Unipolar depression with bipolar family history: Links with the bipolar spectrum

The aim of the present paper was to find if unipolar major depressive disorder (MDD) with bipolar family history could be included in the bipolar spectrum, by comparing it to unipolar MDD without bipolar family history, and to bipolar II disorder, on typical bipolar variables. A sample of 280 consecutive bipolar II outpatients, and a sample of 135 consecutive unipolar MDD outpatients, presenting for major depressive episode (MDE) treatment, were interviewed with the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders (4th edn). Hypomanic symptoms during the MDE were systematically assessed. Clinical variables used to validate the inclusion of unipolar MDD with bipolar family history in the bipolar spectrum were young age of onset, many MDE recurrences, atypical features, and depressive mixed state (DMX; an MDE plus >2 concurrent hypomanic symptoms), following many previous studies reporting that these variables were typical features of bipolar disorders. Means were compared by t‐test and frequencies by χ2 test (stata 7). Two‐tailed P < 0.05 was chosen. Unipolar MDD with bipolar family history was present in 20% of MDD patients. Comparisons among unipolar MDD with bipolar family history (UP+BPFH), unipolar MDD without bipolar family history (UP–BPFH), and bipolar II (BPII), found that UP+BPFH versus UP–BPFH had a significantly lower age, lower age of onset, fewer recurrences, and more DMX; that UP+BPFH versus BPII had no significant differences (apart from recurrences); and that UP–BPFH versus BPII had significantly different age, age of onset, recurrences, atypical features, and DMX. Findings suggest that UP+BPFH shows many bipolar signs, and that it could therefore be included in the bipolar spectrum. Unipolar MDD with bipolar family history had a clinically significant 20.0% frequency in the unipolar MDD sample, supporting the clinical usefulness of this depression subtype. The subtyping of MDD based on bipolar family history could have treatment implications.

Keywords: bipolar II disorder; bipolar spectrum; depressive mixed state; family history; unipolar major depressive disorder

The Diagnostic and Statistical Manual of Mental Disorders (4th edn, revised; DSM‐IV‐TR) classification of bipolar disorders (bipolar I, bipolar II, cyclothymic, and bipolar disorders not otherwise specified) does not include mood disorders with bipolar signs and no history of mania/hypomania.[1] These mood disorders with bipolar signs, not meeting DSM‐IV‐TR diagnostic criteria for bipolar disorders, are included by many authorities in the so‐called 'bipolar spectrum'. The most recent definitions of the bipolar spectrum are by Akiskal and Pinto,[2] Ghaemi et al.[3] and Angst et al.[4] Akiskal and Pinto's bipolar spectrum included bipolar I (mania), bipolar I½ (major depressive episode (MDE) and protracted hypomania), bipolar II (MDE and hypomania), bipolar II½ (MDE and cyclothymic disorder), bipolar III (antidepressant‐associated hypomania), bipolar III½ ('bipolarity masked and unmasked by stimulant abuse'), and bipolar IV (MDE and hyperthymic temperament). Ghaemi et al. proposed the new diagnosis of bipolar spectrum disorder.[3] This diagnosis included a unipolar major depressive disorder (MDD; no history of mania/hypomania) plus bipolar signs, including family history of bipolar disorder and antidepressant‐induced mania/hypomania (the most important ones, according to the authors), many MDE, atypical features of depression, and onset of the first MDE before age 25. Akiskal reported that bipolar family history was the most important validating criterion of the bipolar spectrum.[[5]] Family history was also the most important criterion for validating a psychiatric diagnosis according to Robins and Guze.[7] The soft bipolar spectrum of Angst et al. included bipolar II, minor bipolar disorders (hypomania and mild depressions), and hypomania only (no depression).[4] A new list of mood disorders that could be included in the bipolar spectrum has been recently reported, including depression with bipolar family history ('pseudo‐unipolar depression').[6] In the same paper by Akiskal clinical validating features for including mood disorders in the bipolar spectrum were listed, including bipolar family history, many MDE, young age of onset, and depressive mixed state (DMX: an MDE plus some concurrent hypomanic symptoms not meeting full criteria for hypomania).[6] Classic features often reported to distinguish bipolar versus unipolar disorders were young age of onset, many recurrences, atypical features of depression, and bipolar family history.[[5], [8]] Recent studies supported the bipolar nature also of DMX.[[11]]

