Glyn Lewis

Childhood interleukin-6, C-reactive protein and atopic disorders as risk factors for hypomanic symptoms in young adulthood: a longitudinal birth cohort study

Background There are no existing longitudinal studies of inflammatory markers and atopic disorders in childhood and risk of hypomanic symptoms in adulthood. This study examined if childhood: (1) serum interleukin-6 (IL-6) and C-reactive protein (CRP); and (2) asthma and/or eczema are associated with features of hypomania in young adulthood. Method Participants in the Avon Longitudinal Study of Parents and Children, a prospective general population UK birth cohort, had non-fasting blood samples for IL-6 and CRP measurement at the age of 9 years (n = 4645), and parents answered a question about doctor-diagnosed atopic illness before the age of 10 years (n = 7809). These participants completed the Hypomania Checklist at age 22 years (n = 3361). Results After adjusting for age, sex, ethnicity, socio-economic status, past psychological and behavioural problems, body mass index and maternal postnatal depression, participants in the top third of IL-6 values at 9 years, compared with the bottom third, had an increased risk of hypomanic symptoms by age 22 years [adjusted odds ratio 1.77, 95% confidence interval (CI) 1.10–2.85, p < 0.001]. Higher IL-6 levels in childhood were associated with adult hypomania features in a dose–response fashion. After further adjustment for depression at the age of 18 years this association remained (adjusted odds ratio 1.70, 95% CI 1.03–2.81, p = 0.038). There was no evidence of an association of hypomanic symptoms with CRP levels, asthma or eczema in childhood. Conclusions Higher levels of systemic inflammatory marker IL-6 in childhood were associated with hypomanic symptoms in young adulthood, suggesting that inflammation may play a role in the pathophysiology of mania. Inflammatory pathways may be suitable targets for the prevention and intervention for bipolar disorder.

Commentary on Griffith-Lenderinget?al. (2013): Cross-lagging cannabis and psychosis vulnerability [Editorial Material]

Deciding whether an exposure causes a disease is a difficult and uncertain task in epidemiology. That association is not causation is well known, yet Griffith-Lendering and colleagues [1] claim that their study ‘suggests that there is a bidirectional causal association between [cannabis use and psychosis vulnerability]’. This is a strong claim to make based upon the data presented. TRAILS (Tracking Adolescents’ Individual Lives Survey) is a cohort study of adolescents recruited at age 11 from schools in the northern area of the Netherlands who have now been followed-up to age 25 years. The data used in the paper were from 13, 16 and 19 years, and both cannabis use and ‘psychosis vulnerability’ were collected at all time-points. The analysis that is presented is a cross-lagged model using structural equation modelling. Their results indicated that psychosis vulnerability at both 13 and 16 years was associated with later cannabis use, while there was evidence that cannabis use at 16 was associated with psychosis vulnerability at 19, but no evidence of an association between cannabis use at 13 and psychosis vulnerability at 16 years. The association between cannabis use and psychotic symptoms is well established, although it is worth noting that the authors do not find that early cannabis use is more likely to lead to psychosis than later use, at least at these ages [2]. Of greater note, perhaps, is the possibility that psychotic symptoms may lead to cannabis use. The term ‘psychosis vulnerability’ helps to avoid making too close a link between self-administered scales and psychotic illness. There is now a great deal of evidence that supports a continuum of psychotic experiences, from the large number of people who endorse items about ‘strange thoughts’ or report magical thinking in response to paper-and-pencil tests to the relatively few who develop psychotic illness and are seen in mental health services. The authors do not report the frequency of the psychotic experiences they are studying, but as they used a self-administered questionnaire they are likely to be very common. For a clinician, psychotic experiences reported by up to 30 or 40% of the population would appear to have very little to do with the patients they see who present with psychotic illnesses. If interviewers are used to decide whether the respondents have delusions or hallucinations, the prevalence of psychotic phenomena drops quite markedly and becomes relatively uncommon [3]. The nature of the psychotic outcome may have some implications for the results presented here. There may be common features between psychosis vulnerability and psychotic illness, but there must also be some features that are not shared and perhaps it is these that are associated with later cannabis use. For example, psychotic symptoms in adolescents are associated with both internalizing and externalizing behaviour [4]. Could other behaviour that is associated with psychotic vulnerability be leading to cannabis use rather than psychosis vulnerability itself? The association between cannabis and psychosis (and psychotic symptoms) is well described, but it is still impossible to be certain that the relationship is causal [5]. The main reason for this uncertainty is the possibility of residual confounding; that people who use cannabis differ in terms of characteristics that, in turn, lead to psychosis. There is good evidence that the relationship between cannabis and psychosis is confounded as adjustment for confounders reduces the size of the association. Those investigators who adjusted more comprehensively for confounders also found that the association was reduced somewhat more. This also implies that individuals who develop psychosis have characteristics that, in turn, would lead to cannabis use, so the opposite relationship, that people with psychosis might use cannabis, is also likely to be confounded and this could lead to bidirectional associations, as found by the authors. Although, in the current study, the authors adjusted for gender, parental psychopathology, tobacco and alcohol use it is plausible, perhaps likely, that there is some residual confounding and this could lead to the reported ‘bidirectional’ associations. What is not clear from previous research is the extent to which the association between psychosis and later cannabis use is explained by confounding. For example, in the Ferdinand study [6] bidirectional associations are reported but the authors did not adjust for confounders. The analysis uses a ‘cross-lagged’ structural equation model. In principle, these powerful models allow for the simultaneous associations between the variables at all times, but do not circumvent the need for careful interpretation. For example, gender and parental psychopathology were introduced only before the 13-year timepoint and these adjustments may not be working as effectively for the associations reported between 16 and 19 years. There is evidence from both cohort studies and clinical observation that cannabis can lead to relapse of psychotic illness [7]. Acute intoxication with tetrahydrocannabinol also leads to psychotic symptoms [8]. At present there is evidence to support the possible harmful effects of COMMENTARY bs_bs_banner