The Increasing Use of Polypharmacotherapy
`for Refractory Mood Disorders: 22 Years of Study
`
`Mark A. Frye, M.D.; Terence A. Ketter, M.D.,;
`Gabriele S. Leverich, L.C.S.W.; Teresa Huggins, Ph.D.;
`Caprice Lantz, M.A.; Kirk D. Denicoff, M.D.; and Robert M. Post, M.D.
`
`
`
`Background: Few studies have approached
`the subject of polypharmacotherapy systemati-
`cally. This retrospective review of 178 patients
`with refractory bipolar disorder or unipolar de-
`pression (Research Diagnostic Criteria or DSM-
`IlI-R criteria) discharged from the National Insti-
`tute of Mental Health (NIMH) Biological
`Psychiatry Branch between 1974 and 1996 was
`conducted to assess the degree and efficacy of
`“add-on” pharmacotherapy.
`Method; Following completion of formal
`structured blinded research protocols, patients
`entered a treatment phase (often again on a blind
`basis) in which all agents available in the commu-
`nity could be utilized. Each patient’s retrospective
`life chart and all prospective double-blind nurse-
`and self-rated NIMH data were reviewed. The
`overall degree of improvement at discharge was
`assessed by rating on the Clinical Global Impres-
`sions scale (CGI) as modified for bipolar illness
`(CGI-BP).
`Results: A 78% improvement rate (moderate
`or marked on the CGI) was achieved at the time
`of discharge. There wasa significant relationship
`between numberof medications utilized at dis-
`charge as a function of discharge date (r = 0.45,
`p <.0001). The percentages of patients dis-
`charged on treatment with 3 or more medications
`were 3.3% (1974-1979), 9.3% (1980-1984),
`34.9% (1985-1989), and 43.8% (1990-1995). No
`correlation was found between polypharmacy and
`age (r = —0.03, p = .66). Patients more recently
`discharged from the NIMH hadan earlier age at
`illness onset, more lifetime weeks depressed, and
`a higher rate of rapid cycling than patients in the
`earlier cohorts.
`Conclusion: Increasing numbers of medica-
`tions in more recent NIMH cohorts were required
`to achieve the same degree of improvement at
`hospital discharge. More systematic approaches
`to the complex regimens required for treatment of
`patients with refractory mood disorder are clearly
`needed,
`
`(J Clin Psychiatry 2000;61:9-15)
`
`
`
`Received May 28, 1998; accepted July 28, 1999. From the Biological
`Psychiatry Branch, National Institute of Mental Health, Bethesda, Md.(all
`authors);
`the Department of Psychiatry and Biobehavioral Sciences,
`UCLA School of Medicine, Los Angeles (Dr. Frye}; and the Departmentof
`Psychiatry and Behavioral Sciences, Stanford University School of
`Medicine, Stanford, Calif. (Dr. Ketter).
`Presented in part at
`the 149th annual meeting of the American
`Psychiatric Association, May 6, 1996, NewYork, N.Y.
`Reprint requests to: Robert M. Post, M.D., Chief, Biological Psychiatry
`Branch, NIMH, Bldg.
`10, Room 3N212, 10 Center Dr. MSC 1272,
`Bethesda, MD 20892-1272.
`
`“The true polypharmacy is the skillful combination of
`remedies.”
`—Sir William Osler!
`
`Pe with refractory mood disorder, particularly
`
`manic-depressive illness, are commonly treated
`with multiple medications. Circa 1996, these medications
`primarily included lithium and the anticonvulsants carba-
`mazepine and divalproex sodium, as supplementedbycal-
`cium channelblockers, conventional antidepressants, ben-
`zodiazepines, antipsychotics, and thyroid hormone.
`Lithium carbonate has clearly been the gold standard
`for treatment of acute mania and bipolar depression and
`for long-term maintenance treatment.? However,the fail-
`ure rate for lithium in acute mania has been reportedin re-
`cent reviews to be 50% or higher.** For example, in the
`largest double-blind, randomized study of lithium and val-
`proateto date,° the response rate (defined as a 50% reduc-
`tion in manic severity) to either of these agents at the end
`of the 3-week study period was only 50%. Furthermore,
`patients who have somebipolar subtypes have particularly
`poor responsesto lithium, including those with mixed or
`dysphoric mania,’ rapid cycling,’ a depression-mania-well
`interval sequence,* and substance abuse comorbidity.?""'
