Clinical Psychopharmacology Seminar

Treatment of Refractory Depression

Original Author(s): Bruce Alexander, Pharm.D.
Latest Reviser(s): Mitchell Barnett, Pharm.D., M.S., and Paul J. Perry, Ph.D.
Creation Date: 1996
Last Revision Date: May 2005
Peer Review Status: Internally Peer Reviewed

INTRODUCTION

Treatment refractory, or treatment resistant depression is a significant public health concern, with increased levels of general and mental health related costs found to be directly related to the degree of treatment refractory depression (Russell et al, 2004). This handout will discuss the pharmacologic management of treatment refractory depression (TRD). Acute and prophylactic treatments are reviewed in subsequent monographs. The use of the term "depression" will be used generically unless otherwise specified (i.e., bipolar disorder, depression; major depression, single episode; dysthymia, atypical depression).

The term "treatment-resistant" depression will refer to patients that have failed an adequate dose and time trial of > 2 antidepressants. Specific subtypes excluded (see monograph on pharmacologic management of melancholic depression) all antidepressants are effective treatment for 60-70% of moderately to severely depressed patients (Morris and Beck 1974, Rogers and Clay 1975, Amsterdam and Hornig-Rohan 1996). By contrast, 30-40% of such patients improve with placebo. Thus, one third of patients have a specific drug response (the drug-placebo difference), one third improve with nonspecific treatment (the placebo response), and one third do not respond to a first trial of an antidepressant.

Predictors of Non-response

Nelson et al 1994 in a non-blind design studied the predictors of non-response in a group of 68 depressed patients (DSM III) that had failed to respond to a one-week hospitalization without drug treatment. The mean HAMD was 24 on admission and 20 after one week of hospitalization. Patients were dosed prospectively based on a desipramine serum level obtained 24 hours after a test dose. A dose was chosen that would achieve a desipramine serum level of 160 ng/ml within 1 week. The minimum effective desipramine concentration was 115 ng/ml, which 45/50 patients achieved. Eight patients dropped out for administrative reasons and 10 did so because of side effects. These patients tended to be younger and female than the patients that completed treatment. Overall, 30/50 (70%) responded (HAMD score <12). The four variables that accounted for 47% of the variance were an axis II diagnosis, prior treatment failure, delusions and age < 35 years. No other variables were significantly correlated with non-response. For patients with two or more predictors, only 25% responded, while in those with one or no variables, 91% responded.

Two studies looked at the association between response rates for patients with major depression vs. patients with major depression plus a personality disorder. Pfohl et al (1984) reported that only 16% (n=41) of patients with an axis II diagnosis responded to treatment vs. 50% (n=37) controls. Black et al 1988 reported that of patients with major depressive disorder that 60% (n=91) without and 42% (n=32) patients with an axis II diagnosis, respectively were recovered at hospital discharge (p<0.025). A recent chart review study by Sharma et al (2005) found that among 66 patients diagnosed with treatment refractory depression (failure of two trials of antidepressants), that 35% met the criteria for bipolar disorder and 59% met the criteria for a bipolar disorder within one-year. Further, among those patients classified as MDD (n=25), 52% and 80% were classified as bipolar at baseline and follow up, respectively.

Compliance Issues

In a managed care setting, 28% and 44% of patients stopped taking their antidepressant within 1 and 3 months, respectively of the first prescription (Lin et al 1995).

The number of times a patient has to take a medication during the day significantly affects their compliance. Over 70% of depressed patients failed to take 25% to 50% of their prescribed dosage on a qid regimen (Ayd 1972). However, when changed to a once daily schedule, only 7% of the patients missed doses. Therefore, complicance considerations in a TRD patient should include dosing frequency.

Response Criteria

A widely used definition of antidepressant drug response in research protocols is >50% reduction in the Hamilton Rating Scale for Depression (HRSD) (Endicott et al 1981). A more recent recommendation which is more restrictive, but requires patients to approach a "cure" is the achievement of a post-HRSD score <7 (Zimmerman et al 1985). An argument can be made for including both sets of data in antidepressant response reports.

Potential Treatment Strategies

Treatment refractory literature may be divided into 4 major categories.

OPTIMIZATION

Dose Effect

Two studies have demonstrated that imipramine or desipramine 300 mg/d was superior to 150 mg/d, respectively, to achieve an antidepressant response (Watt et al 1972, Simpson et al 1976).

In another study, forty-one patients, twenty-five women, average age 40 years who had failed to respond to an eight week open trial of fluoxetine 20 mg were followed (Fava et al 1994). Patients were randomly assigned to fluoxetine 40-60 mg/day, fluoxetine 20 mg/day plus desipramine 25-50 mg, or fluoxetine 20 mg plus lithium 300-600 mg/day. Those that received high-dose fluoxetine had a higher proportion of responders (8/15, 53%) than either the fluoxetine + desipramine (3/12, 25%) or fluoxetine + lithium (4/14, 29%)(ns). Among the partial responders, more patients responded to high-dose fluoxetine (5/6, 83%) than in the fluoxetine + desipramine (2/5, 40%) and fluoxetine +lithium (1/8, 13%) (p=0.03). Among the non-responders, 50% responded to fluoxetine + lithium vs. 33% given high-dose fluoxetine vs. 14% of fluoxetine + desipramine (ns).

Fava et al (2002) was unable to replicate this finding in a RCT of 101 outpatients with major depression who were either partial responders or nonresponders to an 8–week trial of fluoxetine 20 mg/d. Patients were randomly assigned to high-dose fluoxetine (HDF) 40-60 mg/day, fluoxetine 20 mg/day plus desipramine 25-50 mg (F+D), or fluoxetine 20 mg plus lithium (F+Li) 300-600 mg/day. There was no significant difference in the response rates between the three treatments regardless of whether the data was grouped as partial responders, non-responders, or total patients; however the trend of response always favored high dose fluoxetine.

TCA Serum Levels

That serum level monitoring might lead to an improved response rate in the treatment of depression has been suggested (Perry et al 1987, Glassman et al 1977). (see table 1). The remaining TCAs and other antidepressants have inadequate data to make conclusions about therapeutic ranges. This topic is extensively reviewed in the monograph dealing with antidepressant dosing.

Glassman et al (1977) reported that 40 to 50% of depressive patients did not achieve recommended serum levels with imipramine 200 to 225 mg/d. In this study, of the 13/16 patients that did not respond to levels <180 ng/ml, 7 responded promptly when a level >200 ng/ml was achieved.

Duration Effect

TCA and MAOI. In an early British Medical Research Council study (1965), 52% of patients treated for 4 weeks with a TCA responded. In 1973, the WPA Symposium on Therapy-Resistant Depression recommended that an adequate course of treatment was imipramine 150 mg/d or its equivalent for 4 to 6 weeks (Ayd 1983).

Quitkin et al (1984) suggests that a 5 to 6 week trial is necessary to establish responsivity. However, their conclusion was criticized because of what was deemed too slow of a dose titration schedule (Glassman and Roose 1990). Glassman and Roose (1990) indicated that 4 weeks is an adequate trial if initial maximum doses are achieved in 3 to 6 days. However, additional data agree with Quitkin et al that 6 weeks are required for most patients to respond to adequate doses of a TCA or MAOI (Georgotas et al 1989).

Two studies of the same design have investigated the effect of duration of treatment on response rate (Dornseif et al 1989, Schweizer et al 1990). Patients were treated with fluoxetine 20 mg/d for 3 weeks. The non-responders were randomized to receive an additional 5 weeks of fluoxetine 20 mg/d, or to have their dose increased to 60 mg/d. After the additional 5 weeks, 50% of patients in both groups responded. Thus, a therapeutic trial of fluoxetine could be considered to be 6 to 8 weeks.

A more recent placebo-controlled study by Quitkin et al (1996) was reported on 693 patients treated prospectively with a variety of antidepressant regimens. A response criterion was a sustained score on the CGI of much improved or very much improved. All patients were rated weekly. They concluded that those patients with some improvement in depressive signs/symptoms by week 4 should have their antidepressant regimen continued until week 6. This was based on the observation that 39% (versus 8% of placebo) of patients at week 4 who had minimal improvement were responders at week 6. However, if at week 4 the patient has had no improvement, then the treatment should be changed. This was based on the observation that for drug and placebo only 13% and 6%, respectively went on to respond at week 6.

