The Role of Psychosocial Stress in the Onset and Progression of Bipolar Disorder and its Comorbidities: The Need for Earlier and Alternative Modes of Therapeutic Intervention

Bipolar illness with early age of onset, by itself, appears to be an extremely robust risk factor for an overall adverse outcome (Birmaher et al., 2006; Carlson, Bromet, & Sievers, 2000; Cate Carter, Mundo, Parikh, & Kennedy, 2003; Engstrom, Brandstrom, Sigvardsson, Cloninger, & Nylander, 2003; Ernst & Goldberg, 2004; Othman, Bailly, Bouden, Rufo, & Halayem, 2005; Perlis et al., 2004; Strober et al., 1988; Suppes et al., 2001). This risk factor is readily understandable from a developmental perspective, because the early mood and behavioral dysfunction may interfere with essential aspects of one’s social, relational, and educational roles and opportunities. Tasks at the neural-developmental level, including emotion regulation and modulation and the ability to exert higher levels of cortical and cognitive control over activity in lower centers may be seriously impaired as well.

The observations of illness-related interference with normal development is particularly troubling from the perspective noted here that an early age of onset of bipolar illness is associated with a greater number of years’ delay until first treatment in both outpatient clinical (Leverich & Post, in press) and epidemiological population samples (Wang et al., 2005). Thus, earlier recognition, diagnosis, and treatment of early onset bipolar disorder may alter what otherwise carries a poor prognosis in adolescence and early and late adulthood.

As illustrated in Figure 1, there appear to be potent gene-environment interactions between childhood stressors of physical abuse and sexual abuse and a positive family history of mood disorders (in first-degree relatives) on the age of onset of bipolar illness. Those who have both vulnerability factors have the earliest age of onset, compared with those who have either one or neither (Garno et al., 2005; Leverich, McElroy, et al., 2002). Obviously, in these types of studies, one cannot dissect the role of familial (nongenetic) factors from those relating to specific genetic ones, and here we only imply that the familial association may or may not have genetic determinants.

The important cross-fostering studies of Meaney and associates in rodents (Francis, Diorio, Liu, & Meaney, 1999), as well as the elegant studies of Insel et al. in several animal species (Francis, Szegda, Campbell, Martin, & Insel, 2003; Insel, 1997; Winslow, Noble, Lyons, Sterk, & Insel, 2003) should make one extremely cautious in ascribing what appear to be genetic predispositions to genes, as opposed to familial/environmental influences that can themselves determine lasting neurobiological and behavioral traits. The study of Francis and associates (Francis et al., 2003) indicates that both in utero environmental influences, as well as those in the neonatal environment, may each be important alone and in combination in helping to determine various behavioral and neurobiological characteristics. Traits modified by cross-fostering can be transmitted to the next generation via changes in DNA methylation (Meaney, personal communication, 2005). Tsankova et al. (2006) reported that chronic social defeat stress in mice and chronic treatment with antidepressants induced opposite effects on brain-derived neurotrophic factor (BDNF). Defeat stress downregulated BDNF by repressive histone methylation, whereas chronic imipramine reversed this via increasing histone dicetylation. Thus, chromatin remodeling may provide a basis for lasting effects of stress and their reversal during antidepressant treatment.

As in the preclinical studies, the early occurrence of stressors in humans may in some instances have more profound effects than those occurring later in life. In our clinical sample (Leverich, McElroy, et al., 2002; Leverich & Post, in press), those with trauma in childhood had a more adverse course of illness than those who experienced these traumas only in adulthood.

Obviously, the quality of the stressor as well as its frequency and severity may all cumulatively impact short-term and long-term behavioral and biochemical adaptations. There may also be special difficulties if the abuser is within one’s own family (Trickett, Reiffman, Horowitz, & Putnam, 1997), and the usual supportive role of this individual is transformed into one of fear and foreboding. Similarly, in the studies of Leverich et al. (2003; Leverich, Perez, et al., 2002), single or rare instances of physical abuse had a smaller impact on suicidality than individuals who reported experiencing these traumas more frequently (Figure 2, top panels). In contrast, it appears that even rare instances of childhood sexual abuse are associated with a highly significant increased incidence of subsequent medically serious suicide attempts and a “dose”-response relationship is less evident in this domain (Figure 2, bottom left) than in physical or sexual abuse later in life during adolescence (Figure 2, bottom right). Moreover, experiencing both types of childhood adversity increases the incidence of suicide attempts more than either stress alone (Figure 3).

