The neurobiology of duration of untreated psychosis: a comprehensive review

Abstract

Duration of untreated psychosis (DUP) is defined as the time from the onset of psychotic symptoms until the first treatment. Studies have shown that longer DUP is associated with poorer response rates to antipsychotic medications and impaired cognition, yet the neurobiologic correlates of DUP are poorly understood. Moreover, it has been hypothesized that untreated psychosis may be neurotoxic. Here, we conducted a comprehensive review of studies that have examined the neurobiology of DUP. Specifically, we included studies that evaluated DUP using a range of neurobiologic and imaging techniques and identified 83 articles that met inclusion and exclusion criteria. Overall, 27 out of the total 83 studies (32.5%) reported a significant neurobiological correlate with DUP. These results provide evidence against the notion of psychosis as structurally or functionally neurotoxic on a global scale and suggest that specific regions of the brain, such as temporal regions, may be more vulnerable to the effects of DUP. It is also possible that current methodologies lack the resolution needed to more accurately examine the effects of DUP on the brain, such as effects on synaptic density. Newer methodologies, such as MR scanners with stronger magnets, PET imaging with newer ligands capable of measuring subcellular structures (e.g., the PET ligand [¹¹C]UCB-J) may be better able to capture these limited neuropathologic processes. Lastly, to ensure robust and replicable results, future studies of DUP should be adequately powered and specifically designed to test for the effects of DUP on localized brain structure and function with careful attention paid to potential confounds and methodological issues.

INTRODUCTION

Psychotic disorders are a debilitating group of disorders that most often begin in the late second to early third decades of life [1]. An early, seminal review by McGlashan highlighted that most deterioration occurs in the first few years of the illness [2]. This was followed shortly by another important review underscoring the long-term benefits of early treatment with antipsychotic medications [3]. Over the past several decades, a wealth of evidence has become available which supports the importance of early treatment for psychotic disorders, and in particular for limiting the duration of untreated psychosis (DUP), or the time from the onset of psychotic symptoms until first treatment. These include studies showing that longer DUP is associated with worse positive symptoms [4, 5], negative symptoms [4–6], and, on a number of measures, functioning [4, 5]. Some studies have also suggested increased relapse rates or worse response rates to antipsychotic medications in individuals with longer DUP [5]. Studies of the effects of DUP on neurocognition have not demonstrated global relationships, though some specific cognitive functions may be adversely affected by longer DUP [7–9]. One study even reported that longer DUP may be associated with an increased risk for psychotic homicide [10]. Many of these outcomes have also been observed in low- and middle-income countries [11] and become worse over time, further confirming the dynamic nature of the first-episode time period, and psychosis in general [12].

While early intervention programs, such as Treatment and Intervention in Psychosis Study [13, 14] and Recovery After an Initial Schizophrenia Episode [15–18], have been modestly successful at reducing DUP or improving outcomes after the first episode of psychosis, a recent meta-analysis showed that there was overall no benefit of these interventions on reducing DUP [19]. Furthermore, the primary treatment for psychotic disorders remains antipsychotic medications, which significantly improve positive symptoms and prevent relapse, but do not address negative or cognitive symptoms or prevent illness onset [20]. Taken together, these data suggest the need to develop more targeted approaches to the treatment of early psychosis, and in particular to modify the deleterious effects of longer DUP. Efforts to develop treatments that substantially alter the course of psychotic disorders have been hampered by a lack of understanding of the neurobiology and pathogenesis of DUP.

The pathogenesis of psychosis, and in particular schizophrenia, was from very early on thought to be progressive in nature [21, 22]. Decades of research have revealed that the pathogenesis of psychotic disorders likely involves early life abnormalities in brain development that interact with aberrant pathophysiological processes during puberty which together lead to the expression of psychotic illness [23–25]. Aberrant synaptic pruning [25], sensitization of the dopaminergic system [26], and neurobiological adaptations to chronic N-methyl-D-aspartate receptor dysfunction [27] are possible pathophysiological mechanisms that produce limited neurodegeneration early in the course of psychotic illness [28]. Whether psychosis itself is neurotoxic remains a matter of debate [29] and various hypotheses for how untreated psychosis could impact brain function have been proposed. Dopaminergic hyperactivity leading to a progressive reduction in regional brain volumes [30] and oxidative injury due to persistent catecholaminergic activity and prolonged activation of the hypothalamic–pituitary–adrenal axis [31] provide possible explanations for how chronic psychosis could be neurotoxic. In addition, glutamate-mediated excitotoxicity may also contribute to these effects through neuronal overstimulation that leads to an excessive influx of calcium and subsequent excitotoxicity and, ultimately, cell death via apoptosis [24].

Clinically oriented studies of DUP and neurotoxicity to date do not support a clear neurotoxic effect of psychosis [29], though the substantial methodological limitations of the available studies limit any definitive conclusions. While these aforementioned studies were mostly not designed to directly address the topic of DUP and neurotoxicity, a number of studies from the previous several decades have examined the neurobiologic correlates of DUP. Anderson et al. published a review of 48 neuroimaging studies that examined DUP in 2015 [32]. They highlighted the substantial limitations of the studies that they reviewed and concluded that while there was limited evidence for an association between brain structure and DUP, they were unable to draw any strong conclusions about a relationship between DUP and brain structure, or lack thereof.