`Although the mood-stabilizing anticonvulsants provide
`treatment alternatives, some patients fail to respond acutely
`or prophylactically to carbamazepine and divalproex so-
`dium, and others can gradually develop loss of efficacy or
`tolerance,’ Although early observations of the acute effi-
`cacy of a new generation of potential mood-stabilizing
`anticonvulsants (gabapentin and lamotrigine) look prom-
`ising,'*'> further controlled studies are encouraged.
`Often, polypharmacotherapy is needed for maximalsta-
`bilization. This is not infrequent clinical practice in the
`
`J Clin Psychiatry 61:1, January 2000
`
`1 of 7
`
`Alkermes, Ex. 1029
`
`1 of 7
`
`Alkermes, Ex. 1029
`
`

`

`Frye et al.
`
`METHOD
`
`of Mental Health (NIMH), between 1974 and 1995 was
`conducted to assess the degree and efficacy of “add-on”
`polypharmacotherapyutilized on a double-blind basis.
`
`medical managementoftuberculosis,'® congestive heart
`failure,'’ autoimmunedisorders,'* multiple sclerosis,'” ac-
`quired immunodeficiency syndrome,” immunosuppres-
`sion in transplantation,”' and refractory epilepsy.” In fact,
`lamotrigine and gabapentin received their recent U.S. Food
`and Drug Administration (FDA) approval utilizing “add-
`on” designs.’ Furthermore,
`lamotrigine and valproate
`Thereferral base for the BPB, NIMH, is refractory uni-
`have what appears to be pharmacodynamic synergy with
`polar and bipolar patients (meeting Research Diagnostic
`anticonvulsanteffects” and enhanced efficacy for mood
`Criteria or, more recently, DSM-III-R criteria) who have
`stabilization.”**’ However,fewclinical studiesin affective
`failed usual therapies in the community and who wish to
`disorder have incorporated this design, and no FDA ap-
`participate in intensive neurobiological evaluation and
`clinical treatment utilizing double-blind protocols.
`provals have been based on such adjunctive clinicaltrials.
`from
`Patients sequentially discharged from the Unit
`The most common mode of polypharmacotherapy (or
`1974 to 1996 were included in the analysis if they com-
`complex combination therapy) utilized in psychiatry is
`“add-on” or adjunctive pharmacotherapy. The prevalence
`pleted one or more double-blind monotherapy protocols
`and associated research procedures and wished to remain
`of this practice has been reported to exist in 28%to 75%
`of diverse patient populations and study designs andis re-
`on the Unit to begin a more clinically based treatment
`viewed extensively by Rapp and Kaplan*® and Gardoset
`phase of hospitalization. This was, again, often conducted
`al.” More recent reviews, both in bipolarillness**' and
`on a blind basis utilizing alternative and add-on medica-
`unipolar depression,” continue to suggest that this is a
`tion in an attemptto further treat and acutely stabilize their
`refractory affective illness. There was no standard clinical
`very commonpractice.
`In the National Ambulatory Medical Care survey, psy-
`algorithm for the study group; each patient’s case was
`evaluated separately taking into account
`initial double-
`chiatric practice, compared with other medical specialties,
`blind, protocol-driven monotherapy, drug trial responsive-
`waspredictive of greater use of polypharmacotherapy, and
`manic patients were 4 times more likely to receive mul-
`ness, past drug trials prior to NIMH, and clinical tolerabil-
`tiple medications than nonpsychiatric patients.’ When
`ity. The vast majority of patients had experienced multiple
`unsuccessful clinical trials prior to their NIMH admission.
`lithium is prescribed, greater than 70% ofthe timeit is uti-
`lized in combination strategies.’ This polypharmaco-
`Patients gave informed consent for a placebo period of
`therapy has been reported to be more prevalent in women
`evaluation and for each of the protocol medications under
`and to increase with age.** Furthermore, somepointpreva-
`clinical research investigation. This first included a major
`monotherapyresearchfocuson piribedil*' and pimozide,”
`lence studies have noted an increasing rate of poly-
`pharmacotherapy in more recent study cohorts.***’ The
`and then,
`sequentially, carbamazepine,“ valproate,”
`study by Hallin et al.” noted single-agent therapy (48%
`thyrotropin-releasing hormone (TRH), nimodipine,” la-
`motrigine,*’ and gabapentin.*’ Patients almost always con-
`lithium) in 57% of 240 bipolar patients in 1989; in con-
`trast, single-agent therapy (26% lithium) was used in only
`tinued taking double-blind medications throughout
`the
`hospitalization; i.e., neither they nor the nursing staff were
`37% of 190 patients 5 years later in 1994, Neither of these
`aware of when they were on active medication or the se-
`studies assessed baseline severity of illness or the degree
`quence of new monotherapies and then “add-on” medica-
`of improvement on combination treatment.
