1. American Psychiatric Association. (1980): Diagnostic and Statistical Manual of Mental Disorders, 3rd. ed. American Psychiatric Association, Washington, D.C. 2. Angst, J. (1985): In: Drug Treatment of Neurotic Disorders: Focus on Alprazolam, edited by M. H. Lader and H. C. Davies, pp. 70-75. Churchill Livingstone, New York. 3. Barlow, D. H., Di Nardo, P. A., Vermicyea, В . В ., et al. (1986): J. New. Ment. Dis., 174:63-72. 4. Beck-Friis, J., Kjellman, В ., Aperia, F„ et al. (1985): Acta Psychiatr. Scand, 71:319-330. 5. Beckman, H., and Goodwin, F. K. (1980): Neuropsychobiology, 6:91-100. 6. Bremer, A., Charney, D. S., and Heninger, G. R. (1984): Arch. Gen. Psychiatry, 41:1129-1135. 7. Brown, G. L., Goodwin, F. K., and Bunney, W. E. (1982): In: Serotonin in Biological Psychiatry, edited by B. T. Ho, J. С Schoolar, and E. Usdin, pp. 287-307. Raven Press, New York. 8. Brown, W. A., and Shuey, I. (1980): Arch. Gen. Psychiatry, 37:737-751. 9. Buchsbaum, M. S., Mayer, D. J., and Murphy, A. L. (1977): Acta Psychiatr. Scand, 56:69-79. 10. Bunney, W. E., Fawcett, J. A., Davies, J. M., et al. (1969): Arch. Gen. Psychiatry, 21:138-150. 11. Carrol, B. J. (1979): In: Lithium: Controversies and Unresolved Issues, edited by Т . В . Cooper, S. Gershon, N. S. Kline, and M. Schou, pp. 171-197. Excerpta Medica, Amsterdam. 12. Cassano, G. B. (1983): In: The Benzodiazepines: From Molecular Biology to Clinical Practice, edited by E. Costa, pp. 287-293. Raven Press, New York. 13. Cazzullo, С L., Sacchetti, E., Allaria, E., et al. (1982): In: Typical and Atypical Antidepressants. Clinical Practice, edited by E. Costa and G. Racagni, pp. 237-247. Raven Press, New York. 14. Cazzullo, С . L„ Sacchetti, E., Vita, A., et al. (1984): IRCS Med. Set., 12:917-918. 15. Cazzullo, С L., Smeraldi, E., and Sacchetti, E. (1979): Prog. Neuropsychopharmacol. Biol. Psy­chiatry, 3:25-38. 16. Charney, D. S., and Nelson, J. С (1981): Am. J. Psychiatry, 138:328-333. 17. Conte, G., Vita, A., Alciati, A., et al. (1984): In: Current Trends in Lithium and Rubidium Therapy, edited by G. U. Corsini, pp. 211-221. MTP Press, Lancaster. 18. Covi, L., and Lipman, R. S. (1984): Clin. NeuropharmacoL 7S 1:924-925. 19. Fawcett, J. A., Maas, J. W., and Dekirmenjan, H. (1972): Arch. Gen. Psychiatry, 26:246-251. 20. Feighner, J. P., Robins, E., Guze, S., et al. (1972): Arch. Gen. Psychiatry, 26:57-63. 21. Freud, S. (1894): Collected Works, I. Hogarth Press and Institute of Psychoanalysis, London. 22. Glassman, A. H., Kantor, S. J., and Shostak, M. (1975): Am. J. Psychiatry, 132:716-719. 23. Goldberg, D., Rickels, K, Downing, R., and Hesbacher, P. (1976): Br. J. Psychiatry, 129:61-67. 24. Goldberg, D., and Simpson, N. (1986): In: Drug Treatment of Neurotic Disorders: Focus on Alprazolam, edited by M. H. Lader and H. С Davies, pp. 76-82. Churchill-Livingstone, New York. 25. Goodwin, F. K., Cowdry, R. W., and Webster, M. H. (1978): In: Psychopharmacology: A Gen­eration of Progress, edited by M. A. Lipton, A. DiMascio, and K. F. Killam, pp. 1277-1288. Raven Press, New York. 26. Hamilton, M. (1960): Neurol. Neurosurg. Psychiatry, 23:56-62. 27. Hamilton, M. (1983): Br. J. Clin. Pharmacol., 15:165S-169S. 28. Invernizzi, G, and Sacchetti, E. (1985): In: Proceedings of the IVth National Congress of the Italian Society of Biological Psychiatry (Pavia, Italy, 19-20 October 1985), edited by G. Nappi, P. Pinelli, and D. Kemali, pp. 75-80. Emi Ras, Pavia, Italy. 29. Kraepelin, E. (1921): Manic Depressive Insanity and Paranoia. E. & S. Livingstone, Edinburgh. 30. Krauthammer, C, and Klerman, G. (1979): In: Manic Illness, edited by B. Shopshin, pp. 11-28. Raven Press, New York. 31. Maas, J. W. (1983): MHPG: Basic Mechanisms and Psychopathology. Academic Press, New York. 32. Maas, J. W., Fawcett, J. A., and Dekirmenjian, H. (1968): Arch. Gen. Psychiatry, 19:129-134. 33. Mendels, J., Weinstein, N.. and Cochrane, С (1972): Arch. Gen. Psychiatry, 27:649-653. 