- Email: [email protected]
A New Modifiable Risk Factor for Schizophrenia?
Commentary
Biological Psychiatry
A New Modifiable Risk Factor for Schizophrenia? Henning Tiemeier and Tim I.M. Korevaar Schizophrenia is a neurodevelopmental disorder associated with a very substantial disease burden. When teaching medical students psychiatric epidemiology, I often ask which potentially modifiable environmental risk factors for schizophrenia researchers have identified. In contrast to other mental disorders such as depression, my list of answers comprises only a handful of exposures: viral infections and toxins during pregnancy, inflammation, and psychotropic drugs such as cannabis; social adversity as encountered by immigrants; and older paternal age. Thus, there is an intense interest to identify more risk factors for schizophrenia that are amenable to intervention. With their article published in this issue of Biological Psychiatry, Gyllenberg et al. (1) may have extended this short list. In a large birth cohort, the authors showed that the occurrence of schizophrenia is related to prenatal exposure to abnormal maternal thyroid functioning. This study focuses on hypothyroxinemia. This is not a medical condition you will have heard about in medical school. Persons with hypothyroxinemia are asymptomatic, and this subclinical thyroid hormone deficiency is defined biochemically by low free thyroxine (fT4) concentrations with normal serum thyroid-stimulating hormone concentrations (2). So should prenatal exposure to hypothyroxinemia be added to the list of potentially modifiable environmental risk factors for schizophrenia? The authors carefully avoid suggesting clinical implications in their article. Yet, it is an intriguing possibility arising from this excellent study. Gyllenberg et al. (1) used a powerful design to test the relationship of hypothyroxinemia with schizophrenia: a matched nested case-control study including 1010 subjects with schizophrenia and 1010 control subjects. Admittedly my favorite epidemiologic design, a nested case-control study denotes a prospective study embedded in an ongoing cohort. It is prospective, as the thyroid assessment in stored material cannot be influenced by schizophrenia development. The present study was embedded in the Finnish Maternity Cohort of .1 million Finns born after 1983. Cohort participants can be linked to the Hospital and Outpatient Discharge Registry, which allows schizophrenia ascertainment. The cohort has the unique feature that blood serum drawn from many women in early pregnancy is available for later biomarker analyses. In the current sample, stored serum was available in two thirds of all mothers with offspring that developed schizophrenia. The investigators matched the schizophrenia subjects 1:1 with control subjects and not 1:2 or higher. Funding and blood samples are used very efficiently, perhaps at the price of some statistical power. Nevertheless, the current study was well powered (results would survive a Bonferroni correction), and all measurements were specifically performed for this project. The study result is unlikely to be a chance finding.
The validity of findings can be discussed briefly. It is not easy to imagine how selection or information bias could explain the results. It seems unlikely that the schizophrenia cases without stored serum were systematically different. However, confounding remains a possibility. Data from our group showed that maternal body mass index and smoking are the most important determinants of hypothyroxinemia (3). As is often the case in registry studies, only limited confounders are available, and these were not adjusted for by Gyllenberg et al. The association disappeared when analyses were corrected for birth weight and preterm birth. However, the authors considered these variables mediators rather than confounders. This is very debatable because smoking and body mass index strongly influence birth weight and could confound a mediated effect. However, the observed association between hypothyroxinemia and schizophrenia can be judged a finding likely to be valid. Gyllenberg et al. argue convincingly that their findings are biologically plausible and help the reader infer causality. They cite studies demonstrating that numerous neurodevelopmental processes are regulated by thyroid hormone–dependent genes. In the future, researchers should consider conducting Mendelian randomization studies to test possible causal effects of maternal thyroid hormones on schizophrenia. Mendelian randomization uses the genetic variants that are associated with a risk factor as an instrumental variable. However, there are challenges to be overcome before genetic associations with thyroid parameters can be used to study the role of hypothyroxinemia in neurodevelopment. First, the current genetic studies explain too little of the variance of thyroid hormones (thyroid-stimulating hormone, 5.6%; fT4, 2.30%) (4). Second, genetic information from the mother would be needed to model the instrumental variable, and genetic information from the offspring might be needed to control if some genetic variations also have mental health effects later in life. For the time being, the consistency of the present findings in other nonclinical human studies may be the most convincing argument for causality. A replication study of the association with schizophrenia will need to be done in the future. Given the uniqueness of the data available in the Finnish Maternity Cohort, it is hard to imagine how such a study could be designed in other countries. However, hypothyroxinemia was previously related to other neurodevelopmental disorders, such as autism and attention-deficit/ hyperactivity disorder (5). Early life risk factors, such as exposure to low maternal folate levels during pregnancy, are not very specific causes for one neurodevelopmental disorder only. Scientists have long accepted the fact that many environmental factors convey risks for many disorders; there
