This assumes that biomarkers related to symptom severity worsen in some way immediately before the onset of schizophrenia. However, even if this is not the case, it could be that a different combination of symptom-related biomarkers is helpful in predicting conversion.
Ideally a biomarker could be used to identify single individuals as approaching schizophrenia with enough confidence to justify beginning treatment before an official diagnosis. Although biomarkers need to have a high degree of predictive accuracy, there is also the issue of specificity. Individuals with bipolar disorder may experience psychosis, but unlike schizophrenia, it is not necessary for diagnosis. This overlap in psychosis means that a successful biomarker needs to be sufficiently sensitive to predict onset of psychosis andbe specific to SSD (Insel et al., 2011).
As previously discussed, biomarkers of sensory memory, particularly the auditory MMN, have been singled out for being reliably smaller in chronic schizophrenia compared to neurotypical individuals (Naatanen et al., 2015; Umbricht & Krljes, 2005). A recent large-sample study found typical MMN in unaffected siblings of people with schizophrenia, leading them to propose that reduced MMN may be closely linked with the onset of psychosis (Donaldson et al., 2021).However, efforts to date have found that the auditory MMN is only slightly smaller in early course schizophrenia (Lho et al., 2020; Salisbury et al., 2007) and is not reliably reduced in individuals at their first-episode of schizophrenia (Erickson et al., 2016; Haigh, Coffman, et al., 2017), or in first-degree unaffected relatives (Magno et al., 2008). Furthermore, longitudinal studies following individuals who are at-risk of developing schizophrenia have yet to detect robust reductions in auditory MMN that provide any confidence of impending conversion to psychosis (Atkinson et al., 2017; Erickson et al., 2016; Koshiyama et al., 2017).
However, the emphasis on a single biomarker to predict psychosis onset may be the issue. In the at-risk population, two recent meta-analyses point to WM performance as a predictor of developing psychosis. For instance, re-evaluating findings relying on the MATRICS Consensus Cognitive Battery identified the factors of overall cognition, attention, processing speed, and WM as significantly predictive of developing psychosis (Zheng et al., 2018). This is broadly consistent with an earlier meta-analysis that selected WM and visual learning to identify individuals likely to develop psychosis (De Herdt et al., 2013). In individuals with schizotypy, meta-analysis again pointed to a general deficit across verbal and visuospatial WM as a significant predictor of developing psychosis (Siddi et al., 2017). Multifactorial analyses of multiple biomarkers have improved specificity of predicting who is at-risk of developing schizophrenia later in life (Seidman, Shapiro, et al., 2016) and with improvements in machine learning, more sophisticated tools are available to detect even subtle patterns of functional abnormalities in SSD (Tai et al., 2019).