34. Montgomery, S. A., and Montgomery, D. B. (1982): In: Typical and Atypical Antidepressants: Clinical Practice, edited by E. Costa and G. Racagni, pp. 347-358. Raven Press, New York. 35. Nasrallah, H. A., McCalley-Whitters, M., and Chapman, S. (1984): Am. J. Psychiatry, 141:919. 36. Nelson, J. C, and Bowers, M. B. (1978): Arch. Gen. Psychiatry, 35:1321-1328. 37. Nelson, W. H., Khan, A., and Orr, W. W. (1984): J. Affective Disord., 6:297-306. 38. Owen, F., Cross, A. J., Crow, T. J., et al. (1983): Lancet, ii:1286. 39. Raskin, A., Schulterbrandt, J. G., Boothe, H., et al. (1972): In: Recent Advances in the Psycho-biology of the Depressive Illnesses, edited by T. A. Williams, M. M. Katz, and J. A. Shield, pp. 315-329. U.S. Government Printing Office, Washington, D.C. 40. Roose, S. P., Glassman, A. H., Walsh, В . Т ., et al. (1983): Am. J. Psychiatry, 140:1159-1162. 41. Rosenbaum, A. H., Toshihiko, M., Schatzberg, A. P., et al. (1983): Am. J. Psychiatry, 140:314-318. 42. Sacchetti, E., Allaria, E., Negri, F., et al. (1979): Biol. Psychiatry, 14:473-484. 43. Sacchetti, E., Conte, G, Pennati, A., et al. (1985): J. Psychiatr. Res., 19:579-586. 44. Sacchetti, E., and Smeraldi, E. (1980): In: La Depressione come Problema Psicobiologico, edited by E. Smeraldi and E. Sacchetti, pp. 49-79. Edi Ermes, Milan. 45. Sacchetti, E., Vita, A., Conte, G., et al. (1984): In: Current Trends in Lithium and Rubidium Therapy, edited by G U. Corsini, pp. 165-187. MTP Press, Lancaster. 46. Sacchetti, E., Vita, A., Conte, G, et al. (1986): IRCS Med. Scl, 14:407-412. 47. Schatzberg, A. F., Orsulak, P. J., Rosenbaum, A. H., et al. (1983): In: Treatment of Depression: Old Controversies and New Approaches, edited by P. J. Clayton and J. E. Barrett, pp. 53-59. Raven Press, New York. 48. Schulsinger, F„ Kety, S. S., Rosenthal, D., and Wender, P. H. (1979): In: Origins, Prevention and Treatment of Affective Disorders, edited by M. Schou and E. Stromgren, pp. 277-287. Ac­ademic Press London. 49. Smeraldi, E., Gasperini, M., Macciardi, F., et al. (1983): J. Psychiatr. Res., 17:309-317. 50. Stanley, M. (1983): Lancet, i:214. 51. Targum, S. D., Rosen, L. N., DeLisi, L. E., et al. (1983): Biol. Psychiatry, 18:329-336. 52. Traskman, L., Asberg, M., Bertilsson, L., et al. (1981): Arch. Gen. Psychiatry, 38:831-836. 53. Vita, A., Sacchetti, E., Conte, G, et al. (1985): L’Encephale, 11:71-77. 54. Weissmann, M. M., Leckmann, J. F., Merikangas, K. R., et al. (1984): Arch. Gen. Psychiatry, 41:845-852. 55. Winokur, G. (1978): In: Psychopharmacology: A Generation of Progress, edited by M. A. Lipton, A. DiMascio, and K. F. Killam, pp. 1213-1221. Raven Press, New York. 56. Zis, A. P., Grof, P., and Goodwin, F. K. (1979): In: Lithium: Controversies and Unresolved Issues, edited by T. B. Cooper, S. Gershon, N. S. Kline, and M. Schou, pp. 381-398. Excerpta Medica, Amsterdam. /> (more…)

Differences in individual responses to psychopharmacological therapy may be assumed to be phenotypical indicators of distinct underlying pathogenetic dis­orders if it is also certain that these differences are not reflecting pharmacokinetic factors. This assumption has been the basis of many significant efforts to verify whether patients who do and do not respond to various antidepressants or lithium salts can be classified into actually distinct subgroups of the major affective dis­orders. As far as lithium therapy is concerned, the consensus is that lithium responders have more relatives with major affective disorders than do nonresponders, thus suggesting that familial factors are important in distinguishing between the two subgroups (11,15). The differences between the two, however, do not appear to be limited to this factor alone. We are beginning to gather evidence (even though further studies will be required) that lithium nonresponders appear to be more likely to be high urinary MHPG excretors and to have ventricular enlargement; further, they are more likely to have had a late disease onset, a