Those patients with a history of childhood physical or sexual trauma go on to have a more adverse course of bipolar illness compared with outpatients without such an early adverse history. In the studies of Leverich et al. (Leverich, McElroy, et al., 2002; Leverich & Post, in press), this more adverse course was manifest both by self-report retrospectively prior to SFBN entry at approximately average age 40, as well as after SFBN entry, as assessed prospectively. Retrospectively, there were more episodes and a greater degree of cycling in those with early abuse history (Table 1). Prospectively, there was an increased percent time ill in those with early abuse history compared with those without such early abuse histories (Table 2) as well as an increased incidence of medically severe suicide attempts.

It is striking that those with early environmental adversity (childhood physical or sexual abuse) developed more Axis I, II, and III comorbidities than their comparison clinic members (Leverich, McElroy, et al., 2002; Leverich, Perez, et al., 2002; Tables 3 and 4). There was a higher incidence of anxiety and eating disorder comorbidities as well as substance abuse in those with early abuse. Each of these comorbidities is likely to be consequential in its own right; in several studies, anxiety comorbidity in the context of bipolar illness is in itself a poor prognosis factor (Boylan et al., 2004; McElroy et al., 2001; Simon et al., 2004).

Similarly, comorbid alcoholism and other substance abuse carry a variety of serious risks, not only in the clinical domain, but also based on the fact that stimulant abuse, for example, engenders a variety of neurobiological effects at the level of alterations in gene expression (Post & Weiss, 1988) that appear to interact adversely and either be additive or potentiative with many of the same brain abnormalities associated with bipolar illness (Post, Speer, Hough, & Xing, 2003; Table 5). Thus, it appears that one is adding a new set of neurobiological alterations and vulnerabilities with such substance abuse. Individuals with primary substance abuse are already at a high risk for a generally poor outcome as well as higher degrees of noncompliance and increased risk of relapse. When these risks are combined with the difficulties inherent in bipolar illness treatment, the potential adverse interactions are extraordinary (Krishnan, 2005; Sonne, Brady, & Morton, 1994; Strakowski, DelBello, Fleck, & Arndt, 2000).

Severe stressors (such as those described here) may thus play a unique role in the vulnerability to the onset of bipolar illness itself and the onset of substance abuse comorbidity. In addition, it also appears likely that stressors in adulthood can influence either type of relapse, that is, into new affective episodes in those who are in relatively stable remission and into renewed substance abuse in those who have been abstinent. The preclinical data in animals strongly support the idea that early stressors may convey a life-long vulnerability to subsequent alcohol or other substance abuse compared with littermate control animals that do not experience these stressors (Huot, Thrivikraman, Meaney, & Plotsky, 2001; Meaney, Brake, & Gratton, 2002; Table 6). In some instances, the differential effect of stress as a function of degree of control over the stressor is also apparent in the subsequent adoption (or not) of substance self-administration as an adult.

The occurrence of early childhood physical or sexual abuse also appears to be associated with an increased occurrence of negative (as opposed to positive) life events in adulthood (Leverich, McElroy, et al., 2002; Table 7). This increased incidence of stressors in those with the experience of early adversity is subject to a variety of interpretations.

One interpretation is that the early occurrence of adversity may put an individual at increased risk for exposure to negative situations and negative outcomes, that is, the accumulation of negative life events. This risk appears to be selective to the occurrence of later negative life events, as the number of later positive life events is equal in those with, versus without, early adversity (Leverich, McElroy, et al., 2002).

The increased number of adverse life events in adulthood reported by people who were abused early in life confounds the interpretation of whether there is stress sensitization or not, that is, increased reactivity to stressors later in life. Considerable evidence in unipolar depression supports stress sensitization, particularly over the first several episode occurrences (Kendler et al., 2000, 2001). It is postulated that following the occurrence of sufficient numbers of “triggered” affective episodes, lesser degrees of stress, the anticipation of stress, or even a lack of stress could still be associated with later episodes (Post, 1992). Thus, if stress sensitization were occurring in the context of an increasing number of negative life events, the individual might be at particularly increased risk of having a subsequent episode based both on the number of stressors and their increased likelihood of triggering an episode when they do occur, that is, the sensitization effect (Post, 2004c; Post, Ketter, Speer, Leverich, & Weiss, 2000). The occurrence of stressors in patients late in their course of episode recurrence cannot be used to refute the stress sensitization hypothesis; evidence against it would be derived from the observation that stressors did not increase the risk of subsequent episodes compared with periods of time without stressors.

