Address for correspondence: Ezra Susser, Department of Epidemiology, Mailman School of Public Health, Columbia University, 722 West 168th Street, Room 1508, New York, NY 10032. Voice: +1-212-342-2133; fax: +1-212-342-2286. firstname.lastname@example.org
It is well established that malnutrition in children has adverse effects on development. Only recently, however, has it become possible to examine the full scope of adverse effects of malnutrition across the life course, which would include latent effects of fetal or childhood malnutrition on health and disease in adult life. We review here a series of studies which have linked early prenatal famine to the risk of schizophrenia in the offspring. Thus we aim to draw attention to the need to look beyond the concurrent effects of malnutrition and consider also the effects that may become apparent decades later.
In most settings malnutrition is strongly associated with poverty, and even in the exceptional circumstances in which an entire population is affected by malnutrition, the poor tend to be the most affected.1,2 An extensive body of work has examined the effects of prenatal and childhood malnutrition on child development and health (see Delisle2a for a review). Also, researchers have related specific micronutrients to the risk of specific neurodevelopmental disorders in children, for example, folate supplements reduce the risk of neural tube defects, and low iodine intake can cause cretinism.3–5
The impact of prenatal malnutrition may not, however, be limited to health and mental health effects that are evident at birth or in childhood. Recent work has provided support for the view that there are also latent effects which become evident in adult life.6,7 This work suggests that we need to extend our perspective to encompass health throughout the life course.8 By limiting our purview to reproductive and child outcomes, we limit our understanding of optimal prenatal and childhood nutrition, and of the potential benefits of improved nutrition. The potential effects on adult health should also be considered and cannot be inferred from the health effects detectable in early life.
A vast literature has now established that early life experiences can have an impact on adult health.9 For example, low birthweight is related to increased risk of cardiovascular disease. Sometimes this work is interpreted loosely as demonstrating that prenatal maternal nutrition influences offspring adult health. However, birthweight is probably not a good index of maternal nutrition during pregnancy. Studies based on the Dutch famine described below suggest that, to the extent that birthweight does reflect maternal nutrition, it reflects mainly nutritional intake during the last trimester of gestation.2,10
There have, in fact, been very few studies with direct measures of both maternal prenatal nutrition and adult health in offspring. The central studies in this field have been follow-ups of the birth cohorts exposed and unexposed to prenatal famine during the Dutch Hunger Winter of 1944–1945. These studies find some evidence of effects of prenatal famine on obesity and insulin resistance, but as yet, the effects are modest and are not consistent across studies.11–15 The results are in general strongest for mental disorders, and especially for schizophrenia.16–21 More recently, a second series of studies has emerged based on the Chinese famine of 1959–1961.22,23 These studies have replicated the Dutch result for schizophrenia. There is also a report of an effect of prenatal famine on socioeconomic status in adult life.24
In light of the state of the evidence, therefore, we focus this chapter primarily on the evidence for latent effects of prenatal nutrition on schizophrenia. Establishing one clear example of a latent effect on a major adult disorder opens the door to considering a range of other potential effects. Then we discuss animal models of prenatal malnutrition, and what they might contribute to understanding the biological mechanisms by which prenatal famine could be linked to this latent effect.
Prenatal Famine and Schizophrenia
Schizophrenia is a disorder characterized by psychotic symptoms, such as delusions and hallucinations, as well as deficits in other domains, such as motivation and affect. As the disorder is often associated with long-term disability, it ranks among the top 10 causes of disability in the WHO classification.25 Currently most investigators consider schizophrenia to be a neurodevelopmental disorder, in the sense that it has some early origins in brain development, but the disorder is not diagnosed until adolescence or adulthood. Typically, the full syndrome required for diagnosis emerges between age 16–29 years. In many cases, however, certain signs or symptoms are apparent long before the full syndrome, and in other cases the syndrome does not emerge until mid-life or even late-life.
The hypothesis that prenatal nutrition may be related to schizophrenia dates back to the mid-20th century.26 It was not tested, however, until the end of the 20th century.17 The first test was based around the historical circumstance of the Dutch Hunger Winter of 1944–1945.2,27 This famine was precipitated by a Nazi blockade of the occupied region of Holland in October 1944 and exacerbated by the severe winter which soon followed. The food shortage was most severe in the occupied cities of western Holland; in the rural areas there was more access to supplementary food.
