Schizophrenia Research Forum - A Catalyst for Creative Thinking

Evidence Mounts for the Maternal Inflammation Hypothesis of Schizophrenia

July 15, 2014. New data from a Finnish birth cohort link higher serum levels of the maternal inflammatory marker C-reactive protein during pregnancy to an elevated risk of schizophrenia in offspring. The study, led by Alan Brown of Columbia University in New York City and published online June 27 in the American Journal of Psychiatry, provides a big boost to the maternal immune activation/inflammation hypothesis of schizophrenia.

The idea that maternal infection during pregnancy plays a role in the baby's risk of later developing schizophrenia originated from studies reporting an association between epidemics of infections such as influenza and increased incidence of the illness in offspring born shortly after (see SRF related conference report; SRF conference report). More recently, birth cohort studies utilizing serological samples collected during pregnancy have demonstrated that maternal infection with a number of pathogens, including influenza, Toxoplasma gondii, and rubella, is associated with schizophrenia (see SRF related news report; SRF news report).

The diversity of associated pathogens suggests that maternal inflammation, more generally, may be the culprit behind this association, a hypothesis backed by both preclinical studies and findings of increased proinflammatory cytokines in mothers whose children later developed schizophrenia (Brown and Derkits, 2010). In addition, autoimmune diseases characterized by chronic inflammation have also been associated with risk for the illness (see SRF related news report).

Inflammation measurement
In the current study, first author Sarah Canetta and colleagues investigated the link between general inflammation during pregnancy and the offspring's risk of schizophrenia by measuring levels of the inflammatory marker C-reactive protein, which helps mark dead cells and select pathogens for degradation.

The authors used subjects from the Finnish Prenatal Studies, a birth cohort of all individuals born in Finland from 1983 to 1998. Within the larger study, the Finnish Prenatal Study of Schizophrenia includes 1,514 cases of schizophrenia and schizoaffective disorder identified from the country’s hospital discharge and outpatient registry. Of these, 777 had enough maternal sera for the C-reactive protein assay. Each case was matched to a control subject based on sex and residence in Finland.

The researchers found a significant association between increasing levels of maternal C-reactive protein, measured using an immunoassay, and a subsequent diagnosis of schizophrenia (odds ratio = 1.12). An examination of the data revealed that this effect was not due to outliers, as more cases than controls were present in the highest 50 percent of protein values, while the opposite was true in the lowest 50 percent.

To further investigate the relationship between maternal C-reactive protein levels and schizophrenia, the authors analyzed a number of known health and demographic variables obtained from the Finnish Population Registry. Several variables (such as increased maternal age, increased number of earlier births, and rural birth) were significantly associated with elevated levels of C-reactive protein. Others (including urbanicity, province of birth, and singleton births) were significantly associated with schizophrenia, though none were associated with both maternal C-reactive protein and schizophrenia. When the variables were included as co-variates in a second statistical model that adjusted for their effects, the strength of the association between maternal C-reactive protein level and schizophrenia increased (odds ratio = 1.28).

Inflammation mechanism
Canetta and colleagues suggest a few different possibilities for the mechanism underlying the link between maternal inflammation and schizophrenia. Elevated C-reactive protein may reflect increased levels of the inflammatory cytokine interleukin 6 (which leads to the production of C-reactive protein), which animal studies have shown is responsible for producing schizophrenia-associated behaviors such as abnormal sensorimotor gating in adult offspring (Smith et al., 2007). Unfortunately, interleukin 6 levels could not be measured in the current study due to an insufficient amount of maternal sera.

However, the authors also noted that the elevated maternal C-reactive protein levels may not be specific to schizophrenia, and hypothesized that after maternal immune activation “primes” the developing brain for later emergence of psychiatric disorders, “interaction with specific genetic or environmental insults … might then determine the specificity of the later disorder.” In support of this idea, the researchers have recently reported similar elevations in maternal C-reactive protein in members of the Finnish Birth Studies who went on to develop autism (Brown et al., 2014). Canetta and colleagues also noted that, although they have not yet examined the relationship between C-reactive protein and bipolar disorder or depression, the literature suggests that maternal infection and inflammation may play a role in these disorders as well.