SEE CORRESPONDING ARTICLE ON PAGE 962
950 & 2016 Society of Biological Psychiatry Biological Psychiatry June 15, 2016; 79:950–951 www.sobp.org/journal
http://dx.doi.org/10.1016/j.biopsych.2016.04.004 ISSN: 0006-3223
Biological Psychiatry
Commentary
is pleiotropy for environment and for genes. Thus, causality can, albeit carefully, be inferred. Before hypothyroxinemia can be nominated for anybody’s list of potentially modifiable schizophrenia risk factors, the prospects of intervention must be considered. In areas with marked iodine deficiency, hypothyroxinemia is closely related to iodine intake. Although severe iodine deficiency is on the verge of being eliminated according to the World Health Organization, millions of women of childbearing age have an insufficient iodine intake (6). Pregnant women are a particular high-risk group because the iodine requirement increases in pregnancy. Overt maternal thyroid deficiency as a result of insufficient iodine intake is associated with irreversible neurodevelopmental changes in the offspring (i.e., cretinism). Work from de Escobar et al. (7) showed the mechanism: in the offspring of rats fed diets with low iodine, the neuronal migration in the hippocampus was affected by low thyroid hormone levels in the brain. Yet, mild to moderate iodine deficiency, as commonly occurring in some Western countries including Finland, has also been related to cognitive impairment measured by IQ and reading ability (8), but the relationship with thyroid hormones has not yet been studied. Maternal iodine supplementation as a public health intervention in moderately iodine-deficient areas may improve cognitive performance of the offspring, but randomized controlled studies with long-term outcomes are lacking. Hypothyroxinemia has also been related to neurodevelopmental outcomes in iodine-sufficient countries such as The Netherlands (5). This finding highlights a lack of the study by Gyllenberg et al. as well as most other studies: iodine, one major presumed cause of hypothyroxinemia, was not assessed. We need more studies from different countries to show the extent to which iodine underlies, or modifies, the effects of hypothyroxinemia in pregnant women—and preferably neurodevelopmental outcomes. Once occurring, hypothyroxinemia can potentially be treated by thyroid hormones. Clinical endocrinologists have advocated randomized trials to demonstrate if screening and treatment for hypothyroxinemia in pregnancy can improve the neurodevelopmental outcome of the offspring. However, neither the choice of the thyroid parameter nor the neurodevelopmental outcome is trivial. Hypothyroxinemia is defined by a combination of laboratory-specific thyroid-stimulating hormone and fT4 cutoff levels. The percentile cutoff of 5%–10% for fT4 used to establish hypothyroxinemia has the implication that, by definition, hypothyroxinemia is always found in 5% or 10% of women. The lack of an absolute cutoff level makes the life of the observational researcher easy and predictable, but complicates prevention and intervention both in practice and in research. Endocrinologists can start to define cross-national evidence-based cutoffs only if they can standardize laboratory methods. A single neurodevelopmental trial of prenatal screening and treatment for subclinical hypothyroidism (many with hypothyroxinemia) has been completed (9). No benefit of treatment for IQ at age 3 years was observed. Possibly, the intervention started too late (median gestational age at start was 13 weeks), as early gestation is key for neurodevelopment. Alternatively, the outcome choice was problematic, as IQ measures at age 3 years are unreliable and poorly predictive of later outcomes. Yet, IQ is probably the best outcome choice, as autism or schizophrenia trials would mean that tens of thousands would have to be treated and many more screened. However, the observed effects of hypothyroxinemia on IQ are small (5) (3–4 IQ
points), although one ongoing trial is powered to detect larger effect on IQ (10). Hence, more very large trials are called for before we can conclude that hypothyroxinemia should be screened for in all pregnant women. In conclusion, it is too early to intervene, but with their elegant study using large Finnish databases, the authors have identified a novel intriguing and potentially modifiable risk factor for schizophrenia. Several seminal studies on infections during pregnancy have already been published using the Finnish Maternity Cohort. The reader may imagine what exciting studies will come next using this resource to answer pressing questions on the relationship of pregnancy exposure with childhood outcomes. It is likely that this team of Finnish and American epidemiologists will add more modifiable environmental exposures to the list of risk factors for schizophrenia and other neurodevelopmental outcomes.