history of delusional depression, suicidal behavior, short episode cycling, and drug-induced mood switches (11,17,43,46). In addition, they seem to have qualitatively distinct pat­terns of response to tricyclic antidepressants (45). We compared the outcome to lithium prophylaxis of 14 definitely anxious depressive patients with that of 18 definitely nonanxious major depressives. The differences between the responsiveness of the two groups to long-term lithium treatment were very striking. None of the definitely anxious patients had been satisfactorily stabilized, whereas over 22% of the definitely nonanxious, 13 of the total of 18, had clearly benefited from the treatment (x 2 = 14.2; p < 0.001). Undoubtedly it would be both reasonable and easy to say that the real strength of the association between a poor lithium response and the presence of a definitely anxious pattern of depression is really weaker than that observed in this particular patient sample. Despite this, however, the results obtained still seem promising, not only from the theoretical viewpoint, but also from the clinical one. Knowing that it is possible to have a reliable early screening of major depression with different lithium treatment prospects, we can more readily plan alternative in­dividualized stabilizing programs on a rational, less hit-or-miss basis. A sizable body of evidence indicates that age at disease onset is a good index for recognizing distinct subtypes of major affective disorder. As compared to patients with early onset, those with late onset have been shown to have: higher MHPG excretion (5,13,42,53); lower alpha-2-adrenoceptor density upon platelets (43); enlarged ventricular size (14,28); a poorer response to long-term lithium therapy (45,46); fewer relatives with major affective disorders (49,55); lowered chances for manic episodes (30); an increased recurrence rate (56); and, when bipolar, higher chances for drug-induced switches into mania (17). In view of the above, it seemed worthwhile to us to explore whether there is a relationship between age at onset and definite anxiety in major depression. Among the 88 patients for whom it was possible to securely ascertain age at onset, those classified as definitely anxious had their first depressive episode later than those who were definitely nonanxious (39.8 ± 11.9 vs. 33.5 ± 12 years; Student Mest = 2.41, p < 0.02). Further, the dichotomy between early and late onset patients (cut-off age: 40 years) provided a satisfactory basis for distinguishing between major depressives with anxiety symptoms from those without anxiety. As many as 27 of the 35 late onset patients and only 22 of the 49 early onset patients were anxious (X 2 = 8.7; p < 0.003). These results do not appear to be mere indirect effects of age or polarity: the anxious and nonanxious major depressives were of comparable age (51.5 ± 10.3 vs. 50.3 ± 11.1 years), and there was an analogous incidence of bipolar disorders among them (34% and 53%, respectively; x 2 = 3.07, p = NS). Delusions On the whole, research on delusional depression strongly points to the pos­sibility that this term actually does denote and circumscribe a specific subtype of the major affective disorders. This not only is characterized by extremely virulent symptoms (evidence compatible with the hypothesis of a severity con­tinuum in depression), but also by specific biological and pharmacological fea­tures. As compared to nondelusional depressives, those with delusions not only are at higher risk for suicidal behavior (40), but also have greater chance for both abnormal dilation of the lateral cerebral ventricles (28,51) and a dysfunction of hypothalamic-pituitary-adrenal axis activity (16). Further, delusional depres­sives show a poor response to tricyclic antidepressants (TAD) given alone, benefit from TAD neuroleptics cotherapies or electroconvulsive therapy (22,36,37), and do not become adequately stabilized by lithium therapy (46). This