The objective of the current study was to conduct a comprehensive review of all studies that have examined the neurobiology of DUP. In addition to reviewing studies that have been published since Anderson et al.’s review [32], we included all studies meeting our inclusion criteria that examine the neurobiology of DUP, including computed tomography (CT), magnetic resonance imaging (MRI), functional magnetic resonance imaging (fMRI), magnetic resonance spectroscopy (MRS), near-infrared spectroscopy (NIRS), electroencephalogram (EEG), and positron emission tomography (PET) studies. In our discussion and synthesis, we particularly focus on the implications of our findings for future studies on the neurobiology of DUP.

METHODS

We searched the PubMed database on March 5, 2021, using the following broad search terms: (duration of untreated) and (illness or psychosis or psychotic or schizophrenia or schizoaffective). This search resulted in 1416 results when limited to English. We reviewed the titles, abstracts, and/or full texts of all 1416 articles and identified 63 that examined the neurobiology of DUP. We then excluded 5 review articles, 7 articles focused on peripheral biomarkers, and 1 article that did not use research methodologies to examine neurobiological lesions.

In addition to the 50 articles identified by our search, we examined the bibliographies of relevant studies and review articles and identified another 33 articles that we included in our review, for a total of 83 articles included in this review (Tables 1–3). We included studies of individuals with any category of psychosis (e.g., schizophrenia, affective psychosis, delusional disorder, substance-induced psychotic disorder, etc.) and any method that directly examines the brain (MRI, CT, PET, MRS, NIRS, EEG). We did not impose any restrictions on date. Articles and data were extracted in duplicate by two of the authors (AWZ and RRG). The following data were extracted for each article: author, publication year, sample size, diagnosis, clinical severity, treatment setting, mean age, gender, antipsychotic exposure, treatment duration, mean length of DUP, imaging modality, magnet strength, and brain regions with significant or non-significant associations. We separately conducted an assessment of study quality and risk of bias for each study using a modified version of the Newcastle-Ottawa Scale [33]. See Fig. 1 for a PRISMA flow diagram.

Table 1.

PET, NIRS, MRS, and EEG studies of DUP.

Author	Year	n	Dx	Clinical severity and/or phase of illness	Mean age (SD) years	% Male	AP N/F/M	AP Tx length	Mean DUP/DUI (SD) months	Imaging measure	Magnet strength	Result
Hafizi	2017	19	FEP	FEP; mildly symptomatic based on the PANSS	27.5 (6.7)	63	14/5/0	<4 weeks	33.6 (40.1) duration of illness which is DUP only for the AP naive patients	PET-[18F]-FEPPA	NA	No relationship between DUI/DUP and inflammation in whole brain, hippocampus, DLPFC, temporal lobe, total gray matter or MPFC
Watts	2020	27	FEP	Psychosis; mildly symptomatic based on the PANSS	25.1 (5.3)	81	8/8/11	NR	NR	PET-11C-CURB	NA	Longer DUP was related to higher fatty acid amide hydrolase in the brain including regions such as amygdala, hippocampus, striatum, DLPFC and MPFC, and ACC
Chou	2014	62	SCZ	SCZ; inpatients and outpatients; mildly symptomatic based on the PANSS	26.3 (9.0) SDUP 31.3 (8.6) LDUP	52	5/0/56	1.7 (1.4) SDUP 58.4 (36.8) LDUP	13.0 (21.0) SDUP 6.9 (10.9) LDUP	NIRS	NA	In the group with a long duration of treatment (>6 months), DUP was negatively associated with activity in the left inferior frontal gyrus, left middle frontal gyrus, left postcentral gyrus, right precentral gyrus, bilateral superior temporal gyrus, and bilateral middle temporal gyrus. There were no relationships between DUP and cortical activity in the short duration of treatment (<6 months) group
Chou	2015	28	SCZ	First-episode SCZ; inpatient or outpatient; mildly symptomatic based on the PANSS	30.8 (6.1)	54	0/0/28	<12 weeks	20.7 (27.9)	NIRS	NA	No relationship between DUP and frontotemporal activity during verbal fluency tests
Nishimura	2014	73	SCZ	Chronic SCZ; outpatients or inpatients; mildly to moderately symptomatic based on the PANSS	38.3 (11.4) GG group; 36.1 (13.7) GA group; 38.3 (13.6) AA group	47	NR	NR	19.8 (32.9) in the GG group; 12.1 (18.0) in the GA group; 48.3 (42.0) in the AA group	NIRS	NA	No relationship between DUP and activation in frontal cortex during verbal fluency test in any EGR3 genotype (GG/GA/AA) group
Briend	2020	54	FEP	First episode; inpatient, outpatient, or emergency room; mild symptoms based on the BPRS	23.9 (6.2)	65	NR	<5 days	4.7 (8.5)	MRS/PRESS and volumetry	3T	Longer DUP associated with smaller left whole hippocampus volumes as well as subregions (tail, CA1, subiculum, presubiculum, molecular layer). Longer DUP also associated with higher glutamate + glutamine levels in the hippocampus (left).
Bryll	2020	40	SCZ	Inpatient and acutely decompensated moderately symptomatic based on the PANSS	22.7 (7.4)	55	NR	NR	0.8 (0.9)	MRS/PRESS	1.5T	Inverse relationship between DUP and glucose in the right frontal lobe and a positive relationship between DUP and choline in the anterior cingulate
Galinska	2009	30	SCZ	First treatment SCZ; moderately symptomatic based on the PANSS	22.5 (3.6)		1/0/29	12–224 days	7.3 (11.6)	MRS/PRESS	1.5T	No relationship between DUP and metabolites in left frontal lobe, left thalamus, or left temporal lobe
Jayakumar	2006	12	SCZ	First-episode SCZ; outpatients; moderately symptomatic based on the PANSS	28.7 (8.8)	83	12/0/0	NA	47.3 (41.1)	Volumetry/31P MRS	1.5T	No relationship between caudate volumes and DUP and no relationship between DUP and 31P metabolites
Shirayama	2010	19	SCZ	Outpatients with SCZ; mildly symptomatic based on the BPRS	30.5 (5.6)	63	NR/NR/18	NR	31.2 (28.8)	MRS/PRESS	3T	DUP was found to be positively related to NAA/(GPC + PC) ratio and negatively related to the (GPC + PC)/(Cr + PCr) ratio in the medial prefrontal cortex
Theberge	2004	19	SCZ	First-episode SCZ	25 (8)	74	19/0/0	NA	28 (26)	MRS/STEAM	4T	DUP was positively correlated with choline, but not NAA, in left anterior cingulate and left thalamus
Wang	2005	18	SCZ	First-episode SCZ; moderately symptomatic based on the BPRS	30.0 (8.6) SDUP 29.6 (6.2) LDUP	61	18/0/0	NA	5.5 (4.0) SDUP 61.1 (43.5) LDUP	EEG	NA	Neither P300 amplitude nor latency differed between patient groups with short (<2 years) or long (≥2 years) DUP
Manchanda	2005	122	FEP	FEP; inpatients and outpatients	25.1 (8.0)	76	NR	NR	17.1 in the group with normal EEGs; 17.2 in the group with essentially normal EEGs; 13.2 in the group with abnormal EEGs	EEG	NA	No relationships between EEG abnormalities and DUP
Manchanda	2008	117	FEP	FEP; inpatients and outpatients; mildly symptomatic based on the SAPS and SANS	NR	74	NR	NR	18.8 (27.8)	EEG	NA	No relationships between EEG abnormalities and DUP
Manchanda	2014	103	FEP	FEP; mildly symptomatic based on the SAPS and SANS	NR	77	NR	NR	NR	EEG	NA	No relationships between EEG and DUP