Major depressive disorder with bipolar family history could be included in the bipolar spectrum if it had at least some of the bipolar signs reported in the previous section. It could be clinically useful to study this subtype of depression if it were found to be not uncommon in clinical practice. Major depressive disorder with bipolar family history does not include the bipolar (especially the bipolar II[[3], [5]],[8]) misdiagnosed as MDD (at least 40%) because of the common clinical mistake (and also for the clinical difficulty) of overlooking prior hypomanic episodes, or the problem of MDD reclassified to bipolar disorder when mania/hypomania develops later.[8] One problem increasing the frequency of MDD misdiagnoses is related to the DSM‐IV‐TR itself, requiring a minimum duration of hypomania of 4 days for the diagnosis of bipolar II, which is a cut‐off not based on data while a cut‐off of 2 days is supported by data.[[5], [15]],[16] Another problem is related to antidepressant‐associated hypomania, which is not classified under bipolar II in DSM‐IV‐TR, while follow‐up studies found that most of these patients later had spontaneous hypomania or mania.[[2], [5]],[16]

A subgroup of unipolar with bipolar family history was reported in studies conducted before the 1990s (before DSM‐IV).[[5], [8]] This unipolar subgroup was found to have bipolar signs such as young age of onset, many recurrences, hypersomnic‐retarded depression (but also agitated depression), pharmacological hypomania, and lithium response. Limitations of these studies were the lack of formal diagnostic criteria in some of them, outdated definitions of bipolar II, tertiary‐care settings (meaning samples of more severe patients), and interviews by non‐clinicians.[8]

The present study on unipolar MDD with bipolar family history has several advantages compared to these older studies on the same topic. Modern diagnostic criteria (DSM‐IV) were used, a modern definition of bipolar II disorder was followed, a new family history interview was used[17] that can assess bipolar I and bipolar II, the study was conducted in a non‐tertiary care setting, the interview was conducted by a clinician trained in bipolar II diagnosis and using the structured interview for DSM‐IV,[18] and DMX was used as a new bipolar sign.

The purpose of the present study was to find if unipolar MDD with bipolar family history could be included in the bipolar spectrum.

The present study was conducted by the author, a senior clinical and mood disorder research psychiatrist, at the Outpatient Psychiatry Private Center in Ravenna and Forli, Italy (a University of California at San Diego (USA) collaborating center). Private practice is more representative of mood disorder patients in Italy (maybe also in the USA[19]), where national mental health services and university centers usually treat the most severe patients. Most individuals can be seen by a private psychiatrist (reducing a possible selection bias related to income). Many authorities now believe that mood disorder patients in tertiary‐care centers are not representative of the patients usually treated in clinical practice.[[2], [20]] The author's interrater agreement was checked. Some just interviewed MDE patients (n = 15) were recalled in 48 h, and re‐interviewed by a second psychiatrist (blind to the author's assessment results), from the same department and with a busy private practice (in order to carry out the test with a psychiatrist visiting private practice patients, who were the patients included in the present study). In the past that psychiatrist had also done reseach studies with the author. The psychiatrist, before the assessment of patients, was trained by the author about study interview methods and the diagnosis of bipolar II disorder. The author checked the quality of the training by taking part in some interviews with the second psychiatrist. These patients were not included in the analysis of interrater agreement. The kappa‐statistic results were the following: agreement, 86.6%; expected agreement, 50.2%; kappa = 0.73, z = 2.8, and P = 0.0023, showing substantial (0.61–0.80) agreement (according to the stata 7 reference manual, the statistical software used in the present study.