picture of delusional depression aroused our interest in attempting to determine whether definite anxiety and delusion in major depressives are inter­twined phenomena. In the group of 68 patients studied, only 3 of the 31 (9.6%) nonanxious depressives were also unerringly delusional. Clear delusional states, however, were found in 12 of the 37 (32%) definitely anxious patients (x 2 = 5.08, p = 0.023). In the latter group, the relative risk of mood-congruent delusions was 4.5 times higher than that in the former. These data clearly indicate that the anxiety-delusions association in major depression is not merely fortuitous, despite the high incidence of false positives (37%), which is at least partially secondary to an overrepresentation of anxious versus delusional cases in the total study group.

The results of the various studies described now need to be discussed within a general framework. The first thing to highlight is that the associations were obtained from relatively small groups of highly selected patients in naturalistic settings. Consequently, supplementary controlled studies that also include less extreme cases will be necessary before generalizations become possible. Further, even in the case of the apparently more straightforward associations, it is evident that we are dealing with partial associations since there is a considerable degree of overlap between definitely anxious and definitely nonanxious major de-pressives. Because our work was specifically addressed to definite anxiety in major depression and not to the general occurrence of anxiety symptoms, extreme caution should guide comparisons with the findings from studies based on dif­ferent theoretical and experimental approaches. Despite these limitations, the accumulating biochemical, pharmacological, and clinical data in this area already appear to support the hypothesis that definite morbid anxious patterns phenotypically aggregate a particular subtype of de-pressives who are quite likely to have specific identifiable biological dysfunctions. Major depressives who also have morbid anxious symptoms were found to have a relatively high urinary MHPG excretion, poor response to lithium therapy, a late disease onset, and a history of delusions and suicidal behavior. As compared to this type of patient, the prototypical major depressive, relatively protected from anxiety, has a rather low MHPG excretion, no affective relapses during long-term lithium treatment, an early disease onset, and no history of suicide attempts or delusional symptoms. The clinical picture of anxious depressives is not really surprising. Its similarity to the classical Kraepelinian descriptions of involutional melancholia (29) im­mediately comes to mind. Further, it is worth noting that a similar symptom-grouping resulted from an earlier attempt (39) to differentiate discrete subgroups of depressed patients by factor analysis: one of these subgroups was characterized by high psychic and somatic anxiety scores, suicidal ideation, psychotic features, and a late disease onset. On the other hand, the fact that our definite morbid anxiety patients generally were the most impaired can also be viewed as a specific case of prior clinical evidence indicating that the more severe the depression, the more marked is the patient’s anxiety level (3,23,24,33). Consequently, our findings also can broadly represent further evidence supporting the hypothesis that anxiety and depression are two correlated dimensions (24). This is true, however, only if we limit our­selves to a purely descriptive clinical context. Our current knowledge of the biological correlates of the clinical aspects we found to be associated to anxiety symptoms, together with the findings on MHPG excretion and lithium therapy among anxious depressives, points to a substantial pathogenetic discontinuity among subgroups of depressed patients who appear to have a varying vulnerability to concomitant