Dx diagnosis, Tx treatment, AP N/F/M antipsychotic naive/free/medicated, FEP first episode of psychosis, SCZ schizophrenia, SAD schizoaffective disorder, SPD schizophreniform disorder, PNOS psychosis NOS, SDUP short duration of untreated psychosis, LDUP long duration of untreated psychosis, DUP duration of untreated psychosis, CT computed tomography, MRI magnetic resonance imaging, fMRI functional magnetic resonance imaging, rsfMRI resting-state functional magnetic resonance imaging, NIRS near-infrared spectroscopy, MRS magnetic resonance spectroscopy, STEAM stimulated echo acquisition mode, PRESS point-resolved spectroscopy, DTI diffusion tensor imaging, FA fractional anisotropy, EEG electroencephalogram, PET positron emission tomography, T Tesla, NA not applicable, NR not reported, RD responder, NRD non-responder, BPRS Brief Psychiatric Rating Scale, PANSS Positive and Negative Syndrome Scale.

Table 3.

Structural imaging studies.

Author	Year	n	Dx	Clinical severity and/or phase of illness	Mean age (SD) years	% Male	AP N/F/M	AP Tx length	Mean DUP/DUI (SD) months	Imaging measure	Magnet strength	Result
Bangalore	2009	82	FEP	FEP; inpatient or Outpatient	23.7 (6.5)	74	77/0/5	1–2 days	30.4 (39.9); DUI 49.8 (50.6)	Volumetry	1.5T	DUI was inversely related to gray matter density in the left fusiform gyrus extending into the left lingual gyrus, left declive, and right parahippocampal gyrus.
Behere	2009	14	SCZ	Mildly symptomatic based on SANS and SAPS	27.9 (6.2)	64	14/0/0	NA	27.6 (28.7)	Volumetry	3T	No relationship between DUP and orbitofrontal cortex volumes
Boonstra	2011	57	SCZ	Mildly to moderately symptomatic based on the PANSS	24.7 (5.8)	89	36/0/0	NA	DUI 73.1 (73.9)	Volumetry	1.5T	No relationship between DUP and cerebral gray or white matter, total brain volume, cerebellar volume, or ventricular volume at baseline or change in these measures over 5 years
Cahn	2002	20	SCZ	First-episode SCZ; mildly to moderately symptomatic based on the PANSS	27.6 (6.4)	80	20/0/0	NA	48.1 (61.1)	Volumetry	1.5T	No relation between total brain, frontal lobe, gray/white matter, cerebellar, hippocampal, thalamic, caudate, lateral, and third ventricular volumes at DUP
Crespo-Facorro	2007b	61	FEP	FEP; outpatient or inpatient; moderately symptomatic based on the SANS and SAPS	27.8 (7.4)	71	0/0/61	<6 weeks	10.1 (17.1)	Volumetry	1.5T	No relationship between DUP and thalamic volumes in correlational analyses and inversely related in a regression analysis
Crespo-Facorro	2007a	76	FEP	FEP; outpatient or inpatient; moderately symptomatic based on the SANS and SAPS	27.7 (6.8)	70	0/0/76	<6 weeks	11.1 (18.1)	Volumetry	1.5T	Inverse relationship between DUP and caudate volume
Crespo-Facorro	2009	142	FEP	FEP; outpatient or inpatient; mildly to moderately symptomatic based on the SANS and SAPS	30.7 (9.0) SCZ group 28.5 (9.2) SPD group 28.4 (6.3) PNOS group	62	0/0/142	<6 weeks	17.6 (35.4) SCZ group 4.9 (7.1) SPD group 6.4 (19.6) PNOS group	Volumetry	1.5T	No relationships between DUP and whole-brain gray matter, cortical lobe gray matter volumes, lateral ventricular volume, caudate, putamen, and whole-brain white matter volumes
Crespo-Facorro	2010	142	FEP	FEP; outpatient or inpatient; mildly to moderately symptomatic based on the SANS and SAPS	29.7 (NR)	62	0/0/142	<6 weeks	12.5 (28.8)	Volumetry	1.5T	No relationship between DUP and insular volumes or cortical surface
Crespo-Facorro	2011	142	FEP	FEP; outpatient or inpatient; mildly to moderately symptomatic based on the SANS and SAPS	29.7 (8.7)	62	0/0/142	<6 weeks	12.5 (28.8)	Volumetry	1.5T	No relationship between DUP and total cortical thickness or thickness of each cortical lobe
Duggal	2005	30	FEP	First-episode SCZ; inpatients or outpatients; moderately symptomatic based on the SANS, SAPS, and BPRS	31.4 (5.5) in males; 25.9 (8.7) in females	50	30/0/0	NA	32.9 (43.6) in males; 42.9 (32.0) in females	Volumetry	1.5T	No relationship between DUP and insular volumes in medication naive participants with SCZ
Ebdrup	2010	38	SCZ	First-episode SCZ; moderately symptomatic based on the PANSS	26.2 (5.4)	68	38/0/0	NA	DUI 41.5 (51.1)	Volumetry	3T	No relationship between DUP and striatal or ventricular volumes, but a trend inverse relationship between DUP and hippocampal volumes
Emsley	2017	23	SCZ and SPD	First-episode SCZ or SPD; inpatients and outpatients	25.3 (6.5)	78	23/0/0	NA	10.6 (11.3) at the time of baseline scan	Volumetry	3T	DUP did not affect change in cortical volume after a year of treatment with antipsychotic medications
Fannon	2000	37	SCZ, SAD, or SPD	FEP; inpatients or outpatients	24.2 (4.9)	70	13/NR/NR	<12 weeks	6.6 (6.1)	Volumetry	1.5T	No relationships between DUP and whole-brain volume, cortical gray matter, temporal lobe volume, or ventricular (lateral or third) volumes