Three remarkable features of this famine made it possible to study its effects on schizophrenia in adulthood. First, the food rations distributed to the population were documented. Although individuals found ways to supplement the official ration, the caloric content of the ration was strongly correlated with intake. When rations fell, food intake fell. Second, the peak period of famine was of short duration. The famine ended abruptly in May 1945, when the Allied troops liberated western Holland. In the last months of the famine, the ration fell to extremely low levels, and supplementary food was increasingly difficult to find. In addition, the population was nutritionally depleted. Thus the period of most severe starvation was approximately from February 1945 until liberation in early May. The increased severity in these last months is reflected in data on mortality, fertility, and birthweight.2 Third, information could be obtained on schizophrenia admissions in adulthood for the individuals born in the famine cities of western Holland before, during, and after the famine. The Dutch national psychiatric registry recorded hospital admissions for specific diagnoses from 1970 onward. Taken together, these features made it possible to define birth cohorts exposed to famine at specific periods of gestation, and to test whether the exposure was linked to an increased risk of schizophrenia.
Although the result for schizophrenia emerged through a series of studies,16 we summarize here the main findings in a single figure. In Figure 1, we define the birth cohort of October 16–December 31, 1945 as severely exposed to famine in early gestation. Based on their birth dates, we can infer that the vast majority of this birth cohort was conceived or in early gestation during the peak of the famine. This inference is supported by the excess of neural tube defects and other congenital anomalies of the central nervous system in this birth cohort (Fig. 1). It is also supported by the drop in the birth rate, which reached a nadir in September 1945 and remained low until the end of 1945 (Fig. 1); the decline in fertility correlated strongly with rations around the time of conception in these Dutch birth cohorts.2Figure 1 also illustrates that this exposed birth cohort had a sharply increased risk of schizophrenia in adulthood, a result based on the national psychiatric registry data. Finally, the figure illustrates that among males there was an excess of Schizoid Personality Disorder diagnosed at age 18 in the same exposed birth cohort.19,28 This result derived from military induction data: all males born 1944–1946 were subject to a military draft at age 18 years, and the induction examination included a psychiatric assessment. Current diagnostic practice would probably classify the individuals with Schizoid Personality as having a “Schizophrenia Spectrum” personality disorder; familial aggregation studies and other evidence suggest that “Schizophrenia Spectrum” personality disorders are etiologically related to schizophrenia.29
In sum, the results from the Dutch famine studies suggest that in the birth cohort conceived or in early gestation at the height of the famine, there was a sharp increase in neurodevelopmental disorders at birth, in adolescence, and in adulthood. It appears then, that the same exposure led to different neurodevelopmental disorders at different points in the life course. These disorders may also share other features, for instance, neural tube defects and schizophrenia have both been associated with a genetic variant in the folate pathway, MTHR677TT,30 and as noted above, schizophrenia and schizophrenia spectrum personality disorders share a genetic diathesis. At this point, however, the possibility of an interrelationship among all these three disorders is intriguing but not established.
Although these data offered fairly compelling evidence for a link between prenatal malnutrition and the risk of schizophrenia in offspring, the number of exposed cases was modest, and a single study is rarely sufficient basis for a causal inference. Also, other interpretations were plausible. For instance, during periods of famine people often resort to food substitutes—such as tulip bulbs in the Dutch Hunger Winter—which might be toxic to the developing brain. Although starvation would have led to the ingestion of these food substitutes, a causal pathway mediated by ingestion of toxic food substitutes would have different implications for the pathophysiology and ultimately for preventive interventions.
It proved difficult, however, to find another historical circumstance in which this result could be tested. Famines are common, but the documentation of psychiatric outcomes in a defined population for decades after a famine is rare indeed. It was nearly a decade before the finding was replicated, and this study was done by an independent group.22 The study was done in the Wuhu region of Anhui Province, China, and based on the devastating famine which afflicted China following the Great Leap Forward. In Wuhu, the peak of this famine was in 1959 and 1960. The key data available in Wuhu were the number of births in each year, and the number of people born in these years who subsequently received outpatient or inpatient treatment for schizophrenia. The authors were able to demarcate a district which was served by the same psychiatric hospital over the main decades of risk for schizophrenia in the relevant birth cohorts. Also, the population of this district was remarkably stable over these decades, in part because of tradition, and in part because changes in district of residence were controlled.