anxiety. Obviously, the research that has been undertaken to date is not sufficient for allowing a definitive answer to the question of whether the biological differences found really reflect distinct subtypes of major depression or if they are only indicative of a few main typologies. We shall have definite answers when we conduct studies that test, through stepwise, cluster, or multivariate techniques, whether we are observing first- or second-order associations or we are looking at additive or interactional effects. Yet, there is some indirect evidence that we can address at this juncture. A relatively high MHPG excretion was associated with a poor response to lithium therapy (45,46) and with a late disease onset (5,13,42,53). In turn, a poor lithium prognosis was associated with a late disease onset (45,46) and with delusional depression (46) and suicidal behavior (45,46). Further, delusional depression was associated with an increased probability of suicide attempts (40). Conse­quently, it seems reasonable to assume that many of the variables found to be associated with a definitely anxious major depression are indices (at different phenotypical levels) of a few main typologies. If further study definitively confirms this hypothesis, then the clinical category anxious depression might take on novel and stimulating meaning for the bio­logically oriented researcher as well, now often thinking of it as a label that the clinician sometimes abuses. ACKNOWLEDGMENTS This research was supported in part by the C.N.R. grant no. 86.01935.56. The authors thank Dr. June Shmelzer La Rosa for her precious assistance in revising the manuscript.

Independent information from epidemiological, clinical, biological, and phar­macological research impels formulation of the hypothesis that suicidal tendency represents a behavioral characteristic typically concentrated in specific subgroups of major depressives. Epidemiologically, the tendency for suicide attempters to repeat self-destructive acts is a rule rather than an exception. Clinically, suicide proneness has been shown to be associated with delusional depression (40). When major depressive suicide attempters were biologically compared to nonattempters, it was found that the patients in the former group had: a different genetic makeup (48); critically lowered CSF concentrations of 5-HIAA (7,52) and possibly HVA (34); increased chances for higher 17-OH-corticosteroid urinary concentration (10); lowered MAO platelet activity (9); higher melatonin plasma levels (4); ce­rebral ventricular enlargement (28,35); and augmenting average evoked responses (2). Further, in post-mortem studies of specific brain areas suicide attempters present either subnormal 5-HT concentrations (7) or supranormal densities of 5-HT 2 receptors (38,50). Pharmacologically, a history of suicidal behavior seems to have predictive value for a poor response to lithium therapy (45,46). Assuming suicide proneness as a marker of a distinct subgroup of major af­fective disorders, we analyzed the distribution of autoaggressive antecedents among definitely nonanxious major depressives. Previous suicidal behavior was indeed specific to the anxious depressives. Fourteen of the 15 patients composing our group of suicide attempters also were definitely anxious depressives whereas the relationship between anxiety-no anx­iety in the group of 60 nonattempters was exactly 1 (x 2 = 7.59, p = 0.006; odds ratio = 14). Since the information on suicidal behavior was collected after the cases were classified as anxious or nonanxious, the observed association cannot now help us to deal with the issue of whether suicidal behavior represents a consequence of collapsing anxious defenses. What we can say is that anxiety and suicidal behavior may have generally common or similar pathogenetic determinants and a partially interchangeable functional significance, for example, the patient’s attempt to alleviate his suffering from depression.

