Four decades ago, a major controversy raged in the United Kingdom (1). Was depression one thing or many? The unifiers were led by Aubrey Lewis, powerful head of Maudsley Hospital. The dividers were led by Martin Roth, a follower of Karl Jaspers. Following his teacher Adolf Meyer, Lewis took a pragmatic approach to diagnosis; if clinical differences did not make a difference in practice, then there was no difference. Roth emphasized psychopathology: if clinical syndromes could be shown to be different in phenomenology, then they are different.
The terms of the debate were endogenous versus exogenous depression: the first biologically and the second environmentally caused. Exogenous depression was synonymous with neurotic or reactive depression – associated with anxiety and mood reactivity, and highly sensitive to psychosocial stressors. Endogenous depression was melancholic, unreactive in mood, and non-anxious. Epidemiological studies in the 1960s and 1970s found that such divisions did not imply differences in outcomes; Lewis won, and his ideas were crowned posthumously in 1980 with DSM-III: all kinds of depression were lumped in one label – the almighty major depressive disorder (MDD).
Four decades later, we can pass a new judgment on the British debate. The accompanying discussion paper (2) raises the issue with the recent STAR*D study, as we have also done (3), augmented by a recent genetic meta-analysis of serotonin polymorphism and a mathematical analysis of possible combinations of depressive symptoms.
Of these, the mathematical factorial argument is weakest. It seems misapplied, implying that no serious disease can have more than one or two symptoms. One can think of many medical diseases – pneumonias, AIDS, syphilis, lupus, coronary artery disease – with multitudes of symptoms. To be a disease does not mean having a single symptom. The genetic meta-analysis argument is stronger, and the STAR*D long-term results stronger still. The low long-term remission rates of STAR*D are sobering, despite the reasonably good acute treatment response rates. Clearly, if MDD is one entity, then it is not very responsive to treatments. It is very possible that the low long-term response rates reflect a wide variation in treatment response given the various subgroups of depression.
Maybe Martin Roth had it right, but not in the exact terms of the old UK debate. The distinction is not between biological versus non-biological depressions. Most depressive conditions (non-bipolar) can be shown to be about equally genetic and environmental (4). It is not their etiology that distinguishes them; rather, in the tradition of Jaspers, it is their psychopathology, and, in the tradition of Kraepelin, it is their outcomes. Neurotic depression has a completely different psychopathological picture than melancholia; they may also differ markedly in treatment response, melancholia being more and neurotic depression less responsive to at least some antidepressants (like tricyclic agents or electroconvulsive therapy). And there may be a third subgroup, mixed depression – where manic and depression symptoms combine with marked irritability and psychomotor agitation (5) – which is also antidepressant non-responsive and neuroleptic responsive.
The old debate is not over.