Conclusion

It is important for comparison purposes that studies of "TRD" include specific details about dose, tapering, serum levels, and duration of treatment as well as diagnostic subtypes.

It is generally recommended that a patient should receive at least 4 weeks of an antidepressant at maximal tolerated doses before being considered a treatment failure (Wager and Klein 1988, Shelton 1999). Doses up to 400 mg/d should be considered in a patient that tolerates lower doses.

SUBSTITUTION

Tricyclic Antidepressants

TCAs have been classified as primary noradrenergic or serotonergic agents. This classification system has lead authors to speculate that if a patient fails treatment with one type of TCA agent that he may respond to the other type (Stern et al 1980). However, there is evidence that this practice only produces a response rate of 20% to 30% (Nierenberg and White 1990, Shelton 1999). No information is available on the combined use of TCAs.

One controlled study reported that 65% (11/17) of patients that failed a paroxetine trial responded to a 6 week trial of imipramine (Peselow et al 1989).

Monoamine Oxidase Inhibitors (MAOI)

A review of 10 studies reported that 62% (138/221) of patients that had failed a treatment trial of a TCA responded to a MAOI (Thase and Rush 1995). Following is a description of a selected group of reports.

Quitkin et al (1981) reported, retrospectively, that 76% of 40 patients with bipolar depression treated with a first or second-generation agent responded favorably to treatment. The authors indicated that 6 of 8 non-responders to standard antidepressants responded to a MAOI. They recommended MAOI, not another type of "cyclic" antidepressant, be used as second-line treatment when ECT is deemed unsuitable in non-responders.

Georgotas et al (1983) found that 13/20 (65%) patients responded to an average phenelzine dose of 48 mg/d (80% MAOI inhibition) in a 2 to 7 week open trial. All were TCA-resistant elderly patients and their final HRSD scores were <10.

Pare (1985) reported that 15/26 (58%) DSM-III-diagnosed patients with depression, who had failed successive four-week trials of a first and second-generation antidepressant, respectively, responded to a trial of the MAOI tranylcypromine.

Roose's et al (1986) reported all patients who failed to respond to an adequate trial of TCAs responded to MAOIs or ECT.

A double blind, placebo-controlled study reported that 17/26 (65%) TCA-failures improved with the MAOI phenelzine 72 + 8 mg/d (McGrath et al 1987).

Nolen et al (1988) in an open randomized trial studied 67 patients that failed > imipramine 150 mg/d for 4 weeks. Serum concentrations were not reported. Forty-five (58%) of the patients responded to a subsequent trial of tranylcypromine.

Amsterdam and Berwish (1989) in a case series of 14 patients that failed an average of 5 different treatments reported that all patients responded to high doses of tranylcypromine. Average doses were 114 mg/d (+ 14).

The report of several patients treated long term (i.e., > 6 weeks) with MAOIs developing tolerance to the antidepressant effect is disturbing (Mann 1983). Despite increases in daily doses of MAOI and stable MAOI inhibition, the patient's HRSD scores returned to pretreatment levels. The number of patients who develop tolerance to MAOIs is not known and appropriate management guidelines are not available.

Second-Generation Antidepressants

Bupropion. Thirty TCA-refractory patients were randomized to placebo (n=11) or bupropion (n=19). Bupropion 300-750 mg/d (modal dose 600-750 mg/d) was significantly superior to placebo on both the HRSD and CGI scales (Stern et al 1983). Twelve of 19 patients (63%) were rated as much or very much improved. Note: current dosing guideline state a maximum of 450 mg/d may be prescribed.

An open study reported bupropion was effective in treating 54% of patients refractory to primary treatments (Johnston et al 1991).

Nefazodone. In one open study 11/20 (55%) of patients that had failed multiple antidepressant achieved a CGI rating of much or very much improved after being treated with a mean nefazodone dose of 288 mg/d (median 300 mg/d) (Sajatovic et al 1999).

SSRIs. Overall response rates for switching from a TCA to a SSRI have ranged from 43% to 75% in outpatient studies (Thase and Rush 1995, Amsterdam and Hornig-Rohan 1996). One example was a fluoxetine study conducted by Amsterdam et al (1994) in 43 treatment-resistant patients who received fluoxetine 20 mg for five weeks. Patients were defined as responders if their HAMD scores dropped greater than 50% and the final score was <7. The response rate was 56%.

A more common practice today is switching among SSRIs. For inpatients converted from fluoxetine to sertraline in a retrospective review, only 26% were considered responders (Zarate et al 1996). In another open trial, 63% of outpatients switched from sertraline to fluoxetine responded (Thase et al 1997). In another open report of switching from one SSRI to another the overall success rate ranged from 40 to 67% (average 51%) (Joffe et al 1996). There were no consistent drug differences, though the numbers in each group were small.

Trazodone. Cole et al (1981) treated 25 patients that failed TCA with trazodone 100 to 600 mg/d in an open 2 to 4 week trial. Overall, HRSD scores decreased from 24 to 15. At least 10 patients of the 22 that tolerated trazodone demonstrated a >50% reduction in their score within 4 weeks. An additional 4 patients responded at 2 to 3 months for a total improvement rate of 63%. However, 7/14 patients that initially responded relapsed within a 6 to 24 month follow up period.

Venlafaxine. A mixed sample of 84 depressed inpatient and outpatients participated in a venlafaxine study (Nierenberg et al 1994). Patients were defined as refractory if their HRSD scores were > 20, failed to respond to 3 or more trials from two of three antidepressant classes (one must be a TCA, ECT may be substituted for one antidepressant trial), and also failed an augmentation trial with T3, lithium or carbamazepine. A partial response was defined as a 50% decrease in the HRSD and MADRS and a 2 on the CGI. A full response was defined as < 8 on the HRSD, < 12 on MADRS and a 1 on CGI. The mean dose of venlafaxine was 245±99.3 mg/day. There were 14 dropouts due to lack of efficacy (n=4), adverse drug reactions (n=1), lack of efficacy and adverse drug reactions (n=3), suicide (n=1), other (n=5). At three months, 33% met the criteria for a partial or full response. At six months 15% still met the response criteria and at nine months eight percent of the patients were still meeting the response criteria. A more recent study in 53 elderly patients who had failed paroxetine plus psychotherapy compared the efficacy of switching to venlafaxine XR versus augmentation of paroxetine with either: lithium, buproprion SR, or nortriptyline (Whyte 2004). While the rates of response (Hamilton-17 <10 for 3 weeks) were similar (42-45%), the authors noted much better tolerance with venlafaxine XR compared to each of the augmentation regimens.

Carbamazepine.

Post et al (1986) reported in a double blind, placebo-controlled trial that in 35 bipolar depressed patients most had some improvement with carbamazepine. Interestingly, in the on-off-on design most patients failed to relapse on placebo.

Conclusion

Most reports of substitution of one antidepressant for another are not blinded or controlled. Therefore, the true success of this strategy is unknown. Generally, 50 to 60% of patients will respond when switching antidepressant classes. The success when switching between SSRIs averages 45% in the small number of uncontrolled trials. This has led at least one author to recommend switching to another antidepressant class, not another SSRI if a patient fails an adequate trial of one SSRI (Shelton 1999).

AUGMENTATION

Lithium Augmentation

There have been 27 prospective clinical studies of lithium augmentation published since 1981 encompassing a total of 803 patients. Of these studies, 10 were double blind, placebo controlled trials, 4 were randomized comparator trials, and 13 were open comparison trials, in addition there have been over 22 case reports (Schopf 1989, Rouillon and Gorwood 1998, Bauer 2003). Selected open comparisons and controlled trials are detailed below. The results of the controlled trials appear in Table 2. See Nierenberg and White (1990), Rouillon and Gorwood 1998, and Bauer 2003 for a complete review. A suggested mechanism of action indicates that lithium produces a rapid increase in presynaptic synthesis and/or release of serotonin. This may augment the antidepressant effect.

