The study from Christina Dalman and colleagues at the Karolinska Institute is very thought provoking. Antibodies to gliadin (a wheat-related protein) and casein (a component of milk) were assessed in neonatal dried blood spots from individuals with nonaffective psychosis versus healthy matched controls. The antibodies extracted from the dried blood spots were (almost certainly) maternally derived. The study found that high concentrations of IgG directed at gliadin (but not casein) were associated with an increased risk of nonaffective psychosis. The authors also undertook a careful set of analyses in order to explore potential confounding factors. In the discussion, the authors explore a wide range of possible mechanisms that could underpin their findings. These will certainly help guide future research.
The study contributes to the growing body of convergent evidence linking altered immune function and risk of psychotic disorders (e.g., prenatal immune activation in animal models, findings from GWAS related to the MHC, etc.). Immune systems and the brain share key...
The study contributes to the growing body of convergent evidence linking altered immune function and risk of psychotic disorders (e.g., prenatal immune activation in animal models, findings from GWAS related to the MHC, etc.). Immune systems and the brain share key features—both sense the environment and both learn. From a phylogenetic perspective, acquired immunity is a more recent innovation compared to brains (e.g., Drosophila has a very efficient, if tiny, brain, but no acquired immunity). Thus, we should not be surprised to find that the immune system recycles genes and proteins that are also involved in brain development and adult function. Often, these genes and proteins carry immune-related names (e.g., major histocompatibility complex); we should not mistakenly assume that these labels exclusively index their functions.
The study used a large, Swedish biobank of neonatal dried blood samples. These samples provide the research community with access to a range of analytes (e.g., DNA, antibodies to infectious agents, vitamin D, cytokine, and other bio-immune markers). The international biomedical research community owes a great debt to our colleagues in the Nordic countries. Access to linked health registers and biobanks from these nations has been an essential research tool for epidemiology. Just as the PCR machine has led to major advances in molecular genetics, linked health registers and biobanks are now an indispensable tool for epidemiology. If only the Nobel Prize committee would accept nominations for this category of research!
I concur with the comments of John McGrath last week on the paper by Karlsson et al. and its potential implications. This Swedish study supports the view that early immune response may be involved in the origins of schizophrenia. In this instance, the immune response was specific to a nutritional factor. The result from a single observational study such as this one cannot be considered definitive, and the explanation is necessarily speculative. Nonetheless, this work gains credence from previous work on celiac disease and gluten sensitivity.
I would like to add, however, two points. First, I think that in the schizophrenia field, we need to be much more cautious than we are at present, in our uptake and speculations based on single observational studies (I think the authors of the paper would agree with this). Second, I would like to convey some considerable optimism for the future of this line of work.
First, as one of the originators of the use of archived blood samples in studies of schizophrenia (Susser et al., 2000), I am only too well aware of the limitations of...
First, as one of the originators of the use of archived blood samples in studies of schizophrenia (Susser et al., 2000), I am only too well aware of the limitations of these measures and of related approaches such as use of dried blood spots from neonates. It is often said that confidence in the data increases with the distance from it. That is because the study investigators know about the many questions that remain unresolved in their data but cannot all be detailed in publications. The best one can do is express appropriate caution, which I think the authors of this paper do quite well. What often follows, however, is a less cautious uptake by other investigators and by the public, generating speculation on each new finding reported. This is by no means a problem limited to these measures or to epidemiologic studies (as opposed, say, to genetic studies). But it behooves us as a field to always be on guard for the possibility that small samples, imprecise measures, and design features may account for initial findings which generate a new round of speculation, and are later refuted or found to actually have more subtle or different implications than the authors intended.
In regard specifically to epidemiologic research on the role of immune factors in schizophrenia, we still have very limited evidence from the studies of prenatal sera, because these have thus far been based on small samples, have not been very consistent with one another, have not been tested as mediators of the relation of prenatal infectious exposures to schizophrenia, and also for other reasons too numerous to detail here. We also have only limited evidence thus far from dried blood spots, this being to my knowledge the first such study focused on immune response per se (as opposed to an immune response that is a proxy for previous infection).
So why the optimism about the future? There are many ways that could, and I think will, be developed to enhance the validity of such findings, and to interrelate them with findings from neurodevelopmental neuroscience. I shall mention only three here. One is to further validate the detection methods for immune factors in archived prenatal or perinatal serum or blood spots, which in schizophrenia research have necessarily been stored for long periods of time. This is basic work that has been difficult to fund in the past, but with the increasing use of long-archived samples (not only in schizophrenia research), it is now being initiated in earnest by several research groups. Another is to combine different cohorts for collaborative studies by consortia, and to post data on accessible websites, to permit much larger studies to be conducted, and to facilitate replication by others, as well as transparency. In this arena, epidemiologic studies of schizophrenia lag behind genetic studies, but there have been some early efforts (e.g., Opler et al., 2008; Abel et al., 2010), and the precedent set by genetics will push epidemiologists toward this approach. Among the several long-term advocates of this approach are E. Fuller Torrey, Robert Yolken, Stephen Buka, and myself. We did not succeed in persuading the current generation of schizophrenia researchers to implement it, but I believe that the next generation of researchers will. The increasing use of dried blood spots, which are available on very large populations, the proliferation of pregnancy cohorts with prenatal and perinatal biospecimens, and the recent formation of consortia to examine prenatal and perinatal antecedents of autism should all help catalyze these efforts. Finally, we now have improved methods that are increasingly used to rigorously test causal relations suggested by population-based studies when randomized trials are not feasible (e.g., Smith, 2008; Donovan and Susser, 2011; Relton and Davey Smith, 2012). With these and other advances, we will soon be much better able to separate the wheat from the chaff, and to identify the results that justify substantial investment in further epidemiologic studies, as well as related neuroscience, to elucidate potential mechanisms.
Abel KM, Wicks S, Susser E, Dalman C, Pedersen M, Mortensen PB, Webb RT. Birth weight and adult mental disorder: is risk confined to the smallest babies? Archives of General Psychiatry 67(9):923-930, 2010. Abstract
Smith GD. Assessing intrauterine influences on offspring health outcomes: can epidemiological studies yield robust findings? Basic Clin Pharmacol Toxicol 102: 245-256. (2008). Abstract
Donovan S and Susser E. Commentary: Advent of Sibling Designs. International Journal of Epidemiology 40: 345-349, 2011. Abstract
Opler MG, Buka SL, Groeger J, McKeague I, Wei C, Factor-Litvak P, Bresnahan M, Graziano J, Goldstein JM, Seidman LJ, Brown AS, Susser E. Prenatal exposure to lead, delta-aminolevulinic acid, and schizophrenia: further evidence. Environmental Health Perspectives 116(11):1586-90, 2008. Abstract
Relton CL, Davey Smith G. Two-Step epigenetic Mendelian Randomization: a strategy for establishing the causal role of epigenetic processes in pathways to disease Int J Epidemiol 41; 161-176, 2012. Abstract
Susser E, Schaefer C, Brown A, Begg M, Wyatt RJ. The design of the prenatal determinants of schizophrenia (PDS). Schizophrenia Bulletin 26(2):257-273, 2000.