Not only is there an increased number of episodes and overall morbidity observed in the retrospective course of illness in those with early adversity compared with those without, but also this continues prospectively. When these patients were treated naturalistically by experts in the field, either with monotherapy or with a wide range of treatment modalities in complex combination (Kupka et al., 2005; Nolen et al., 2004; Post et al., 2003), they appeared to be less responsive to this prospective treatment, as rated by clinicians, compared with those without such adversity (Leverich, McElroy, et al., 2002). This is particularly important in the area of increased duration of time depressed compared with time euthymic.

It is also clear that there are multiple potential contributors and confounds to this less favorable outcome because those with early adversity have an earlier age of onset, more prior episodes, and more comorbidities, each of which themselves have been associated with a generally more adverse course of illness. Although it may be possible to statistically correct for these highly intercorrelated events and diverse impacts, it is equally possible that from a clinical perspective, it is the accumulation and interaction of these multiple adverse risk factors (including familial history, stressors, earlier onset, greater number and faster recurrence of episodes, and more anxiety and substance abuse comorbidity) that lead to increasing vulnerability to episode recurrence and general treatment resistance. It is likely from a clinical, if not statistical, perspective that attempting to prevent or reduce the impact of one or more of these factors (i.e., early adversity, more stressors, more episodes, or more comorbidities) may eventually be important in helping patients achieve better outcomes.

Kendler and colleagues (Kendler, Myers, & Prescott, 2005; Kendler, Thornton, & Prescott, 2001) have described gender differences in the types of stressors that appear particularly pertinent in the occurrence of unipolar depressive episodes. These investigators found that losses in the area of a woman’s social network and support system were particularly pertinent to depression onsets. In contrast, job-related difficulties and divorce or separation appeared to be particularly pertinent for depression onset in men. Other factors for women included an increased incidence of housing problems, loss of a confidante, and illness in women in one’s social network. Men also had an increased incidence of job loss, legal problems, and robbery.

Based on patient questionnaire data (Leverich, Perez, et al., 2002), we had the opportunity to explore gender differences in psychosocial stressors. The stressors were rated at two different time points in the illness. The first time point, at illness onset, obviously involves a great deal of potential memory obfuscation based on the long duration of recall necessary for these adults who, at SFBN entry, were on average 40 years old with a duration of illness of about 20 years. However, at the second time point, stressors were rated in relationship to the onset of the most recent episode, which would involve much more immediate recall and therefore much less chance for inaccuracy.

There were interesting gender differences at both illness onset and the most recent episode in the types of stressors that were significantly related to making a suicide attempt (Table 8). Men who made suicide attempts (compared with those who did not) experienced a wide variety of negative life events including four in the category of loss of social supports, three in the area of job and legal difficulties, and one in medical and health care access, particularly at the time of bipolar illness onset.

In contrast, in females who made a serious suicide attempt, there were fewer stressors at illness onset, but many more prior to the last affective episode before SFBN entry. This suggests the accumulation of many psychosocial stressors that were significantly related to making a serious suicide attempt and generally occurred after illness onset and became more problematic prior to the last affective episode. Of interest, the development of difficulties in the area of meeting mul social role demands (prior to the most recent episode) was highly significantly related to women (but not men) with bipolar illness having made a suicide attempt. Others in the areas of (a) loss of social supports; (b) financial problems; and (c) medical problems, health care coverage, and access were also (but less significantly) related to women having made a suicide attempt.

In contrast, in men, a host of negative environmental events were already involved, even at illness onset in those who would subsequently make a suicide attempt. Thus, the data in Table 8 suggest that the buildup of multiple stressors during the illness could relate to suicide attempts in women, while men who subsequently made suicide attempts had these stressors and negative life events even before the illness started.