TCAs

Open trials

de Montigny et al (1983) in a non-blinded study reported on 39 patients with unipolar depression who did not respond to three weeks of TCA therapy. In 74% of the observations, patients demonstrated >50% improvement within 48 hours of adding lithium at a dose of 900 mg/d. For the patients who required the presence of both lithium and the TCA to respond, the subsequent removal of lithium usually resulted in a relapse.

Nine cases of lithium addition to an antidepressant drug regimen in nonresponding depressed patients was reported by Louie and Meltzer (1984). The response to lithium addition was more variable than the previously described reports. Two patients showed sustained improvement, two showed transient improvement but then relapsed, two other cases with bipolar histories became manic, and three did not respond at all. Kushnir (1986) reported that the combination of low-dose lithium and antidepressants produced rapid, lasting remission of depression in five physically ill geriatric patients.

In a large case series, Price et al (1986) reported that 47/84 (56%) of patients improved with lithium 900 to 1500 mg/d (0.5 to 1.3 mEq/L) when it was added for at least 10 days (avg 3 weeks). HRSD scores decreased from 31 to 23. It is important to note that a HRSD score of 23 means that the patient still has significant pathology. In a follow up report to that study, Nierenberg et al (1990) reported that after an average treatment time of 29 months in 66 patients, 32 (49%) were considered to have a good outcome, 15 (23%) a fair response, and 29% a poor outcome.

Twenty-one elderly patients (age range, 64 to 88 years) with unipolar, nonpsychotic major depression who failed to respond to a six week course of nortriptyline (mean dose=77±18 mg/day), fluoxetine (mean dose=35±9 mg/day), or phenelzine (53±9 mg/day) were given an open label trial of lithium augmentation for two weeks (0.5-1 mEq/L) (Flint et al 1994). Response was defined as a HRSD score of <8. There was an over response rate of 24%. Six of 15 nortriptyline and all four fluoxetine patients experienced a neurologic/neuromuscular adverse effect.

Controlled trials

Van Putten and Sanders (1975) reported improvement in 35 refractory patients augmented with lithium. Placebo substitution was performed in 5 patients and all relapsed.

A double-blind study by Kantor et al (1986) could not demonstrate a lithium augmentation effect. Seven patients who did not respond to at least 21 days of antidepressant treatment were randomly assigned to lithium or placebo added to their standard treatment. Four patients received lithium and three received placebo. The combination was continued for 48 hours, after which six of the seven patients showed no significant improvement. The remaining patient, who had received lithium, improved markedly over the 48 hour period. However, the patient relapsed within one week. They concluded that no evidence exists for a consistent therapeutic effect of lithium adjunct therapy in antidepressant-resistant depression. However, the chance for a type II statistical error in this study was very high because of the small numbers of patients treated and the short (48 hour) time frame.

Heninger et al (1983) noted similar findings in a double-blind, placebo-controlled study of 15 depressed patients (14 UAD, 1 BAD). All patients met DSM III criteria and were refractory to at least 21 days of TCA therapy. Lithium doses of 900-1200 mg/d (0.5-1.1 mEq/L) produced "significant and clinically meaningful improvement" by the 7-12th day of lithium augmentation. There was no correlation between serum lithium levels and treatment response.

Stein and Bernadt (1987) reported that 61% of 26 patient treated with lithium augmentation responded compared to 2/10 (20%) patients receiving placebo.

Zusky et al (1988) added lithium or placebo for 2 weeks to 16 patients unsuccessfully treated for 9 months with a TCA. Half of the patients treated with lithium demonstrated partial improvement, but there was no overall difference between the 2 groups. The likelihood of a type II error in this study was also very great.

A randomized, double blind, placebo-controlled study of 2 weeks duration was conducted by Joffe et al (1993) to directly compare lithium and liothyronine as augmentation strategies. Fifty outpatients, with unipolar, nonpsychotic major depression who failed to respond to desipramine (n=45) or imipramine (n=5), 2.5 mg/kg/day x 5 weeks and HRSD scores >16 were included. Patients received either a dose of 900 mg/day of lithium or liothyronine 37.5 mcg/day or placebo in addition to the antidepressant. The lithium plasma concentration was adjusted after one week to >0.55 mEq/L. Response criteria was defined as a 50% or more reduction in the HAMD and a final score less than 10. Ten of 17 (59%) subjects responded to liothyronine, 9/17(53%) responded to lithium and three of 16 (19%) responded to placebo.

SSRIs

Only fluoxetine and citalopram have been studied in placebo-controlled trials (Katona et al 1995, Baumann et al 1996) (see table 2 for results).

In the Fava et al (1994) study presented earlier, of the partial responders only 13% of the fluoxetine plus lithium patients responded. However, among the non-responders 50% responded to fluoxetine plus lithium.

Meta-analysis

Bauer and Dopfmer performed a meta-analysis of 9 placebo-controlled double-blind studies that tested the efficacy of lithium augmentation of antidepressants in partial- and non-responding patients with major depression. Importantly, they aggregated the data for the 3 best studies whose inclusion criteria required a minimum dose of lithium carbonate 800 mg/d or a lithium level of greater than or equal to 0.5 mEq/L. The pooled odds ratio of response during lithium augmentation versus placebo was 3.3 (95% CI 1.5-7.5). The corresponding relative response rate was 2.1 (95% CI 10-44%), and the number of patients needed to be treated to obtain one more responder was 3.7. Inclusion of the six other studies with less aggressive exposure to lithium actually improved the statistical estimates. The authors concluded that lithium augmentation is the first choice treatment procedure for depressed patients who fail to respond to antidepressant monotherapy.

TCA and SSRI Literature Summary

• Overall response rate in 22 open reports was 168/237 (71%).
• Of the 168 patients in uncontrolled trials that responded, 61 (36%) responded within 48 hours. The reported response time may take as long as 21 days (Price et al 1986).
• Response was not correlated with lithium levels; however an attempt to keep lithium levels >0.3 mEq/L should be made.
• Typical doses used were 600 to 1200 mg/d.
• Antidepressant levels are not affected by lithium.
• There is no relationship between response and subtype of depression (e.g., bipolar vs unipolar).
• Duration of treatment of the combination in responders is not well defined. In some patients lithium was discontinued after several weeks without relapse of symptoms, while in other cases patients relapsed. The combination might be continued for several months in responders before a trial discontinuation is considered.

MAOIs

Nelson and Byck (1982) reported that all three patients who had not responded to phenelzine 75-90 mg/d had a significant improvement within two to four days after the addition of lithium 600 to 900 mg/d.

Carbamazepine

Kramlinger and Post (1989) added lithium to 15 depressed patients (13 bipolar) who had not responded to carbamazepine treatment alone. With serum lithium levels of 0.7 to 1.2 mEq/L, 8 (53%) of patients responded.

Antidepressant/Antipsychotic

Price et al (1983) reported that 5/6 DSM III diagnosed patients with MDD or bipolar affective disorder and mood-congruent psychotic features that had failed treatment with a TCA/typical antipsychotic combination responded to lithium addition. Three of the five responders had a moderate to marked improvement within 7 days. The remaining two responded after three weeks.

Nelson and Mazure (1986) retrospectively studied 21 psychotically depressed patients who had failed treatment with a combined antidepressant- antipsychotic drug regimen who then received lithium augmentation treatment. The result was compared to the response rate of 15 patients treated with ECT. Lithium augmentation was effective in 8/9 bipolar patients but in only 3/12 patients with unipolar depression. ECT was effective in 9/15 with unipolar depression. They concluded that lithium augmentation was the treatment of choice in non-responding psychotic depression inpatients with bipolar illness, but ECT was indicated in non-responding unipolar patients.

Lithium Conclusion

Approximately 50% of treatment refractory depressive patients will respond when lithium is added to their antidepressant therapy. It remains unclear whether this response is due to a synergistic effect, or due to the antidepressant effect of lithium itself. The level of evidence for the efficacy of lithium augmentation is high compared to other alternatives, therefore lithium should be considered a first line treatment choice in patients with depression that does not adequately respond to standard antidepressant therapy (Bauer et al, 2003).

