Skip to main content
NCBI home page
Proceedings of the National Academy of Sciences of the United States of America logoLink to Proceedings of the National Academy of Sciences of the United States of America
. 2011 Apr 25;108(22):9262–9267. doi: 10.1073/pnas.1104836108

Antidepressant effects of selective serotonin reuptake inhibitors (SSRIs) are attenuated by antiinflammatory drugs in mice and humans

Jennifer L Warner-Schmidt a,1, Kimberly E Vanover b, Emily Y Chen a, John J Marshall a, Paul Greengard a,1
PMCID: PMC3107316  PMID: 21518864

Abstract

Antiinflammatory drugs achieve their therapeutic actions at least in part by regulation of cytokine formation. A “cytokine hypothesis” of depression is supported by the observation that depressed individuals have elevated plasma levels of certain cytokines compared with healthy controls. Here we investigated a possible interaction between antidepressant agents and antiinflammatory agents on antidepressant-induced behaviors and on p11, a biochemical marker of depressive-like states and antidepressant responses. We found that widely used antiinflammatory drugs antagonize both biochemical and behavioral responses to selective serotonin reuptake inhibitors (SSRIs). In contrast to the levels detected in serum, we found that frontal cortical levels of certain cytokines (e.g., TNFα and IFNγ) were increased by serotonergic antidepressants and that these effects were inhibited by antiinflammatory agents. The antagonistic effect of antiinflammatory agents on antidepressant-induced behaviors was confirmed by analysis of a dataset from a large-scale real-world human study, “sequenced treatment alternatives to relieve depression” (STAR*D), underscoring the clinical significance of our findings. Our data indicate that clinicians should carefully balance the therapeutic benefits of antiinflammatory agents versus the potentially negative consequences of antagonizing the therapeutic efficacy of antidepressant agents in patients suffering from depression.

Keywords: citalopram, fluoxetine, S100A10

Mood disorders including major depressive disorder (MDD) affect as many as one in five individuals and are the most prevalent psychiatric conditions (1). Approximately one-third of patients suffering from MDD are refractory to any kind of antidepressant treatment including selective serotonin reuptake inhibitors (SSRIs), tricyclic antidepressants (TCAs), monoamine oxidase inhibitors (MAOIs), and electroconvulsive therapy (ECT) (2).

The hypothesis that cytokines play a role in depression is based in part on the observations that (i) many patients undergoing treatment involving IFNα or interleukin-2 develop depressive symptoms; (ii) sickness behavior induced by the endotoxin lipopolysaccharide (LPS) or interleukin-1 shares some common features with depression; (iii) several cytokines can activate the hypothalamic–pituitary–adrenal (HPA) axis, which is commonly activated in depressed individuals; and (iv) some cytokines regulate brain norepinephrine or serotonin systems, which are linked to MDD and its treatment (3–6).

p11, a member of the S100 family of proteins, is a key regulator of depressive-like states and antidepressant responses in rodent models (7–9). p11, also called S100A10, is a small acidic protein that interacts with specific serotonin receptors to regulate their trafficking and influence their localization at the cell surface (7, 9). This action affects the excitability of the cells and leads to profound behavioral responses. p11 knockout (KO) mice display a depressive-like phenotype and p11 overexpressing mice display antidepressant-like responses in classical behavioral paradigms, including the tail suspension and forced swim tests (7). Genetically “helpless” mice, which exhibit some symptoms of depression, as well as humans suffering with MDD, have reduced levels of p11 mRNA and protein in the cerebral cortex and striatum (7, 10). In rodents, three classes of antidepressants (SSRIs, TCAs, and ECT) increase p11 levels in the cerebral cortex and hippocampus (7, 8). Here we report an investigation of antiinflammatory drugs and of antidepressant agents on p11 expression and antidepressant behavior.

The “sequenced treatment alternatives to relieve depression” (STAR*D) clinical trial was a large, real-world study of treatment-resistant depression that evaluated a series of treatments and clinical outcomes. Many patients continue to suffer residual symptoms after weeks of treatment with a single agent and even after trying many different antidepressants and combination therapies. The STAR*D study found that only 36.8% of patients exhibited remission after treatment with the SSRI, citalopram (CIT), and found a cumulative remission rate of 67% after multiple treatments were attempted (11). The underlying factors contributing to treatment resistance remain unclear. We used the STAR*D dataset to determine whether nonsteroidal antiinflammatory drugs (NSAIDs) could play a role in the treatment outcome of depressed individuals taking SSRIs.

Here we test a model in which SSRI antidepressants increase brain levels of certain cytokines, which in turn regulate p11 levels and ultimately control the behavioral response to an SSRI antidepressant (Fig. 1). Importantly, we have identified an antagonism by NSAIDs of SSRI responsiveness, which is likely mediated through the action of certain cytokines and p11 in the brain. We believe that this antagonism contributes, in part, to the high resistance rates to SSRIs seen in MDD. We believe that reduced use of NSAIDs by physicians in severely depressed patients being treated with SSRIs would significantly improve positive outcomes from this major class of antidepressant.

Fig. 1.

Fig. 1.