`Peselowet al.,*’ in a large naturalistic study of lithium
`tion trials including conventional antidepressants, benzo-
`diazepines, neuroleptics, and thyroid hormones. Only in
`prophylaxis for patients with bipolar illness (N = 305), re-
`the last year of the study (1995) were patients aware of
`ported that the probability of remaining euthymic was
`when they were in a given 6-week phase (1, 2, or 3) com-
`85%, 52%, and 37% after 1, 3, and 5 years of lithium
`paring lamotrigine, gabapentin, and placebo on a random-
`monotherapy, respectively. For an affective relapse, pa-
`ized crossover basis.'*
`tients were treated with lithium plus an adjunctive mood
`stabilizer, antidepressant, neuroleptic, or benzodiazepine;
`Double-blind ratings consisted of twice-daily nurses’
`ratings by consensus using the 15-point Bunney-Hamburg
`70% of these patients did better on combination treatment
`Rating Scale for depression, mania, anger, anxiety, and
`and obtained greater protection against subsequentrelapse
`psychosis” andaself-rating of mood utilizing a 100-mm
`when compared with the initial course of lithium mono-
`therapy.
`visual analog scale. These daily longitudinal measures
`Few studies have approached the subject of poly-
`were depicted graphically for comparison with prior course
`pharmacotherapy systematically, and it is the purpose of
`ofillness, as assessed by the retrospective life chart method
`(NIMH-LCM).* The NIMH-LCMallowsfor systematic
`this article to initiate such a discussion. This retrospective
`quantification of a numberof course-of-illness variables
`review of 178 patients with refractory mood disorderdis-
`including age at time offirst symptoms, duration ofillness,
`charged from the 3-West Clinical Center Research Unit of
`and past hospitalizations, Notall of the 178 patients had a
`the Biological Psychiatry Branch (BPB), National Institute
`
`10
`
`J Clin Psychiatry 61:1, January 2000
`
`2 of 7
`
`Alkermes, Ex. 1029
`
`2 of 7
`
`Alkermes, Ex. 1029
`
`

`

`Polypharmacotherapy for Refractory Mood Disorders
`
`Figure 1. Increasing Polypharmacotherapy in More Recent
`National Institute of Mental Health (NIMH) Discharges
`
`§ Bipolar Patient
`® Unipolar Patient
`
`57
`
`a
`
`r=0.45
`p<.0001
`N=178
`
`os
`
`a6
`
`sa ©
`
`
`NumberofMedications
`
`reflected in age at onset
`retrospective life chart as
`(N = 160) andlifetime weeks of depression (N = 138) cor-
`relationsto discharge date.
`The demographic profile of the 178 patients included
`131 with bipolar disorder (76, bipolar I; 50, bipolar II;
`5, schizoaffective bipolar type) and 47 with recurrent
`unipolar depression (5 schizoaffective depressed type);
`108 female and 70 male patients were included. The
`mean + SD patient age at discharge was 40 + 12.8 years;
`women were older
`(42.1+12.3°
`years)
`than men
`(36.9 + 12.9 years, p< .008). There was no correlation
`between age and discharge date (N=178, r=0.04,
`p < .64), suggesting that age at discharge remainedrela-
`tively constant over the 22-year period. The mean + SD
`duration ofprior illness was 17.3 + 11.2 years (N = 160).
`The NIMH-LCMretrospective clinical demographics
`examined by gender revealed a longer duration of illness
`for women (mean+SD=19.8+10.9 years, N=99)
`than for men (13.34 10.5 years, N=61; p< .0002), a
`greater number of hospitalizations for depression for
`r correlations between discharge date and other NIMH-
`LCM variables were calculated. A stepwise multiple re-
`women (mean + SD =4.5+5.83, N=91) than for men
`gression analysis was performed on these variables using
`(2.5 + 3.53, N = 55; p < .02), and a greater numberoflife-
`time weeks of depression for women than for men (Mann-
`number of medicationsat discharge as the dependentvari-
`able. Unipolar and bipolar subgroups were analyzed sepa-
`Whitney U mean rank = 57.8 for women versus 41.3 for
`men, N = 103; U = 820.5, p < .01). When age, duration of
`rately, but when no major differences were observed, they
`were combined for the sake of brevity of presentation.