Thyroid Augmentation

The correlation between thyroid function and depressive disorders has long been observed. Numerous studies have suggested a large spectrum of thyroid function abnormalities, although these abnormalities do not meet criteria for clinical thyroid disease. Examples include: increased thyroxine (T4) levels, decreased circulating triiodothyronine (T3) levels, or a blunted thyrotropin (TSH) response to thyrotropin releasing hormone (TRH). (Iosifescu et al 2001) It should be note that thyroid hormone used alone has no significant antidepressant effect in euthyroid depressed patients. Augmentation with thyroid hormone involves the addition of triiodothyronine (T3) 5 to 50 mcg/d or levothyroxine (T4) 0.1 mg/d.

There are over 20 published reports on thyroid augmentation of antidepressant drugs (Extein and Gold 1988, Nierenberg and White 1990, Aronson et al 1996). Although very few used a HRSD score <10 to define response and even fewer were double-blinded.

Tricyclic Antidepressants

Two early controlled studies (Gitlin et al, 1987, Joffe et al 1993) examining thyroid supplementation with TCAs found conflicting results. The first study, involved 16 MDD patients who were unresponsive to 4 weeks of imipramine (mean dose 206 ± 54 mg day). Patients received either triiodothyronine (T3, Cytomel®) 25 mcg or placebo for 2 weeks, followed by the other treatment (cross-over design). The authors found no evidence of a beneficial effect for triiodothyronine on either HRSD or global improvement scales. In the second study, fifty outpatients (males and females), with MDD who had failed desipramine (mean dose=201mg/day) or imipramine (mean dose =200mg/day) were randomized to receive triiodothyronine (mean dose=37.5 mcg/day), lithium (mean, dose=935.5mg/day), or placebo. Both triiodothyronine and lithium were more effective than placebo in reducing scores (HRSD) with no difference in augmentation effect between the two. Further, when response was defined as a 50% or more reduction in HRSD scores and a final score < 10, ten of 17 (59%) responded to triiodothyronine, nine of 17 (53%) responded to lithium, and three of 16 (19%) responded to placebo, suggesting both triiodothyronine and lithium may be effective in MDD refractory to TCAs.

Further evidence that MDD patients receiving TCA may benefit from triiodothyronine comes from a recent meta-analysis (Altshuler et al 2001) of 6 studies (n=125, percentage of females in studies 53%-100%). Instead of focusing on the addition of triiodothyronine after “TCA treatment failure”, the authors examined time until improvement (HDRS <8) while on TCA (amitriptyline or imipramine) therapy compared to time until improvement while on TCA therapy with triiodothyronine (25mcg/day-50mcg/day). It was found that TCA patients reached a score of 8 or less in 17.4± 9.7 days, compared to 24.8± 6.2 days in patients receiving only TCA therapy (P<.01), further suggesting a synergistic effect of triiodothyronine and TCAs.

See tables 3 and 4 for additional results of TCAs with thyroid augmentation (Extein and Gold 1988, Nierenberg and White 1990, Aronson et al 1996).

Following are guidelines derived from the above reports:

• Onset of response typically was 3 to 14 days. Some patients improve within 1 to 3 days.
• Both males and females responded equally well.
• If the patient does not respond within 21 days, thyroid hormone should be discontinued.
• No guidelines exist on duration of combined treatment if a patient responds. The drug might be discontinued in several months to determine a continuing need.
• The patient should be monitored for signs/symptoms of hyperthyroidism.
• Thyroid supplementation did not increase TCA side effects or serum levels.
• Three of 4 studies that reported that the presence of subclinical hypothyroidism, determined by administering TRH, did not predict a response to thyroid supplementation.

More controlled trials with a larger number of patients needs to be performed.

SSRIs

SSRIs have become the current first line choice for MDD. Case reports over the last several years have suggested that thyroid augmentation may be useful with SSRIs, however to date, only one controlled study has been performed. (Appelhof et al 2004). In the 8 week, randomized double blinded study, 113 patients with MDD (HRSD >16) were randomized to receive triiodothyronine 25mcg/day, 50mcg/day, or placebo, along with paroxetine (titrated up to 30mg/day from 10mg/day over two weeks). Results showed no significant advantage in triiodothyronine treated patients (25mcg/day, 50mcg/day, 25mcg&50mcg combined vs. placebo) among any of the observed measures (HRSD 50% response, HRSD remission (<10), MADRS, or BDI). However it should be noted that patients were not screened for refractory MDD with paroxetine, or any other antidepressant agent, prior to the study. Thus further clinical trials involving truly “SSRI refractory” patients are needed to determine the usefulness of triiodothyronine in SSRI refractory patients.

Mirtazepine

Among 26 patients (SSRI n=23, venlafaxine n=3, bupropion n=1) with major depression (DSM-IV) who had not responded to adequate antidepressant monotherapy, Carpenter et al (2002) augmented the patients with either mirtazepine 15-30 mg/d (n=11) or placebo (n=15). The observed partial response (50% decrease in HAMD and much or very much improved on CGI) was 64% for the active drug, and 20% for placebo group, (p=0.046). However, the remission rates of 45% and 13% for the active drug and placebo groups respectively, did not demonstrate a difference from placebo in the combination therapy.

Monoamine Oxidase Inhibitor/Antipsychotic

A woman with psychotic unipolar depression responded with the addition of T3 to phenelzine in combination with thiothixene (Hullett and Bidder 1983). She responded within 2 weeks after having been hospitalized or institutionalized for 3 years. She was a DST suppresser but responded subnormally to TRH.

Buspirone

The efficacy of buspirone augmentation was studied in twenty-five patients with MDD who had failed several prior treatments (each for five weeks), including desipramine 2.5 mg/kg/d, phenelzine 60-90 mg/d, tranylcypromine 50 to 80 mg/d, and lithium augmentation >0.6 mEq/L (Joffe and Schuller 1993). These patients also failed fluvoxamine (mean 150 mg/day) or fluoxetine (42.5 mg/day) trials after five weeks. Buspirone 20-50 mg/day for three weeks was added. Response criteria were defined as a CGI of 2 or 3. Seventeen of 25 (68%) patients responded.

In a more recent study, the response rate for buspirone and placebo after two weeks of augmentation was 51% and 47% (p = ns), respectively (Landen et al 1998). The author's were unable to explain the lack of response, though concurrent psychological counseling may have had an influence.

Reserpine

Haskovec and Rysanek (1967) in an open study reported that 14/15 depressed patients that failed 10 weeks of imipramine with the last week's dose of 300 mg/d responded to reserpine 7.5 to 10 mg/d for 2 days. Eight cases relapsed within 3 weeks, while 6 did not.

In a double-blind, placebo-controlled trial of 14 patients treated with imipramine >150 mg/d for at least 3 weeks, 8/8 patients that had reserpine 5 mg i.m. daily added for 2 days responded (Hopkinson and Kenny 1975). This compared to a response of 1/6 patients treated with saline injections. The improvement in the reserpine-treated cases continued for at least 3 months. Two additional controlled trials (Amsterdam and Berwish 1987; Price et al 1987) which treated a total of 17 patients reported that only 2 patients had a response to reserpine 6 to 10 mg i.m. daily. These patients had failed 3 to 4 weeks of TCA treatment.

Conclusion

Augmentation with lithium or thyroid might be considered after a patient has failed an adequate trial of a second antidepressant (Shelton 1999).

COMBINATION THERAPY

There are few controlled trials directly assessing the efficacy of combining various classes of antidepressants in the management of treatment-resistant depression.

SSRIs + Pindolol

The theory behind the use of the 5-HT1a antagonist pindolol in SSRI resistant therapy is related to the observation that resistance to SSRIs is related to insufficient desensitization of 5-HT1a autoreceptors. This desensitization allows for continued SSRI-mediated feedback inhibition of pre-synaptic serotonin release, and normally occurs after 2 or more weeks of antidepressant exposure. This two-week time frame correlates with the delay in antidepressant response. Although in theory the blockade of pre-synaptic 5-HT1a receptors with pindolol should circumvent feedback inhibition, attempts at enhancing antidepressant responses in SSRI refractory patients with its use have failed to produce consistent results.