Open in a new tab

Schematic representation of the model tested in the current study. Here we suggest that antidepressants, specifically SSRI antidepressants, increase brain levels of certain cytokines, which in turn increase levels of p11, the effect of which produces behavioral antidepressant responses. Each step in this pathway can be antagonized by NSAID coadministration.

Results

Effects of SSRIs and NSAIDs on Certain Cytokines and p11.

We measured mouse brain levels of cytokines using a bead-based ELISA following chronic treatment with the SSRI citalopram in the presence or absence of ibuprofen (IBU) cotreatment (SI Materials and Methods). We focus on the frontal cortex, a brain area that is strongly linked to antidepressant responses in mice and humans (12, 13). Results identified cytokines that fell into one of three major categories: (i) cytokines that were increased by citalopram, the effect of which was abolished by IBU cotreatment (Fig. 2A); (ii) cytokines that were increased by citalopram, the effect of which was not affected by IBU cotreatment (Fig. 2A); or (iii) cytokines that were not changed by either citalopram or IBU (IL-1a, IL-4, and IL-13). IBU reduced plasma levels of both citalopram and its metabolite didesmethyl citalopram (ddCIT) compared with mice that received citalopram alone (CIT: 1508.36 ± 282.7 versus 537.03 ± 65.8 pg/μL, P < 0.05; ddCIT: 128.0 ± 21.0 versus 34.5 ± 10.8 pg/μL, P < 0.01).

Fig. 2.

Fig. 2.

Open in a new tab

Effect of antidepressants and NSAIDs on cytokine levels and p11 in the mouse frontal cortex. (A) Mouse frontal cortex samples from animals treated with vehicle (VEH), citalopram (CIT), ibuprofen (IBU), or both IBU and CIT were analyzed for levels of cytokines. Group 1 cytokines were increased by citalopram, the effect of which was abolished by ibuprofen cotreatment. Group 2 cytokines were increased by citalopram, the effect of which was not affected by ibuprofen. (B) Western blot analysis of p11 protein or actin loading control in the frontal cortex of mice receiving a chronic selective serotonin reuptake inhibitor (citalopram or fluoxetine) or a tricyclic antidepressant (desipramine), alone or in combination with ibuprofen or acetylsalicylic acid. Representative blots (Upper); quantification of five to six mice per group (Lower). All data are presented as means ± SEM. Statistically significant effects of antidepressants (#P < 0.05) or NSAIDs (*P < 0.05) are noted.

We have shown previously that p11 is increased in the mouse frontal cortex by multiple classes of antidepressants (7, 8). Here we investigated whether antiinflammatory agents alone or in combination with antidepressants regulated p11 levels. Interestingly, coadministration of either ibuprofen (IBU) or another NSAID, acetylsalicylic acid (ASA), with antidepressants for 14 d blocked the increase in p11 caused by two different SSRIs, citalopram or fluoxetine (Fig. 2B). The tricyclic antidepressant (TCA) desipramine induced a small increase in p11, and this increase was not significantly affected by IBU or ASA [Fig. 2B, interaction NSAID × antidepressant: F(6, 52) = 4.48, P < 0.01]. Taken together, these experiments show that SSRI antidepressants increase brain levels of certain cytokines and p11, the effect of which is abolished by NSAID cotreatment.

Antidepressant-Induced Increases in p11 Levels Are Cytokine Dependent.

Two of the cytokines that were regulated by both citalopram and ibuprofen, namely, IFNγ and TNFα, were further studied as possible mediators of the inhibitory effects of NSAIDs on SSRI-induced p11 levels described above. First, we investigated whether IFNγ or TNFα signaling is required for antidepressant-induced increases in p11, using IFNGR1 or TNFR1 KO mice. Western blotting analysis of frontal cortex from IFNGR1 KO, TNFR1 KO, or WT control mice treated with chronic citalopram revealed that both IFNGR1 and TNFR1 signaling are necessary for the increase in p11 by citalopram. Citalopram significantly increased p11 in WT, but not IFNGR1 KO or TNFR1 KO mice [Fig. 3A, interaction genotype × treatment F(1, 22) = 6.22, P < 0.05].

Fig. 3.

Fig. 3.

Open in a new tab

IFNγ and TNFα are necessary and sufficient for antidepressant-induced increases in p11 levels. (A) Western blot analysis of p11 protein or actin loading control in the frontal cortex of WT, IFNGR1 KO, or TNFR1 KO mice treated for 3 wk with citalopram. Representative blots (Upper); quantification of 8–10 mice per group (Lower). All data are presented as means ± SEM. Statistically significant effects of antidepressants (##P < 0.01) or genotype (*P < 0.05) are noted. (B) Western blot analysis of p11 or actin loading control reveals that acute i.p. injection of IFNγ (10 μg/kg bodyweight) or TNFα (10 μg/kg bodyweight) significantly increases p11 protein in mouse cortex compared with vehicle injected controls. Representative blots (Upper); quantification of 5–6 mice per group (Lower). All data are presented as means ± SEM. (*P < 0.05).

