Rank I: Visual domain

Reduced visual (symbol) span (a measure of visual attention and working memory), aged speed of visual processing, and deficit in distance vision on the right, were found to achieve biomarker status. When pooled together these biomarkers demonstrated the greatest strength on ROC analysis. On translational analysis, processing in the visual domain was found to be influenced by oxidative stress (HPL) and low levels of vitamin D, along with dysfunctions in overall nutrition-related biochemistry. Such prominent visual findings in schizophrenia and schizoaffective disorder highlight the need for more detailed visual assessment and correction of visual deficits for first episode psychosis patients, particularly since visual impairments have already been reported in psychosis [35], in both medicated and medication-free schizophrenic subjects [36,37]. With respect to the deficit findings related to the right distance vision in schizophrenia, it is known that right parietal cortex and right frontal cortex exert a stronger driving influence on visual areas of the left hemisphere [38].

The role of vitamin D in visual deficit is understandable in terms of its role in promoting glutamate decarboxylase (GAD 67) expression and therefore GABA production [39]. It also has a role in Glutathione production, with low levels of glutathione found in vitamin D deficient animals [40]. A reciprocal mechanism also exists between elevated catecholamines and low vitamin D levels, with catecholamines priming parathyroid hormone, to release calcium (Ca ++) ions from neuronal cells - an effect that is directly opposed by the action of vitamin D, theoretically leading to increased vitamin D utilization and chronically low reserves of this vitamin [41,42]. Taken together, these effects of vitamin D over-utilisation by combating glutaminergic and Ca++ excitotoxic damage and oxidative stress in retinal and visual pathways [43] very-likely set in motion a cycle of elevated arousal via the hypothalamic pituitary adrenal (HPA) pathway, resulting in still more noradrenaline release (with excessive noradrenaline release known to also disrupt sensory processing pathways [44]).

Thus in a chronic state, continued draw-down on the adrenal to produce even more catecholamines, may result in further overutilization of vitamins and eventually adrenal fatigue, with reduced neural transmission that further hinders visual and auditory sensory processing performance [45]. To summarise: primary nutritional-biochemical-oxidative stress imposes sensory processing strain, resulting in secondary HPA activation, resulting in mandated dopamine and catecholamine synthesis that then remotely draws down on vitamin D reserves. This demonstrates the potential for a vicious cycle of poor nutrition → low Vitamin D → low glutathione → increased oxidative stress and reduced GAD 67 expression, with low GABA and neuronal calcium neuro-excitotoxicity, followed by visual pathway damage → sensory strain → HPA axis activation → elevated catecholamines → further increased vitamin D utilisation → back to decreased vitamin D availability with low glutathione and high calcium excitotoxic neuronal damage explaining the underlying pathology of the hidden visual deficits in schizophrenia. Further research in this area is required [46].

Rank II: Auditory domain

The auditory domain demonstrated the second strongest results on ROC analysis.

This domain contained biomarkers of reverse digit span (auditory (verbal) working memory), competing words (intra-cerebral dichotic listening), and delayed speed of auditory processing. This domain was also significantly correlated with the elevated catecholamine domain, implying that deficits in this domain also cause increased activation of the HPA, resulting in further elevated catecholamine release. This domain was the most strongly related to nutritional-biochemical abnormalities such as low folate, low vitamin D and high vitamin B12, with low vitamin B6 demonstrating a strong specific translational relationship.

The finding of low reverse digit span (impaired auditory working memory) has been previously documented for schizophrenia [47]. This project also utilised the (SCAN-3) test [48] as a measure of auditory processing disorder and an unsuspected finding was the prevalence of impaired competing word deficit (representing dichotic listening disorder) in the participants with schizophrenia and schizoaffective disorder. Subsequently, we found that dichotic listening deficits have also been previously reported in schizophrenia [49] and that reduced auditory speed of processing has been reported in chronic, non-paranoid schizophrenics [50]. Other literature indicates that reduced auditory processing speed is implicit to these disorders themselves and not attributable to anti-psychotic medication [51,52].