Goff	2018	57	SCZ or SPD	First-episode SCZ or SPD; inpatients and outpatients; mildly to moderately symptomatic based on the PANSS	25.5 (7.3)	46	57/0/0	NA	5.9 (5.2) at baseline	Volumetry	3T	No relationship between DUP and left or right hippocampal volumetric integrity (HVI) at baseline. DUP was correlated with left but not right HVI annualized reduction from baseline to 8-week follow-up
Gunduz	2002	51	SCZ	First-episode SCZ, SAD, or SPD	24.5 (5.0)	73	36/0/15	10.7 (11.5) days	22.0 (36.6)	Volumetry	1.5T	No relationship between DUP and basal ganglia regions of interest (caudate, putamen, nucleus accumbens subcommissural limbic forebrain)
Guo	2013	57	SCZ	Inpatients with SCZ; mildly to moderately symptomatic based on the PANSS	25.1 (6.3) SDUP 25.7 (6.7) LDUP	59 SDUP 53 LDUP	57/0/0	NA	1.7 (0.3) SDUP; 14.0 (2.6) LDUP	Volumetry	1.5T	Inverse relationships between DUP and right parahippocampal gyrus, right superior temporal gyrus, left fusiform gyrus, left middle temporal gyrus, and right superior frontal gyrus
Gutierrez-Galve et al.	2010	37	SCZ or SAD	First-episode SCZ or SAD	26.8 (8.8)	68	0/0/37	<384 days	10.4 (22.8)	Morphometry	1.5T	No relationship between DUP and cortical thickness or surface area in frontal or temporal regions
Haukvik	2016	79	FEP	FEP; mildly symptomatic based on the PANSS	27.6 (7.7)	66	52 of 138 patients at baseline and 42 of 79 at follow were taking antipsychotic medications	NR	31.7 (49.5)	Volumetry	1.5T	No relationship between a wide variety of cortical and subcortical volumes and DUP
Hietala	2003	14	SCZ or SAD	First admission and episode SCZ	29.9 (7.3)	NR	14/0/0	NR	36 median	Volumetry	1.5T	Relationship between longer DUP and less left temporal gray matter volume
Ho	2003	156	SCZ, SAD, or SPD	First-episode SCZ, SAD, or SPD; moderately symptomatic based on the SANS and SAPS	25.8 (8.4)	62	124/NR/NR	<3 months for full group	17.3 (33.9) for full group	Volumetry	1.5T	No relationships between DUP and any gray or white matter of cortical lobes, or cerebellar or CSF volumes
Ho	2005	105	SCZ, SAD, or SPD	First-episode SCZ, SAD, or SPD; moderately symptomatic based on the SANS and SAPS	26.6 (9.0)	61	S4/NR/NR	<3 months	14.0 (27.5)	Volumetry	1.5T	No relationships between DUP and hippocampal volumes
Hoff	2000	50	SCZ, SPD, SAD, PNOS	FEP, inpatients	27.4 (7.0)	64	0/0/50	<1 month	11.4 (16.2)	Volumetry	NR	No relationships between DUP and cerebral hemisphere, lateral ventricular, or temporal lobe volumes
Ichimiya	2001	20	SCZ or SPD	SCZ or SPD	28.3 (6.9)	100	20/0/0	NA	61.2 (62.4)	Volumetry	1.5T	No relationship between DUP and cerebellar vermal volume
Jayakumar	2005	18	SCZ	First-episode SCZ; mildly to moderately symptomatic based on the PANSS	24.9 (6.3)	50	18/0/0	NA	10.3 (5.1)	Volumetry	1.5T	No relationship between DUP and overall gray matter, white matter, CSF, and intracranial volumes
John	2008	23	SCZ or SPD	Recent onset (<5 years) SCZ or SPD; outpatient	30.1 (5.8)	44	23/0/0	NA	16.3 (14.8)	Morphometry	1.5T	No relationship between the area of whole corpus callosum or its regions and DUP
Joyal	2003	18	SCZ or SAD	First-episode SCZ or SAD; mildly to moderately symptomatic based on the PANSS	28.0 (7.0)	61	18/0/0	NA	41.0 (34.0)	Volumetry	1.5T	No relationship between amygdaloid volume and DUP
Keshavan	1998a	25	FEP	FEP	25.4 (8.3) SCZ group 21.8 (5.4) Non-SCZ psychosis group	64	25/0/0	NA	37.9 (47.1) SCZ group 23.9 (27.6) Non-SCZ psychosis group	Volumetry	1.5T	DUP was inversely related to the volume of the left superior temporal gyrus, primarily in males
Keshavan	1998b	25	SCZ or SAD	FEP	27.2 (8.9) SCZ group 23.2 (0.7) Non-SCZ psychosis group	60	25/0/0	NA	70.3 (62.8) SCZ group 36.2 (37.1) Non-SCZ psychosis group	Volumetry	1.5T	No relationship between caudate or putamen volumes and DUP
Keshavan	2002	31	SCZ, SAD, or SPD	First-episode SCZ, SAD, SPD; inpatient or outpatient; mildly to moderately symptomatic based on the BPRS	24.2 (8.1)	65	31/0/0	NA	63.6 (65.5)	Morphometry	1.5T	No relationship between the area of whole corpus callosum or its regions and DUP
Lappin	2006	81	FEP	FEP; outpatient or inpatient	26.8 (7.8)	61	NR/NR/69	NR	2.6 (3.5)	Volumetry	1.5T	Longer DUP was associated with gray matter reductions in left middle and inferior temporal, left occipital, and left fusiform cortices, and with gray matter excess of the left basal ganglia. No relationships between DUP and whole gray or white matter volumes or CSF volumes
Madsen	1999	21	SCZ	First-episode SCZ or SPD; inpatients	33 (20–41) median RD group 27 (19–40) median NRD group	NR	NR	<“a few weeks”	5 (1–120) median RD group 60 (1–120) median NRD group	CT Volumetry	NA	DUP predicted greater frontal sulcal enlargement over 5 years
