We collected information on first- and second-degree relatives of 112 subjects (35 men and 77 women), diagnosed as having Major Depression, Recurrent (UP) (15 men and 41 women), and Bipolar Disorders (BP) (20 men and 36 women), according to DSM-III criteria (1). All the probands were patients being followed at the Institute of Clinical Psychiatry of the University of Milan. These patients represented an overlapping subset of those discussed in previous reports (4,14). Naturally, only subjects for whom complete information about second- degree relatives and tricyclic antidepressant drag treatment outcome was available could be included in the study. Each patient participated in at least one clinical trial, although most of them took part in several trials with different antidepressant drags of the classical tricyclic type only (chlorimipramine, imipramine, amitriptyline, and nortriptyline). Each trial consisted of 4 weeks of treatment with full dosages (3,12). Responders were the patients who always showed improvement under the tricyclic treatment. This group consisted of 63 probands, 18 men (UP = 6; BP = 12) and 45 women (UP = 29; BP = 16). We classified as nonresponders the patients who did not improve irrespective of the administered drag; they were 17 probands, 9 men (UP = 4; BP = 5) and 8 women (UP = 2; BP = 6). We defined as poor responders the patients who only showed a slight improvement and/or the patients who had poor outcome on treatment with one of the abovementioned tricyclics and a good outcome on treatment with another one. This last group consisted of 32 probands, 8 men (UP = 5; BP = 3) and 24 women (UP = 10; BP = 14). The demographic and clinical information on the families of our patients such as sex, present age, presence of secondary cases of Affective Disorders, and age at onset was collected by direct interviews with the patients and their relatives. Data on secondary cases in families were confirmed by hospital records, in instances in which the subjects previously had been admitted to a psychiatric institute, or by asking the psychiatrists for additional information. Relatives considered to have Major Affective Disorders were those meeting the DSM-III criteria for Major Depression, Recurrent, and Bipolar Disorder. Affective Spectrum Disorders were diagnosed in cases in which the relative met the DSM-III criteria for Dysthymic Disorder, Cyclothymic Disorder, or Atypical Depression. Since we were studying the segregation of susceptibility to Affective Disorders and not the segregation of the disorders themselves, we sought to identify families susceptible to Affective Disorders, regardless of their phenotypical manifestations (both Major Affective Disorders and Affective Spectrum Disorders were included). We submitted our data to segregation analyses, performed by means of Lalouel and Morton’s POINTER program (7). For purposes of the analysis, the nuclear family unit was broadened to include family members who have led to its ascertainment. This procedure introduces the use of the POINTER concept, that is, a relative of extreme phenotype (an affected relative) outside the nuclear family to which it points. The program assumes that each component of a nuclear family (father, mother, and their common offspring) may have only one pointer, a given pointer being assigned only to the closest component of the nuclear family. Information about a pointer includes its phenotype and relationship to the family members. We adopted conditional likelihoods to calculate the likelihoods of sibships in nuclear families conditioned to pointers as well as on parents. The threshold model requires population prevalences so that liability thresholds can be defined. We incorporated separate prevalences for males and females. Since the probability of illness changes with age and assuming that age of onset is accurately recorded and does not affect ascertainment, we formulated a continuous liability scale defined by 10 age categories, five for males and five for females. We subdivided the total population prevalence of the disorder according to the proportion of onset in these five classes of age (ranging between 10 and 14, 15 and 25, 26 and 37, 38 and 49, and more than 50 years). We assumed uniform risk distribution within each liability class. The analyses were performed on the SPERRY UNI VAC 1100/90 Computer of the University of Milan.