Although it is possible that elevated C-reactive protein is not a causative factor in the illness, “this finding provides the most robust evidence to date that maternal inflammation may play a significant role in schizophrenia, with possible implications for identifying preventive strategies and pathogenic mechanisms in schizophrenia and other neurodevelopmental disorders,” conclude the authors.—Allison A. Curley.

Canetta S, Sourander A, Surcel HM, Hinkka-Yli-Salomäki S, Leiviskä J, Kellendonk C, McKeague IW, Brown AS. Elevated Maternal C-Reactive Protein and Increased Risk of Schizophrenia in a National Birth Cohort. Am J Psychiatry. 2014 Jun 27. Abstract

Comments on News and Primary Papers
Comment by:  Stephen Marder, SRF Advisor
Submitted 10 July 2014
Posted 11 July 2014

Accumulating evidence indicates that activation of the maternal immune system—from infectious and non-infectious sources—increases the risk of schizophrenia. This interesting study using data from the Finnish Prenatal Study of Schizophrenia measured C-reactive protein, a general marker of inflammation, in maternal serum from 777 schizophrenia subjects and an equal number of controls. The importance of this study derives from the size of the sample and the strength of the relationship between immune activation and schizophrenia risk.

View all comments by Stephen MarderComment by:  Chris Carter
Submitted 16 July 2014
Posted 16 July 2014
  I recommend the Primary Papers

If the development of schizophrenia depends upon maternal infection, perhaps a large number of susceptibility genes (possibly related to infection susceptibility and the immune system) would be concentrated in maternal alleles, with other maternal/paternal genes contributing later in life, and in different ways, for the offspring.

Have there been , or are there programmed, any genome-wide association studies on non-schizophrenic mothers of schizophrenic patients? A comparison of maternal/paternal genetic donation might perhaps be a useful means of dissecting out the various pathways leading to disease.

View all comments by Chris Carter

Comments on Related News

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  David GoldmanColin Hodgkinson
Submitted 20 July 2014
Posted 21 July 2014

The fact that schizophrenia is moderately to highly heritable is yesterday’s news; however, this genetic study, because of its large magnitude and with >100 genome-wide significant loci, is a watershed event in the discovery of the genetic variation that must be responsible for this inheritance. There are limitations. The principal limitation is that not one of the findings reported in this paper meets the standard demanded in medical genetics. As reviewed by Flint and Munafo, and following the emphasis of the paper, the focus is on the statistical findings rather than the identification and validation of any one of the >100 genome-wide significant loci at the level of the functional nucleotide difference and how that translates into behavior.

In fact, the principle finding of this study is that no common coding sequence variant accounts for any large fraction of the genetic liability to schizophrenia. It is interesting to speculate on the nature of the genetic variation that causes schizophrenia based on the loci discovered here, which are responsible for an important but still small part of that genetic liability. However, the conclusion that the variants are regulatory in nature will have to await a more complete accounting of the genes and loci involved, and the actual identification of the loci responsible. This is different than pointing to significant associations to SNPs outside of coding regions and to lack of associations to SNPs within coding regions.

Despite the cold water thrown on the eight hundred candidate genes previously implicated in schizophrenia, all of which Flint and Munafo label as “of dubious value,” it is actually critically important that more than a few were among the genome-wide significant loci. Otherwise, and for example if genetic variation at no dopamine or glutamate gene was found to be important, one might doubt the validity of this study. We shouldn’t aggregate or characterize the other association results as if we really understand them or know why they did not generate signals in this genome-wide association study. Clearly some are false positives. Some involve VNTR loci that are not even captured by SNP arrays. Some may be valid in particular populations but not others where the functional variant is absent. Some may depend on the study of particular phenotypes that are not the schizophrenia diagnosis itself but are associated with the disease (for example COMT and cognitive phenotypes) or that can be conflated with schizophrenia (for example DISC1 and schizoaffective disorder).

This genome-wide association study is a starting point for studies on more than 100 genes to elucidate their roles in schizophrenia but it is also a challenge to all of us interested in the biology of schizophrenia. It is best to keep an open mind about the genes involved in schizophrenia and the types of alleles at these genes until some of those functional alleles have been verified.

View all comments by David Goldman
View all comments by Colin Hodgkinson

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  Francis McMahon, SRF Advisor
Submitted 22 July 2014
Posted 22 July 2014

Those of us who attended the annual meeting of the International Society of Psychiatric Genetics in Boston last October were electrified by the striking findings of the Psychiatric Genomics Consortium, reporting over a hundred genome-wide significant genetic marker associations in the largest ever genome-wide association study of schizophrenia. After what seemed like a long wait, this landmark work has now appeared in the journal Nature.

