Bipolar Depressive Subtypes, Other Clinical Factors, and Treatment Response
Although pharmacotherapy with lithium is an effective treatment for many patients with bipolar disorder, Maj and colleagues (1989) reported that even an initially complete response to lithium, defined as no episodes during the first 2 years of treatment, did not predict response over the next 5 years. Among the 25 patients sustaining treatment over a 5-year period, 40% relapsed and another 38% of the original sample dropped out. The range of lithium’s prophylactic effectiveness, once broadly asserted, is now bounded by caveats that increasingly narrow the claims made for lithium maintenance therapy. Possible factors that may reduce the efficacy of lithium are mixed episodes, rapid cycling, comorbidity, and demographic factors.

Mixed Episodes
Several investigators (Himmelhoch and Garfinkel 1986; Prien et al. 1988; Secunda et al. 1985) reported a substantially poorer outcome for lithium therapy in patients with mixed episodes (dysphoric mania, depressive mania) compared with patients with pure affective episodes. The outcome of anticonvulsant therapy for mixed mania is also less robust than the response of patients with pure manic episodes to these agents. Patients who meet DSM-IV criteria for mixed episodes are very difficult to treat and demonstrate a similar but poor response to lithium, anticonvulsants, or placebo (Swann et al. 1997). Interestingly, however, mania associated with subsyndromal depressive features appears to respond better to anticonvulsants than to lithium.

Rapid Cycling
Rapid-cycling bipolar disorder has traditionally been defined as 4 or more episodes of mania or depression per year, although Bauer and Whybrow (1995) suggested changing the criterion to 12 or more episodes per year. Several studies demonstrated reduced efficacy of lithium in patients with rapid-cycling bipolar illness (Dilsaver et al. 1993; Dunner 1979; Kukopulos et al. 1983). As is typical of these studies, Kukopulos et al. (1983) found good response to lithium prophylaxis in only 16% of rapid-cycling patients compared with a 57% response rate in continuously cycling patients with low cycle frequency. Maj et al. (1998) reported that no rapid-cycling patients treated with lithium remained episode-free over a 5-year follow-up period.

Bowden et al. (1999a) reported the outcome of the open phase of an ongoing double-blind study in which lamotrigine was added to the treatment regimen of rapid-cycling bipolar disorder. More than 50% of the sample remained stable while taking lamotrigine monotherapy for at least 2 weeks after tapering all other concomitant treatments.

Comorbid Substance Abuse
It is estimated that more than 60% of patients with bipolar disorder meet criteria for substance abuse at some point during their lives, which exceeds even the rate for patients with unipolar depression (Regier et al. 1990). This relatively high rate of comorbid substance abuse is associated with the occurrence of mixed episodes and rapid cycling (Himmelhoch and Garfinkel 1986) and therefore with relative resistance to treatment.

Comorbid Anxiety
Comorbid anxiety disorders are also common with bipolar disorder (Hammen et al. 1990; Sachs et al. 1993; Weissman et al. 1984). Early-onset bipolar disorder is associated with chronicity and resistance to treatment with lithium (Black et al. 1988), as well as with childhood anxiety disorders (avoidant, overanxious, separation), panic disorder, or substance abuse (Sachs et al. 1993).

Comorbid Personality Disorders
The rate of Axis II disorders in bipolar disorder remains unclear in part because the criteria for bipolar illness often overlap with criteria for personality disorders such as borderline personality disorder. Charney et al. (1981) found that in their study 23% of patients met the criteria for one or more personality disorders, compared with 61% of unipolar nonmelancholic depressed patients. The presence of comorbid Axis II disorders likely contributes much frustration for patients and care providers. However, the only study to date with long-term follow-up (McGlashan 1986) revealed no difference in treatment outcome between bipolar patients with a personality disorder and those without. Although Axis II disorders did not appear to directly affect outcome, the presence of a personality disorder was associated with a high risk for alcohol abuse in this sample.

Comorbid Attention-Deficit/Hyperactivity Disorder
High rates of comorbid ADHD have been reported among children who meet criteria for bipolar disorder (Biederman et al. 1996). The children with ADHD and bipolar disorder frequently had mixed acute episodes in the context of a chronic course of illness. This finding has potential implications for treatment-resistant bipolar disorder comorbid with ADHD across the life cycle.

Comorbid General Medical Illness
As in unipolar depression, unrecognized comorbid medical conditions can also contribute to treatment resistance in bipolar disorder. The lack of recognition of medical illness either caused or exacerbated psychiatric symptoms in up to 46% of psychiatric inpatients (Hall et al. 1981). Clinicians should therefore always consider the possibility of medical comorbidity contributing to resistance in patients with bipolar disorder and gather medical history in a systematic fashion before embarking on further treatment algorithms. Routine laboratory tests together with careful medical history taking represent adequate initial medical evaluations of these patients.

Other Predictive Factors
Children, adolescents (Hsu 1986), elderly persons (Stone 1989), and persons with onset of the disorder before age 18 (Black et al. 1988) appear less likely to respond to lithium. Additional characteristics predictive of poor response to lithium include rapid cycling (Maj et al. 1999), the occurrence of three or more prior episodes (Gelenberg et al. 1989), and a family history negative for affective disorder.