The Chinese famine was long lasting and the Wuhu data on birth rates were available for years rather than months or weeks. Following the result for the Dutch study, however, it could be hypothesized that the schizophrenia risk should peak in those birth years in which the birth rate dropped. (The Dutch famine results in Fig. 1 indicate that the schizophrenia risk peaked shortly after the nadir in the monthly birth rate.) This is exactly what was observed (Fig. 2). Although the measure of exposure was less precise, the numbers were much larger, making these two studies quite complementary.
Two studies with concordant results substantially strengthened the hypothesis that prenatal malnutrition had latent effects on adult schizophrenia. The lead investigators of the two studies (the current authors) now joined together in the completion of a third study which could further solidify (or undermine) the hypothesis. This was conducted in the region of Liuzhou, in southern China, using essentially the same design as the Wuhu study. Again the results were concordant, that is, the risk of schizophrenia peaked in the annual birth cohorts in which the birth rate dropped.23
Some special features of this third study added strength to the hypothesis. It was based on a very large number of people, even larger than the Wuhu study. It was conducted in a region of China which differed in customs, ethnic diversity, and historical famine experience from the Wuhu region. Finally, and most important, it permitted us to clearly differentiate the impact in urban and rural areas. This was important because, in contrast to Holland, the famine in China affected mainly the rural areas. Urban residents received rations and generally suffered little or no starvation, while the rural population suffered starvation on a massive scale. Therefore, an increased risk of schizophrenia due to prenatal malnutrition should be evident in the rural not the urban area. This is what we observed.23
The two Chinese studies, together with the earlier Dutch study, make a very compelling case that prenatal exposure to famine increases risk of schizophrenia and possibly other forms of major mental illness in later life.
The mechanisms by which prenatal nutritional stress may produce increased risk of schizophrenia have been discussed elsewhere.7,31–34 There are several routes by which nutritional deficiency may directly affect growth of the developing fetal brain. One possibility is that effects are mediated through the hypothalamic-pituitary-adrenal axis. Another possibility is that nutritional deficiency, especially of micronutrients involved in the folate pathway, may indirectly affect brain development by interfering with DNA stability and/or expression.35 The folate pathway plays a key role in DNA synthesis, methylation, and repair. The rate of de novo mutations may be increased, and this could give rise to a schizophrenia phenotype in later life. The increased mutation rate may be global in nature or restricted to unstable genomic regions, such as those where extensive copy-number variation, is reported. Epigenetic regulation of genes critical for brain development may also be affected.36
Prenatal starvation might also influence genetic selection in a number of ways. First, genetic factors might influence which mothers and fathers are able to conceive a child under conditions of starvation. Second perhaps schizophrenia risk alleles are preferentially retained in times of famine so that more individuals at risk come to term and survive postnatally. An intriguing example of a fertility advantage associated with a disadvantageous phenotype is a large 900 kb inversion polymorphism on 17q. It is liable to delete and cause mental retardation but has been selected for in European populations because of its fertility/fecundity advantage.37 Third, nutritional stress may alter molecular regulatory mechanisms and release previously accumulated but unexpressed variation, a phenomenon sometimes known as the Waddington effect.38,39 Environmentally sensitive chaperone proteins are known to buffer phenotypic variation.39 Heat shock protein (Hsp) 90, in particular, affects signaling pathways that underlie many threshold traits. Both Hsp 70 and 90 are elevated in vertebrates in response to starvation during early development.40 Prenatal nutritional stress could therefore shift the threshold for phenotypic expressivity of a trait, such as schizophrenia. These and other potential physiologic and molecular responses to prenatal nutritional stress are, of course, not mutually exclusive.