`illness, and lifetime weeks depressed were controlled for,
`genderdifference remained significant (p < .007, p < .05,
`Meansare reported including + standard deviation. Group
`differences were compared using the Student t test except
`respectively). No significant differences in age at dis-
`charge (p = .89) or duration of illness (p = .7) by bipolar
`where extreme outliers required the use of the Mann-
`WhitneyUtest.
`versus unipolar subtype were found.
`The overall rating of improvement at discharge was
`made by the Clinical Global Impressions scale (CGI) as
`modified for bipolar illness (CGI-BP),” which allows for
`rating of degree of clinical improvement in depression,
`mania, and overall illness. Moreover, it addresses many of
`the specific criticisms leveled at the CGI relating to the
`types of ratings, technical and scaling problems, defini-
`tion of time domain of the rating, and confounding of
`clinical response with tolerability and side effects. The
`main measure was degreeofoverall clinical improvement
`at discharge, as assessed from the patients’ most appropri-
`ate worst phase ofillness, which almost invariably was
`during their period of baseline evaluation while receiving
`placebo. Degree of
`improvement was classified as
`marked (essentially complete remission), moderate (dis-
`tinct clinically important improvement, but some symp-
`toms
`remain), mild (slight but
`insufficient
`to affect
`patient’s basic clinical status or
`functioning), or no
`change or similar degrees of worsening, as described in
`detail elsewhere.”
`Degree ofclinical global response achieved at dis-
`charge was subsequently analyzed asit related to the ma-
`jor demographic and course-of-illness variables available
`from the NIMH-LCM. Statistical analysis was conducted
`using SPSS software (version 8.0; Chicago,IIl.). Pearson
`
`o-4
`1971
`
`adTe Td
`1976
`1981
`1986
`1991
`1996
`Discharge Date
`
`RESULTS
`
`The overall response rate as measured by CGIscore at
`the time of hospital discharge was 78% (35% marked im-
`provement, 43% moderate improvement). Of the 21%
`whowere nonresponders, 16% were mildly improved, 4%
`showed no change, and 1% were mildly to moderately
`worse. The analysis of CGI versus date of discharge
`yielded a weak positive correlation (r= 0.20, p < .007),
`showing that the patients more recently discharged from
`the NIMH were,if anything, slightly more improved than
`those studied earlier.
`There was a highly significant positive relationship be-
`tween increased numberof discharge medications and the
`more recent discharge date (r = 0.45, p < .0001; N = 178;
`Figure 1). There was no correlation between the degree
`of polypharmacotherapy and age at discharge (r = —0.03,
`p = .66). Length of hospital stay correlated with discharge
`year (r = 0.432, p < .0001) and the number of discharge
`medications (r= 0.34, p <.0001); i.e., more recent pa-
`tients had longer hospitalizations, most likely related to
`the greater number of blinded monotherapy and add-on
`trials available and offered to patients who remained un-
`improvedin their clinical treatment phase of their hospi-
`
`J Clin Psychiatry 61:1, January 2000
`
`ll
`
`3 of 7
`
`Alkermes, Ex. 1029
`
`3 of 7
`
`Alkermes, Ex. 1029
`
`

`

`Frye et al.
