While only a limited amount of work has been conducted in this field, we found growing epidemiologic evidence that, not only can mood and anxiety disorders be differentiated by symptom syndromes and trajectories, but that the factors associated with these disorders may vary between these subtypes.
Most population-based epidemiologic research investigating heterogeneity in the symptom syndromes of common mood and anxiety disorders has focused on major depressive disorder, with only sparse work relating to ADHD, bipolar disorder, panic attack, PTSD, or social phobia. This work suggests that depression may not be a homogenous disorder characterized by a single phenotype (Figure 1A), but a heterogeneous constellation of potentially distinct subtypes of symptom syndromes (Figure 1C). For depression, the syndromic subtype most commonly identified in this review was atypical depression, a subtype defined by symptoms including increased appetite and weight gain and greater chronicity. Although the validity of atypical depression has been debated since the subtype was codified in the DSM-IV in 1994 [5, 68, 69], three latent class analyses provided support for an atypical subtype that was distinct in terms of its patterns and severity of symptoms [22, 24, 25]. However, these analyses did not support the presence of mood reactivity as a necessary symptom for the diagnosis of atypical depression, a finding corroborated by recent clinical work . There was less evidence for other clinical subtypes of depression, including melancholic, seasonal, and vascular depression [26, 36, 37]. For example, an anhedonic latent class exhibiting a greater loss of interest was identified by one study ; however, other work showed that these symptoms were non-specific . Additional epidemiologic replication of these clinical subtypes is needed. Additionally, although cross-sectional and longitudinal research has identified high rates of comorbidity of mood and anxiety disorders, few studies of symptom syndromes explored symptoms of more than one disorder. These are some of the limitations that will have to be addressed for a new typology to emerge.
Familial aggregation studies suggest that heterogeneity in the symptom syndromes of depression and ADHD are partly explained by genetic similarities [24, 32]. However, the particular genetic, sociodemographic, psychological, social, or environmental characteristics that explain observed heterogeneity in symptoms is unclear; many of these characteristics may lie along the same etiologic pathway. Future research will have to address a number of issues. First, because most studies were cross-sectional and started in adulthood, it was impossible to distinguish associations that may reflect a causal pathway from those that may be spurious. Longitudinal assessments that establish a temporal structure between risk factor and phenotype will help to understand observed associations. Second, most of the associations between exposures and subtypes were non-specific in nature. For example, analysis of the National Comorbidity Survey showed that depressive atypicality, a subtype based on patterns of symptoms, was associated with interpersonal dependency, reduced self esteem and stressful life events [22, 24]. Similarly, analysis of symptom subtypes in the Baltimore Epidemiologic Catchment Area study found more severe subtypes were associated with female gender and family history but not stressful life events, while mild or moderate cases were associated with family history and stressful events, but not female gender . Although it is plausible that one exposure may be associated with multiple subtypes, further phenotypic characterization of these disorders will increasingly help disaggregate these relations and facilitate identification of the specific genetic, personal, and social factors associated with distinct symptom subtypes .
We found strong evidence of heterogeneity in the longitudinal trajectories of common mood and anxiety and disorders. There are three potential explanations for these patterns. First, heterogeneity in trajectories may be the result of having distinct symptom subtypes in the population, each of which may have a distinct longitudinal course (Figure 1C). In this case, heterogeneity in symptom syndromes may be spuriously confused as heterogeneity in the longitudinal trajectories of mood and anxiety disorders. Second, there may be true intra-individual heterogeneity in the longitudinal course of a single clinical disorder because of differential exposure to genetic, personal, or environmental factors. Therefore, a homogenous clinical disorder may present as multiple phenotypes (Figure 1B). Third, there may be intra-individual heterogeneity in the longitudinal course of multiple symptom subtypes (Figure 1D). Overall, our review found stronger evidence for intra-individual heterogeneity in the longitudinal course of a single disorder than heterogeneity resulting from having distinct clinical subtypes in the population. Several studies conducted among children and adolescents found evidence of distinct depressive trajectories characterized by different levels of symptom severity. In general, the most prevalent trajectories were stable patterns of consistently low to moderate levels of symptoms [39, 40, 42, 45]; however, up to one-quarter of some samples were assigned to classes characterized by persistent levels of severe symptomatology [42, 43].
Only sparse research has investigated intra-individual heterogeneity in the longitudinal course of multiple symptom subtypes. In that study, Angst and colleagues (2002) found that atypical depression was characterized by greater longitudinal chronicity . Studies that examined trajectories of mood and anxiety disorders combined were also very limited, although both clinical and epidemiologic evidence suggests that may individuals experience frequent transitions between symptoms of these disorders over time. It is possible that individuals who experience symptoms of both disorders over a lifetime share characteristics that distinguish them from individuals who only experience symptoms of one specific disorder. This seems an area worthy of further investigation.
