6.

Owens MJ, Nemeroff CB. Role of serotonin in the pathophysiology of depression: focus on the serotonin transporter.

Clin Chem.

1994;40:288-295.

7.

Aghajanian GK, Sanders-Bush E. Serotonin. In: Davis KL, Charney D, Coyle JT, Nemiroff C, eds.

Neuropsycho-pharmacology: The Fifth Generation of Progress

. Philadelphia: Lippincott Williams & Wilkins; 2002. Available at:

http://www.acnp.org/default.aspx?Page=5thGenerationChapters

. Accessed August 22, 2007.

8.

Lemonde S, Turecki G, Bakish D, et al. Impaired repression at a 5-hydroxytryptamine 1A receptor gene polymorphism associated with major depression and suicide.

J Neurosci.

2003;23:8788-8799.

9.

Albert PR. A functional promoter polymorphism of the 5-HT1A receptor gene: association with depression and completed suicide.

Biol Psychiatry.

2004;55:46S.

10.

Lemonde S, Du L, Bakish D, et al. Association of the C(1019)G 5-HT1A functional promoter polymorphism with antidepressant response.

Int J Neuropsychopharmacol.

2004;7:501-506.

11.

Serretti A, Artioli P, Lorenzi C, et al. The C(-1019)G polymorphism of the 5-HT1A gene promoter and antidepressant response in mood disorders: preliminary findings.

Int J Neuropsychopharmacol.

2004;7:453-460.

12.

Arias B, Catalan R, Gasto C, et al. Evidence for a combined genetic effect of the 5-HT(1A) receptor and serotonin transporter genes in the clinical outcome of major depressive patients treated with citalopram.

J Psycho-pharmacol.

2005;19:166-172.

13.

Blier P, de Montigny C, Chaput Y. A role for the serotonin system in the mechanism of action of antidepressant treatments: preclinical evidence.

J Clin Psychiatry.

1990;51:14-20.

14.

Parsey RV, Oquendo MA, Ogden RT, et al. Altered serotonin 1A binding in major depression: a [carbonyl-C-11]WAY100635 positron emission tomography study.

Biol Psychiatry.

2006;59:106-113.

15.

Rudnick G, Clark J. From synapse to vesicle: the reuptake and storage of biogenic amine neurotransmitters.

Biochim Biophys Acta.

1993;1144:249-263.

16.

Parsey RV, Olvet DM, Oquendo MA, et al. Higher 5-HT1A receptor binding potential during a major depressive episode predicts poor treatment response: preliminary data from a naturalistic study.

Neuropsycho-pharmacology.

2006;31:1745-1749.

17.

Parsey RV, Hastings RS, Oquendo MA, et al. Lower serotonin transporter binding potential in the human brain during major depressive episodes.

Am J Psychiatry.

2006;163:52-58.

18.

Kugaya A, Sanacora G, Staley JK, et al. Brain serotonin transporter availability predicts treatment response to selective serotonin reuptake inhibitors.

Biol Psychiatry.

2004;56:497-502.

19.

Lesch KP, Gutknecht L. Pharmacogenetics of the serotonin transporter.

Prog Neuropsychopharmacol Biol Psychiatry.

2005;29:1062-1073.

20.

Kraft JB, Peters EJ, Slager SL, et al. Analysis of association between the serotonin transporter and antidepressant response in a large clinical sample.

Biol Psychiatry.

2007;61:734-742.

21.

Parsey RV, Hastings RS, Oquendo MA, et al. Effect of a triallelic functional polymorphism of the serotonin-transporter-linked promoter region on expression of serotonin transporter in the human brain.

Am J Psychiatry.

2006;163:48-51.

22.

Bondy B. Pharmacogenomics in depression and antidepressants.

Dialogues Clin Neurosci.

2005;7:223-230.

23.

Motsinger AA, Ritchie MD, Dobrin SE. Clinical applications of whole-genome association studies: future applications at the bedside.

Expert Rev Mol Diagn.

2006; 6:551-565.

24.

Mayberg HS, Brannan SK, Mahurin RK, et al. Cingulate function in depression: a potential predictor of treatment response.

Neuroreport.

1997;8:1057-1061.

25.

Saxena S, Brody AL, Ho ML, et al. Differential brain metabolic predictors of response to paroxetine in obsessive-compulsive disorder versus major depression.

Am J Psychiatry.

2003;160:522-532.

26.

Brody AL, Saxena S, Silverman DH, et al. Brain metabolic changes in major depressive disorder from pre- to post-treatment with paroxetine.

Psychiatry Res.

1999;91: 127-139.

27.

Davidson RJ, Irwin W, Anderle MJ, Kalin NH. The neural substrates of affective processing in depressed patients treated with venlafaxine.

Am J Psychiatry.

2003; 160:64-75.

28.

Chen CH, Ridler K, Suckling J, et al. Brain imaging correlates of depressive symptom severity and predictors of symptom improvement after antidepressant treatment.

Biol Psychiatry.

Jan 8, 2007; [Epub ahead of print].

29.

Siegle GJ, Carter CS, Thase ME. Use of FMRI to predict recovery from unipolar depression with cognitive behavior therapy.

Am J Psychiatry.

2006;163:735-738.

30.

Ghashghaei HT, Barbas H. Pathways for emotion: interactions of prefrontal and anterior temporal pathways in the amygdala of the rhesus monkey.

Neuroscience.

2002;115:1261-1279.

31.

Reiss PT.

Regression With Signals and Images as Predictors

. New York: Doctoral thesis in the Department of Biostatistics, Columbia University; 2006.