`
`
`Table 1. Adjunctive Medication to Primary Mood Stabilizer at NIMH Discharge
`
`Lithium
`(N = 55)
`Unipolar Bipolar
`(N=9)
`(N=46)
`
`Carbamazepine
`(N = 49)
`Unipolar Bipolar
`(N=11)
`(N=38)
`
`Divalproex Sodium
`(N = 12)
`Unipolar Bipolar
`(N=0)
`(N= 12)
`
`Calcium
`Channel Blocker
`(N =7)*
`Unipolar Bipolar
`(N=1)
`(N=6)
`
`Total
`(N = 123)
`Unipolar Bipolar
`(N=21) (N= 102)
`
`Adjunctive
`Medication, N (%)
`Any adjunctive
`
`medication 1(100)=2.(33)5(56) 12(26) 6(55) 23(61) 0 12 (100) 12(57) 49 (48)
`
`
`
`
`
`
`
`
`0
`4 (33)
`0
`2 (33)
`5 (24)
`18 (18)
`6 (16)
`2 (18)
`6 (13)
`3 (33)
`Antidepressant
`0
`1 (8)
`0
`0
`1 (5)
`10 (10)
`Neuroleptic
`1 (11)
`4 (9)
`0
`5 (13)
`0
`0
`0
`0
`0
`3 (3)
`Benzodiazepine
`0
`1 (2)
`0
`2 (5)
`Calcium channel
`0
`1 (9)
`1 (2)
`0
`blocker
`2 (2)
`1 (5)
`0
`0
`
`13 (13)
`4(19)
`0
`0
`3 (33)
`2(4)
`1 (9)
`4(11)
`Thyroid
`
`“A total of 10 patients were discharged on treatment with a calcium channel blocker, only 3 of them as an adjunctive medication.
`
`0
`0
`
`1 (8)
`7 (38)
`
`
`
`Figure 2. Earlier Onset of Symptoms in More Recent NIMH
`Cohorts
`
`@ Bipolar Patient
`® Unipolar Patient
`
`8
`
`.
`.
`
`a
`
`r=-0.30
`p<.0001
`N= 160
`
`s
`
`” °
`
`60-
`
`e
`8
`
`5
`
`8
`"se
`°
`e
`
`E 50-
`=
`E
`an
`a 40
`%
`
`=aoa<
`
`
`
`ee
`1971
`1976
`1981
`1986
`1991
`1996
`Discharge Date
`
`neuroleptics, while used as necessary when other agents
`were not effective, did not constitute a major portion
`of the discharge medications for this group of refractory
`patients.
`A number of NIMH-LCM retrospective demographic
`variables that might be related to the increasing use of
`polypharmacy were examined. When the age at onset of
`first mood symptomswasassessed, a significant decrease
`over time (r = —0.30, p < .0001; N = 160) was noted;i-e.,
`the patients more recently discharged from the NIMHre-
`ported earlier symptom onset than patients who were dis-
`charged earlier (Figure 2). Secondly, over the 22-year pe-
`riod,
`there was a concomitant pattern of increasing
`duration of lifetime weeks of depression experienced
`prior to NIMHhospitalization as a function of discharge
`date (r = 0.28, p < .001; N = 138).
`There was also an increase in the percentage of rapid
`cyclers over the study period. In the 1970s, rapid cyclers
`constituted 30% of the population; in the 1980s, 56%; and
`in the 1990s, 70% (x? = 14.66, p< .001). The rapid cy-
`clers, in comparison with non-rapid cyclers, had a longer
`
`talization following the protocol-driven research phase.
`However, the partial correlation between discharge date
`and number of medications, controlling for length of
`hospital stay, was still significant (r=0.36, df= 175,
`p < .001). The significance remained when controlling for
`rapid-cycling status (r = 0.32, df= 159, p < .001). More-
`over, for statistical confirmation, a stepwise multiple re-
`gression analysis was performed evaluating the continu-
`ous variables such as age at discharge,ageatillness onset,
`past hospitalizations for depression, length of NIMH hos-
`pitalization, and discharge date with numberof discharge
`medications as the dependent variable. The only variable
`showing significance was discharge date (R* = 0.182,
`t= 4.60, p < .0001).
`the mean number of
`By arbitrarily defined epochs,
`discharge medications
`for 1974-1979 was
`1.5;
`for
`1980-1984, 1.5; for 1985-1989, 2.5; and for 1990-1995,
`3.0. The percentages of patients discharged on treatment
`with 3 or more medications in these same epochs were
`3.3%, 9.3%, 34.9%, and 43.8%, respectively. The degree
`of clinical global improvement on the CGI-BP was not
`correlated with the number of discharge medications
`(r= 0.09, p < .26; N = 178).