Recent randomized control trials have looked at the use of pindolol in SSRI refractory patients and have obtained different results. To illustrate, a recent study of 9 outpatients (8 males) who had been on paroxetine 40mg in the morning for at least two months, had failed at least two other antidepressants, and had baseline HDRS scores >21, were randomly assigned to receive pindolol 7.5mg/daily (n=4) in the morning or placebo (n=4) for four weeks. Outcomes measured included the Hamilton Depression Scale (HAM-D), the Hamilton Anxiety Scale (HAM-A), the Bech-Rafaelsen melancholia Scale, and the Zung Depression Inventory. Subjects receiving pindolol had significant improvement (P<.05) in all four measures starting at week two which continued through week four. (Sokolski et al 2003).

Conversely, a second study involving 42 outpatients who had failed to demonstrate an adequate response to fluoxetine, paroxetine, or sertraline showed no benefit with pindolol. To be enrolled in the study, patients were require to have received at least moderate doses (fluoxetine 20mg/day, paroxetine 40mg/day, or sertraline 50mg/day) for at least six weeks, and have baseline HAM-D scores above 25. Patients were maintained on their SSRI therapy and were randomly assigned to receive either pindolol 2.5mg three times daily or placebo for three weeks. At the end of three weeks, the placebo patients were switched to pindolol, while the patients previously receiving pindolol were maintained on pindolol for an additional three weeks. Outcome measures included the HAM-D and the Beck Depression Inventory (BDI). At three weeks, no differences were observed between the groups among either the HAM-D or the BDI. Further, no significant differences were observed at 6 weeks. (Perry et al 2004).

To better explore the heterogeneous results from the published studies, Ballesteros et al conducted a meta-analysis of nine published randomized control studies of pindolol with SSRIs. The authors incorporated results for 594 patients (n=298 pindolol, n=296 placebo). SSRI included in the studies included fluoxetine (n=4), paroxetine (n=3), fluvoxamine (n=1), several SSRIs (n=1). Individually, four studies supported (P<.05) use of pindolol with SSRIs, while five did not support its use. Overall, use of pindolol was favored at two weeks (OR=2.8, 95% CI=1.4-5,7; P<.01), but not at weeks four to six. The authors concluded that the efficacy of pindolol may be limited to the first two weeks of treatment. (Ballesteros and Callado 2004). Finally, it was hypothesized that there may be a genetic component accounting for differences in patient response to pindolol. (Segrave and Nathan 2005).

TCA + SSRIs

Rosenthal et al (1991) reported in an open study that 10 refractory patients responded within 24 hours to 2 weeks when fluoxetine was added to heterocyclic antidepressants.

In the controlled trial of Fava et al (1994) (study design presented earlier) among the partial responders and non-responders, fluoxetine + desipramine produced a response in only 40% (2/5) and 14% (1/7) of patients, respectively.

Thirty consecutive outpatients with recurrent major depression experiencing a current episode (DSM-III-R criteria) were treated with the addition of fluoxetine to their antidepressant regimen (Weilburg et al 1989). Fifteen patients carried an Axis II diagnosis. Seventeen patients were receiving a single antidepressant, while 13 were receiving a combination of agents (e.g., antidepressant plus lithium, a benzodiazepine, or carbamazepine). The mean maximum antidepressant dose expressed in imipramine equivalents was 213 mg/d (range 37.5-400 mg/d). The average duration of treatment was 11.7 months (range 1-36 months). The duration of the current episode was 1 year (n=20), 3 to 12 months (n=5), 1 to 3 months (n=5).

The patients were rated retrospectively as responders (clear improvement) or non-responders (depressive symptoms continued). Twenty-six of 30 (83%) patients were considered responders. The time to response was 1 to 3 days (n=4), 4 to 7 days (n=6), 1 to 3 weeks (n=12), and >4 weeks (n=4). Therefore, approximately 2/3 of responders responded within 3 weeks of adding fluoxetine. Twelve responders had their original antidepressant medication discontinued after the improvement on fluoxetine. Eight (66%) patients relapsed on fluoxetine alone, but all 8 recovered with restarting the initial treatment. In all cases the antidepressant doses were lowered after the addition of fluoxetine because of a drug-drug interaction. No serious adverse effects were reported. A systematic measurement of antidepressant serum levels was not performed before and after the addition of fluoxetine; therefore, it is possible the improvement may have been due to an elevation of antidepressant serum levels.

A recent report of 13 patients with DSM-III-R nonpsychotic major depression studied patients unsuccessfully treated with an individual trial of either imipramine or desipramine and fluoxetine (Levitt et al 1999). The patients then received the combination of fluoxetine and imipramine or desipramine for 3 weeks. Based on HRSD and CGI scores, the patients that had the best response (>40% reduction in the HRSD) had significantly higher TCA levels then the non-responders. These findings suggest that the improvement with the combination of fluoxetine and imipramine or desipramine may be related to the plasma levels of the TCA.

TCA + MAOIs

A controlled investigation of combined treatment with a TCA plus a MAOI in nonresponding patients is not available (Stern and Mendels 1981, Murphy et al 1984).

In a follow-up study, twelve patients treated with a combination of amitriptyline and isocarboxazide were evaluated over a 3-year period (Berlanga and Ortega-Soto 1995). Every 6 months an attempt was made to discontinue the MAOI and continue only amitriptyline. At the end of the study, 4 patients maintained their response with amitriptyline, 6 required both amitriptyline and isocarboxazide, and 2 relapsed.

Currently, there are no criteria indicating which patients should receive combined TCA-MAOI treatment. If the combination is going to be used, it is recommended that the MAOI and TCA are started simultaneously or the MAOI a few days after the TCA, and that a lower dose be used then when the drugs are used individually. It is suggested that amitriptyline and phenelzine may be the safest combination.

TCAs + Antipsychotics

Antipsychotic.A review of the literature published prior to 1979 reveals a combined response rate of 22% for TCA alone and 83% for the combination (Nelson et al 1986). Three studies treated these patients with antipsychotics alone and reported a 96% response. Conversely, a 1980 study (Kaskey et al) reported that only 2/6 (33%) patients responded to an antipsychotic alone. In the same study, 7/7 (100%) patients treated with the combination responded, while only 3/10 (33%) patients treated with an antidepressant alone improved. However, as antipsychotics alone are not generally considered appropriate single treatment for depression, the question involves antidepressants alone vs. combination treatment.

In a prospective, double-blind investigation Spiker et al (1985) reported that amitriptyline was effective in only 41% of a group of delusional depressives, whereas the amitriptyline/perphenazine combination was effective in 78%. However, after reanalyzing the data controlling for amitriptyline serum concentration, 64% of patients treated with amitriptyline alone responded if their combined serum concentration was >250 ng/ml. This response rate was not significantly different from the response rate of combination treatment.

Additionally, two studies involving 56 delusional depressives and 58 non-delusional depressives concluded that response was not significantly different between single antidepressant treatment and the antidepressant-antipsychotic combination (Lykouras et al 1986, Nelson et al 1986). One study, in fact, reported that delusional depressives were more likely to respond to an antidepressant alone than non-delusional depressive patients, although they required twice as long to respond (i.e., > 3 weeks).

Finally, Barbee et al (2004) found in a retrospective chart review of 30 unipolar depressed patients who had failed multiple antidepressants and also failed the augmentation with at least one atypical antipsychotic, that aripiprazole may be effective. Specifically, 14/30 (47%) of the patients were rated much or very much improved.

Howarth and Grace (1985) reported that only 32% of delusional depressed patients responded to an antidepressant alone after 3 weeks, but 62% responded after 9 weeks of high-dose TCA treatment.

TCA + Bupropion

Lam et al (2004) treated 61 patients with depression refractory to at least one trial with citalopram, bupropion-SR, or citalopram with bupropion-SR. Patients were treated by adding citalopram or bupropion-SR, or by switching to the other anti-depressant using an open label design. Patients were assessed at baseline and at 6 weeks of treatment with the 29-item Structured Interview Guide for the Hamilton Depression Rating Scale, Seasonal Affective Disorders Version (SIGH-SAD). The combination therapy (n=32) was superior to monotherapy switch, SAD change score (-14.8 vs. –10.1, respectively, P<.05) and the proportion of patients in clinical remission (28% vs. 7%, P<.05). There were no significant differences in the proportion of patients who had side effects, or in the severity of side effects.