Malla	2002	114	SCZ or SPD	First-episode SCZ or SPD; inpatients and outpatients; mildly to moderately symptomatic based on the SANS and SAPS	27.0 (9.9)	69	73% unmedicated	NR	26.4 (42)	CT Volumetry	NA	No relationship between DUP and CT volumes
Malla	2011	80	FEP	FEP; inpatients and outpatients; mildly symptomatic based on the PANSS	23.4 (3.5) SDUP 24.3 (3.7) LDUP	73	NR/NR/73	NR	1.8 (1.3) SDUP 26.5 (39.8) LDUP	Volumetry	1.5T	Short-DUP patients (<18 weeks) showed larger volumes in whole-brain gray matter, bilateral medial frontal gyri, bilateral rectal gyri, left frontal subgyral, left postcentral gyrus, and right superior parietal lobule than long DUP (>18 weeks) patients. Long DUP patients displayed larger volumes in the left inferior parietal lobule. No differences in whole-brain white matter volume
Penttila	2010	46	SCZ	Psychosis	NR	59	NR	NR	7.6 mean; 4.8 median	Volumetry	1.5T	No relationships between DUP and gray matter or white matter volume or CSF. DUP was inversely associated with right limbic area (hippocampus)
Rapp	2017	23	FEP	FEP; mildly to moderately symptomatic based on the BPRS	27.2 (6.5)	74	14/0/9	<3 weeks	11.3 (16.6)	Volumetry	1.5T	No relationship between DUP and hippocampal, pituitary, or gray matter volume when adjusted for age, sex, and whole-brain volume; a positive relationship between pituitary volume and DUP when adjusting for only whole-brain volume
Roiz-Santianez	2010b	118	SCZ or SPD	First-episode SCZ or SPD	30.0 (9.1)	61	0/0/118	<25 weeks	13.7 (30.2)	Volumetry	1.5T	No relationship between DUP and insular thickness
Roiz-Santianez	2010a	80	SCZ	First-episode SCZ	NR	60	0/0/80	4.4 (3.7) mean weeks	17.8 (35.8) at intake to program (not at time of scan)	Volumetry	1.5T	No relationship between DUP and temporal pole volume
Roiz-Santianez	2011	141	FEP	FEP; inpatients, outpatients, emergency room referrals; mildly to moderately symptomatic based on the BPRS, SANS, and SAPS	29.6 (8.6)	62	0/0/141	NR	12.6 (28.9)	Volumetry	1.5T	No relationship between DUP and straight gyrus volume or surface area
Smith	2012	58	SCZ, SAD, or SPD	First-episode SCZ, SAD, or SPD	20.6 (4.9) val/val group 20.6 (4.5) met group	66	36/0/22	15-day median	13.0 median val/val group 9.7 median met group	Volumetry	1.5T	No relationship between DUP and hippocampal volume
Takahashi	2007	38	SCZ	SCZ inpatients and outpatients with an illness duration less than 5 years; mildly symptomatic based on the SAPS and SANS	24.1 (4.3)	53	NR/NR/37	11.8 (15.7) months	6.6 (10.7)	Volumetry	1.5T	Individuals with longer DUP had smaller left planum temporale; no relationships between DUP and prefrontal volumes
Takayanagi	2010	42	SCZ	First-episode SCZ; inpatients; mildly to moderately symptomatic based on the BPRS	28.6 (5.9) males; 29.7 (5.7) females	57	0/0/42	57.4 days	7.7 (10.3) males; 9.9 (13.6) females	Volumetry	1.5T	No relationship between DUP and orbitofrontal volumes
Tauscher-Wisniewski	2005	37	FEP	FEP	25.5 (5.8)	59	37/0/0	NA	24.4 (31.6)	Volumetry	1.5T	No relationship between DUP and caudate volume when age is used as a covariate
Venkatasubramanian	2003	15	SCZ	SCZ; outpatients; moderately to severely symptomatic based on the PANSS	31 (11)	47	15/0/0	NA	48 (range 6–144)	Volumetry	1.5T	No relationship between DUP and caudate volumes
Venkatasubramanian	2008	51	SCZ	SCZ; moderately symptomatic based on the PANSS	30.2 (7.9)	57	51/0/0	NA	37.8 (36.5)	Volumetry	1.5T	No relationship between DUP and thickness of the left medial orbitofrontal cortex and the remaining volumes of bilateral pars triangularis, bilateral lateral orbitofrontal cortex, or left medial OFC and the thickness of right medial OFC
Venkatasubramanian	2010	66	SCZ	SCZ; moderately symptomatic based on the PANSS	28.6 (7.9)	61	66/0/0	NA	38.2 (35.1)	Morphometry	1.5T	No relationship between DUP and corpus callosum areas
Whitworth	1998	41	SCZ	First-episode SCZ; inpatient	24.5 (4.7)	100	NR	<1 week	8.3 (17.7)	Volumetry	1.5T	No relationship between DUP and total brain volume, each hemisphere, hippocampal volume, or amygdala volume
Xiao	2015	128	SCZ	First-episode SCZ; moderately to severely symptomatic based on the PANSS	24.3 (8.1)	39	128/0/0	NA	11.5 (22.0)	Volumetry	3T	No relationship between DUP and cortical thickness
Yao	2014	64	SCZ	First-episode SCZ; moderately to severely symptomatic based on the PANSS	24.2 (8.6)	44	64/0/0	NA	8.6 (14.3)	Volumetry	3T	No relationship between white matter volume and DUP in left posterior limb of the internal capsule, the right posterior limb of the internal capsule, and right subgyral frontal white matter near the precentral gyrus
Zanetti	2008	129	FEP	FEP	29.0 (range 18–50)	55	NR/NR/76	NR	22.8 (47.8)	White matter hyperintensities	1.5T	No relationship between DUP and white matter hyperintensities