Studies undertaken during the late sixties (19,32) report a reduction in urinary MHPG excretion in depressed patients as compared to normal controls. This peripheral index undoubtedly is one of the most frequently used tools in research on the catecholaminergic correlates of affective disorders (31,43,44,53). Even if there is still some uncertainty about this measure, it seems sufficiently reliable for us to use it as a basis for subdividing major affective disorders into two fundamental subtypes, the relatively low and high groups. We ourselves found a bimodal MHPG excretion distribution among 153 drug-free patients with major affective disorders (Conte et al., unpublished report). Furthermore, and more importantly, relatively low and high excretors differ in age at onset (5,13,42,53), the number of platelet alpha-2-adrenoceptors (43), urinary Cortisol levels (41), cerebral ventricular size (53), clinical responsiveness to both antidepressant drugs (31,44,47,53) and long-term lithium treatment (45,46), and, possibly, a proclivity to develop mania (31). As for the MHPG relationship with symptoms, the emerging picture is that none of the latter in itself is sufficiently able to influence MHPG excretion so that the division of the patients into low and high excretors is appreciably affected, despite the fact that a slight effect on MHPG cannot always be excluded (31,42,44). This general conclusion also appears to be applicable to anxiety. As previously noted, studies focused on anxiety symptoms may inevitably unify mere situational phenomena with well-established traits. At the same time, it is extremely difficult to take into account the possibility that these traits sometimes can be effectively masked by positive chance circumstances and thus elude clinical detection. Con­sequently, studies of the relationship between MHPG excretion and anxiety in depression have not addressed the issue of whether the biological phenotypes of low and high excretors indicate a markedly different susceptibility to anxiety. With the latter as the investigational framework, we compared the MHPG excretion of 29 definitely anxious patients (14 males, 15 females) with major affective disorders with that of 16 definitely nonanxious ones (4 males, 12 fe­males). Since our general sample median MHPG value for males tended to shift to the right of that for the females, we used sex as a supplementary variable in the two-way ANOVA analysis of the present data. This indicates that being both male and definitely anxious positively, but independently, affects MHPG excre­tion (sex effect: F (1,41 ) = 1535, p = 0.001; anxiety effect: F (1,4 i) = 6AS,p = 0.016; interaction: F (l,41) = 0.03, p = NS). Independent nonparametric analyses after division of the patients into defi­nitely low and definitely high MHPG excretor groups (cut-off points: the median values of 1,650 Mg/24 hr for males and 1,119 /ug/24 hr for females) clearly indicate that the association between high MHPG excretion and anxiety is a strong one. The finding that 88.5% of the 26 high excretors (23 patients) were definitely anxious and that 32% of the 19 low excretors (6 patients) were definitely non-anxious means that the relative risk of anxiety is fully 16.6 times greater in the former than in the latter biochemically defined group. Thus, anxiety traits unrelated to the chance circumstances surrounding any particular episode are not randomly distributed among patients with major de- pressive disorders. On the contrary, they are strongly associated with readily distinguishable biochemical patterns.

Institute of Psychiatry, University of Milan, School of Medicine, 20122 Milan, Italy
In 1894 Freud (21) proposed the separation of anxiety neurosis from neuras­thenia. Since then, there has been intensive theoretical and extensive experimental work on the relationship between anxiety and depression. Nonetheless, we still cannot firmly conclude whether anxiety and depressive disorders are distinct entities. Our clinical experience, however, has unquestionably demonstrated that symptoms of anxiety very frequently color the picture of depressive disorders, so much that the label "anxious depression" is widely used in spite of the fact that it is not a category recognized by most of the latest diagnostic systems. The presence of symptoms of both anxiety and depression in clinically de­pressed patients definitely does not appear to be confined exclusively to depressive neurosis or to other minor types of depression. Anxiety is also present in patients that are readily diagnosed as having either some of the traditionally defined disorders, such as manic-depressive psychosis or endogenous depression, or the operationally defined major affective disorders. A few examples will suffice to confirm the presence of anxiety in major depres­sion. Factor analysis (12) has shown that, at the very least, the anxiety and depressive factors loaded equally in over 65% of the cases of a sample of patients with endogenous depression. Similarly, a recent study (2) demonstrated that as many as 25% of the patients with major depression and 10% of those with brief recurrent depressions had an overlapping anxiety disorder as well. On the other hand, an ever-increasing body of data from genetics, biochemistry, neuroendocrinology, neuromorphology, and clinical psychopharmacology (8,25,31,43,46,47,53,55) indicates that major depression is not a homogeneous disorder, but subsumes a number of biologically and, perhaps, etiopathogenetic distinct subforms. In order to understand the presence of anxious symptomatology in major depressives it seems particularly important to look at this problem in the light of the inherent biological heterogeneity of major depression. Although prelim­inary, recent reports (6,54) of different familial incidence of major depression among the first-degree relatives of depressive patients with and without associated panic attacks or agoraphobia suggest that specific anxious symptoms are pref­erentially clustered in distinct subtypes of major depression. In order to explore this hypothesis, we undertook a series of exploratory anal­yses specifically designed to test whether some biochemical, pharmacological, and clinical variables, usually singled out as putative markers of the heterogeneity of major affective disorders, could be useful in screening major depressive patients who would consistently present anxiety symptoms. This chapter summarizes the findings from these preliminary studies, soma of which are still underway.