To examine whether IFNγ or TNFα was sufficient to increase p11 in vivo, mice were injected with recombinant IFNγ, TNFα, or vehicle and euthanized 4 h later. Western blot analysis revealed that both cytokines significantly increased p11 protein in the frontal cortex compared with vehicle injected controls [Fig. 3B, F(2, 16) = 8.519, P < 0.01].

Immunohistochemical detection of IFNγ receptor 1 (IFNGR1), p11, and the neuronal marker NeuN demonstrated that neurons in layer 5 of the mouse cortex express both p11 and IFNGR1 (Fig. S1). TNF receptor 1 (TNFR1) was also coexpressed with p11 in cortical neurons (Fig. S1). These data support the idea that these cytokines may regulate p11 levels.

NSAIDs Prevent the Antidepressant-Like Effect of SSRIs in Classical Behavioral Paradigms.

Because p11 has been shown to be both necessary and sufficient for behavioral antidepressant responses (7–9) and IBU potently inhibited antidepressant-induced increases in p11 (Fig. 2B), we examined the possibility that IBU might inhibit the behavioral response to antidepressant drugs. We tested various classes of antidepressants including SSRIs (citalopram and fluoxetine), TCAs (imipramine and desipramine), a MAOI (tranylcypromine), and an atypical antidepressant (bupropion) in two well-established mouse models of depression: the tail suspension test (TST) and the forced swim test (FST). All antidepressants tested significantly reduced immobility time in both the TST [Fig. 4A, interaction antidepressant × NSAID: F(4, 95) = 6.11, P < 0.001] and the FST [Fig. 4B, interaction antidepressant × NSAID: F(6, 106) = 5.07, P < 0.001]. IBU significantly attenuated the antidepressant-like effects of SSRIs in both tests (Fig. 4 A and B). IBU was less effective in altering the behavioral response to TCAs and failed to alter the behavioral response to other classes of antidepressant drugs.

Fig. 4.

Fig. 4.

Open in a new tab

Effects of antidepressants and NSAIDs on behavioral responses. NSAIDs and other analgesics attenuate the behavioral response to SSRIs. IBU (5–7 d) diminished the behavioral response to the selective serotonin reuptake inhibitors citalopram (CIT) and fluoxetine (FLX), was less effective in altering behavioral responses to the tricyclic antidepressants imipramine (IMI) and desipramine (DMI), and did not affect responses to other classes of antidepressants, including the monoamine oxidase inhibitor tranylcipromine (TCP) and the atypical antidepressant bupropion (BUP) in two tests of antidepressant activity, the tail suspension test (A) and forced swim test (B). Mice receiving 5–7 d of nonsteroidal antiinflammatory drugs ibuprofen (IBU), naproxen (NPX), acetylsalicylic acid (ASA), or the analgesic acetaminophen (ACE) showed diminished response to citalopram (CIT) in the tail suspension test (C) and forced swim test (D). There was no response to chronic citalopram treatment when ibuprofen was coadministered before testing in the tail suspension test (E) or the novelty suppressed feeding test (F). All data are presented as means ± SEM. Statistically significant effects of antidepressants (#P < 0.05) or NSAIDs/analgesics (*P < 0.05, **P < 0.01) are noted. n = 8–16 per group.

To examine the specificity of the effect of IBU on SSRI-induced behavioral changes, we tested the effect of three different NSAIDs and an analgesic on the behavioral response to citalopram. Mice were treated with IBU (1 mg/mL), naproxen (2 mg/mL), acetylsalicylic acid (3 mg/mL), or acetaminophen (3 mg/mL) in their drinking water for 5–7 d and received a single injection of citalopram (20 mg/kg, i.p.) or saline before testing in the TST or FST. All of the drugs tested significantly blocked the antidepressant effect of citalopram on immobility time in both tests (Fig. 4 C and D) compared with mice that received citalopram alone [TST interaction antidepressant × NSAID: F(6, 129) = 3.25, P < 0.01; FST interaction antidepressant × NSAID: F(6, 106) = 5.07, P < 0.005].

To determine whether ibuprofen inhibits the behavioral response to chronic SSRI administration, we coadministered citalopram and ibuprofen for 2 wk before testing mice in the novelty suppressed feeding (NSF) test or TST. In both tests, ibuprofen blocked the behavioral antidepressant-like response to citalopram [Fig. 4 E and F, NSF interaction antidepressant × NSAID: F(1, 56) = 4.23, P < 0.05; TST interaction antidepressant × NSAID: F(1, 56) = 4.23, P < 0.05]. There were no differences among groups tested in home cage feeding (vehicle 0.08 g ± 0.024; citalopram 0.1 ± 0.033; ibuprofen 0.08 ± 0.02; ibuprofen and citalopram 0.08 ± 0.02). Citalopram or ibuprofen and citalopram significantly increased bodyweight (vehicle 23.3g ± 0.27; citalopram 24.7 ± 0.23; ibuprofen 23.4 ± 0.35; ibuprofen and citalopram 25.0 ± 0.20, P < 0.01).

Effects of Cytokines and p11 on Behavioral Responses.