A sample of 280 consecutive bipolar II outpatients and a sample of 135 consecutive unipolar MDD outpatients, presenting spontaneously for MDE treatment, were included in the last 3 years. Informed consent was obtained after the procedure had been fully explained. Only patients presenting without psychopharmacotherapy for at least 2 weeks before intake assessment were included, in order to avoid the inclusion of antidepressant‐induced mixed states.[2] Current substance abuse and severe personality disorder patients were not included because this could confound the diagnosis of bipolar II and mixed states.[16] Clinically significant general medical illness and dementia patients were not included. Patients were interviewed by the author during the first visit with the Structured Clinical Interview for DSM‐IV Axis I Disorders–Clinician Version (SCID‐CV)[17] and the Global Assessment of Functioning scale.[1] Bipolar I and bipolar II family history of patients' first‐degree relatives was investigated with the structured Family History Screen[17] by interviewing the patient and often also one close relative. The SCID‐CV is partly semistructured and based on clinical evaluation. Clinicians using semistructured interviews made more correct diagnoses of bipolar II than interviewers using structured interviews.[[8], [23]],[24] Patients were systematically SCID‐CV interviewed for history of manic/hypomanic episodes and for hypomanic symptoms during the index MDE. Hypomanic symptoms during the MDE had to last at least 1 week, appeared during the MDE, and were present at the time of the interview. The SCID‐CV structured question on racing thoughts was supplemented by the Koukopoulos and Koukopoulos definition of crowded thoughts (head continuously full of ideas that the patient is unable to stop)[25] because it was found easier to understand by patients, and to broaden the assessment of mental overactivity. The SCID‐CV skip‐out instruction of the stem question about past hypomanic mood was not followed because it was shown that systematic assessment of all past hypomanic symptoms increased the frequency of bipolar II diagnoses.[26] History of mania/hypomania was always investigated soon after the diagnosis of MDE before the assessment of study variables, in order to avoid a possible bias related to the knowledge of indicators of bipolar disorders. Depressive mixed state was defined as an MDE plus ≥3 concurrent hypomanic symptoms, following Akiskal and Benazzi's definition.[13] The DSM‐IV 4 day minimum duration of hypomania for bipolar II diagnosis (a cut‐off not based on data[27]) was not followed. Instead, at least 2 days of hypomania were required, on the basis of previous studies supporting this cut‐off.[[16], [28]] Most present study bipolar II patients had had more than one hypomania (increasing reliability).[16] Often, family members or close friends supplemented the clinical information during the interview (increasing reliability).[16]

Clinical variables used to validate the inclusion of unipolar MDD with bipolar family history in the bipolar spectrum were young age of onset of the first MDE, many MDE recurrences (>4), DSM‐IV atypical features of MDE, and DMX (MDE + >2 concurrent hypomanic symptoms), following previous studies reporting that these variables were typical features of bipolar disorders.[[3], [5], [8], [26], [29], [34]]

Means were compared by t‐test and frequencies by χ2 test.[36]stata Statistical Software, Release 7, was used (Stata, College Station, TX, USA; 2001). A two‐tailed P < 0.05 was chosen as α level.

Unipolar MDD with bipolar family history was present in 27 MDD patients (27/135, 20.0%). Bipolar family history was present in 53.7% of bipolar II patients. Comparisons among unipolar MDD with bipolar family history (UP+BPFH), unipolar MDD without bipolar family history (UP–BPFH), and bipolar II (BPII), on typical bipolar variables, are presented in Tables 1,2. The UP+BPFH versus UP–BPFH had significantly lower age, lower age of onset, fewer recurrences, and more DMX. The UP+BPFH versus BPII had no significant differences, apart from recurrences; UP–BPFH versus BPII had significantly different age, age of onset, recurrences, atypical features, and DMX.

1 . Comparisons among UP+BPFH, UP–BPFH, and BPII MDE on typical bipolar variables by t‐test or χ 2 test

1 MDD, major depressive disorder; UP+BPFH, unipolar MDD with bipolar family history; UP–BPFH, unipolar MDD without bipolar family history; BPII, bipolar II; GAF, Global Assessment of Functioning scale; MDE, major depressive episode.