Low folate has been associated with sudden sensorineural hearing loss and folate supplementation assists hearing in older adults [53,54]. The damaging effects of Vitamin D lack on neural circuitry and sensory end organ function has already been described in the visual domain section above and this damage may well apply to auditory sense organs and neural circuitry as well. Low vitamin B6 was specifically related to dysfunction in the auditory domain, which is understandable, since this vitamin is a rate-limiting cofactor for tyrosine hydrolase and for the dopa-decarboxylase enzyme (aromatic L-amino acid decarboxylase) which converts L-DOPA into dopamine [55,56]. In a setting of delayed auditory function and HPA activation we theorise that there is an excessive need for the brain cells and adrenal gland to synthesise dopamine. Draw-down on vitamin B6 reserves for dopamine synthesis could be predicted to be high and since dopamine is a key activating neurotransmitter for the ear and auditory pathway neurotransmission, it is understandable that deficiency due to over-utilisation of this vitamin disturbs auditory activation and neural transmission [57].

Rank III: Catecholamine domain

This domain demonstrated a strong correlate with the Symptom Intensity Rating (SIR) index and also produced a strong result on ROC analysis - findings consistent catecholamine elevation being integral to symptom-formation and with the monoamine hypothesis of schizophrenia [58]. However, in contrast to the dopaminergic hypothesis of schizophrenia [59], within the catecholamine domain, elevated noradrenaline levels predominated over elevated adrenaline levels and levels of both generally predominated over dopamine levels. Both dopamine and noradrenaline have been found to be raised in the blood of medicated and non-medicated patients with schizophrenia [60] and elevated noradrenaline levels have been reported in cerebral spinal fluid of both drug-free and neuroleptic-medicated schizophrenic patients [61]. Sustained, elevated levels of noradrenaline and adrenaline have also been reported in post-traumatic stress disorder [62], which is thought to be a traumatic response to inner confusion and anxiety caused by thought disorganisation and hallucinatory experiences [63]. We therefore theorise that this finding of peripherally elevated urinary catecholamines may not only cause sensory processing strain, but may also be reflective of dual visual and auditory sensory strain becoming a subliminal dynamic that is also an important component of secondary HPA activation, leading to further sensory signal disruption and psychosis-formation in schizophrenia and schizoaffective disorder. This interpretation partly fits with the 1990 findings of Adler LE, et al. [64] and also with the finding of stress-related pituitary enlargement in treated and untreated patients with schizophrenia [65-67]. It raises further important issues in the relationship to evolutionary and developmental origins of schizophrenia [68]. Raised catecholamines are also reported to suppress immune system responses [69,70], a finding that fits with reported reduced microglial activation in schizophrenia [71]. Noradrenalin and adrenaline elevation appear associated with mixed behavioural symptoms of anxiety and vigilant aggression that co-occur together with negative symptoms [72] and it is understandable that draw-down on both vitamin D and B6 (and even on folate reserves), may occur under the HPA axis mandate to synthesise more catecholamines as a secondary response to primary auditory or visual sensory strain. Evidence also exists that whilst low levels of Noradrenaline positively modulate cortical neurotransmission, at greatly increased levels, Noradrenaline neurotransmission is disrupted [72-74] which can be considered as a basis for fronto-temporal dysconnectivity, attentional disorganisation with impaired frontal cognitive function [75,76], temporal isolation and psychosis.

High Cu/Zn ratio and low vitamin D levels within the overall nutritional-biochemical domain exert strong translational effects upon the elevated catecholamine domain. This is to be expected, since copper is a cofactor for noradrenaline synthesis from dopamine and reports exist of high copper and low zinc associated with noradrenaline excess in schizophrenia [77,78]. Animal models also demonstrate vitamin D deficiency in relationship to elevated brain dopamine and noradrenaline levels [42]. The means by which vitamin D reserves may be over-utilised through combating the effects of catecholamine-inducted parathyroid hormone activation, has already been described [40,41].