Cytokines, their receptors, and p11 are expressed by neurons, glia, and endothelial cells. To determine whether the effects of antidepressants, IFNγ, or TNFα on immobility required neuronal p11 expression, we generated a conditional p11 KO mouse by crossing a mouse with lox-P sites flanking exon 2 of the p11 gene with a mouse expressing the Cre recombinase under the CAMK2α promoter (14, 15). The CAMK2α promoter was chosen because it is expressed only in neurons of the forebrain. The resulting p11 KO mouse lacked p11 only in cells that expressed the CAMK2α-Cre. Immunohistochemical analysis of the p11 KO mice showed a complete lack of p11 protein expression in the neurons of the cortex and hippocampus, and lessened expression in the striatum (Fig. S2). Citalopram had no effect on TST immobility in p11 KO mice (Fig. 5A). In contrast, p11 KO mice responded normally to a tricyclic antidepressant, desipramine, underscoring the specificity of our results for serotonergic antidepressants (Fig. 5A).

Fig. 5.

Fig. 5.

Open in a new tab

Effects of cytokines and p11 on behavioral responses. (A) CAMK2α conditional p11 KO mice show reduced immobility in response to desipramine, but do not respond to citalopram in the tail suspension test. (B) Immobility is significantly reduced by acute injection of IFNγ or TNFα in wild-type mice, but not in CAMK2α-conditional p11 KO mice (n = 13–24 per group). All data are presented as means ± SEM. *P < 0.05, **P < 0.01. (C and D) Behavioral analysis of WT, IFNGR1 KO, or TNFR1 KO mice treated for 14 d with citalopram in the tail suspension test (C) or the novelty suppressed feeding test (D). All data are presented as means ± SEM. Statistically significant effects of citalopram (#P < 0.05) or genotype (*P < 0.05) are noted.

Acute injection of IFNγ (10 μg/kg) or TNFα (10 μg/kg) before testing in the TST significantly reduced immobility in wild-type mice, consistent with an antidepressant-like effect [Fig. 5B, interaction genotype × treatment F(2, 64) = 3.55, P < 0.05]. A lower dose of IFNγ (1 μg/kg) or TNFα (1 μg/kg) had no effect on TST immobility (IFNγ 107.9% ± 9.6, n = 6; TNFα 98.6% ± 7.3, n = 5). IFNγ or TNFα had no effect on TST immobility in p11 KO mice [Fig. 5B, interaction genotype × treatment F(2, 77) = 5.18, P < 0.01], suggesting that the antidepressant-like actions of IFNγ and TNFα are mediated by p11 in neurons of the cortex.

We tested whether behavioral responses to chronic SSRI treatment were absent in IFNGR1 KO or TNFR1 KO mice. The effect of citalopram on TST immobility was abolished in IFNGR1 KO mice, but not in TNFR1 KO mice [Fig. 5C, interaction genotype × treatment F(2, 71) = 93.32, P < 0.0001]. Citalopram significantly reduced the latency to feed in WT mice but not in IFNGR1 KO or TNFR1 KO mice in the NSF test [Fig. 5D, interaction genotype × treatment F(4, 107) = 8.39, P < 0.001]. Both KO mouse lines responded normally to desipramine in the TST (WT: 61.03% ± 9.9; IFNGR1 KO: 71.56% ± 9.3; TNFR1 KO: 65.4% ± 8.1, percentage of vehicle control group ± SEM), demonstrating the specificity for serotonergic antidepressants.

NSAIDs and Other Analgesics Inhibit SSRI Efficacy in a Clinical Population.

The TST and FST are two well-established rodent behavioral paradigms that predict antidepressant efficacy in humans (16, 17). To determine whether NSAIDs influence SSRI efficacy in clinically depressed individuals, we took advantage of the STAR*D dataset. The first level of this large scale clinical trial investigated remission rates in depressed patients taking citalopram for 12 wk (Materials and Methods). Contingency tables and analyses were performed on 1,546 human subjects for which there were concomitant medication data and a data point at week 12 for presence/absence of clinical “remission” from depressive symptoms (Table 1). These data show that 182 subjects were in remission at the end of 12 wk of treatment with citalopram and had taken an NSAID at least once during those 12 wk. There were 628 subjects in remission who had not taken an NSAID. There were 227 subjects who were treatment resistant (i.e., did not experience remission) and had taken an NSAID at least once during the 12 wk of treatment. Finally, there were 509 subjects who were treatment resistant and had not taken any NSAID.

Table 1.

Effects of NSAIDs and other analgesics on treatment response to citalopram in humans

Remission No Remission remission rate (%) P value
NSAID 182 227 44.5 P = 0.0002
No NSAID 628 509 55.2
Analgesic 52 88 37.1 P = 0.0002
No analgesic 758 648 53.9
Both (NSAID and analgesic) 23 40 36.5 P = 0.0138
None 787 696 53.1
Either (NSAID or analgesic) 211 275 43.4 P < 0.0001
None 599 461 56.5
Vitamins 44 31 58.7 P = 0.2874
No vitamins 766 705 52.1
Open in a new tab

Number of patients in remission or not in remission at the end of 12 wk of treatment with citalopram shown as a function of presence or absence of various concomitant treatments. Data were analyzed using Fisher's exact test; P < 0.05 indicated a statistically significant relationship between the concomitant medication and the treatment outcome.