Dx diagnosis, Tx treatment, AP N/F/M antipsychotic naive/free/medicated, FEP first episode of psychosis, SCZ schizophrenia, SAD schizoaffective disorder, SPD schizophreniform disorder, PNOS psychosis NOS, SDUP short duration of untreated psychosis, LDUP long duration of untreated psychosis, DUP duration of untreated psychosis, CT computed tomography, MRI magnetic resonance imaging, fMRI functional magnetic resonance imaging, rsfMRI resting-state functional magnetic resonance imaging, NIRS near-infrared spectroscopy, MRS magnetic resonance spectroscopy, STEAM stimulated echo acquisition mode, PRESS point-resolved spectroscopy, DTIL diffusion tensor imaging, FA fractional anisotropy, EEG electroencephalogram, PET positron emission tomography, T Tesla, NA not applicable, NR not reported, RD responder, NRD non-responder, BPRS Brief Psychiatric Rating Scale, PANSS Positive and Negative Syndrome Scale.

Fig. 1. PRISMA figure.

Flow chart showing the identification and review of articles included in this review.

RESULTS

PET

A few PET studies have examined relationships between neurobiology and DUP (Table 1). In a recent study, investigators examined the relationship between the brain’s endocannabinoid system and DUP using the fatty acid amide hydrolase radioligand [¹¹C]CURB in 27 patients with first-episode psychosis (FEP) (16 of whom were antipsychotic free). They found that longer DUP was related to higher fatty acid amide hydrolase in the brain including in regions such as the amygdala, hippocampus, striatum, dorsolateral prefrontal cortex (DLPFC), medial prefrontal cortex (MPFC), and anterior cingulate cortex (ACC) [34]. The same group examined inflammation in 19 antipsychotic free individuals with FEP (14 of whom were antipsychotic naive) using [¹⁸F]FEPPA [35]. They observed no relationships between DUP or duration of illness (DUI) and inflammation in whole brain, hippocampus, DLPFC, temporal lobe, total gray matter, or MPFC.