Could prenatal starvation also lead to adult neuropsychiatric and behavioral abnormalities being overrepresented in the second generation of offspring? It is a completely open question at present. However sufficient time has now elapsed since the Chinese famine to start exploring this hypothesis. It is a real possibility. The germ line of F2 is present in the developing fetus when F0 is pregnant with F1 and so the germ line of F2 is itself directly physically exposed to the effects of malnutrition in F0. Such effects are illustrated by coat color in agouti mice where color of F2 is influenced by diet of F0.41 In this case the transmission is epigenetic. However direct physical damage from malnutrition is another possibility.
Clinical field studies of individuals with schizophrenia and their families, both exposed and unexposed to famine, are now being undertaken by the authors in collaboration with Chinese colleagues. The aim is to test some of the molecular hypotheses proposed above. Nonhuman animal studies of prenatal malnutrition and effects on brain and behavior are still essential. These studies depend on the fact that the general sequences of steps and events in brain development and differentiation across mammalian species are remarkably similar. There are, however, important differences between humans and other mammalian species, in particular in regard to extent of prenatal and postnatal brain development. No animal model can therefore be a perfect substitute for the study of human brain development itself.
Studies looking at prenatal malnutrition and long-term behavioral effects in rodents are still limited in number. Early studies in mice demonstrated that a low protein (6–8% casein) diet fed for 1–2 months prior to conception and through gestation resulted in reduced cerebral weight and protein content of newborn. DNA content was normal, suggesting that there was no reduction in cell number.42 This was later independently confirmed by quantitative analysis of cellular densities as well as an estimate of the cortical volume. The total number of neurons was never smaller than in the controls.43,44 Later work, much of it conducted by Janina Galler, Peter Morgane, and colleagues, has been performed on rats. The prenatal diets given before and after conception to both male and female rats are designed to try explore the more subtle abnormalities of brain and behavior we now think are associated with human prenatal under-nutrition. The results suggest that prenatal protein malnutrition affects the brain generally but the impact is greatest on the limbic system. Many subtle derangements are present, often requiring quantitative anatomical, functional, and behavioral studies to detect.45 One of the most interesting findings is diminished growth and arborization of serotoninergic neurons and decreased serotoninergic nerve terminals in the hippocampus. This has recently been followed up by dual probe microdialysis assays of extracellular serotonin and dopamine in hippocampus and medial prefrontal cortex of prenatally malnourished rats.46 Extracellular dopamine is significantly reduced in prefrontal cortex of malnourished rats, raising the possibility that dopamine innervation of prefrontal cortex is abnormal in prenatally malnourished animals. These animals also differ in the response of their serotonin and dopamine systems to stress.
These and similar studies suggest that the long-term brain and behavioral phenotypes associated with prenatal nutritional and other forms of prenatal stress in rodents can be subtle. Some of the neurobiology and behaviors, as discussed above, have features that resemble a schizophrenia-like phenotype, but so little is known in rodents as to what the expected phenotype should look like that one has to be cautious. Similarly, we do not know whether prenatal protein deficiency is the most appropriate animal model. It is tempting to think that deficiency of one specific nutrient or micronutrient may hold the key. Folate have been given most attention, but insufficiency of any single essential amino acid may hinder fetal protein synthesis and affect brain development. Methionine has a crucial role in the folate pathway, and deficiency could lead to effects similar to those of a deficiency of folate itself.47 Tryptophan is required for the synthesis of serotonin. Could tryptophan deficiency affect the serotonergic system in hippocampus? Only by applying the latest quantitative techniques and performing detailed studies of the brain and behavioral phenotypes associated with global and specific nutritional deficiencies can some of these interesting questions be answered. Recently, such studies have been initiated by numerous research groups, and the forthcoming results should soon produce a far more robust literature.48
The example of schizophrenia demonstrates that prenatal malnutrition is associated with latent effects on adult health and disease. The challenge now is to understand the causal pathways that account for these latent effects. Until we do, we cannot be entirely sure that the relationship is causal and cannot use these findings to tailor interventions to prevent schizophrenia or other diseases. This is the focus of our current work. The rapid advent of new genomic technologies will enable us to test some of the hypothesized pathways among the large numbers of persons who were exposed to the Chinese famine. Also, in the past 20 years large randomized trials and quasi-experimental interventions have been conducted with early prenatal nutritional supplements, and in coming years the follow-up of the offspring from these trials will permit testing of other hypothesized pathways.