Acknowledgments and Disclosures The authors report no biomedical financial interests or potential conflicts of interest.
Article Information From the Departments of Child and Adolescent Psychiatry (HT), Epidemiology (HT), and Internal Medicine (TIMK) and Rotterdam Thyroid Center (TIMK), Erasmus Medical Center, Rotterdam, The Netherlands. Address correspondence to Henning Tiemeier, MD, PhD, Department of Child and Adolescent Psychiatry, Erasmus Medical Center, PO Box 2060, CB Rotterdam, Netherlands; E-mail: [email protected]. Received Apr 4, 2016; accepted Apr 6, 2016.
References 1.
2. 3.
4.
5.
6.
7. 8.
9.
10.
Gyllenberg D, Sourander A, Surcel H-M, Hinkka-Yli-Salomäki S, McKeague IW, Brown AS (2016): Hypothyroxinemia during gestation and offspring schizophrenia in a national birth cohort. Biol Psychiatry 79:962–970. Negro R, Soldin OP, Obregon MJ, Stagnaro-Green A (2011): Hypothyroxinemia and pregnancy. Endocr Pract 17:422–429. Korevaar T, Medici M, Ghassabian A, Visser E, Tiemeier H, Visser TJ, Peeters RP (2014): Which basic patient characteristics are predictors of thyroid dysfunction during pregnancy? Presentation no. SAT-0547. Presented at Endocrine Society’s 96th Annual Meeting and Expo, June 21–24, Chicago, Illinois. Porcu E, Medici M, Pistis G, Volpato CB, Wilson SG, Cappola AR, et al. (2013): A meta-analysis of thyroid-related traits reveals novel loci and gender-specific differences in the regulation of thyroid function. PLoS Genet 9:e1003266. Ghassabian A, Henrichs J, Tiemeier H (2014): Impact of mild thyroid hormone deficiency in pregnancy on cognitive function in children: Lessons from the Generation R Study. Best Pract Res Clin Endocrinol Metab 28:221–232. Zimmermann MB, Gizak M, Abbott K, Andersson M, Lazarus JH (2015): Iodine deficiency in pregnant women in Europe. Lancet Diabetes Endocrinol 3:672–674. de Escobar GM, Obregon MJ, del Rey FE (2004): Role of thyroid hormone during early brain development. Eur J Endocrinol 151:U25–U37. Bath SC, Steer CD, Golding J, Emmett P, Rayman MP (2013): Effect of inadequate iodine status in UK pregnant women on cognitive outcomes in their children: Results from the Avon Longitudinal Study of Parents and Children (ALSPAC). Lancet 382:331–337. Lazarus JH, Bestwick JP, Channon S, Paradice R, Maina A, Rees R, et al. (2012): Antenatal thyroid screening and childhood cognitive function. N Engl J Med 366:493–501. Casey B, de Veciana M (2014): Thyroid screening in pregnancy. Am J Obstet Gynecol 211; 351–353.e351.