In this study low vitamin B6 was specifically related to high dopamine levels, which is understandable since B6 is required as a cofactor at two enzyme levels as a cofactor for dopamine synthesis by tyrosine hydroxylase and for conversion of L DOPA into dopamine by the enzyme DOPA decarboxylase. The latter reaction is extremely important since L-DOPA is the only peripherally synthesized catecholamine capable of entering the blood brain barrier so its synthesis is the rate-limiting step for noradrenalin synthesis within the brain [55,56]. Apart from low vitamin B6 from poor nutritional intake in poorly functioning schizophrenic states [79], low vitamin B6 findings may be due to increased utilization of B6 under HPA imperative to increase L-Dopa and dopamine synthesis, triggered by sensory-strain-HPA-system-activation in order to manufacture more noradrenaline [80]. Given such an HPA imperative to synthesise more noradrenaline, the numerous research findings of elevated dopamine in schizophrenia, leading to the dopamine hypothesis of this condition [59] may just be reflective of dopamine’s role as a necessary precursor of noradrenaline.

Rank IV: HPL/creatinine domain

Hydroxyhemopyrroline-2-one (HPL) emerged as a legitimate biomarker on ROC analysis. In a similar manner to translational relationships for catecholamines, HPL received strong translational influence from high Cu/Zn ratio and low vitamin D levels. Pyrrole products are ubiquitous in tissue biochemistry and HPL’s presence in urine is linked to oxidative stress and to low levels of vitamin B6 and zinc [81], which can further deplete B6 reserves and elevate free copper levels, respectively. Urinary HPL is thought to arise as an abnormal product of haeme metabolism induced by haemeoxidase under conditions of extreme oxidative stress where the porphyrin moiety of haemoglobin becomes degraded or where there is abnormal porphyrin synthesis [82-85]. When injected into rats and mice, HPL causes reduced activity, head twitching and backward locomotion (retreat) behaviour [86] and patients with elevated HPL levels report symptoms of anxiety, depression, low stress-tolerance and mood swings. HPL may also be stress and catecholamine-related and as such, elevated levels may be disorder-non-specific [87]. The structure of the HPL molecule has belatedly been discovered [88] and these findings invite further investigation of this oxidative, haeme-related molecule and its reported psychiatric effects.

Rank V: Nutritional-biochemical domain

biochemical/biochemical biomarkers such as elevated free copper: zinc ratio, low vitamin B6 activation, low vitamin D and low folate levels, which on the “divide by three” test failed to exert sufficient individual strength to gain firm diagnostic association, were still able to demonstrate sufficient cumulative strength in terms of odds ratio of diagnostic association to have significant correlation with symptom severity.

While red cell zinc failed to produce an independent finding on ROC analysis, elevated serum free copper, in ratio with red cell zinc, was a biomarker of schizophrenia or schizo-affective disorder, confirming the hair-analysis findings of Ghanem et al. in 2009 [77]. Elevated free copper contributes to neuronal damage in circuits responsible for sensory-processing [89] and copper levels may also be influenced by reduced ceruloplasmin (copper binding protein) levels which has been linked to psychosis [90]. Free copper is redox active in the body and also contributes to reactive oxygen species and oxidative stress [91]. It also blocks the action of cystathione beta synthase (CBS) (Figure 1), an enzyme located upstream of glutathione production, preventing the formation of this major antioxidant for the brain [92]. Copper also binds to cholesterol producing an oxidative product 17-hydroxy cholesterol that is toxic to neurons [93]. Given these multiple neurotoxic and oxidative stress-related effects of elevated free copper, it is understandable that it can play a strong role in degradation of neurones in sensory-processing circuitry.