2 . T/χ 2 test results

2 GAF, Global Assessment of Functioning scale; MDE, major depressive episode.

Unipolar MDD with bipolar family history had a clinically significant 20.0% frequency in the unipolar MDD sample, supporting the clinical usefulness of this proposed depression subtype. Unipolar MDD with bipolar family history had significantly lower age of onset, fewer recurrences, and more DMX (bipolar variables) versus UP–BPFH. Unipolar MDD with bipolar family history versus BPII had no significant differences on bipolar (onset, atypical features, DMX) variables, apart from recurrences. Unipolar MDD without bipolar family history versus BPII comparisons were all significantly different on all bipolar (onset, recurrences, atypical features, DMX) variables. These findings suggest that UP+BPFH shows many bipolar signs and that it is close to BPII. It could therefore be included in the bipolar spectrum. Age of onset is an important variable supporting the subtyping of mood disorders[[6], [10]],[37] and DMX was found to be a marker of bipolar II disorder.[[11]] Family history is an important variable supporting diagnostic validity and subtyping of mood disorders[[5]] Previous family studies supported the inclusion of unipolar with bipolar family history in the bipolar spectrum.[[5]],[[8], [38]] Winokur et al. found that unipolar with bipolar family history had a lower age of onset than unipolar without bipolar family history.[41] A recent review by Joffe et al. found clinical, course of illness, family history, biology, and treatment response data supporting a link between bipolar depression and unipolar depression.[42] The present study frequency of bipolar disorders among first‐degree relatives of bipolar II patients was in line with previous studies on the family history of bipolar patients.[[43]] Findings of previous studies on the same topic[[5], [8]] were replicated in the present study by using updated and improved study methods. As in the studies conducted before the 1990s, unipolar with bipolar family history was found to have a different response to tricyclic antidepressants and to lithium compared to unipolar without bipolar family history,[[46]] the present study findings may have important implications for the treatment of MDD and for treatment‐resistant MDD.[[3], [5]] It was found that MDD failure to many antidepressants could be a bipolar sign,[[3], [6]] a finding that would shift the treatment toward the use of mood stabilizers. Akiskal and Mallya reported on a subgroup of antidepressant treatment‐resistant MDD responding to mood stabilizers,[51] and Koukopoulos and Koukopoulos reported on agitated MDD made worse by antidepressants and responding to antipsychotics.[25] A bipolar family history could be found in at least some treatment‐resistant MDD, and could be a marker suggesting different treatment approaches. Because MDD has a probability of responding to a first antidepressant of approximately 50%, and nearly 30% of MDD may lose response with time,[52] subtyping of MDD beyond DSM‐IV‐TR is required to improve treatment outcome of MDD. Bipolar depression misdiagnosed as MDD could have negative effects by antidepressants, such as induction of mixed states and switching to mania/hypomania.[53] Bipolar family history could be used to subtype MDD. Surprisingly, family history is never included as a criterion in the classification of mood disorders in DSM‐IV‐TR. Studies with the new antidepressants and mood stabilizers are required to support the subtyping of MDD on the basis of bipolar family history, in order to show if this subtyping may have treatment impact. Limitations of the present study should be discussed. Use of a single interviewer limited the validity of the findings. However, an interviewer's bias is unlikely because the present study variables were part of a much larger set of variables systematically assessed and recorded during the first visit (for MDE) of each new patient during the last years. Therefore, the present study goal was not known to the author when the data were collected. The author's interrater agreement was found to be substantial. History of mania/hypomania was always investigated soon after the diagnosis of MDE before the assessment of study variables, avoiding a possible bias related to the knowledge of indicators of bipolar disorders. The study sample features had typical bipolar features, supporting the validity of the interview. The interview was conducted by a clinician studying and treating mood disorders for a long time, using validated structured/semistructured interviews, information from key informants, and systematically probing for past hypomania. The SCID‐CV is partly semistructured and based on clinical evaluation (not on simple yes/no answers to its structured questions). Trained clinicians using semistructured interviews made more correct diagnoses of bipolar II compared to structured interviewing[23] and more valid assessment of mood disorders than structured interviews by non‐medical interviewers.[24] These study features may have reduced study limitations.[[8], [16]] Some advantages of the present study could be the inclusion only of outpatients, no concurrent psychopharmacotherapy at assessment, no substance abuse disorders, inclusion of a large sample of bipolar II disorder patients, and a non‐academic, non‐tertiary care setting.