Because no Bonferoni corrections were applied in this exploratory analysis, these data remain to be further replicated and confirmed in other studies. Nonetheless, they suggest the likely possibility that the impact of different kinds of psychosocial stressors associated with bipolar patients making a serious suicide attempt may be, in part, gender related, both at illness onset and in those preceding the most recent episode. As having made a serious suicide attempt is itself a marker of a more severe course of bipolar illness as noted above, many of these stressors could also reflect vulnerability to illness progression.

The data on the most recent episode are also of interest in relation to the general findings in the literature that episodes later in the illness are likely to be more spontaneous, whereas earlier episodes are likely to be triggered by psychosocial stressors. From the stress sensitization perspective (Post, 2004c; Post et al., 2000, 2003; Post, Leverich, Xing, & Weiss, 2001; Post, Weiss, Leverich, Smith, & Zhang, 2001), when stressors do occur in later episodes, there is increased vulnerability to stressors of small magnitude, or hypothetically, as suggested by Kraepelin (1921), even those that are just anticipated. In this regard, the findings may suggest the importance of building support networks for prevention of episode recurrence as well as suicide attempts in women, while in men, employment, legal, and medical adversities, in addition to loss of social supports, may be more important targets for therapeutic intervention, even from the first onset of illness. Many of these same factors may also have an impact on the occurrence of comorbidities. As outlined in Table 5, stressors can both predispose to the adoption of problematic substance abuse and also be involved in precipitating relapses in those who have abstained. It is noteworthy that women with bipolar illness are seven times more likely to suffer from alcohol abuse than women in the general population (Frye et al., 2003).

Hierarchical logistic regression, using all of the variables that were significantly related to suicide attempts in the univariate analysis in the entire patient group, also identified these different variables separately in women and in men that added a significant odds ratio for making a suicide attempt (Table 9). In women, a history of early abuse remained a significant risk factor initially after all of the other variables were entered in Models 2– 4. More than four hospitalizations for depression, Cluster B Axis II comorbidities, and physical abuse as an adult, as well as problems meeting multiple social role demands, were also risk factors for a suicide attempt.

Conversely, in males, a family history of drug abuse, occupational problems, and loss of social support showed significant odds ratios for the risk of a suicide attempt. Only having greater than four hospitalizations for depression was a common risk factor for suicide attempts in both men and women, but with a substantially higher odds ratio for women than for men.

In the case of a bilineal (both parents) positive family history for affective illness (with at least one parent bipolar), that is, a high genetic and familial risk, there is an approximately 70% chance that the offspring of these parents will develop a unipolar or bipolar affective disorder over their lifetime (Gershon et al., 1982; Lapalme, Hodgins, & LaRoche, 1997). In the context of this extraordinarily high risk, it would appear warranted to begin to intervene at the first appearance of subthreshold symptoms even before the onset of a full diagnosis. Such early intervention would not be primary prevention, that is, presymptomatic, but such secondary prevention would be a more conservative threshold for considering treatment in those at highest risk because of a bilineal positive family history.

Without formal studies, clinicians and family members seeking such early interventions at the point of emerging symptoms not meeting full diagnostic thresholds would be doing so in the absence of good data in the hope that a proven agent for treatment of a full-blown episode would also be effective in the initial stages of illness development and prevention. This may or may not prove to be a valid assumption. For example, it had been assumed that the highly effective antiseizure drug phenytoin would prevent the development of seizures following neurosurgery or head trauma, yet repeated controlled studies have indicated that this is not the case, and other drugs needed to be explored for this purpose of primary prevention or antiepileptogenesis (Temkin, Jarell, & Anderson, 2001). In parallel in the laboratory, phenytoin, carbamazepine, and lamotrigine do not work for the early prevention of the development of full-blown amygdala-kindled seizures in animals, whereas valproate, diazepam, and levetiracetam are effective in both the early and the middle full-blown phases (Post, 2004a).

Such pharmacological dissociations are likely to occur in the realm of the early development (prevention) versus acute treatment of a full-blown affective episode as well, although one cannot assume that the therapeutic profile of a drug would carry over from kindled seizures to affective illness. Lithium, which is effective in full-blown affective episodes, lacks efficacy in any stage of amygdala-kindled seizure evolution. Ultimately, for the prevention of the development of full-blown affective episodes, only empirical data will suffice. Such data will not be available for many years because these types of studies in bipolar illness have not yet been initiated (as they have been already for many years in schizophrenia; Post & Kowatch, 2006).