Zinc plays a reciprocal role to copper since it competes for metallothionein binding protein and acts as a defence against copper by inhibiting its absorption from the gastrointestinal tract [94]. Copper is a cofactor for the enzyme which synthesises noradrenaline from dopamine (Figure 1), which explains why low zinc facilitates high free copper which is then associated with elevated norepinephrine [78]. Animal models also suggest that zinc facilitates copper/zinc superoxide dismutase in blocking cell apoptosis signalling [95] and that zinc deficiency impairs fatty acid and myelin composition [96,97].

This research found that schizophrenic and schizo-affective patients may be extremely sensitive to inadequate levels of activated B6, since its ROC determined cut-off value occurred within the normal laboratory reference range. This is theoretically due to the many co-factor roles of B6 across biochemical pathways related to monoamine synthesis and prevention of oxidative stress. Indeed, inadequate levels of activated B6 may be reflective of its increased utilisation in these multiple biochemistry pathways. For instance, in the initial stages of dopamine synthesis, the enzyme tyrosine hydroxylase requires B6 as a cofactor [98] and downstream to this, L-Dopa is the only catecholamine precursor to enter the central nervous system being converted into dopamine by the enzyme DOPA decarboxylase, with vitamin B₆ as a required cofactor for this reaction (Figure 1), [55,56]. In the transulfuration pathway (Figure 1) the enzymes cystathione beta synthase (CBS) and cystathionase also require vitamin B6 for the synthesis of the brain antioxidant glutathione via [99]. Such B6 over-utilization with relative deficiency may then inhibit the action of the enzymes CBS and S-adenosylhomocysteine hydrolase (SAHH) [100], leading to back-ups in the transmethylation cycle and methylation cycle with elevation of homocysteine and S-adenosyl homocysteine (SAH) [101], respectively. Indeed, when such effects coincide with free copper toxicity and further CBS inhibition [89,91], glutathione synthesis may then be drastically reduced and free radicals damage neuronal circuitry made vulnerable in this setting of oxidative stress. In such a setting, high copper also promotes further noradrenaline synthesis from dopamine [78] whilst elevated SAH contributes to raised catecholamine levels by directly inhibiting cathechol-o-methyl-transferase (COMT), [102-104]. With inhibition of this COMT enzyme that is responsible for the final step of catecholamine metabolism (Figure 1), catecholamine intermediate metabolic products accumulate and break-away metabolism of backed-up intermediate-metabolites result in abnormal neurotoxic products and further cytotoxic neuronal consequences [105].

In this study methyltetrahydrofolate reductase (MTHFR 677 T) polymorphism demonstrated a positive significant correlate for high homocysteine levels, however MTHFR polymorphism, along with plasma homocysteine, urine HVA and MHMA were not successful as biomarkers on ROC analysis. Though weak enzyme dynamics due to thermo-lability of the MTHFR 677 T polymorphism has received much research attention with regard to schizophrenia and homocysteine elevation [106], there is evidence that contradicts this finding [107]. The MTHFR polymorphism is linked to deficiency of its product 5-methyltetrahydrofolate (5-MTHF) and in settings where enzymes in other homocysteine metabolism pathways are also inhibited, backed-up, elevated homocysteine may occur (Figure 1). Such findings have been reported in schizophrenia [106] and it may be that our ROC finding of elevated vitamin B12 represents a compensatory reaction of unknown mechanism designed to assist methionine reconstitution from homocysteine. In a setting of MTHFR polymorphism homozygosity with lack of 5-MTHF product, elevated vitamin B12 may be a necessary compensation in order to sustain methylation cycle precursors for maintenance of S–adenosyl methionine (SAMe) levels (Figure 1).

Low red cell folate was also a biomarker in this study and this finding has been previously reported in schizophrenia [108] where poor diet may convey vulnerability for low methylation and low SAMe (Figure 1), [109]. SAMe also stabilises CBS in the transulfuration pathway, protecting this pathway against glutathione depletion and therefore against oxidative stress [110].