TCA + CNS Stimulant

Wharton et al (1971) treated seven patients with delusional unipolar depression who were refractory to TCA (imipramine 150 to 225 mg/day n=6, nortriptyline 75 mg/day n=1) with methylphenidate.. Five of the seven showed marked improvement with the combination. In three of the patients, TCA blood levels were measured and were found to increase significantly after methylphenidate was added. This interaction was believed to be due to inhibition of liver hydroxylating enzymes. However, the exact mechanism of the combination effectiveness is unknown.

TCA + MAOI + CNS Stimulants

In an uncontrolled case series, Feighner et al (1985) reported the successful treatment of "treatment resistant" depression with the addition of a CNS stimulant (i.e., d-amphetamine, methylphenidate) to a MAOI or to a combination of TCA and MAOI. The literature on combinations of CNS stimulants and MAOIs indicates that hypertensive and hyperthermic crises may occur when high doses are given. The author's state, "Our clinical experience is that these hypertensive crises are, at most, infrequent, and that this combination is often effective." It would appear that with many pharmacologic and non-pharmacologic treatment options available, this combination would be infrequently utilized and demand careful considering before intitating.

Methylphenidate in doses ranging from 10 to 40 mg/day was used to augment SSRI treatment in a case series of five patients. Self-reported symptom reduction was achieved rapidly in all cases (Stoll et al 1996).

Primary Mood Stabilizers + MAOIs

There are no controlled trials of combining lithium or anticonvulsants plus antidepressants in the treatment of refractory patients (Dietrich and Emrich 1998). Open reports include adding anticonvulsants to TCAs, MAOIs, SSRIs, trazodone, and bupropion. In total these reports only involve 57 patients.

Ten inpatients (7 bipolar and 3 unipolar) of which eight had also failed carbamazepine, lithium and combination, received in an open design phenelzine (mean dose 68 mg/day) or tranylcypromine (mean dose 55 mg/day) in addition to carbamazepine, lithium or the combination (Ketter et al 1995). Four of the ten patients improved substantially and became euthymic, allowing discharge from the hospital on the carbamazepine ± lithium plus MAOI combination. All ten patients tolerated the combination.

L-tryptophan

Though widely studied in the treatment of depression alone and in combination with standard antidepressants, little information is available in the management of refractory patients (Nierenberg and White 1990). L-tryptophan was pulled from the market when patients died from eosinophilia myalgia syndrome (Milburn et al 1991).

Conclusion

It is recommended that some combinations not be used at all (e.g., SSRI + CNS stimulants or SSRIs + MAOIs) or only if the patient has failed multiple single drug trials, including the two drugs to be used together.

CONCLUSION

A study of 431 patients with depression reported that 93% recovered over a 10-year period, with 67% recovering with 1 year (Mueller et al 1996). Of those that remained depressed for 5 years continuously, 38% recovered between years 5 and 10. This information suggests that the outcome of treatment of depression is generally positive; however, many patients may require long-term intervention.

It is important to develop a systematic approach to the treatment-resistant depressed patient. The following are general guidelines:

• Consider contributing medical conditions that may contribute to or cause the depressed state.
• Use treatment appropriate to the subtype of depression (e.g., nortriptyline in melancholia).
• Ensure adequate antidepressant dosage (or if available serum levels) and duration of treatment (at least equal to 4 weeks and 6 preferred).
• If a patient had a partial response to an antidepressant and doesn't have significant side effects than increase the dose.
• If a patient had no response to an antidepressant than the drug should be discontinued and an antidepressant from another class tried.
• If nonresponsive at this point, then a trial with a different class of antidepressant might be attempted or augmentation with lithium or thyroid might be attempted.
• Electroconvulsive treatment can be considered after any of these steps.
• Don’t forget that many common medications may cause depression, e.g., beta-blockers, oral

contraceptives, narcotics, and calcium channel blockers.

*p < 0.05 in favor of lithium

**p > NS

#crossover design

REFERENCES

Altshuler LL, Bauer M, Frye MA, et al (2001). Does thyroid supplementation accelerate tricyclic antidepressant response? A review and meta-analysis of the literature. Am J Psychiatry 158:1617-22.

Amsterdam JD, Berwish NJ (1987). Treatment of refractory depression with combination reserpine and tricyclic antidepressant therapy. J Clin Psychopharmacol 7:238-42.

Amsterdam JD, Berwish NJ (1989). High dose tranylcypromine therapy for resistant depression. Pharmacopsychiatry 22:21-5.

Amsterdam JD, Maislin G, Potter L (1994). Fluoxetine efficacy in treatment resistant depression. Prog Neuro-Psychopharmacol Biol Psychiatry 18:243-61.

Amsterdam JD, Hornig-Rohan M (1996). Treatment algorithms in treatment-resistant depression. Psychiatric Clin N America 19:371-86.

Appelhof BC, Brouwer JP, Van Dyck R, et al (2004). Triiodothyronine addition to paroxetine in the treatment of major depressive disorder. J Clin Endocrinol Metab 89:6261-76.

Aronson R, Offman HJ, Joffe RT, et al (1996). Triiodothyronine augmentation in the treatment of refractory depression. Arch Gen Psychiatry 53:842-8.

Ayd FJ (1972). Patient compliance. Int Drug Ther Newsletter 7:33-40.

Ayd FJ (1983). Treatment-resistant depression. Int Drug Ther Newsletter 18:25-7.

Ballesteros J, Callado LF (2004). Effectiveness of pindolol plus serotonin uptake inhibitors in depression: a meta-analysis of early and late outcomes from randomised controlled trials. J Affective Disord 79:137-47.

Barbee JG, Conrad, EJ, Jamhour, NJ (2004). Aripiprazole augmentation in treatment-resistant depression. Ann Clin Psychiatry 16:189-94.

Bauer M, Dopfmer S (1999). Lithium augmentation in treatment-resistant depression: Meta-analysis of placebo-controlled studies. J Clin Psychopharmacol 19:427-34.

Bauer M, Mazda A, Baethge, C, et al (2003). Lithium augmentation therapy in refractory depression: clinical evidence and neurobiolgical mechanisms. Can J Psychiatry 48:440-48.

Baumann P, Nil R, Souche A, Montaldi S, et al (1996). A double-blind, placebo-controlled study of citalopram with and without lithium in the treatment of therapy-resistant depressive patients: a clinical, pharmacokinetic, and pharmacogenetic investigation. J Clin Psychopharmacol 16:307-14.

Berlanga C, Ortega-Soto HA (1995). A 3-year followup of a group of treatment-resistant depressed patients with a MAOI/tricyclic combination. J Affective Disord 34:187-92.

Black DW, Bell S, Hulbert J, et al (1988). The importance of Axis II in patients with major depression. A controlled study. J Affective Disord 14:115-22.

Carpenter LL, Ysamin S, Price LH (2002). A double-blind, placebo-controlled study of antidepressant augmentation with mirtazepine. Biol Psychiatry 51:183-8.

Cole JO, Schatzberg AF, Sniffin C, Zolner J, Cole JP (1981). Trazodone in treatment-resistant depression: an open study. J Clin Psychopharmacol 1(Suppl):49-54.

Coryell W (2000). Augmentation strategies for inadequate antidepressant response: a review of placebo-controlled studies. Ann Clin Psychiatry 12:141-46.

Cournoyer G, de Montigny C, Ouellette J, et al (1984). Litium addition in tricyclic resistant unipolar depression: a placebo controlled study. Collegium International of Neuro-Psychopharmacology 14:F17.

de Montigny C, Cournoyer G, Morissette R, et al (1983). Lithium carbonate addition in tricyclic antidepressant-resistant unipolar depression. Arch Gen Psychiatry 40:1327-34.

Dietrich DE, Emrich HM (1998). The use of anticonvulsants to augment antidepressant medication. J Clin Psychiatry 59(Suppl 5):51-8.