The results further demonstrate the highly polygenic nature of schizophrenia risk. The implicated genes represent a large range of biological functions and converge on a few well-known pathways, chiefly FMRP. The most significant individual finding remains the HLA region, adding to the now widely-held idea that immunity plays an important role in the etiology of schizophrenia.

What now? The mapping of functional alleles in individual genes is the logical next step, but will be a major challenge. Some critics will express skepticism that we have learned much more than we knew after the last large GWAS. Clearly the GWAS method works for schizophrenia – but will we now want to study even larger samples? Skeptics might reasonably ask what we will learn from the next 100 markers that we have not already learned from the first 100.

As a field, we should challenge ourselves to run at least one of these findings to ground, establishing the functional risk alleles and using this information to formulate bold new hypotheses about the causes and treatment of schizophrenia. Even one new effective medication that comes out of these findings will make the entire effort worthwhile.

View all comments by Francis McMahon

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  Bryan Roth, SRF Advisor
Submitted 22 July 2014
Posted 22 July 2014

This is indeed a "landmark paper" and one eagerly awaited by the field of psychiatry, and likely medicine in general.

One thing to emphasize (which was nicely stated in Tom Insel's blog) is that in no case was an actual gene identified. As these are all rather large loci which contain both open reading frames (ORFs) as well as non-coding regions (introns and probably other yet to be identified non-coding RNAs), the real work will be to identify the precise mutation(s) associated with the loci.

Additionally, it will be critically important to determine the directionality of the mutation(s) identified for each locus. Thus, before embarking on a drug discovery expedition, it is important to know if the particular mutations augment or inhibit the activity of the particular molecular entity imputed.

If we take DRD2 (D2-dopamine receptor) as an example, it is important to know if the mutations reside in the coding region and, if so, whether they alter expression, signaling, signaling bias, neuronal targeting, and so on. If the mutation(s) are in non-coding regions (introns, promoter regions, non-coding 3'-region), it will be important to understand how this might alter the expression/function of DRD2. For essentially all of the targets imputed to drive a drug discovery program forward, it is essential to know this information.

Thus, for the calcium channels implicated (CACNA1C, CACNB2, and CACNA1I, which encode voltage-gated calcium channel subunits), we need to know how (and if) the mutations ultimately identified affect channel function, as the design of drugs at these targets will depend upon whether we need to augment or inhibit activity.

Finally, as each of these risk alleles has only a minute effect on the overall risk for schizophrenia, it is unknown whether creating a drug which modulates the activity of a single target could ever lead to a population-wide effect on disease progression/outcome.

Nonetheless, these findings are foundational for the field and provide proof for the power of this approach.

View all comments by Bryan Roth

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  Philip Seeman (Disclosure)
Submitted 22 July 2014
Posted 22 July 2014

Of the many DNA regions found to be associated with schizophrenia in this study (Ripke et al., 2014), the only region that is associated with current treatment is the dopamine D2 receptor. The study shows that this DNA region is 50,000 bases away from the D2 gene and is in the DNA promoter region that controls the expression of the D2 gene. Let’s hope that other DNA regions may lead to improved treatment.

These current results support Van Rossum’s long-standing hyper-dopamine transmission theory of schizophrenia. While there are many causes for schizophrenia, it appears that a final common path for clinical signs and symptoms goes through dopamine D2 receptors.


Ripke S. et al. Nature, July 21, 2014. doi 10.1038.nature13595.

View all comments by Philip Seeman

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  Zoran Vukadinovic
Submitted 23 July 2014
Posted 31 July 2014

My brief comment will focus on one of the findings which is replicated in this study. Namely, this is the second GWAS which has reported that the gene encoding the CaV3.3 subtype of T-type calcium channel (CACNA1L/I) is involved in the pathophysiology of schizophrenia. The reference for the first study can be found below (Strange et al., 2012).