Folate and vitamin B12 (cobalamin and methylcobalamin) have important roles in sustaining production of SAMe, which in turn serves to prevent oxidative stress and catecholamine elevation by serving to stabilise CBS and also as a cofactor for catechol-o-methyltransferase’s (COMT) metabolism of catecholamines [111]. Therefore relative unavailability of SAMe due to colluding influences of homozygous MTHFR polymorphism, low folate or low B12 levels may allow oxidative stress and impair COMT activity, explaining our correlate and ROC findings of elevated HPL and elevated catecholamine levels. The effects of oxidative stress and elevated catecholamine levels have been well documented in literature and are also noted in psychotic children [112-114].

In contrast to the above findings and their biochemical interpretations, other research findings have not found notably elevated homocysteine levels in schizophrenia [115]. Indeed in this research project, homocysteine failed to reach ROC biomarker status and this may be noteworthy in the collateral setting of positive ROC’s findings for elevated free copper to zinc ratio. This is because elevated free copper inhibits the homocysteine metabolising enzyme (cystathione-beta-synthetase (CBS) in the transulfuration pathway leading to glutathione synthesis [91]. A further plausible explanation is that since vitamin B6 is required for homocysteine synthesis, low vitamin B6 may accompany low homocysteine. If this is so, then the causal explanation that COMT inhibition in schizophrenia may occur via SAMe unavailability for COMT facilitation [111] is less likely and an alternative explanation is required. Since SAH-hydrolase (SAHH) requires vitamin B6 for S adenosyl homocysteine (SAH) conversion to homocysteine (Figure 1), SAH is more likely to accumulate in a setting of low vitamin B6 [101]. Then in a setting of retrograde, elevated SAH, the remotely situated COMT enzyme which metabolises catecholamines is directly inhibited [103,104]. Taken together, these explanations bring together our ROC biomarker findings of elevated catecholamines, low vitamin B6, high free copper and high vitamin B12, providing a more cohesive dynamic understanding of potential interactions and causes. Finally, the accompanying finding of failed biomarkers for urine HVA and MHMA reflects the tendency for elevated catecholamine intermediate metabolic products to find break-away metabolic pathways in a setting of their inhibited end-stage COMT metabolism. These abnormal metabolic pathways unhappily resulting in neurotoxic intermediate metabolites which can contribute to loss of neuronal density in schizophrenia [105,116].

In summary, visual and auditory sensory processing systems in this study demonstrate peripheral deficits and prematurely aged, delayed processing that appears translationally related to folate, B6 and D vitamin-deficiency, copper-related neural excitotoxicity and oxidative stress. These abnormalities then accumulate to contribute to the fundamental architecture of schizophrenia and schizoaffective disorder. Sensory processing strain may then initiate HPA axis activation, leading to a cycle of elevated catecholamine synthesis with inherent overutilization of vitamin cofactors and further depletion of their body reserves. This then initates a runaway process of catecholamine elevation, which may occur in a setting of S adenosyl homocysteine (SAH) inhibition of the catechol-o-methlyltransferase (COMT) enzyme. Inhibition of this catecholamine-metabolising enzyme then allows to build-up of elevated catecholamines and break-away neurotoxic catecholamine metabolites that inflict further damage on sensory neurones (Figure 3).

Limitations

One focus of this research was to use biomarker discovery to characterise schizophrenia and schizoaffective disorder. To avoid confounding variables many exclusion criteria were imposed which slowed recruitment and resulted in a small discovery data-set. Small sample size, together with non-normal data distribution for many variables, precluded use of principal component analysis, latent variable analysis and multivariate analysis. ROC analysis and Spearman’s analysis were used instead to identify promising biomarkers and their relationships. Sequentially collapsing the data through ROC analysis and then regression analysis, has potential for loss of some data strength. Though case–control study designs have inherent susceptibility to prevalence and selection bias [117,118], they are allowable for a discovery data-set [119] and have a respectable record in discovery research [120] for low period prevalence disorders, such as schizophrenia (0.35 per 1000) [121].