Dornseif BE, Dunlop SR, Potvin JH, et al (1989). Effect of dose escalation after low-dose fluoxetine therapy. Psychopharmacol Bull 25:71-9.

Endicott J, Cohen J, Nee J, Fleiss J, et al (1981). Hamilton depression rating scale. Arch Gen Psychiatry 38:98-103.

Extein IL, Gold MS (1988). Thyroid potentiation of tricyclics. Psychosomatics 29:166-74.

Fava M, Rosenbaum JF, McGrath PJ, et al (1994). Lithium and tricyclic augmentation of fluoxetine treatment for resistant major depression: A double-blind, controlled study. Am J Psychiatry 151:1372-4.

Fava M (2001). New approaches to the treatment of refractory depression. J Clin Psychiatry 61(Suppl 1):26-32.

Fava M, Alpert J, Nierenberg A, et al (2002). Double-blind study of high-dose fluoxetine versus lithium or desipramine augmentation of fluoxetine in partial responders and non-responders to fluoxetine. J Clin Psychopharmacol 22:379-87.

Flint AJ, Rifat SL (1994). A prospective study of lithium augmentation in antidepressant-resistant geriatric depression. J Clin Psychopharmacol 14:353-6.

Feighner JP, Herbstein J, Damlouji N (1985). Combined MAOI, TCA, and direct stimulant therapy of treatment-resistant depression. J Clin Psychiatry 46:206-9.

Flint AJ, Rifat SL (1994). A prospective study of lithium augmentation in antidepressant-resistant geriatric depression. J Clin Psychopharmacol 14:353-6.

Georgotas A, McCue RE, Cooper TB, et al (1989). Factors affecting the delay of antidepressant effect in responders to nortriptyline and phenelzine. Psychiatry Res 28:1-9.

Gitlin MJ, Weiner H, Fairbanks L, et al (1987). Failure of T3 to potentiate tricyclic antidepressant response. J Affect Disord 13:267-72.

Glassman AH, Roose SP (1990). Tricyclic treatment: what is adequate and who is refractory? In: Tasman A, Goldfinger SM, Kaufamn CA, eds. Am Psychiatric Press Rev Psychiatry 9:60-73.

Glassman AH, Perel JM, Shostak M, et al (1977). Clinical implications of imipramine plasma levels for depressive illness. Arch Gen Psychiatry 34:197-204.

Haskovec L, Rysanek K (1967). The action of reserpine in imipramine-resistant depressive patients. Psychopharmacologia (Berlin) 11:18-30.

Heninger GR, Charney DS, Sternberg DE (1983). Lithium carbonate augmentation of antidepressant treatment. Arch Gen Psychiatry 40:1335-42.

Hopkinson G, Kenny F (1975). Treatment with reserpine of patients resistant to tricyclic antidepressants. A double-blind trial. Psychiatria Clinica 8:109-14.

Howarth BG, Grace MGA (1985). Depression, drugs, and delusions. Arch Gen Psychiatry 42:1145-7.

Hullett FJ, Bidder TG (1983). Phenelzine plus triiodothyronine combination in a case of refractory depression. J Nerv Ment Dis 171:318-20.

Joffe RT, Schuller DR (1993). An open study of buspirone augmentation of serotonin reuptake inhibitors in refractory depression. J Clin Psychiatry 54:269-71.

Joffe RT, Singer W, Levitt AJ, et al (1993). A placebo-controlled comparison of lithium and triiodothyronine augmentation of tricyclic antidepressants in unipolar refractory depression. Archives of General Psychiatry 50:387-93.

Joffe RT, Levitt AJ, Sokolov STH, et al (1996). Response to an open trial of a second SSRI in major depression. J Clin Psychiatry 57:114-5.

Johnston JA, Lineberry CG, Ascher JA, et al (1991). A 102-center prospective study of seizures in association with bupropion. J Clin Psychiatry 52:450-6.

Kantor D, McNevin S, Leichner P, et al (1986). The benefit of lithium carbonate adjunct in refractory depression--fact or fiction? Can J Psychiatry 31:416-8.

Kaskey GB, Nasr S, Meltzer HY (1980). Drug treatment in delusional depression. Psychiatry Res 1:267-77.

Katona CL, Abou-saleh MT, Harrison DA, et al (1995). Placebo-controlled trial of lithium augmentation of fluoxetine and lofepramine. Br J Psychiatry 166:80-6.

Ketter TA, Post RM, Parekh PI, et al (1995). Addition of monoamine oxidase inhibitors to carbamazepine: Preliminary evidence of safety and antidepressant efficacy in treatment-resistant depression. J Clin Psychiatry 56:471-5.

Kramlinger KG, Post RM (1989). The addition of lithium to carbamazepine: antidepressant efficacy in treatment-resistant depression. Arch Gen Psychiatry 46:794-800.

Kushnir SL (1986). Lithium-antidepressant combinations in the treatment of depressed, physically ill geriatric patients. Am J Psychiatry 143:378-9.

Lam RW, Hossie H, Solomons K et al (2004). Citalopram and bupropion-SR: combining versus switching in patients with treatment-resistant depression. J Clin Psychiatry 65:337-40.

Landen M, Bjorling G, Agren H, et al (1998). A randomized, double-blind, placebo-controlled trial of buspirone in combination with an SSRI in patients with treatment-refractory depression. J Clin Psychiatry 59:664-8.

Levitt AJ, Joffe RT, Kamil R, et al (1999). Do depressed subjects who have failed both fluoxetine and a tricyclic antidepressant respond to the combination? J Clin Psychiatry 60:613-6.

Lin EHB, Von Korff J, Katon W, et al (1995). The role of the primary care physician in patient's adherence to antidepressant therapy. Med Care 33:67-74.

Louie AK, Meltzer HY (1984). Lithium potentiation of antidepressant treatment. J Clin Psychopharmacol 4:316-21.

Lykouras E, Marriaras D, Christodoulou GN, et al (1986). Delusional depression: phenomenology and response to treatment: a prospective study. Acta Psychiatrica Scand 73:324-9.

Mann JJ (1983). Loss of antidepressant effect with long-term monoamine oxidase inhibitor treatment without loss of monoamine oxidase inhibition. J Clin Psychopharmacol 3:363-6.

McGrath PJ, Stewart JW, Harrison W, et al (1987). Treatment of tricyclic antidepressant depression with a monoamine oxidase inhibitor. Psychopharmacol Bull 23:169-72.

Milburn DS, Myers CW (1991). Tryptophan toxicity: a pharmacoepidemiologic review of eosinophilia-myalgia syndrome. DICP Ann Pharmacother 25:1259-62.

Morris JB, Beck AT (1974). The efficacy of antidepressant drugs. Arch Gen Psychiatry 30:667-74.

Medical Research Council (1965). Clinical trial of the treatment of depressive illness. Brit Med J 1:881-6.

Mueller TI, Keller MB, Leon AC, et al (1996). Recovery after 5 years of unremitting major depressive disorder. Arch Gen Psychiatry 53:794-99.

Murphy DL, Sunderland T, Cohen RM (1984). Monoamine oxidase-inhibiting antidepressants. Psychiatric Clin N America 7:549-62.

Nelson JC, Byck R (1982). Rapid response to lithium in phenelzine non- responders. Br J Psychiatry 141:85-6.

Nelson JC, Mazure CM (1986). Lithium augmentation in psychotic depression refractory to combined drug treatment. Am J Psychiatry 143:363-6.

Nelson JC, Price LH, Jatlow PI (1986). Neuroleptic dose and desipramine concentrations during combined treatment of unipolar delusional depression. Am J Psychiatry 143:1151-4.

Nelson JC, Mazure Cm, Jatlow PI (1994). Characteristics of desipramine-refractory depression. J Clin Psychiatry 55:12-19.

Nierenberg AA, Price LH, Charney DS, et al (1990). After lithium augmentation: a retrospective follow up of patients with antidepressant refractory depression. J Aff Disord 18:167-75.

Nierenberg AA, White K (1990). What next?: A review of pharmacologic strategies for treatment resistant depression. Psychopharmacol Bull 26:429-60.