Childhood adversity (such as physical or sexual abuse) and a positive family history of affective disorder, either as sole risk factors or combined risk factors, yield a highly significant earlier onset of bipolar disorder (Figure 1) compared with those without these vulnerability factors (Garno et al., 2005; Leverich, McElroy, et al., 2002; Perlis et al., 2004). Given the strong inverse relationship between the age of illness onset and the time delay until first treatment (Leverich, McElroy, et al., 2002; Leverich & Post, in press; Wang et al., 2005), early evaluation of the troubled child with these dual risk factors would appear indicated.

The genetics of suicide can run parallel to or separate from the genetics of primary affective illness. As such, a positive family history for suicide appears to be an independent risk factor. Recent study data indicate that the children of parents who have made a suicide attempt have a sixfold increased risk of making an attempt themselves (Brent et al., 2002). It is interesting that in that study familial transmission of suicide attempts was more likely if (a) the adult probands had a history of sexual abuse or (b) the offspring were female and had a mood disorder, substance abuse disorder, increased impulsive aggression, and a history of sexual abuse.

In a more recent study, of the probands with mood disorder who had attempted suicide 23.2% had a first-degree relative with a history of suicidal behavior compared with 13.2% in the families of nonsuicidal individuals. Rates of reported childhood abuse and severity of lifetime aggression were also higher in probands with a family history of suicidal behavior (Mann et al., 2005). Therefore, it would appear appropriate to give special attention to the affectively troubled individual with a positive family history of suicide or suicide attempts.

When a child’s parent has an affective illness, especially depression in the mother, depending on the age of the child and the associated psychological and social support from other individuals in the family, there may be a substantially elevated risk of affective illness and other psychiatric diagnoses of the children in the family (Klein, Lewinsohn, Rohde, Seeley, & Olino, 2005; Rohde, Lewinsohn, Klein, & Seeley, 2005). Similarly, the presence of postpartum depression or psychosis is a considerable risk factor for a difficult outcome not only in the current newborn, but potentially in the other children in the family as well (Bonari et al., 2004).

Drug or alcohol abuse in one of the parents also places the children at considerable increased risk both for affective disorders, as well as for a variety of externalizing behaviors. Particularly when parental substance abuse is combined with any of the risk factors noted above, there may be added risks for the child for affective and other disorders (Strakowski et al., 2000). A number of studies have also indicated that the parental presence of PTSD is associated with an increased risk of this disorder in the offspring (Davidson, Tupler, Wilson, & Connor, 1998; Yehuda, Halligan, & Bierer, 2001).

Clearly, the occurrence of PTSD in a child should lead to early assessment and potential intervention. A variety of other anxiety disorders may be comorbid with or precursors to affective disorders, and when primary anxiety disorder diagnoses are present, the child deserves careful evaluation (Harpold et al., 2005). As noted previously, the comorbid presence of an anxiety disorder in an adult with bipolar illness places that individual at increased risk for a poor outcome compared with those without an anxiety disorder (Boylan et al., 2004; Henry et al., 2003; Simon et al., 2004).

Several other risk factors could be enumerated, but this brief list should be sufficient to indicate some of the already known risk factors and predictors of subsequent childhood and adult affective disorders. The high odds ratios for the development of affective illness and associated later difficulties suggest the clinical importance of considering early specific therapeutic and preventive measures in those at high risk.

Illness education and management have become an integral part of the therapeutics of managing child- and adolescent-onset diabetes, with the view that careful monitoring and patient participation are crucial to maintaining good control of glucose levels and subsequently having a beneficial effect on long-term prognosis (Weinzimer, Doyle, & Tamborlane. 2005). These educational and monitoring processes were widely utilized both before and after the recent definitive evidence that such tight regulation did lessen late-life sequellae such as MI. It would be almost inconceivable that a diabetes patient would receive medication and only a perfunctory set of instructions about injecting oneself with X, Y, or Z doses of insulin and coming back to see the physician in a few months. Yet, parallel limited illness education and instructions too often are given in the treatment of a first or later episode of bipolar illness. It is disappointing that, even after a more serious episode requiring hospitalization for mania or for depression because of suicidality, postdischarge follow-up instructions are often similarly lacking in substance, detail, reinforcement, and ongoing modification as new data on the degree and completeness of an individual’s clinical response become available.