Of those subjects who took an NSAID, 45% were in remission, and 55% were treatment resistant. Of those subjects who did not take an NSAID, the reverse relationship was observed with 55% in remission and 45% being treatment resistant. In other words, a higher percentage of patients were treatment resistant to citalopram if they had taken an NSAID than if they had not taken an NSAID. This relationship was statistically significant (P = 0.0002).

Similar analyses were conducted for other analgesics and similar results were found. Subjects taking other analgesics were less likely to undergo remission (37% in remission) compared with those not taking analgesics (54% in remission, P = 0.0002).

Another analysis was conducted to determine whether the relationship between remission and concomitant medication was strongest for subjects who were taking both NSAIDS and other analgesics. Despite the relatively fewer number of subjects taking both NSAIDs and other analgesics, the relationship appeared to be quite strong with 63% of subjects who were taking both types of medication failing to show remission (P = 0.0138) in contrast to 47% failure for patients taking neither.

Moreover, a contingency table was set up for subjects who had taken either NSAIDs or analgesics. The relationship between taking either of these medications and being more likely to show treatment resistance was highly statistically significant (P < 0.0001).

Importantly, there seemed to be specificity in the relationship between the presence of an NSAID and/or other analgesic with treatment resistance (lack of remission) because there was no relationship between treatment remission and other concomitant medications such as vitamins.

Discussion

Previous work from our laboratory indicated that p11 is a determinant of depressive-like states and antidepressant responses (7–9). Here we provide evidence that antidepressants increase brain levels of certain cytokines, which increase p11 levels, which then induce antidepressant-like behavioral responses (Fig. 1). We have further shown that IFNγ and TNFα are two cytokines that may be involved in this process. Antiinflammatory drugs antagonized both the induction of p11 by and the behavioral response to SSRI antidepressants. We used the STAR*D dataset to confirm our results in a clinical population. Consistent with our mouse studies, we found that human patients reporting concomitant NSAID or other analgesic treatments showed a reduced therapeutic response to citalopram. Concomitant use of NSAIDs may be an important reason for high SSRI treatment resistance rates. We suggest that NSAIDs and other analgesics may potentially interfere with the therapeutic efficacy of SSRIs.

P11 expression is detected in various brain areas including the frontal cortex, hippocampus, striatum, amygdala, and dorsal raphe nucleus (7). Overexpression of p11 in the forebrain mimics the action of an antidepressant (7). Here we show that p11 in forebrain neurons is necessary for the action of an SSRI antidepressant, but not a tricyclic antidepressant, suggesting that SSRI and noradrenergic antidepressants might act through different mechanisms and that p11 is selectively involved in pathways related to SSRI activity.

The antidepressant-like effects of TNFα on p11 and behavior could be mediated by direct effects of TNFα on neurons or indirectly through neurotrophic factors. TNFα increases production by astrocytes of neurotrophic factors, including nerve growth factor (NGF), glia-derived neurotrophic factor (GDNF), and brain-derived neurotrophic factor (BDNF) (18, 19). TNFα also increases BDNF in human cerebral endothelial cells (20), and BDNF (i) increases p11 in the mouse cerebral cortex and (ii) has antidepressant-like and neurogenic effects that require p11 (8, 21). The effects of TNFα on levels of neurotrophins may be an important component of the mechanism by which TNFα produces an antidepressant-like response in rodent models of depression.

Our results further suggest that, like TNFα, IFNγ mediates its antidepressant activity through its action on p11. IFNγ increases p11 through a direct interaction with IFNγ binding sites on the p11 promoter (22). IFNγ increases neurite outgrowth (23), promotes neuronal differentiation (24), and enhances neurogenesis (25), all of which are consistent with antidepressant-like activity (26). However, it has been suggested that IFNγ mediates depression-behaviors caused by immune activation (27). Others have suggested that IFNγ plays a role in depression-like behaviors, but find that IFNγ KO mice respond normally to chronic unpredictable stress in the FST (28). These results are not inconsistent with our findings. It is well established that there is an increase in the level of plasma cytokines in depressed subjects (4). We speculate that the production of these cytokines and their actions in the periphery may be distinct from those local effects observed in brain areas like the frontal cortex. It is also possible that the increased levels of certain cytokines in the periphery of depressed individuals are involved in efforts by the brain to compensate for depression.

Future studies will be necessary to determine the mechanism by which NSAIDs inhibit SSRI efficacy. Acetaminophen is not generally thought to be antiinflammatory, so perhaps the antipyretic actions of the NSAIDs and analgesics is more related to their antagonism of SSRI efficacy. The two drugs could interfere with each other in the periphery, because most NSAIDs do not readily cross the blood brain barrier and because blood levels of citalopram and its metabolite were decreased in mice that also received ibuprofen. We cannot exclude the possibility that NSAIDs are having a direct effect on the interaction between SSRIs and the serotonin transporter, even though the involvement of p11 in antidepressant activity is mediated by neurons in the forebrain.