There was strong evidence that a variety of factors experienced over the lifecourse, ranging from personal to social, may influence trajectories of common mood and anxiety disorders. There was also evidence to suggest that vulnerability to external factors may vary between individuals with differing trajectories. Extended longitudinal studies starting in childhood are needed to distinguish between these effects. The extant literature suggests that certain characteristics may predispose individuals to membership in more severe, versus less severe, longitudinal trajectories of depression. For example, among children, adolescents, and young adults, female gender, poorer school performance, greater exposure to stressful life events, and comorbidities including substance use were all associated with trajectories characterized by greater depressive symptomatology [41–44, 57–60]. Similarly, among adults, increased demands among caregivers were associated with more severe depressive trajectories . Conversely, a number of studies suggest that certain factors, particularly greater access to social and material resources, may act as buffers and predispose individuals to less severe trajectories of depression. Among adolescents and young adults, research showed that stronger social relationships and greater access to social supports predicted membership in trajectories characterized by less severe symptoms of depression [39, 59]. Furthermore, one study showed that greater parental educational attainment was associated with a steeper decline in adolescents' depressive symptoms , suggesting that socioeconomic status may be associated with less severe depressive trajectories. Among adults, greater educational attainment and financial resources were associated with less severe depressive trajectories [46, 63]. As with symptom syndromes, whether environmental factors are the cause of a particular trajectory, or whether individuals with a genetic predisposition to a particular trajectory are more susceptible to specific environmental factors, can only be answered by extended longitudinal studies and the assessment of interaction between genetic and environmental factors [70, 71]
Finally, an alternative way of viewing these patterns was proposed by twin studies and the Baltimore Longitudinal Study of Aging, which both support a trait-state model of depression, where symptom levels can be accounted for by two factors: a level (average or "trait") effect that is highly heritable and reflects underlying vulnerability and a residual ("state") effect that is non-inheritable and reflects occasion specific circumstances [49, 71]. In this theoretical framework, trajectory subtype may be considered a manifestation of trait effects. These studies concluded that attempts to identify environmental determinants of symptoms of depression might best focus on deviations about average levels over multiple assessments.
Further inference about the heterogeneity in symptom subtypes and trajectories of common mood and anxiety disorders over the lifecourse is limited by a number of methodological issues. First, the studies included in this review were population-based and utilized a variety of diagnostic survey instruments that are, to varying degrees, imperfect substitutes for clinician-administered structured interviews. Error in the measurement of the mood and anxiety disorders may influence the validity of individual studies and complicate comparisons between studies. Further information on the validity of common diagnostic instruments can be found elsewhere . Second, in most studies that assessed the characteristics associated with distinct clinical presentations or trajectories, methods of variable selection were not theoretically predicated. This makes it difficult to assess whether a particular characteristic was associated with heterogeneity in mood and anxiety disorders across studies. A multilevel framework for understanding how factors experienced over the life course influence the heterogeneity of common mood and anxiety disorders may facilitate model specification and improve comparability between studies. Third, studies used a number of different methods for defining subtypes. Studies that assessed heterogeneity of clinical presentations either specified criteria a priori or used statistical methods to identify subtypes. In general, disorders recognized as distinct clinical entities by the DSM-IV, including seasonal and atypical depression, were more likely to rely on definitions specified a priori than less commonly studied subtypes. The dichotomy between pre-defined and empirically derived subtypes represents a bias-variance trade-off. While having pre-defined criteria facilitates comparisons between studies, there is ongoing debate about the validity of recognized subtypes [73, 74]. On the other hand, methods such as latent class analysis are more flexible and permit investigation of the number and nature of underlying subgroups in a sample [75, 76]. However, these techniques can be overly sensitive to the data and may complicate comparisons across studies. For example, are the mild depressive classes from two different studies qualitatively similar [22, 25]? This trade-off was not relevant when considering studies assessing heterogeneity in trajectories of mood and anxiety disorders. These studies typically used latent class growth, semi-parametric group-based modeling, or other techniques to identify distinct trajectories according to the inferential goal of the analysis [77, 78]. Fourth, despite the clinical and epidemiologic evidence for frequent comorbidity and transitions between mood and anxiety disorders, more than 90 percent of the studies identified by our review assessed symptoms of depression alone. Comorbidity with, and transitions between, these disorders is likely to be a field particularly worthy of further investigation. Finally, most studies did not distinguish between age and cohort effects.