Multigenerational Positive Family History of Psychiatric Disorders Is Associated With a Poor Prognosis in Bipolar Disorder

Family history and course of illness data were collected from 676 outpatients with bipolar illness who completed a detailed patient questionnaire on admission to the Stanley Foundation Bipolar Treatment Outcome Network (currently the Bipolar Collaborative Network) from 1995 to 2001.^3,5–7 The patients had an average age of 41 years and were from four sites in the United States (Los Angeles, Dallas, Cincinnati, and Bethesda). All patients provided informed consent to participate in the study.

Patients (i.e., probands) were asked whether their parents and maternal and paternal grandparents had or likely had a diagnosis of depression, bipolar disorder, alcohol abuse, drug abuse, suicide attempt requiring medical attention or a completed suicide, or other psychiatric illness (specifically stated as “i.e., anxiety, panic attacks, eating disorders, attention deficit disorder, behavioral problems, obsessive-compulsive disorder, autism, etc.”). A family history of schizophrenia was also elicited, but the number of positives was too small to be considered in the analysis. Because this analysis focused on the generational effect, we combined the data from the father and the mother to reflect the parental generation and any of the four grandparents to represent a positive family history in that generation, similar to the procedure reported by Weissman et al.¹

The patient questionnaire inquired about each patient’s illness characteristics, including a history of childhood abuse (verbal, physical, or sexual); age of onset of bipolar disorder (i.e., first depression associated with dysfunction or first mania or hypomania); a lifetime history of any anxiety disorder; history of substance abuse; rapid cycling (i.e., four or more episodes per year); having had 20 or more episodes; and a progressive course of illness estimated by an increasing severity of manic or depressive episodes or increasing frequency of occurrences prior to network entry.

We only included patients from the United States because we had previously seen much more severe illness characteristics in those patients compared with patients from the Netherlands and Germany, such that including the European patients would have confounded the multigenerational analysis.^3,7 We considered the history of any type of abuse in childhood as positive, because we previously reported a considerable adverse impact of verbal, as well as physical/sexual, abuse.⁸ Early age of onset was established as 18 years of age or younger as reported previously.^3,6,9 Evidence of illness deterioration or progression was assessed by ratings of worsening severity of mania or depression or an increase in frequency of occurrences prior to network entry. A total measure of the number of the seven PPFs, excluding illness progression, was also examined in relationship to the family history.

We performed a chi-square analysis of the effect of zero, one, or two generations positive for each illness characteristic in relationship to each patient’s poor prognosis factors, as well as a personal history of suicide attempt and the measures of illness progression. We performed an additional ad hoc chi-square test on the significance of the difference between one and two generations positive for each of the psychiatric difficulties. Most of the one-generation positives were attributable to a parental positive diagnosis, except for alcoholism, which was 50% attributable to either a parental or grandparental diagnosis. The total number of PPFs was examined with an analysis of variance with Bonferroni post hoc tests comparing zero to one, one to two, and zero to two generations positive.

Table 1 shows the percentage of patients who had zero, one, or two generations of their family who were positive for each of the seven PPFs. The very small percentages of patients with two generations positive for suicide and for drug abuse potentially limits conclusions that can be drawn for these categories because of the small number.

As illustrated in Table 2, compared with family history negative patients, those with one or two generations of a positive family history of depression or bipolar disorder had highly significantly more patients with difficult illness characteristics in most domains and in the total number of PPFs.

The only exception was multigenerationality for depression, which was not strongly related to the bipolar proband’s rapid cycling, number of episodes, or suicide attempts. In the ad hoc test, significantly more patients with early age of onset, anxiety disorder comorbidity, and 20 or more episodes were seen in two generations compared with one generation positive for depression.

Family positivity for bipolar disorder in one or more generations was strongly related to the proband’s history of abuse in childhood, history of substance abuse, rapid cycling, and total number of PPFs but was less strongly associated with anxiety disorder comorbidity, substance abuse, and number of episodes, and was not associated with a history of suicide attempts (Table 2). The significance of the increased incidence of substance abuse and rapid cycling was attributable to the difference between zero and one positive generation, and the presence of a second positive generation for bipolar disorder had no further effect.

A family history of substance abuse had a striking relationship to patients' history of abuse in childhood as well as substance abuse. In patients with a family history of two generations positive for alcohol abuse, 82% had a history of abuse in childhood compared with 65.6% in those with one positive generation (p,0.001). Two generations who were positive for alcohol abuse also had significantly more rapid cycling and 20 or more episodes compared with one positive generation.

In patients with a positive family history of other psychiatric disorders (including anxiety disorders), significant effects were apparent for all seven PPFs (with the exception of proband suicide attempts). However, none of the ad hoc direct comparisons of two generations compared with one generation with positive histories was significant. The same was true for the three measures of illness progression (increasing severity or frequency of episodes), again with the exception of suicide attempts (Table 3).