Analysis of our clinical data strongly suggests that remission rates among depressed individuals may be improved by avoiding certain common over-the-counter medications such as ibuprofen, aspirin, and acetaminophen. Depressed patients treated for 12 wk with citalopram are significantly less likely to show full remission if they are concomitantly taking NSAIDs and/or other analgesics.

The data suggest that treatment with NSAIDs prevents clinical responses to antidepressants. However, it is possible that underlying condition(s) contribute to treatment resistance rather than any one particular mechanism of action of concomitant medication. Indeed, it has been reported that depressed patients with painful physical symptoms took longer to achieve remission from depressive symptoms and were less likely to achieve remission than patients without pain (29). We cannot exclude the possibility that severity of depression and accompanying pain symptoms could be associated with antidepressant treatment resistance. However, in one study, Leuchter and colleagues (29) adjusted statistically for potential confounding factors such as race, sex, ethnicity, and severity of depression at baseline, and report that the statistical significance of the relationship between pain and remission from depression was lost. They concluded that the presence and severity of physical pain are not predictors of poor antidepressant treatment outcome, but that physical pain is associated with some factors that are predictors. Our present data suggest that at least one of the factors associated with physical pain that is a predictor of SSRI treatment outcome is concomitant therapy with NSAIDs and other analgesics. Despite the lack of understanding of causality associated with our clinical data analysis, our animal data strongly suggest that treatment with NSAIDs and/or other analgesics prevents antidepressant action of SSRIs.

Because the clinical analyses were conducted as post hoc analyses, it would be informative to evaluate the effects of NSAIDs and other analgesics on SSRI antidepressant response in a prospective, double-blind, randomized clinical study. Specifically, it will be important to standardize medications to better evaluate their role in determining treatment outcome. In the present study, no adjustments were made for multiple comparisons in the analyses. However, the effects were highly statistically significant such that small statistical adjustments would not affect the overall statistical significance or interpretation of the data. Moreover, the lack of a significant relationship between vitamins and clinical response suggests specificity to the underlying mechanisms by which NSAIDs and other analgesics prevent clinical remission. In addition, medical coding for concomitant medication in the database may not have been consistent across subjects or medications. Individuals taking many prescription medications may have been less likely to report over-the-counter medications, so reports of over-the-counter NSAIDs and analgesics may be underrepresented in the STAR*D dataset. Also, there are no data regarding dose of concomitant medications in the database and there is little information regarding duration of use (i.e., duration is only available for some medications). The utility of the data shown in Table 1 may be limited by not being able to differentiate between subjects taking concomitant medications for a single administration or only rarely and subjects taking the medications chronically. The magnitude of effect for the relationship between medication and treatment resistance may be even greater for those subjects who chronically take NSAIDs and/or other analgesics. Lastly, any subjects who discontinued before week 12 may have discontinued due to lack of efficacy. The percentage of subjects in remission may be overrepresented at week 12. However, many subjects (26%) dropped out of the acute phase of the trial due to nonmedical reasons (2). Therefore, subject discontinuation may or may not have affected the outcome of the present results. Despite these weaknesses, our findings indicate that many common over-the-counter medications greatly affect treatment response to SSRI antidepressants.

NSAIDs have been reported to increase the efficacy of some antidepressant treatments, but those reports focus on tricyclic or noradrenergic antidepressants (30) and not SSRIs. Our data indicate that the antagonism by NSAIDs of antidepressant responses is specific for serotonergic antidepressants. It is important to emphasize that the interaction between antidepressants and antiinflammatory agents appears to be specific to the efficacy of SSRIs and not a general effect on all classes of antidepressants. Furthermore, there is no evidence from our studies that NSAID administration alone has any effect on depressive-like states.

We report here a robust inhibitory effect of NSAIDs on SSRI-induced increases in p11 and on antidepressant-like behaviors in rodents. We confirmed the association in a dataset from a large-scale real-world human study (STAR*D), underscoring the clinical significance of these results. Work is ongoing to understand the cellular and molecular mechanism(s) underlying these effects, but the clinical implications of our findings are clear. With that, we urge the medical community to consider these findings when designing treatment strategies for their patients that include SSRIs.

Materials and Methods

Animals.

Eight- to 10-wk-old male mice were used for all experiments, and were housed four to five per cage with ad libitum access to food and water. C57BL/6 mice were purchased from Charles River Laboratories. p11 KO mice (derived and maintained at The Rockefeller University) were also used (7). CAMK2α−p11 conditional knockout mice were generated for these studies (SI Materials and Methods ). Animal use and procedures were in accordance with the National Institutes of Health guidelines and approved by the institutional animal care and use committees.

Sample Preparation and Western Blotting.

Western blotting was performed using standard procedures as described (9) (SI Materials and Methods).

Behavioral Assays.

The TST, FST, and open field locomotor activity were performed as described (9). Novelty suppressed feeding was performed as described (21).

Clinical Data Analyses.

Please see SI Materials and Methods.

Statistical Analyses.

