• The consequences of immigration have given rise to several epidemiological studies, notably in England and the Netherlands, which found an increased risk of developing schizophrenia in immigrants, whether of the first or second generation 3 . This risk could be partly due to conditions of greater social adversity, but other hypotheses have been put forward, such as the mother’s diminished immunity to the viruses of the host country.

• The psychological effect of an urban environment may also interact with levels of reality distortion and depressive symptoms in people experiencing first-episode psychosis and may affect the syndromal presentation of psychotic disorders 4 . For individuals suffering from psychosis, spending time in an urban environment makes them think more negatively about other people and increases their paranoia 5 .

• The role of disturbances in early parent/child relationships, whether it be early separation, a lack of emotional support and physical contact, or abuse, has also been invoked in the later development of the disorder. Zugno et al. showed that rats subjected to maternal deprivation had an increased risk for schizophrenia-like behavior and cholinergic alteration 6 .

• Toxicity factors, foremost among them, psychoactive substances like cannabis, amphetamines or cocaine, can cause acute episodes punctuating the course of the disorder. In most cases, the toxic agents involved determine to a lesser extent the onset of a chronic disease than they precipitate its onset or interfere with its progressive course, making it worse with more complications and sometimes poorer treatment response 7 . Nonetheless several prospective cohort studies on the general population have shown that subjects who use cannabis had a higher risk of developing a psychotic disorder 8 .

• Role of certain infectious agents: It was during the nineteenth century that the hypothesis of the role of infectious agents in the occurrence of schizophrenia, as well as in bipolar disorder, was first formulated, notably by Jean Esquirol in 1845, followed by Eugen Bleuler in 1911 and Emil Kraepelin in 1919 9 . This hypothesis was in line with the discovery of the infectious nature of the psychiatric manifestations of syphilis which occurred during the eighteenth century. Among the different infectious agents, the influenza virus has been implicated most often. Its pathogenic role was suggested for the first time during the 1918 pandemic: many individuals who contracted the "Spanish" flu developed psychiatric disorders, especially with clinical presentations that resembled schizophrenia, as described at the time by Menninger 10 .

However, the epidemiological starting point for the hypotheses formulated on the role of early environmental vulnerability factors was the demonstration of a seasonal imbalance in the distribution of births of future patients, which is evident in schizophrenia. Many studies have identified, in the Northern Hemisphere, an excess of births of subjects with schizophrenia during the winter and early spring, especially between December and April 11 . On average, it is estimated that in European and North American countries, the proportion of subjects with schizophrenia born in winter or at the start of spring exceeds the figures expected in the general population by about 5 to 15%. The "age-incidence" effect, which concerns all diseases where the incidence increases with the age of subjects, seems too small to account for the observed seasonal peak of births. In addition, the existence of this peak during the austral winter in the Southern Hemisphere, even though it is less evident, runs counter to such a bias 12 . This phenomenon of seasonal imbalance is not specific to schizophrenia, since it is found in other diseases like bipolar disorder 13 or substance or alcohol abuse 14 .

In order to account for the observed seasonal distribution of births, various hypotheses have been put forward, some of which are more satisfactory than others: climate, dietary, obstetrical, chronobiological or infectious. Among these alternatives, two quickly gained traction: the role of obstetrical complications and the infectious hypothesis, falling in line with a neurodevelopmental model of schizophrenia.

• Role of obstetrical complications: obstetrical complications are associated with an increased risk of developing schizophrenia. It is a well-documented yet low risk with an odds ratios adjusted for the effect of exposure being approximately 2. The complications most often involved are bleeding during pregnancy, pre-eclampsia and gestational diabetes, as well as certain delivery complications, such as uterine atony, neonatal asphyxia and hemorrhage 15 . Once again, this association is not specific to schizophrenia, as a higher frequency of obstetrical complications has been found in other psychiatric disorders, such as mood disorders, anorexia nervosa or autism 16 17 .

• Role of the influenza virus: influenza infection during the gestation period is associated with an increased risk for the child of later developing schizophrenia 18 . Once again, this association is not specific to this diagnosis, as it is found, for example, in the predisposition to unipolar mood disorder 19 . The mechanism of action of the influenza virus has not been elucidated, but the leading hypothesis suggests that in certain genetically predisposed mothers, antiviral maternal antibodies could cause an autoimmune reaction which is harmful to the fetal brain 20 .

Several studies have demonstrated an association between prenatal infection and certain functions and morphological brain abnormalities in patients with schizophrenia, whether they be in executive function deficits or an enlarged cavum septum pellucidum, suggestive of dysgenesis 30 31 .

Some studies have found an association between the peak of birth seasonality and certain clinical forms, mainly disorganized subtype 32 , paranoid subtype 33 or with pronounced negative symptomatology 34 . In addition, patients with schizophrenia who were exposed to an elevated risk of influenza infection during pregnancy are more likely to be single, which could indicate poorer social adjustment 35 , and with the disorganized clinical subtype 18 .