Similarly, the total number of PPFs as seen in Table 4 varied with a positive family history for each psychiatric difficulty. Compared with one positive generation, two generations that were positive for depression and alcoholism showed significantly more PPFs.

Our findings suggest that having at least one generation that is positive for many different psychiatric diagnoses compared with no family history conveys vulnerability to many adverse characteristics of bipolar disorder and the total number of PPFs in patients in the network. In many (though not the majority) of instances, having two compared with one generation positive for a given psychiatric problem was associated with an increased likelihood of having a PPF present. For example, if the proband had a family history of two generations that were positive for depression, they showed significantly earlier onset bipolar disorder, more anxiety disorder comorbidity, and 20 or more episodes compared with probands who had only one generation positive for depression.

These findings in patients with bipolar disorder are similar to those of Weissman et al.¹ in unipolar depression, where the additional grandparental positivity conveyed a greater risk for illness in the offspring than parental positivity alone. In that study, among members of the third generation who had both a depressed grandparent and parent, the risk of psychopathology was extremely high: 60% had some diagnosis, which was most often an anxiety disorder or conduct disorder. The incidence of depression in the offspring as a function of number of generations positive for depression was reported as follows: no relatives positive = 8.5%; one parent positive = 7.9%; one grandparent positive = 15.6%; and two generations positive (a parent and grandparent) = 40.5%.

Before discussing the potential implications of our findings, several limitations and caveats must be recognized. The diagnoses of family members were not based on direct interview but were reported by the patients (who themselves had bipolar illness as determined by the Structured Clinical Interview for DSM-IV and retrospective and prospective course of illness measures). Although some have argued that this type of patient report provides a reasonable approximation of family history, others see it as suspect. Thus, the present findings should be taken as preliminary until they are replicated with more formal and interview-based methods. Similarly, age of onset and other illness characteristics were also based on patient report and were therefore subject to memory difficulties, state-dependent effects, and other biases. However, we reported previously that many of these retrospective patient-rated illness variables were correlated with prospectively observed clinician-rated outcomes, suggesting their reliability and validity.^6,10

These findings should also be viewed in the context of our data indicating that both positive parental history^2,3 and grandparental history of mood disorders and other psychiatric difficulties are more prevalent in the United States than in Germany or the Netherlands.⁴ Thus, the generality of our findings is likely limited to patients recruited by academic centers in the United States and should not necessarily be extrapolated to other populations. We chose to examine only the patients in our network who were from the United States to avoid many of the confounders that would have been involved in combining populations with very different backgrounds, illness course characteristics, and social-cultural differences, including access to universal health care. In addition, examining the European patients separately would have severely limited the number for most variables because there were only 292 patients from the three European sites, and these patients had a much lower incidence of a positive family history of most psychiatric disorders.⁴

We believe our patient sample from the United States is representative of bipolar outpatients studied in many other academic sites in many other U.S. cities because the characteristics of our patient population closely mirror those of the patients in the Systematic Treatment Enhancement Program for Bipolar Disorder network, who were drawn from entirely different cities and institutions.^9,11 Another potential limitation of our study is that we did not know whether the proband had direct contact with the grandparents, such that epigenetic environmental effects as well as the genetic effects could have contributed to the findings. We also did not know whether the illness of the parents or grandparents was well or poorly controlled.

Early age of onset was associated with a positive family history for unipolar depression, drug abuse, and other psychiatric difficulties (including anxiety disorders) in addition to a history of bipolar disorder. These findings suggest the possibility that vulnerability to early onset bipolar disorder may emanate from multiple different psychiatric difficulties in the parental generation. Bilinear positive family histories convey a greater vulnerability than unilinear ones,^8,12,13 but we were not able to examine this factor because of the small number of subjects involved. Nonetheless, childhood onset bipolar disorder has consistently been linked to a great genetic loading based on a greater number of relatives who were positive for mood disorders.¹⁴

The probands’ history of abuse in childhood was among the most robust correlates of the presence of a positive family history for a wide variety of psychiatric difficulties. The toxic effect of early stress in childhood on medical and psychiatric illness in adulthood has been amply documented in the general population¹⁵ as well as our patient cohort.^8,16,17 Compared with one generation, a history of alcohol abuse in two generations was also associated with a higher frequency of abuse during the proband’s childhood.