All comparisons were made by ANOVA using Prism 5 software (GraphPad). In experiments composed of more than two groups, data were first analyzed by two-way ANOVA followed by a post hoc Bonferroni test. Statistical significance was set at P < 0.05.

Supplementary Material

Supporting Information
supp_108_22_9262__index.html (929B, html)

Acknowledgments

This work has been supported by The Skirball Foundation, USAMRAA Grants W81XWH-08-1-0111 and W81XWH-09-1-0401, National Institutes of Health (NIH)/National Institute of Mental Health (NIMH) Grant MH074866, NIH/National Institute on Aging Grant AG09464. Plasma analyses were carried out in the MS/Proteomics Resource of the W. M. Keck Foundation Biotechnology Laboratory at Yale University. The clinical data analysis was made possible by limited access datasets distributed from “Sequenced Treatment Alternatives to Relieve Depression” (STAR*D), supported by NIMH Contract N01MH90003 to the University of Texas Southwestern Medical Center. The clinicaltrials.gov identifier is NCT00021528.

Footnotes

The authors declare no conflict of interest.

See Commentary on page 8923.

This article contains supporting information online at www.pnas.org/lookup/suppl/doi:10.1073/pnas.1104836108/-/DCSupplemental.

References

  • 1.Kessler RC, et al. Lifetime prevalence and age-of-onset distributions of mental disorders in the World Health Organization's World Mental Health Survey Initiative. World Psychiatry. 2007;6:168–176. [PMC free article] [PubMed] [Google Scholar]
  • 2.Warden D, Rush AJ, Trivedi MH, Fava M, Wisniewski SR. The STAR*D Project results: A comprehensive review of findings. Curr Psychiatry Rep. 2007;9:449–459. doi: 10.1007/s11920-007-0061-3. [DOI] [PubMed] [Google Scholar]
  • 3.Dunn AJ, Swiergiel AH, de Beaurepaire R. Cytokines as mediators of depression: What can we learn from animal studies? Neurosci Biobehav Rev. 2005;29:891–909. doi: 10.1016/j.neubiorev.2005.03.023. [DOI] [PubMed] [Google Scholar]
  • 4.Irwin MR, Miller AH. Depressive disorders and immunity: 20 years of progress and discovery. Brain Behav Immun. 2007;21:374–383. doi: 10.1016/j.bbi.2007.01.010. [DOI] [PubMed] [Google Scholar]
  • 5.Miller AH, Maletic V, Raison CL. Inflammation and its discontents: The role of cytokines in the pathophysiology of major depression. Biol Psychiatry. 2009;65:732–741. doi: 10.1016/j.biopsych.2008.11.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 6.Khairova RA, Machado-Vieira R, Du J, Manji HK. A potential role for pro-inflammatory cytokines in regulating synaptic plasticity in major depressive disorder. Int J Neuropsychopharmacol. 2009;12:561–578. doi: 10.1017/S1461145709009924. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 7.Svenningsson P, et al. Alterations in 5-HT1B receptor function by p11 in depression-like states. Science. 2006;311:77–80. doi: 10.1126/science.1117571. [DOI] [PubMed] [Google Scholar]
  • 8.Warner-Schmidt JL, et al. A role for p11 in the antidepressant action of brain-derived neurotrophic factor. Biol Psychiatry. 2010;68:528–535. doi: 10.1016/j.biopsych.2010.04.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Warner-Schmidt JL, et al. Role of p11 in cellular and behavioral effects of 5-HT4 receptor stimulation. J Neurosci. 2009;29:1937–1946. doi: 10.1523/JNEUROSCI.5343-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Alexander B, et al. Reversal of depressed behaviors in mice by p11 gene therapy in the nucleus accumbens. Sci Transl Med. 2010;2(54):54ra76. doi: 10.1126/scitranslmed.3001079. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 11.Rush AJ, et al. Acute and longer-term outcomes in depressed outpatients requiring one or several treatment steps: A STAR*D report. Am J Psychiatry. 2006;163:1905–1917. doi: 10.1176/ajp.2006.163.11.1905. [DOI] [PubMed] [Google Scholar]
  • 12.Slattery DA, Neumann I, Cryan JF. Transient inactivation of the infralimbic cortex induces antidepressant-like effects in the rat. J Psychopharmacol. 2010 doi: 10.1177/0269881110368873. in press. [DOI] [PubMed] [Google Scholar]
  • 13.Mayberg HS, et al. Deep brain stimulation for treatment-resistant depression. Neuron. 2005;45:651–660. doi: 10.1016/j.neuron.2005.02.014. [DOI] [PubMed] [Google Scholar]
  • 14.Casanova E, et al. A CamKIIalpha iCre BAC allows brain-specific gene inactivation. Genesis. 2001;31:37–42. doi: 10.1002/gene.1078. [DOI] [PubMed] [Google Scholar]