`Table | summarizes the types of adjunctive discharge
`medication to 4 major mood stabilization treatment
`groups sequentially utilized (lithium, carbamazepine, di-
`valproex, and calcium channel blockers). Over the entire
`study period, for those patients who were discharged on
`treatment with a mood stabilizer (N = 123), adjunctive
`medications included conventional (or unimodal) antide-
`pressants (18.7%),
`thyroid supplementation (13.8%),
`neuroleptics (8.9%), benzodiazepines (2.4%), and cal-
`cium channel blockers (2.4%). In comparison with bi-
`polar patients discharged on treatment with fewer than 2
`medications, bipolar patients who were discharged on
`treatment with 2 or more medications were morelikely to
`be rapid cyclers (Fisher exact test, p = .04). There was no
`difference in monotherapy versus polytherapy (i.e., 22
`medications) by gender, age at discharge, or ageatillness
`onset. Remarkably,
`the unimodal antidepressants and
`
`12
`
`J Clin Psychiatry 61:1, January 2000
`
`4 of 7
`
`Alkermes, Ex . 1029
`
`4 of 7
`
`Alkermes, Ex. 1029
`
`

`

`duration of illness (t= 2.86, df= 116, p=.005), more
`past hospitalizations for depression (t= 2.05, df = 92,
`p = .04), and a trend for a greater number of medications
`at discharge (t = 1.74, df = 119, p = .08).
`
`DISCUSSION
`
`Several assets and liabilities are apparent in the inter-
`pretation of this study. One of manyliabilities of this study
`was that it was conducted in a research setting and im-
`provement was based only on status at discharge; confir-
`mation of more substantial and sustained clinical efficacy
`requires longer-term follow-up, whichis in progress, Sec-
`ondly, there was no precise clinical algorithm under which
`all patients were treated. This option was precluded be-
`cause of the evolution in research focus and potential
`therapeutic agents evaluated over the study period. How-
`ever, after primary research evaluations and monotherapy
`protocols were completed, the general pattern wasto tar-
`get current symptomatology in sequential, blinded mono-
`therapy and then use add-onclinicaltrials in an attempt to
`maximize mood stability on an individualized basis. This
`would be attempted usingall prior information(i.e., retro-
`spective life chart) and the patient’s carefully observed
`course ofillness at the NIMH. Although viewedas a limi-
`tation for developmentofa large-scale algorithm, the gen-
`eral use of sequential add-onclinical trials does, to some
`extent, parallel clinical practice, although in a blinded
`fashion.Finally, the potential implications that this unique
`treatment-refractory,
`increasingly rapid-cycling cohort
`has for the general population of patients in a practiceset-
`ting must be cautiously and conservatively considered.
`One of the study assets was that virtually all of the
`medication trials were conducted on a double-blind basis.
`Secondly, clinical response to an initially blind protocol
`medication was often reconfirmed with a phase of pla-
`cebo substitution and rechallenge with the active agent.
`Thirdly, extensive life chart data were available on each
`patient, so that the likelihood of a placebo response or a
`responseattributable to the natural courseofillness could
`be factored into the clinical and research evaluations in
`further attempting to determine clinical improvementat
`discharge not likely related to spontaneousillness varia-
`tion. Finally,
`the primary research goal of the initial
`phasesof study allowed for a more extended continuation
`of the blind evaluations that would not be possible in most
`clinical settings.
`Despite the liabilities noted above, several preliminary
`conclusions about the increasing need for combination
`treatment at discharge nevertheless can be drawn. The
`relatively low percentage in bipolar and unipolar patients
`of the use of adjunctive antidepressants (bipolar, 18%; uni-
`polar, 24%) and neuroleptics (bipolar, 10%; unipolar, 5%)
`in discharge regimens, despite a highly favorable overall
`improvementrate of 78%, was surprising. This is at vari-
`
`Polypharmacotherapy for Refractory Mood Disorders
`
`ance with mosttraditional clinical treatment units in which
`patients with acute mania or rapid cycling, under pressure
`of short-term hospitalizations, are almost uniformly ex-
`posed to and treated with neuroleptics. A recent review
`noted 40% to 72% of bipolar patients were currently
`treated with an antipsychotic and 90% to 100% hadhis-
`tory of neuroleptic exposure.*’ Secondly, as reviewed by
`Frye etal.,”! typical neuroleptics have significantliability
`for lack of mood stabilization, acute extrapyramidal symp-
`toms, and tardive dyskinesia in bipolar patients. In the
`course of exposing fulminantly manic patients to alterna-
`tive investigatory and now more routinely used agents
`such as carbamazepine and valproate, we uncovered this
`general lack of necessity for neuroleptic use even in this
`highly treatment-refractory rapid-cycling population.