It is not entirely clear how to interpret these generational findings, because both genetic and epigenetic factors could be at play. Illness across multiple generations would be expected to carry a stronger genetic load.¹ However, parental history of ongoing psychiatric illness would also carry the potential for familial environmentally mediated vulnerability and genetic vulnerability, as evidenced by increases in psychiatric diagnoses in offspring of mothers whose depression was treated but in whom remission was not achieved.¹⁸ Environmentally mediated effects could also include a downward spiral of socio-economic status across generations, especially in light of the increased burden of multiple childhood stressors and those related to the course of bipolar illness,^8,16 but there are no measures of socio-economic status that would allow us to examine this possibility adequately. Recent carefully controlled studies in animals also suggest that some behavioral vulnerabilities and abnormalities from a parental history of stress or drug exposure can be conveyed transgenerationally to the offspring, even in the absence of parenting influence or contact, suggesting that some epigenetic traits may be conveyed via germ-line cells, oocytes, and sperm.^19–23

Regardless of how these transgenerational effects are conferred mechanistically, they would appear to have important clinical, societal, and theoretical implications. It is now widely believed that bipolar disorder is polygenic,^24–27 and it is possible we are seeing the contribution of other genetic traits to various aspects of the course of bipolar disorder. Perhaps these other psychiatric conditions need to be better assessed in multiple generations in genetic studies of bipolar illness onset vulnerability, course of illness, and even treatment response.

The presence of a mood disorder or other psychiatric diagnoses in two generations conveys a risk of occurrence of childhood abuse in the range of 70%–90% for bipolar probands, suggesting that this might be an appropriate arena to consider for preventive interventions. Family-focused treatment and related family interventions that focus on enhancing interpersonal communication and minimizing highly expressed emotion may be a good place to start.²⁸

Our findings that three generations (i.e., the adult patients, their parents, and their grandparents) all had a high burden of multiple psychiatric illnesses suggest the need for changes in clinical and public health approaches to bipolar illness in the United States. The multigenerational occurrence of these difficulties suggests that their presence is long-term and persistent and that current conventional clinical management strategies are not sufficient to break this pattern.

Treatment approaches also need to be developed for bipolar patients with the high levels of comorbidity seen here, including anxiety and substance abuse comorbidity. The potential treatments of these comorbidities are almost never directly studied in patients with bipolar disorder, because these patients are typically excluded from industry- and National Institute of Mental Health-sponsored randomized controlled trials, which are usually aimed at studying homogeneous populations with treatment-responsive illness. Thus, the appropriate treatment paradigms for these common yet complex patient characteristics are virtually unstudied and remain obscure. One attempt to study bipolar depressed patients who had a comorbid diagnosis of generalized anxiety disorder and often had a comorbid substance disorder as well failed to demonstrate the efficacy of quetiapine over placebo, even though quetiapine had demonstrated efficacy in patients with the single illnesses of bipolar depression and anxiety disorders.²⁹

In addition to these complex presentations, two thirds of bipolar illness seen in our adults in the United States began in childhood and adolescence.^3,9 Early onset illness is more severe and difficult to treat than adult onset illness^3,9,30,31 and is associated with a longer delay to first treatment, which itself is an additional risk factor for a poor outcome.^6,32 Childhood onset is also associated with greater numbers of mood episodes, which themselves are associated with more treatment resistance, cognitive dysfunction, medical comorbidity, and dementia in old age.³³ There is a dearth of pharmacological treatment studies of childhood onset bipolar disorder,^34–36 which further compounds the problem of an inadequate knowledge base for a complex multigenerational adult disorder. A new research agenda is needed to better inform these gaps in treatment knowledge for patients with bipolar disorder at virtually every stage of illness development and evolution.

Patients also need to be better educated about their illness, the benefits of self-monitoring, and the multiple treatment options available so that they can play a larger role in illness recognition and management. When patient monitoring and management was successfully achieved for diabetes, disease complications such as blindness, heart disease, kidney disease, and limb amputations became less frequent and were delayed in onset for decades or more. Parallels are likely to occur for bipolar disorder. Initial controlled data indicate that, in patients with a first hospitalization for mania, randomization to 2 years of treatment in a specialty clinic yields a much more benign course of bipolar illness than treatment as usual. Even more remarkably, these differences were maintained and amplified over a period of 6 years, even though all of the patients received treatment as usual for the last 4 years of the study.^37,38 A similar process of integrated care and carefully monitoring needs to be fostered for the high proportion of patients with bipolar disorder in the United States who are treatment resistant or at high risk for it, as well as at risk for a decade or two of lost life expectancy compared with the general population.^39,40

In a prior analysis,⁴ we preliminarily observed a higher incidence of multiple psychiatric illnesses in the offspring of our adult bipolar patients from the United States compared with those from Europe. Thus, it would appear that we are and will be seeing a new generation of young patients with complex and difficult-to-treat bipolar disorder, the basis of which appears to have existed for multiple generations. The transgenerational presence and transmission of complex bipolar disorder in the United States is associated with a host of adverse illness characteristics or PPFs. This transgenerational vulnerability now requires better recognition and pathophysiological and therapeutic study and new approaches to treatment and prevention.