An adequate case definition presents particular difficulty when anxious depression is being studied in a biological context. For one thing, the inclusion procedures usually employed in clinical-phenomenological studies did not seem very satisfactory for our experimental purposes. First, the term depression gen­erally is applied in favor of the minor neurotic depressions, that is, the types for which current knowledge of the biological correlates of major depression has as yet to be applied. Similarly, the application of a diagnosis of anxiety often applies to simple anxious moods or to emotional lability, to situations with a paucity of presenting symptoms. This may involve the risk of overestimation of the diagnosis of anxious depression, if it is true that signs of anxiety are present in at least 90% of cases of depression (27). Further, the sporadic presentation of anxiety symptoms as a reaction to fleeting situations is a universal that adds to the difficulties of distinguishing between anxious and nonanxious depression. The application of more stringent inclusion criteria, however, involves the risk that the conclusions from the most recent experimental work will not be completely comparable to those dating farther back. Acknowledging that this Gordian knot could not be undone to complete sat­isfaction, we thought it preferable to use criteria usually employed in biologically oriented research to distinguish between anxious and nonanxious depression. This not only involved a benchmark relative to intensity, but also one of symp-tomatological consistency in the course of a few independent depressive episodes in the same individuals. Consequently, our series of analyses included only those patients who fully met the criteria for both a primary (20) and major affective disorder (1) and for whom a subclassification as definitely anxious or definitely nonanxious was pos­sible. Operationally, the definitely anxious patients were those who scored at least 9 points on the Covi anxiety scale (18) during the course of at least three independent depressive episodes. Definitely nonanxious patients were those with less than 6 Covi scale points when tested during each of three independent de­pressive episodes. None of the patients had a detectable concomitant medical or neurological illness. It was thus possible to evaluate 88 patients (39 males, 49 females; mean age 51.2 ± 10.9 years; 51 with recurrent depression, 37 with bipolar disorder); 50 were definitely anxious, 38 definitely nonanxious. A patient’s inclusion in the different analyses was exclusively based on the possibility of determining whether his or her anxiety proneness was high or low. The following criteria and procedures were also used for different analyses. a. Urinary 3-methoxy-s-hydroxy-phenyl-ethylenglycol (MHPG) was measured (43) during the depressive episodes of hospitalized patients who scored higher than 21 (26) on the Hamilton Rating Scale for Depression, had been drug free for at least 3 weeks previously, and had been on bed rest and fed with a vanillil-mandelic acid (VMA)-free diet for 5 days. MHPG was quantified by gas chro­matography, using 24 hr urine of the last 3 days of the baseline period and calculating the mean. b. Responsiveness to lithium prophylaxis was assessed in those patients who, after at least two depressive episodes in the course of 2 years, were on lithium follow-up therapy (least duration: 24 months; mean: 52 ± 8 months). The dosages were individually adjusted to maintain lithium plasma levels at between 0.5 and 0.9 mEq/1 12 hr after the last lithium administration. Patients were classified (46) as responders if they had had no major affective episode requiring supple­mentary medication in the period between 8 months after starting lithium and the end of the follow-up therapy. Nonresponders were patients who had had two major affective episodes during the same period. Patients who had had minor affective breakdowns or one major episode were excluded, since classification of their responsiveness would have been potentially unreliable. с Age at onset was defined as the patient’s age at the time of his first major affective episode. Previous possible minor episodes or mood swings were not taken into consideration. d. Data on the presence or absence of suicidal behavior and/or delusions du ring depressive episodes were recorded on patients’ charts, having been col­lected on the spot or retrospectively. Patients with overt suicidal thinking but not suicide attempts were excluded. Similarly, our reference regarding delusions was DSM-III (1), with mood congruent delusions as the only one taken into account. Patients were considered delusional if they had experienced delusions in at least two individual episodes and nondelusional if delusions had never been documented in past episodes.