Difference in selection processes may well have been offset by the random nature of success in the recruitment process itself, incurred by participant failure to meet the multiple exclusion criteria combined with a refusal-to-consent ratio of 4 to1, for approached patients and controls alike. Also, the application of this design to only one bracket of psychotic disorders (schizophrenia and schizoaffective disorder) in this project is limiting, since it is not known whether the ROC variables discovered for these conditions are disease specific or also occur in other mental illness states.

The use of the DSM IV as a diagnostic tool for case selection may also be a confounding variable, since diagnoses may range across spectrum and boundaries and thresholds for diagnosis are yet to reach optimal standards for categorisation [122].

A further limitation of this project was that assessors undertaking rating scales and neurophysiological measures, were functioning in a real-world clinic-setting, so were not blind to participants’ patient or control status.

The full effect of confounding variables upon results of this research was also not fully assessed. For instance, although rigorous efforts were made to exclude substance-related diagnoses, it is known that alcohol misuse is often under-reported by patients [123], and has a known association with low folate, high B12 and elevated catecholamine levels [124,125]. It was not possible to control for participant smoking in this project and smoking also has been variably reported to effect monoamine oxidase levels [126], which can theoretically influence monoamine levels. Though fasting biological samples were collected for this study, there was no longer-term control or assessment of dietary intake, which, in principle, could affect reserves of vitamins, minerals and monoamines [127]. Though participant assessments took place within a 4 day window, it is possible that variance in relationship to a patient’s stress or activity state, may have effected monoamine levels and other related-variables, over the period of assessment [128].

Despite biochemical pathway linkages and the strong correlation of elevated catecholamine biomarkers with Symptom Intensity Rating (SIR) as an indicator of clinical severity, the role of medication as a confounding variable in this project cannot be fully excluded. Although there is widespread research evidence that antipsychotic medications increase catecholamine metabolic turnover and reduce dopamine and noradrenaline levels [129], evidence regarding antipsychotics such as Quetiapine is less certain [130]. By blockage of the dopamine-2 receptors and antagonism of alpha 1 adrenergic receptors Quetiapine contributes to suppression of hyperdopaminergic-related positive symptoms in acute schizophrenia [131], 132 and brings about profound hypotension in overdose, from which reversal and rescue can only be achieved by noradrenaline [132]. However this antipsychotic has been contrastingly reported to increase dopamine and noradrenaline levels in the rat cortex [133]. The exact extent of pharmacological action of sodium valproate is also inconclusive. From its action on GABA, and its efficacy in mania treatment, the mechanism of sodium valproate action is presumed to involve reduction of catecholamine levels. Also, since catecholamines normally promote beta oxidation of fatty acids, reduction of catecholamine levels by valproate is proposed as the mechanism for valproate-associated weight gain [134]. However, evidence from rat and ox brain experiments suggests that sodium valproate may increase monoamine levels [135,136] whilst a contrasting trend is reported from mice experiments [133] and a human CSF study indicates that sodium valproate may increase dopamine metabolism [137], theoretically decreasing dopamine levels.

Although the ubiquitous, oxidative-stress related molecule HPL earned its place as a discovery biomarker additional biomarkers for oxidative stress, future candidate markers could be widened to include other markers of oxidative stress. Similarly, other neurotransmitters and measurement of SAMe:SAH levels and COMT activity levels that would further validate putative biochemical linkages.

Elevated urinary catecholamine biomarkers identified in this study cannot be directly related to brain neurotransmitter changes, since fully-synthesized monoamines do not cross the blood–brain barrier even though their precursor substance, L-Dopa is capable of this transition [80]. Spot-urine collection, as a method for peripheral monoamine analysis, has also attracted criticism [138] though it is practical in a psychiatric setting, is successful in paediatric settings [139,140] and is gaining ground as a useful analytic method relating to body biochemistry [141].

Due to all these considerations, the optimal means for replicating and validating this project’s discovery biomarkers is via a large, well-characterized, prospective, multi-site-clinic trial on a single series of consecutive patients, with blinded assessors and an emphasis on collecting data from ultra-high risk medication naïve subjects.