`At the opposite pole, the potentialliability of unimodal
`antidepressants to precipitate acute manic episodes or
`induce cycle acceleration in bipolar patients has been
`noted by manyinvestigators.Again, the vast majority
`of this unusually treatment-refractory population, over-
`represented with rapid cyclers compared with most com-
`munity settings, were able to be discharged without the
`use of conventional antidepressants. When conventional
`antidepressants were used, this was almost always in the
`context of one or more moodstabilizers.
`The
`sequential pharmacotherapy and add-on ap-
`proaches were utilized under the general rubric of using
`new agents with potentially different mechanisms of ac-
`tion, as well as specific targeting of remaining symptoms
`andillness patterns in an attempt to achieve a more robust
`or complete therapeutic effect, as discussed elsewhere.*°
`It is our impression that these at times complex psycho-
`pharmacologic regimens were well tolerated because of
`the use of the principle oftitrating to reach greatesteffi-
`cacy with fewest side effects (rather than targeting spe-
`cific dose or blood level windows). These data thus do not
`directly address important issues regarding the potential
`for added toxicity, teratogenicity, or noncompliance with
`combination treatments.*~**
`Several possibilities could account for the observed
`need for increased multimodal medication regimens used
`at hospital discharge. This could be driven by (1) more
`clinical
`treatments available; (2) more extensive treat-
`ment of patients in the community prior to referral to the
`clinical research programs of the NIMH, suchthat cohorts
`of the more treatment-refractory patients were referred;
`(3) increased severity of illness due to a changing referral
`bias for earlier-onset rapid cyclers; and/or (4) an increas-
`ing severity or refractoriness of illness in the general
`population, such that the need for polypharmacotherapy
`was reflective of a similar trend in the community at
`large. The latter possibility cannot be dismissed altogether
`in light of the evidence for a cohort effect for unipolar and
`bipolar illness*’attributable to a variety of potential
`causes, including genetic anticipation."
`
`J Clin Psychiatry 61:1, January 2000
`
`13
`
`5 of 7
`
`Alkermes, Ex. 1029
`
`5 of 7
`
`Alkermes, Ex. 1029
`
`

`

`Frye et al.
`
`Whattrends in the characteristics of this patient popu-
`lation are consistent with one or more of these possibili-
`ties? Over the study period, patients were admitted with a
`progressively earlier age at illness onset and an increased
`incidence of rapid cycling. Taken together, these data sug-
`gest that a generally more ill and treatment-refractory
`group based on course-of-illness characteristics were ad-
`mitted to the NIMH overthe study period. Perhaps data
`on number of unsuccessful medication trials prior to
`NIMHreferral over the study period would be mosttell-
`ing about past treatment refractoriness. These data are not
`currently available in a systematic fashion, butit is our
`impression that bipolar patients referred to the NIMH
`were generally lithium refractory in the decade of the
`1970s, lithium and carbamazepine refractory in the dec-
`ade of the 1980s, and, in the 1990s, lithium, carbamaze-
`pine, and valproate refractory. This might suggest that pa-
`tients in recent cohorts were more extensively treated in
`the community prior to their referral to a tertiary clinical
`research unit such as the NIMH. Whateverthe basis for
`the referral of the moreill patients, as inferred from their
`prior course-of-illness variables, there was a highly sig-
`nificant relationship of increased numbers of medications
`utilized in the more recently discharged patients (r = 0.45,
`p <.0001; N = 178). This occurred while maintaining or
`slightly enhancing the overall degree of improvement
`achieved at discharge over the study period.
`It is particularly disheartening to note that, aside from
`lithium, there are no FDA-approved agents for long-term
`prophylaxis of bipolar illness. Furthermore, there have
`been no recent attempts at approval of “add-on” drug
`regimens for the large group ofrefractory bipolar patients,
`whereas this has been the mode of approval for the last 4
`anticonvulsants for refractory epilepsy patients. More-
`over, few acute or long-term clinicaltrials in bipolarill-
`ness have been funded by the NIMHinthe past decade”;
`thus, bipolar patients not responsive to lithium and other
`commonly used therapies are treated prophylactically,
`without the benefit and guidance of a systematic clinical
`trials literature.