Given the data on the role of psychosocial stressors in the precipitation and progression of both bipolar illness and its comorbidities and the data supporting the superiority of systematic psychotherapeutic and psychoeducational approaches compared with treatment as usual, a compelling argument at present can be made for enhanced therapeutic maneuvers specifically aimed at these variables (Scott & Colom, 2005).

The task of building a medical team and support network has been expertly enumerated by Miklowitz (2002), and should become essential reading for every patient with bipolar disorder and his or her family members. Inherent in this process is the development of a careful longitudinal monitoring strategy, such that small symptom variations can directly feed back into therapeutic decisions, not only for modified psychopharmacological interventions, but also for additional psychotherapeutic ones as well. We have found that a global or cursory assessment of one’s mood status is not a sufficient means of monitoring the details of an individual’s course of illness that may fluctuate widely in a much higher percentage of patients with bipolar illness than had previously been surmised.

Equally important is an active approach to stress anticipation and management. We know from studies in animals that profound behavioral and neurochemical consequences greatly depend on the “psychological” constructs inherent in the experimental paradigm. That is, an animal given control over the ability to terminate a shock stress does not acquire the same helplessness syndrome or neurochemical abnormalities as its yoked control animal that receives precisely the same amount of physical stressor (Seligman & Beagley, 1975).

Cognitive and behavioral psychotherapy is well suited for helping patients achieve a greater degree of control and, in the process, learning a variety of long-term coping strategies. The data in both unipolar illness, as noted by Fava and colleagues (2004), and an emerging literature in bipolar illness (noted above), also strongly support the importance of such approaches for a variety of positive therapeutic endpoints, including better relapse prevention.

In a related fashion, interpersonal psychotherapy, although less systematically tested in bipolar compared with unipolar illness, has much face validity in dealing with issues of job stress and discord within one’s social network which (as seen above) are important precipitants of both initial affective episodes and their recurrences.

We have discussed the possibility that one should specifically tune the nature of the psychotherapeutic and pharmacotherapeutic interventions to the stage of illness that one is attempting to alter. Ongoing psychoeducational efforts would appear to require modification as a function of stage of illness evolution, severity of symptomatology, and an individual’s basic knowledge or ignorance of the illness, its psychopharmacology, and stress management techniques.

Education about the acute management of a full-blown acute manic or depressive episode must necessarily differ from the educational efforts aimed at monitoring and maintaining a remission. The nuances of self-monitoring and the details of a variety of psychopharmacological approaches that may be required in the management of bipolar illness can be daunting, and would appear to require an ongoing and evolving psychoeducational effort (Post & Leverich, in press). For the vast majority of patients with bipolar illness, this cannot be accomplished in a 20-min medication management visit on a monthly or less frequent basis. More time needs to be allotted to the endeavor and, as needed, supported and enhanced by other clinical members of the treatment team.

More dynamic and interpersonal psychotherapeutic techniques may be pertinent to initial phases of illness when stressors are intrinsically involved, but in the patient who has developed episode automaticity, other approaches would appear indicated. This transition to more spontaneous episode occurrence might be somewhat similar to the transfer of learning and memory mechanisms from those that are associative or declarative and dependent on amygdala and hippocampal systems of the medial-temporal lobe (representational memory), to those involving repeated learning and training where the memory trace appears to reside in neostriatal and, potentially, cerebellar systems (putatively mediating habit memory; Mishkin & Appenzeller, 1987). When affective episodes begin to occur on a more automatic basis, there may be a similar transition in the neural structures mediating their occurrence in a fashion parallel to the transition from representational to habit memory mechanisms.

Such a transition appears to occur in the area of substance use and abuse as well. For example, in psychomotor stimulant administration, the amygdala is crucial for initial associative aspects of environmental context-dependent behavioral sensitization after one high dose of cocaine (Post et al., 1987), but after multiple high doses of cocaine, the amygdala is no longer required and the nucleus ↑ accumbens and other neural substrates appear to become more intimately involved in the behavioral sensitization process.