• Prenatal infection model

  One of the experimental models used to analyze the impact of a prenatal infection consists of injecting the pregnant female rodent with polyriboinosinic-polyribocytidylic acid (Poly I:C), a synthetic analog of double-stranded RNA that stimulates an immune response with cytokine secretion. It has been shown that this exposure is the cause of cognitive, behavioral and pharmacological dysfunction in the neonate as an adult, and that these abnormalities result in disorders found in schizophrenia, such as disorders of sensorimotor gating, selective attention, working memory or hypersensitivity to psychostimulants.

  In addition, the timing of this exposure appears to be paramount, since the abnormalities induced will vary, depending on the stage of pregnancy. Thus, if exposure is late, at the end of gestation, one observes as an adult rodent persistent behavioral abnormalities with learning disabilities 36 , disturbance of visual-spatial working memory 37 and hypersensitivity to the glutamatergic NMDA (N-methyl-D-aspartate) receptor blockade during treatment with the antagonist dizocilpine (MK-801) 37 . They thus involve precisely the dysfunctions that may underpin the negative symptoms found in schizophrenia.

  Bitanihirwe et al. 38 injected mice with Poly I:C on gestation day 17, hence during the late gestational phase. In those neonates grown as adults, deficits in social interaction, anhedonic behavior, and alterations in locomotor behavioral response to acute administration of apomorphine were demonstrated. In addition, male subjects were found to lack cognitive and behavioral flexibility. Lastly, the authors found reduced levels of dopamine and glutamate in the prefrontal cortex (PFC) and hippocampus, as well as decreased levels of γ-aminobutyric acid (GABA) in the hippocampus and of glycine in the PFC. These abnormalities are in favor of a neurodevelopmental model that could account for the occurrence of negative symptoms, with, in addition, an effect correlated with gender.

• Experience of maternal deprivation

  In the neonate rat, long periods of maternal deprivation (from 3 to 24 hours versus 3 to 15 minutes) are associated with hyperactivity of the HPA (hypothalamic-pituitary-adrenal) axis 39 40 , and with a prepulse inhibition (PPI) deficit or lack of reduction in the intensity of startle 41 . The term PPI describes the inhibition of the startle reflex triggered by a violent and unexpected exteroceptive stimulus (pulse) when it is preceded in the short term by a weaker prestimulus (prepulse), too weak to itself induce the reflex. PPI is thus an operational measure of sensorimotor gating, reflecting the capacities of the individual to regulate, to "buffer" the input of different levels of significance and variable chaotic flow of sensory information. Maternal deprivation is also accompanied by cognitive disorders with impaired learning abilities 42 43 .

  Moreover, there is a period of brain development particularly sensitive to the deleterious effects of maternal deprivation: the stress hyporesponsive period, which falls, for the rat, between the 4th and 14th postnatal day. Takase et al. 44 thus showed that among rats subjected to maternal deprivation during this critical period, only the postpubertal males had abnormal hippocampal calcineurin expression associated with a deficit in social interactions and objective memory. Once again, this is an animal model linking negative symptoms and neurodevelopmental impairment.

This model was proposed in light of two observations: that of the high prevalence of childhood abuse in patients with schizophrenia 45 , and that of the similarities between the effects on brain development of the trauma from abuse and certain biological abnormalities found in patients with schizophrenia, whether it be hyperreactivity of the HPA axis or morphological changes to the brain (cerebral atrophy, ventricular enlargement, reversed cerebral asymmetry, hippocampal lesions).

Ross et al. 46 showed that a history of abuse was negatively correlated with negative symptoms. However, there may be two distinct patterns of response to stress:

• Some subjects repeatedly confronted with abuse react through dissociative and positive symptoms: it then involves not so much a process of cerebral atrophy, as dysfunction mediated by dopamine, whence a better response to antidopaminergics 47 .

• Other subjects react through hypervigilance (anxiety, hypersensitivity) that may accentuate the phenomenon known as "pruning", normally occurring at adolescence and corresponding to a decrease in the number of synapses (especially in the neuropil) and to neuronal loss. This excessive neuronal loss could be at the root of the progression toward negative symptoms in adulthood 48 49 .

In this latter model, which associates neurodevelopmental and neurodegenerative processes, the glutamate hypothesis could also play a role. In fact, dysfunction of corticostriatal glutamatergic transmission, by reducing the tonic dopamine release, could be the cause of an increase in the phasic dopamine release linked to stress. This excessive phasic response to stress may induce cerebral damage by increasing excitotoxicity and oxidative stress.

Lastly, still in this mixed model associating a neurodevelopmental approach and a neurodegenerative process, some immune dysfunction could be involved, mainly implicating the role of pro-inflammatory cytokines 50 .