`There are also no controlled studies of the relative mer-
`its of aggressive early-intervention polypharmacotherapy
`as opposed to the late salvage polypharmacotherapy de-
`scribed here. This is in marked contrast to polypharmaco-
`therapy in reducing human immunodeficiencyviral load™
`or the clinical practice of combination chemotherapy for
`malignancy. Polypharmacotherapy may be an underuti-
`lized strategy for achieving maximal moodstability given
`the potential clinical and neurobiological consequences of
`an inadequately treated illness; i.e., increased severity, cy-
`cling, and refractoriness.”
`Althoughthis retrospective study has manylimitations
`(i.e., retrospective review,
`lack of precise algorithm or
`prospective follow-up data, and a treatment-refractory
`population screened for willingness to participate ‘in
`
`it does
`trials research),
`neurobiological and clinical
`impart the great need for further studies in recurrentaf-
`fective illness to clarify the relationship of comorbidities
`and course-of-illness characteristics to the subsequent de-
`gree of polypharmacotherapy and long-termefficacy. In
`managing refractory mood disorders with complex poly-
`pharmacotherapies,tolerability of agents needs to be vig-
`orously monitored to ensure safety and tolerability, espe-
`cially since complex regimens have been associated with
`noncompliance.”
`Given the increasing number ofagents available in the
`bipolar and unipolar pharmacopoeia, controlled study of
`combination therapy (both “add-on” and aggressive “at-
`onset” polypharmacotherapy) could only benefit the devel-
`opmentof optimal and empirically based algorithms for
`achieving maximal mood stabilization in the large popu-
`lation of patients with difficult-to-treat affective illness.
`
`Drug names; carbamazepine (Tegretol and others), divalproex sodium
`(Depakote), gabapentin (Neurontin),
`lamotrigine (Lamictal), nimo-
`dipine (Nimotop), pimozide (Orap).
`
`REFERENCES
`
`ta
`
`|, Bean RB, Osler Sir W. Aphorisms: From His Bedside Teaching and Writ-
`ings. Springfield, Il: Charles C Thomas Publisher; 1951
`. Goodwin FK, Jamison KR. Manic-Depressive Illness. New York, NY:
`Oxford University Press; 1990
`3. Post RM, Ketter TA, Denicoff K, et al. The place of anticonvulsant therapy
`in bipolarillness, Psychopharmacology(Berl) 1996;128:1 15-129
`4. Vestergaard P. Treatment and prevention of mania: a Scandinavian per-
`spective. Neuropsychopharmacology 1992;7:249-259
`5. Bowden CL, and the Depakote Mania Study Group. Efficacy ofdival-
`proex vslithium and placebo in the treatment of mania. JAMA 1994:27]:
`918-924
`6. Himmelhoch JM, Garfinkel ME. Mixed mania: diagnosis and treatment.
`Psychopharmacol Bull 1986;22:613-620
`7. Dunner DL, Fieve RR. Clinical factors in lithium carbonate prophylaxis
`failure. Arch Gen Psychiatry 1974,30:229-233
`8. Faedda GL, Baldessarini RJ, Tohen M, et al. Episode sequence in bipolar
`disorder and response tolithium treatment. Am J Psychiatry 1991;148:
`1237-1239
`9, O'Connell RA, Mayo JA, Flatow L,et al. Outcome ofbipolar disorder on
`long term treatmentwith lithium. Br J Psychiatry }991;159:123-]29
`10. Brady KT, Sonne SC. The relationship between substance abuse and bi-
`polar disorder. J Clin Psychiatry 1995;56(suppl 3):19-24
`11. Tohen M, Waternaux CM, Tsuang MT, et al. Four year follow-up of
`twenty-four first episode manic patients. J Affect Disord 1990;19:79-86
`12. Sporn J, Sachs G. The anticonvulsant lamotrigine in treatment resistant
`manic depressiveillness. J Clin Psychopharmacol 1997;17:185-189
`13. Young LT, Joffe R. Gabapentin in bipolar depression: a case series. Biol
`Psychiatry 1997;42:85 1-853
`14. Calabrese JR, Bowden CL, McElroy SL,et al. Spectrum ofactivity of la-
`motrigine in treatment-refractory bipolar disorder. Am J Psychiatry 1999;
`156:1019-1023
`15, Frye MA, Ketter TA, Kimbrell TA, et al, Gabapentin and lamotrigine
`monotherapy in mood disorder. J Clin Psychophamacol. In press
`16. Control of tube