NIRS

We found three studies that have examined relationships between neurobiology and DUP using NIRS (Table 1). In one study of 73 patients with chronic schizophrenia, Nishimura et al. observed no relationship between DUP and activation in the prefrontal cortex during a verbal fluency task in any early growth response 3 genotype group [36]. In a similar study of 62 patients with schizophrenia within the first 10 years of their illness, nearly all of whom were being treated with antipsychotic medications, investigators observed that DUP was negatively associated with activity in left inferior frontal gyrus, left middle frontal gyrus, left postcentral gyrus, right precentral gyrus, bilateral superior temporal gyrus, and bilateral middle temporal gyrus in the group of patients who had been treated for more than 6 months [37]. There were no relationships between DUP and cortical activity in the short duration of treatment (≤6 months) group. Similar methods were used in a study of NIRS in 28 patients with first-episode schizophrenia, all of whom were receiving antipsychotic medications [38]. The authors observed no relationship between DUP and frontotemporal activity during verbal fluency tests.

MRS

Several studies have used MRS to examine metabolites in the brains of individuals with psychotic disorders and assess their relationship with DUP (Table 1). Bryll et al. reported an inverse relationship between DUP and glucose in the right frontal lobe and a positive relationship between DUP and choline (Cho) in the anterior cingulate of 40 antipsychotic free individuals with schizophrenia who were also experiencing acute decompensation [39]. Another group recently examined DUP using glutamate MRS in 59 antipsychotic naive patients with FEP [40]. They found that those patients with longer DUP (>12 months) had higher glutamate and glutamine (Glx) levels in the hippocampus (left) than did those with short DUP (<12 months). In 19 patients with schizophrenia, nearly all of whom were being treated with antipsychotic medications, Shirayama et al. reported that DUP was positively related to the N-acetyl-l-aspartate (NAA)/glycerophosphorylcholine plus phosphorylcholine (GPC+PC) ratio and negatively related to the (GPC+PC)/creatine plus phosphocreatine (Cr+PCr) ratio in the MPFC [41]. Galinska et al. examined 30 patients with schizophrenia, nearly all of whom were being treated with antipsychotic medications, and found no relationship between DUP and NAA, Glx, Cho or mI in left frontal lobe, left thalamus, or left temporal lobe [42]. Another study of 19 patients with the first-episode, antipsychotic naive schizophrenia examined NAA and choline in left ACC and left thalamus using a 4 Tesla MR scanner [43]. They found that DUP was positively correlated with choline, but not NAA, in left anterior cingulate and left thalamus. Finally, in a 31-phosphorus MRS study of 12 antipsychotic naive patients with schizophrenia, Jayakumar et al. observed no relationships between DUP and ³¹P metabolites [44].

EEG

In one EEG study, Manchanda et al. did not find differences in DUP between individuals with and without EEG abnormalities in a group of 122 patients with FEP, the majority of whom were not taking antipsychotic medications [45] (Table 1). Similar findings were reported by the same group apparently overlapping samples of 117 patients with FEP [46] and 103 patients with FEP [47]. Another group examined 18 antipsychotic naive patients with first-episode schizophrenia [48]. They found that neither P300 amplitude nor latency differed between patient groups with short (<2 years) or long (≥2 years) DUP.

Functional MRI studies

We reviewed a total of nine functional MRI studies including seven resting-state functional MRI (rsfMRI) and two task-based fMRI studies (Table 2). With respect to rsfMRI, Sarpal et al. studied 83 individuals with FEP (37% treatment naive) and found an inverse correlation between DUP and functional connectivity between striatum and several cortical regions including the supramarginal gyrus, middle frontal gyrus, and cingulate gyrus [49]. They also identified a positive relationship between increasing DUP and functional connectivity between the ventral striatum/nucleus accumbens and the subcallosal cortex/orbitofrontal cortex, and medial frontal pole [49]. Although they found no effect of medication status on the length of DUP or connectivity measures, they found that DUP played a mediating role in treatment response through corticostrial connectivity abnormalities [49]. Two additional studies identified abnormalities in connectivity using rsfMRI including a negative correlation with DUP and functional connectivity in the default mode, salience, and central executive networks [50] and select sensory-motor subnetworks [51], though 31% of participants tested positive for cannabis in the former study. In contrast, the remaining rsfMRI studies found no relationship between DUP and alterations of network homogeneity of the default mode network [52], homotopic connectivity across a variety of brain regions [53], gray matter-white matter functional synchrony [54], and amplitude of low-frequency fluctuations across various brain regions and the default mode network [54, 55]. The two task-based fMRI studies we reviewed found no relationship between DUP and differential activation of the DLPFC [56, 57], though one of the studies noted decreased frontostriatal connectivity with the maintenance of increasing working memory load [57].

Table 2.

DTI and fMRI studies of DUP.