This type of transition in the processes and neural structures mediating behavioral sensitization perhaps parallels the clinical difficulties in deconditioning cues associated with relapse to active substance abuse. After a period of acute abstinence a patient may cognitively assume and feel that they are “immune” from associative cues triggering a craving for cocaine or other psychomotor stimulants, but in actuality, they continue to show autonomic and other unconscious reactions to presentations of drug-related cues and related stressors (Childress, McLellan, Ehrman, & O’Brien, 1988; O’Brien, Childress, McLellan, & Ehrman, 1992). Their relapse into active drug use in the face of minor cues or stressors may appear automatic and not even have an active cognitive component, catching them unaware as well.

In this situation of increased automaticity, it would also appear that different techniques, that is, ones that are systematically practiced and overpracticed, are required to further desensitize these now ingrained responses to drug-related cues. Knowing a specific and practical range of behaviors useful in overcoming such breakthrough cravings is of further aid in preventing relapse (Childress, McLellan, Ehrman, & O’Brien, 1987; O’Brien, Childress, McLellan, & Ehrman, 1993). We would submit that dynamic psychotherapies are not as appropriate for this late stage of cocaine relapse prevention as more systematic and practice-based techniques aimed at modifying the habit memory system, including those developed in 12-step programs or similarly constructed therapies. Also, psychotherapeutic work may need to proceed from mourning the loss of the “well self” at the onset of a first bipolar episode to a reevaluation and appropriate setting of short- and long-term goals in the mid phase of illness, to perhaps existential and related psychotherapeutic techniques for a late stage of illness wherein one has to adapt to certain residual problems and disabilities that may no longer be treatment responsive.

A special caveat is warranted in relation to this last point, however. We have repeatedly seen that patients who are apparently al a late and refractory stage of their illness can achieve surprising and often dramatic degrees of clinical improvement once the appropriate therapeutic regimen is instituted. This also appears to be true for patients who appear to have life-long problems with anxiety or PTSD comorbidities, apparent Axis II personality disorders, or associated “character” traits; these can at times almost “magically” disappear when mood and behavior are adequately stabilized with the appropriate psychopharmacological and therapeutic regimens. Thus, it benefits both patient and therapist to continue to assume that notable therapeutic progress is possible at virtually any stage of illness evolution or severity of manifestation.

Here, it is also important to reemphasize the point that the notions of kindling and sensitization that apply to the untreated or inadequately treated illness (Post, 1992, 2004c; Post et al., 2000) do not imply that cycle acceleration, automaticity, and treatment resistance are progressive or invariant outcomes as some have misinterpreted the theory and predictions of outcome. Specific types of treatment can almost always be applied to the manifest illness, although it may take increasingly complex and more heroic measures later than earlier in illness evolution (Frye et al., 2000).

Moreover, patients often come to a psychopharmacological consultation with the statement that they have tried all of the drug approaches to bipolar illness, when in reality they have had only preliminary and cursory exploration of the very wide range of agents now available. This increasingly wide range of therapeutic options for both the primary and comorbid conditions needs to be dealt with in the appropriate psychoeducational framework without overpromising immediate therapeutic benefit or ultimate symptom remission. For example, a drug such as topiramate, which is not effective in the treatment of acute mania in adults, may nonetheless be helpful in comorbid alcohol use, cocaine use, PTSD, eating disorders, and obesity, as reviewed elsewhere (Post, 2004a, 2004b, 2005). Thus, a considerable amount of polypharmacy and polypsychotherapy may be required to match the panoply of bipolar presentations and comorbidities.

In contrast, in a patient who is in the early stages of bipolar illness after only several episodes or is currently in a substantially improved state, we have found the utility of discriminating the notion of achieving “perfection” in the psychopharmacological as opposed to the psychotherapeutic realm. For many with unipolar and bipolar illness, the striving for psychological perfection is problematic and is a target for psychotherapeutic amelioration in the service of a more successful outcome. In contrast, emphasizing that the psychopharmacological goal is for “perfection” in terms of achieving and maintaining a full remission would appear to have some merit as it pertains to hopefulness, compliance, and the required monitoring intensity, such that minor symptoms can be dealt with before they become problematic. Stating that this type of perfection can be an agreed upon goal of long-term maintenance therapy often evokes a very favorable response on the patient’s part, because this has often been their secret goal all along, that is, achieving and maintaining an essential remission, and they are pleased that this is also supported by the treating clinician and physician.