This is an important finding, as CaV3.3 is expressed in the thalamic reticular nucleus (TRN) and has a role in the generation of sleep spindles (Astori et al., 2011), which are markedly reduced in schizophrenia (Ferrarelli and Tononi, 2011). Thus, the finding reported in this GWAS has an important potential pathophysiologic correlate. Sleep spindle reductions in some individuals with schizophrenia may be related to abnormalities of the CaV3.3 subtype the of T-type calcium channel. How this could lead to other deficits in this illness remains to be determined (see Vukadinovic, 2011).

Epidemiologic evidence suggests that cannabis use may be the strongest environmental risk factor for the development of schizophrenia. It is interesting and potentially important that exogenous cannabinoids were found to directly, and independently of, cannabinoid receptors, block T-type calcium channels (including the CaV3.3 subtype) in both cell cultures and neuronal tissues (Ross et al., 2008). This effect may be related to the psychotogenic potential of exogenous cannabinoids (Vukadinovic et al., 2013).

Thus, the now replicated finding that the CACNA1L/I gene may be a susceptibility gene for the development of schizophrenia fits with pathophysiologic evidence of sleep spindle reductions (and possibly TRN deficits) and may, moreover, intersect with a major environmental risk factor (i.e., cannabis use). These three issues may help elucidate the etiology of schizophrenia in at least some patients and should be investigated further.


Strange A., Riley B.P., Spencer C.C.A., Morris D.W., Pirinen M., O’Dushlaine C.T., et al., 2012. Genome-wide association study implicates HLA-C*01:02 as a risk factor at the major histocompatibility complex locus in schizophrenia. Biol Psychiatry . 2012 Oct 15 ; 72(8):620-8. Abstract

Astori S, Wimmer RD, Prosser HM, Corti C, Corsi M, Liaudet N, Volterra A, Franken P, Adelman JP, Lüthi A. The Ca(V)3.3 calcium channel is the major sleep spindle pacemaker in thalamus. Proc Natl Acad Sci U S A . 2011 Aug 16 ; 108(33):13823-8. Abstract

Ferrarelli F, Tononi G. The thalamic reticular nucleus and schizophrenia. Schizophr Bull . 2011 Mar ; 37(2):306-15. Abstract

Vukadinovic Z. Sleep abnormalities in schizophrenia may suggest impaired trans-thalamic cortico-cortical communication: towards a dynamic model of the illness. Eur J Neurosci . 2011 Oct ; 34(7):1031-9. Abstract

Ross HR, Napier I, Connor M. Inhibition of recombinant human T-type calcium channels by Delta9-tetrahydrocannabinol and cannabidiol. J Biol Chem . 2008 Jun 6 ; 283(23):16124-34. Abstract

Vukadinovic Z, Herman MS, Rosenzweig I. Cannabis, psychosis and the thalamus: a theoretical review. Neurosci Biobehav Rev . 2013 May ; 37(4):658-67. Abstract

View all comments by Zoran Vukadinovic

Related News: Bigger Schizophrenia GWAS Reports More Than 100 Hits

Comment by:  Hugo Geerts
Submitted 4 August 2014
Posted 6 August 2014

While this might be a blockbuster breakthrough study from an academic point of view, I would caution that there is still a very long way to go before this could be turned into potential drug targets and discovery programs. Many of the SNPs are contributing very little to the schizophrenia phenotype, suggesting that there might be many different ways to arrive at the same phenotype.

One could say that, for schizophrenia, we are now at a situation in the Alzheimer’s field some 22 years ago when ApoE (major risk factor with an OR of about 4) in sporadic patients and APP in familial patients were identified. Yet no drug program focused on ApoE has entered a Phase 2 POC study, and the (many and expensive) APP clinical programs all have failed to live up to expectations in the clinic so far. It suggests that the one gene-one phenotype hypothesis likely is much too simple. Also, the SNPs listed do not include currently pursued clinical targets such as PDE10 or α7 nAChR.

Under the leadership of Dr. Zaven Khachaturian and the Alzheimer's Association, we have set up a workgroup on how to go from Big (-omics) Data to Smart Data; i.e., how could we generate actionable knowledge (in a drug discovery sense) from all the databases that identify "correlations" or "associations" with certain clinical phenotypes in Alzheimer’s disease? I was wondering if the time would be right to start a similar pre-competitive workgroup initiative in schizophrenia so as to incentivize the pharmaceutical industry by providing it with useful, actionable knowledge about what possible targets are worth pursuing for a large number of patients and how to best affect these targets (i.e., agonism or antagonism).

View all comments by Hugo Geerts