  • 15.Zhang Z, Hofmann C, Casanova E, Schütz G, Lutz B. Generation of a conditional allele of the CBP gene in mouse. Genesis. 2004;40:82–89. doi: 10.1002/gene.20068. [DOI] [PubMed] [Google Scholar]
  • 16.Cryan JF, Mombereau C, Vassout A. The tail suspension test as a model for assessing antidepressant activity: Review of pharmacological and genetic studies in mice. Neurosci Biobehav Rev. 2005;29:571–625. doi: 10.1016/j.neubiorev.2005.03.009. [DOI] [PubMed] [Google Scholar]
  • 17.Cryan JF, Valentino RJ, Lucki I. Assessing substrates underlying the behavioral effects of antidepressants using the modified rat forced swimming test. Neurosci Biobehav Rev. 2005;29:547–569. doi: 10.1016/j.neubiorev.2005.03.008. [DOI] [PubMed] [Google Scholar]
  • 18.Kuno R, et al. The role of TNF-alpha and its receptors in the production of NGF and GDNF by astrocytes. Brain Res. 2006;1116:12–18. doi: 10.1016/j.brainres.2006.07.120. [DOI] [PubMed] [Google Scholar]
  • 19.Saha RN, Liu X, Pahan K. Up-regulation of BDNF in astrocytes by TNF-alpha: A case for the neuroprotective role of cytokine. J Neuroimmune Pharmacol. 2006;1:212–222. doi: 10.1007/s11481-006-9020-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Bayas A, et al. Human cerebral endothelial cells are a potential source for bioactive BDNF. Cytokine. 2002;19:55–58. doi: 10.1006/cyto.2002.0892. [DOI] [PubMed] [Google Scholar]
  • 21.Egeland M, Warner-Schmidt J, Greengard P, Svenningsson P. Neurogenic effects of fluoxetine are attenuated in p11 (S100A10) knockout mice. Biol Psychiatry. 2010;67:1048–1056. doi: 10.1016/j.biopsych.2010.01.024. [DOI] [PubMed] [Google Scholar]
  • 22.Huang XL, et al. Interferon-gamma induces p11 gene and protein expression in human epithelial cells through interferon-gamma-activated sequences in the p11 promoter. J Biol Chem. 2003;278:9298–9308. doi: 10.1074/jbc.m212704200. [DOI] [PubMed] [Google Scholar]
  • 23.Song JH, et al. Interferon gamma induces neurite outgrowth by up-regulation of p35 neuron-specific cyclin-dependent kinase 5 activator via activation of ERK1/2 pathway. J Biol Chem. 2005;280:12896–12901. doi: 10.1074/jbc.M412139200. [DOI] [PubMed] [Google Scholar]
  • 24.Kim SJ, et al. Interferon-gamma promotes differentiation of neural progenitor cells via the JNK pathway. Neurochem Res. 2007;32:1399–1406. doi: 10.1007/s11064-007-9323-z. [DOI] [PubMed] [Google Scholar]
  • 25.Baron R, et al. IFN-gamma enhances neurogenesis in wild-type mice and in a mouse model of Alzheimer's disease. FASEB J. 2008;22:2843–2852. doi: 10.1096/fj.08-105866. [DOI] [PubMed] [Google Scholar]
  • 26.Warner-Schmidt JL, Duman RS. Hippocampal neurogenesis: Opposing effects of stress and antidepressant treatment. Hippocampus. 2006;16:239–249. doi: 10.1002/hipo.20156. [DOI] [PubMed] [Google Scholar]
  • 27.O'Connor JC, et al. Interferon-gamma and tumor necrosis factor-alpha mediate the upregulation of indoleamine 2,3-dioxygenase and the induction of depressive-like behavior in mice in response to bacillus Calmette-Guerin. J Neurosci. 2009;29:4200–4209. doi: 10.1523/JNEUROSCI.5032-08.2009. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 28.Litteljohn D, et al. Interferon-gamma deficiency modifies the effects of a chronic stressor in mice: Implications for psychological pathology. Brain Behav Immun. 2010;24:462–473. doi: 10.1016/j.bbi.2009.12.001. [DOI] [PubMed] [Google Scholar]
  • 29.Leuchter AF, et al. Painful physical symptoms and treatment outcome in major depressive disorder: A STAR*D (Sequenced Treatment Alternatives to Relieve Depression) report. Psychol Med. 2010;40:239–251. doi: 10.1017/S0033291709006035. [DOI] [PubMed] [Google Scholar]
  • 30.Müller N, et al. The cyclooxygenase-2 inhibitor celecoxib has therapeutic effects in major depression: Results of a double-blind, randomized, placebo controlled, add-on pilot study to reboxetine. Mol Psychiatry. 2006;11:680–684. doi: 10.1038/sj.mp.4001805. [DOI] [PubMed] [Google Scholar]
  • 31.Opalka D, et al. Simultaneous quantitation of antibodies to neutralizing epitopes on virus-like particles for human papillomavirus types 6, 11, 16, and 18 by a multiplexed luminex assay. Clin Diagn Lab Immunol. 2003;10:108–115. doi: 10.1128/CDLI.10.1.108-115.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supporting Information
supp_108_22_9262__index.html (929B, html)
1104836108_pnas.201104836SI.pdf (699.4KB, pdf)

Articles from Proceedings of the National Academy of Sciences of the United States of America are provided here courtesy of National Academy of Sciences

RESOURCES