Biological Psychiatry June 15, 2016; 79:950–951 www.sobp.org/journal
951
Biological Psychiatry
A New Modifiable Risk Factor for Schizophrenia? Henning Tiemeier and Tim I.M. Korevaar Schizophrenia is a neurodevelopmental disorder associated with a very substantial disease burden. When teaching medical students psychiatric epidemiology, I often ask which potentially modifiable environmental risk factors for schizophrenia researchers have identified. In contrast to other mental disorders such as depression, my list of answers comprises only a handful of exposures: viral infections and toxins during pregnancy, inflammation, and psychotropic drugs such as cannabis; social adversity as encountered by immigrants; and older paternal age. Thus, there is an intense interest to identify more risk factors for schizophrenia that are amenable to intervention. With their article published in this issue of Biological Psychiatry, Gyllenberg et al. (1) may have extended this short list. In a large birth cohort, the authors showed that the occurrence of schizophrenia is related to prenatal exposure to abnormal maternal thyroid functioning. This study focuses on hypothyroxinemia. This is not a medical condition you will have heard about in medical school. Persons with hypothyroxinemia are asymptomatic, and this subclinical thyroid hormone deficiency is defined biochemically by low free thyroxine (fT4) concentrations with normal serum thyroid-stimulating hormone concentrations (2). So should prenatal exposure to hypothyroxinemia be added to the list of potentially modifiable environmental risk factors for schizophrenia? The authors carefully avoid suggesting clinical implications in their article. Yet, it is an intriguing possibility arising from this excellent study. Gyllenberg et al. (1) used a powerful design to test the relationship of hypothyroxinemia with schizophrenia: a matched nested case-control study including 1010 subjects with schizophrenia and 1010 control subjects. Admittedly my favorite epidemiologic design, a nested case-control study denotes a prospective study embedded in an ongoing cohort. It is prospective, as the thyroid assessment in stored material cannot be influenced by schizophrenia development. The present study was embedded in the Finnish Maternity Cohort of .1 million Finns born after 1983. Cohort participants can be linked to the Hospital and Outpatient Discharge Registry, which allows schizophrenia ascertainment. The cohort has the unique feature that blood serum drawn from many women in early pregnancy is available for later biomarker analyses. In the current sample, stored serum was available in two thirds of all mothers with offspring that developed schizophrenia. The investigators matched the schizophrenia subjects 1:1 with control subjects and not 1:2 or higher. Funding and blood samples are used very efficiently, perhaps at the price of some statistical power. Nevertheless, the current study was well powered (results would survive a Bonferroni correction), and all measurements were specifically performed for this project. The study result is unlikely to be a chance finding.
The validity of findings can be discussed briefly. It is not easy to imagine how selection or information bias could explain the results. It seems unlikely that the schizophrenia cases without stored serum were systematically different. However, confounding remains a possibility. Data from our group showed that maternal body mass index and smoking are the most important determinants of hypothyroxinemia (3). As is often the case in registry studies, only limited confounders are available, and these were not adjusted for by Gyllenberg et al. The association disappeared when analyses were corrected for birth weight and preterm birth. However, the authors considered these variables mediators rather than confounders. This is very debatable because smoking and body mass index strongly influence birth weight and could confound a mediated effect. However, the observed association between hypothyroxinemia and schizophrenia can be judged a finding likely to be valid. Gyllenberg et al. argue convincingly that their findings are biologically plausible and help the reader infer causality. They cite studies demonstrating that numerous neurodevelopmental processes are regulated by thyroid hormone–dependent genes. In the future, researchers should consider conducting Mendelian randomization studies to test possible causal effects of maternal thyroid hormones on schizophrenia. Mendelian randomization uses the genetic variants that are associated with a risk factor as an instrumental variable. However, there are challenges to be overcome before genetic associations with thyroid parameters can be used to study the role of hypothyroxinemia in neurodevelopment. First, the current genetic studies explain too little of the variance of thyroid hormones (thyroid-stimulating hormone, 5.6%; fT4, 2.30%) (4). Second, genetic information from the mother would be needed to model the instrumental variable, and genetic information from the offspring might be needed to control if some genetic variations also have mental health effects later in life. For the time being, the consistency of the present findings in other nonclinical human studies may be the most convincing argument for causality. A replication study of the association with schizophrenia will need to be done in the future. Given the uniqueness of the data available in the Finnish Maternity Cohort, it is hard to imagine how such a study could be designed in other countries. However, hypothyroxinemia was previously related to other neurodevelopmental disorders, such as autism and attention-deficit/ hyperactivity disorder (5). Early life risk factors, such as exposure to low maternal folate levels during pregnancy, are not very specific causes for one neurodevelopmental disorder only. Scientists have long accepted the fact that many environmental factors convey risks for many disorders; there