Nierenberg AA, Feighner JP, Ruldolph R, et al (1994). Venlafaxine for treatment-resistant unipolar depression. J Clinical Psychopharmacol 14:419-23.

Nolen WA, Van dePutte JJ, Dijken WA, et al (1988). L-5HTP in depression resistant to reuptake inhibitors: an open comparative study with tranylcypromine. Br J Psychiatry 147:16-22.

Pare CMB (1985). The present status of monoamine oxidase inhibitors. Br J Psychiatry 146:576-84.

Perry PJ, Pfohl BM, Holstad SG (1987). The relationship between antidepressant response and tricyclic antidepressant plasma concentrations: a retrospective analysis of the literature using logistic regression analysis. Clin Pharmacokinetics 13:381-92.

Peselow ED, Filippi AM, Goodnick P, et al (1989). The sort- and long-term efficacy of paroxetine HCL, G: data from a double-blind crossover study and from a year-long trial vs imipramine and placebo. Psychopharmacol Bull 25:272-6.

Pfohl B, Stangel D, Zimmerman M (1984). The implications of DSM-III personality disorders for patients with major depression. J Affective Disord 7:309-318.

Post RM, Uhde TW, Roy-Byrne PP, Joffe RT (1986). Antidepressant effects of carbamazepine. Am J Psychiatry 143:29-34.

Price LH, Conwell Y, Nelson JC (1983). Lithium augmentation of combined neuroleptic-tricyclic treatment of delusional depression. Am J Psychiatry 140:318-22.

Price LH, Charney DS, Heninger GR (1986). Variability of response to lithium augmentation in refractory depression. Am J Psychiatry 143:1387-92.

Price LH, Charney DS, Heninger GR (1987). Reserpine augmentation of desipramine in refractory depression: clinical and neurobiological effects . Psychopharmacol (Berlin) 92:431-7.

Quitkin FM, McGrath P, Liebowitz MR, et al (1981). Monoamine oxidase inhibitors in bipolar endogenous depressives. J Clin Psychopharmacol 1:70-4.

Quitkin FM, Rabkin JG, Ross D, McGrath PJ (1984). Duration of antidepressant drug treatment. Arch Gen Psychiatry 41:238-45.

Quitkin FM, McGrath PJ, Stewart JW, et al (1996). Chronogical Milestones to Guide Drug Change. Arch Gen Psychiatry 53:785-92.

Rogers SC, Clay PM (1975). A statistical review of controlled trials of imipramine and placebo in the treatment of depressive illness. Br J Psychiatry 127:599-608.

Roose, SP, Glassman, AH, Walsh BT, (1986). Tricyclic non-responders: phenomenology and treatment. Am J Psychiatry 143:345-8.

Rosenthal JS, Kasawan MJ, Hemlock C, et al (1991). Fluoxetine enhancement of heterocyclic antidepressants. In Amsterdam JD (ed): Refractory depression, New York, Raven Press: 105-8.

Rouillon F, Gorwood P (1998). The use of lithium to augment antidepressant medication. J Clin Psychiatry 59(Suppl 5):32-9.

Russel JM, Hawkins K, Ozminkowski RJ et al (2004). The cost consequences of treatment-resistant depression. J Clin Psychiatry 65:341-7.

Sajatovic M, DiGiovanni S, Fuller M, et al (1999). Nefazodone therapy in-patients with treatment-resistant or treatment-intolerant depression and high psychiatric comorbidity. Clin Therapeutics 21:733-40.

Schopf J (1989). Treatment of depressions resistant to tricyclic antidepressants, related drugs or MAO-inhibitors by lithium addition: review of the literature. Pharmacopsychiatry 22:174-82.

Schwarcz G, Halaris A, Baxter L, et al (1984). Normal thyroid function in desipramine non-responders converted to responders by the addition of l-triiodothyronine. Am J Psychiatry 141:1614-6.

Schweizer E, Rickels K, Amsterdam JD, et al (1990). What constitutes an adequate antidepressant trial for fluoxetine? J Clin Psychiatry 51:8-11. Hum Psychopharmacol Clin Exp 20:163-74.

Segrave R, Nathan PJ (2005). Pindolol augmentation of selective serotonin reuptake inhibitor: Accounting for the variability of results of placebo-controlled double-blind studies in patients with major depression.

Sharma V, Mustaq, K, Smith A (2005). A closer look at treatment resistant depression: is it due to a bipolar diathesis? J Affect Disord 84:241-57.

Shelton RC (1999). Treatment options for refractory depression. J Clin Psychiatry 60(Suppl 4):57-61.

Shelton RC (2003). The use of antidepressant in novel combination therapies. J Clin Psychiatry 64(Suppl 2):14-18.

Simpson GM, Lees JH, Cuculic Z, et al (1976). Two dosages of imipramine in hospitalized endogenous and neurotic depressives. Arch Gen Psychiatry 33:1093-102.

Sokolski KN, Conney JC, Brown BJ, et al (2003). Once-daily high-dose pindolol for SSRI-refractory depression. Psychiatry Research 125:81-86.

Spiker DG, Weiss JC, Dealy RS, et al (1985). The pharmacological treatment of delusional depression. Am J Psychiatry 142:430-6.

Stein GS, Bernadt M (1987). A double-blind trial of very low lithium in tricyclic resistant depression. Paper presented at 2nd British Lithium Congress. Wolverhampton, England, September 6-9, 1987.

Stern SL, Rush AJ, Mendels J (1980). Toward a rational pharmacotherapy of depression. Am J Psychiatry 137:545-52.

Stern SL, Mendels J (1981). Drug combinations in the treatment of refractory depression: a review. J Clin Psychiatry 42:368-73.

Stern WC, Harto-Truaz N, Bauer N (1983). Efficacy of bupropion in tricyclic-resistant or intolerant patients. J Clin Psychiatry 44:148-52.

Stoll AL, Pillay SS, Diamond L, et al (1996). Methylphenidate augmentation of serotonin selective reuptake inhibitors: A case series. J Clin Psychiatry 57:72-6.

Thase ME, Kupfer DJ, Frank E, et al (1989). Treatment of imipramine-resistant recurrent depression: I. An open clinical trial of adjunctive L-triiodothyronine (letter). J Clin Psychiatry 50:10.

Thase ME, Rush AJ (1995). Treatment-resistant depression. In: Bloom FE, Kupfer DJ, eds. Psychopharmacology: The Fourth Generation of Progress. New York, NY: Raven Press: 1081-97.

Thase ME, Blomgren Sl, Birkett MA, et al (1997). Fluoxetine treatment of patients with major depressive disorder who failed initial treatment with sertraline. J Clin Psychiatry 58:16-21.

van Putten T, Sanders DG (1975). Lithium in treatment failures. J Nerv Ment Dis 161:255-64.

Wager SG, Klein DF (1988). Treatment refractory patients: affective disorders. Psychopharmacol Bull 24:69-74.

Watt DC, Crammer JL, Elkes A (1972). Metabolism, anticholinergic effects and therapeutic effects on outcome of desmethylimipramine in depressive illness. Psychological Med 2:397-405.

Weilburg JB, Rosenbaum JF, Biederman, et al (1989). Fluoxetine added to non-MAOI antidepressants converts non-responders to responders: a preliminary report. J Clin Psychiatry 50:447-9.

Wharton RN, Perel JM, Dayton PG, et al (1971). A potential clinical use for methylphenidate with tricyclic antidepressants. Am J Psychiatry 127:55-61.

Whyte EM, Basinski J, Farhi P, et al (2004). Geriatric depression treatment in nonresponders to selective serotonin reuptake inhibitors. J Clin Psychiatry 65:1634-41.

Zarate CA, Kando JC, Tohen M, et al (1996). Does intolerance or lack of response with fluoxetine predict the same will happen with sertraline? J of Clinical Psychiatry 57:67-71.

Zimmerman M, Coryell W, Pfohl B (1985). The treatment validity of DSM-III melancholic subtyping. Psychiatry Res 16:37-43.

Zusky PM, Biederman J, Rosenbaum JF, et al (1980). Adjunct low dose lithium carbonate in treatment-resistant depression: a placebo-controlled study. J Clin Psychopharmacol 8:120-4.