SEE CORRESPONDING ARTICLE ON PAGE 962
950 & 2016 Society of Biological Psychiatry Biological Psychiatry June 15, 2016; 79:950–951 www.sobp.org/journal
http://dx.doi.org/10.1016/j.biopsych.2016.04.004 ISSN: 0006-3223
Biological Psychiatry
Commentary
is pleiotropy for environment and for genes. Thus, causality can, albeit carefully, be inferred. Before hypothyroxinemia can be nominated for anybody’s list of potentially modifiable schizophrenia risk factors, the prospects of intervention must be considered. In areas with marked iodine deficiency, hypothyroxinemia is closely related to iodine intake. Although severe iodine deficiency is on the verge of being eliminated according to the World Health Organization, millions of women of childbearing age have an insufficient iodine intake (6). Pregnant women are a particular high-risk group because the iodine requirement increases in pregnancy. Overt maternal thyroid deficiency as a result of insufficient iodine intake is associated with irreversible neurodevelopmental changes in the offspring (i.e., cretinism). Work from de Escobar et al. (7) showed the mechanism: in the offspring of rats fed diets with low iodine, the neuronal migration in the hippocampus was affected by low thyroid hormone levels in the brain. Yet, mild to moderate iodine deficiency, as commonly occurring in some Western countries including Finland, has also been related to cognitive impairment measured by IQ and reading ability (8), but the relationship with thyroid hormones has not yet been studied. Maternal iodine supplementation as a public health intervention in moderately iodine-deficient areas may improve cognitive performance of the offspring, but randomized controlled studies with long-term outcomes are lacking. Hypothyroxinemia has also been related to neurodevelopmental outcomes in iodine-sufficient countries such as The Netherlands (5). This finding highlights a lack of the study by Gyllenberg et al. as well as most other studies: iodine, one major presumed cause of hypothyroxinemia, was not assessed. We need more studies from different countries to show the extent to which iodine underlies, or modifies, the effects of hypothyroxinemia in pregnant women—and preferably neurodevelopmental outcomes. Once occurring, hypothyroxinemia can potentially be treated by thyroid hormones. Clinical endocrinologists have advocated randomized trials to demonstrate if screening and treatment for hypothyroxinemia in pregnancy can improve the neurodevelopmental outcome of the offspring. However, neither the choice of the thyroid parameter nor the neurodevelopmental outcome is trivial. Hypothyroxinemia is defined by a combination of laboratory-specific thyroid-stimulating hormone and fT4 cutoff levels. The percentile cutoff of 5%–10% for fT4 used to establish hypothyroxinemia has the implication that, by definition, hypothyroxinemia is always found in 5% or 10% of women. The lack of an absolute cutoff level makes the life of the observational researcher easy and predictable, but complicates prevention and intervention both in practice and in research. Endocrinologists can start to define cross-national evidence-based cutoffs only if they can standardize laboratory methods. A single neurodevelopmental trial of prenatal screening and treatment for subclinical hypothyroidism (many with hypothyroxinemia) has been completed (9). No benefit of treatment for IQ at age 3 years was observed. Possibly, the intervention started too late (median gestational age at start was 13 weeks), as early gestation is key for neurodevelopment. Alternatively, the outcome choice was problematic, as IQ measures at age 3 years are unreliable and poorly predictive of later outcomes. Yet, IQ is probably the best outcome choice, as autism or schizophrenia trials would mean that tens of thousands would have to be treated and many more screened. However, the observed effects of hypothyroxinemia on IQ are small (5) (3–4 IQ
points), although one ongoing trial is powered to detect larger effect on IQ (10). Hence, more very large trials are called for before we can conclude that hypothyroxinemia should be screened for in all pregnant women. In conclusion, it is too early to intervene, but with their elegant study using large Finnish databases, the authors have identified a novel intriguing and potentially modifiable risk factor for schizophrenia. Several seminal studies on infections during pregnancy have already been published using the Finnish Maternity Cohort. The reader may imagine what exciting studies will come next using this resource to answer pressing questions on the relationship of pregnancy exposure with childhood outcomes. It is likely that this team of Finnish and American epidemiologists will add more modifiable environmental exposures to the list of risk factors for schizophrenia and other neurodevelopmental outcomes.
Acknowledgments and Disclosures The authors report no biomedical financial interests or potential conflicts of interest.
Article Information From the Departments of Child and Adolescent Psychiatry (HT), Epidemiology (HT), and Internal Medicine (TIMK) and Rotterdam Thyroid Center (TIMK), Erasmus Medical Center, Rotterdam, The Netherlands. Address correspondence to Henning Tiemeier, MD, PhD, Department of Child and Adolescent Psychiatry, Erasmus Medical Center, PO Box 2060, CB Rotterdam, Netherlands; E-mail: [email protected]. Received Apr 4, 2016; accepted Apr 6, 2016.
References 1.
2. 3.
4.
5.
6.
7. 8.
9.
10.
Gyllenberg D, Sourander A, Surcel H-M, Hinkka-Yli-Salomäki S, McKeague IW, Brown AS (2016): Hypothyroxinemia during gestation and offspring schizophrenia in a national birth cohort. Biol Psychiatry 79:962–970. Negro R, Soldin OP, Obregon MJ, Stagnaro-Green A (2011): Hypothyroxinemia and pregnancy. Endocr Pract 17:422–429. Korevaar T, Medici M, Ghassabian A, Visser E, Tiemeier H, Visser TJ, Peeters RP (2014): Which basic patient characteristics are predictors of thyroid dysfunction during pregnancy? Presentation no. SAT-0547. Presented at Endocrine Society’s 96th Annual Meeting and Expo, June 21–24, Chicago, Illinois. Porcu E, Medici M, Pistis G, Volpato CB, Wilson SG, Cappola AR, et al. (2013): A meta-analysis of thyroid-related traits reveals novel loci and gender-specific differences in the regulation of thyroid function. PLoS Genet 9:e1003266. Ghassabian A, Henrichs J, Tiemeier H (2014): Impact of mild thyroid hormone deficiency in pregnancy on cognitive function in children: Lessons from the Generation R Study. Best Pract Res Clin Endocrinol Metab 28:221–232. Zimmermann MB, Gizak M, Abbott K, Andersson M, Lazarus JH (2015): Iodine deficiency in pregnant women in Europe. Lancet Diabetes Endocrinol 3:672–674. de Escobar GM, Obregon MJ, del Rey FE (2004): Role of thyroid hormone during early brain development. Eur J Endocrinol 151:U25–U37. Bath SC, Steer CD, Golding J, Emmett P, Rayman MP (2013): Effect of inadequate iodine status in UK pregnant women on cognitive outcomes in their children: Results from the Avon Longitudinal Study of Parents and Children (ALSPAC). Lancet 382:331–337. Lazarus JH, Bestwick JP, Channon S, Paradice R, Maina A, Rees R, et al. (2012): Antenatal thyroid screening and childhood cognitive function. N Engl J Med 366:493–501. Casey B, de Veciana M (2014): Thyroid screening in pregnancy. Am J Obstet Gynecol 211; 351–353.e351.
Biological Psychiatry June 15, 2016; 79:950–951 www.sobp.org/journal
951