Author	Year	n	Dx	Clinical severity and/or phase of illness	Mean age (SD) years	% Male	AP N/F/M	AP Tx length	Mean DUP/DUI (SD) months	Imaging measure	Magnet strength	Result
Collinson	2014	113	SCZ	65 first-episode SCZ; 48 chronic SCZ; all in a state psychiatric hospital; minimal symptoms based on the PANSS	29.3 (7.0) in the first-episode group; 37.0 (9.0) in the chronic group	74	0/0/113	NR	18.6 (25.1) first-episode group; 14.2 (14.6) in the chronic group	DTI	3T	No relationship between DUP and corpus callosum volume or fractional anisotropy measures in corpus callosum
Filippi	2014	43	SCZ	First-contact SCZ; moderately to severely symptomatic based on the PANSS	29.3 (7.4)	56	43/0/0	NA	7.9 (9.7)	DTI	1.5T	Negative relationship between DUP and MD values of the cerebral peduncles, pons and medulla oblongata, middle cingulum, parahippocampal tracts and inferior longitudinal fasciculi bilaterally, and right posterior limb of the internal capsule, right thalamic radiations, right inferior fronto-occipital, and right uncinate fasciculi as well as a positive relationship between DUP and FA values of the posterior limb of the internal capsule, superior longitudinal and inferior fronto-occipital fasciculi bilaterally, and the right uncinate fasciculi
Guo	2012	20	SCZ	First-episode paranoid SCZ; mildly to moderately symptomatic based on the PANSS	24.0 (4.9)	45	20/0/0	NA	6.6 (3.1)	DTI	1.5T	No relationship between DUP and FA in right superior longitudinal fasciculus II, the right fornix, the right internal capsule, and the right external capsule
Kraguljac	2020	66	FEP	FEP; inpatients, outpatients, emergency rooms; moderately symptomatic based on the BPRS	23.8 (6.2)	65	55/NR/NR	<5 days	20.6 (38.9)	DWI	3T	DUP was negatively correlated with whole-brain fractional anisotropy and axial diffusivity compared to controls. No relationship between DUP and mean and radial diffusivity
Lee	2018	95	FEP	FEP; moderately symptomatic based on the SANS and SAPS	33.5 (11.9)	33	NR	8.2 (9.4) days	3.3 (5.0)	DTI	NR	Fractional anisotropy values of left tapetum (part of splenium of corpus callosum) were significantly inversely correlated with DUP
Liu	2013	17	SCZ	Chronic SCZ	38.5 (3.9)	41	17/0/0	NA	184.9 (76.0)	DTI	3T	No relationships between DUP and FA in left inferior longitudinal fasciculus (ILF) and left inferior fronto-occipital fasciculus (IFOF)
Liu	2021	153	SCZ	SCZ; inpatients or outpatients; moderately symptomatic based on the PANSS	27.1 (11.4)	48	153/0/0	NA	41.2 (88.4)	DTI, rsfMRI	3T	No significant associations between DUP and amplitude of low-frequency fluctuations, gray matter-white matter functional synchrony, and fractional anisotropy
Luck	2010	32	SCZ	First-episode SCZ; mildly to moderately symptomatic based on the PANSS	23.6 (0.7)	69	NR/NR/27	157.3 (22.0) days	2.4 (1.1)	DTI	1.5T	No relationships between DUP and FA in left or right fornix
Luck	2011	44	FEP	FEP	23.3 (0.5)	70	NR/NR/36	NR	3.5 (1.5)	DTI	1.5T	No relationships between DUP and uncinate or superior longitudinal fasciculi
Mandl	2013	16	FEP	First episode; mildly symptomatic based on the PANSS	23.4 (3.5)	81	16/0/0	NA	22.2 (35.6)	DTI	1.5T	There was a positive relationship between a combination of fractional anisotropy, mean diffusivity, and magnetization transfer ratio in the genu of the corpus callosum
Zeng	2021	56	SCZ	First-episode SCZ; moderately symptomatic based on the PANSS	24.2 (7.7) good outcome patients; 24.4 (10.0) bad outcome patients	45	56/0/0	NA	1.75 (1, 12; IQR) good outcome patients; 6 (1, 23; IQR) bad outcome patients	DTI	3T	No relationship between DUP and fractional anisotropy and DUP did not aid in separate good versus poor outcome patients
Guo	2014b	49	SCZ	Inpatients with first-episode paranoid SCZ; moderately symptomatic based on the PANSS	22.7 (4.6)	61	49/0/0	NA	22.5 (6.7)	rsfMRI	3T	No relationship between DUP and alterations of the network homogeneity of the default mode network
Guo	2014a	49	SCZ	Inpatients with first-episode paranoid SCZ; moderately symptomatic based on the PANSS	22.7 (4.6)	61	49/0/0	NA	22.5 (6.7)	rsfMRI	3T	No relationship between DUP and voxel-mirrored homotopic connectivity in the precuneus, the precentral gyrus, the superior temporal gyrus, the middle occipital gyrus, and the fusiform gyrus/cerebellum lobule VI
Manivannan	2019	37	SCZ, SAD, SPD, PNOS	FEP; mildly symptomatic based on the BPRS	22.2 (5.1)	70	11/0/26	<2 months	34.7	fMRI	3T	No relationship between DUP and visuospatial working memory performance or activation in the dorsolateral prefrontal cortex (DLPFC), DUP was associated with decreased frontostriatal functional connectivity in a cluster of voxels within the DLPFC
Maximo	2020	55	FEP	FEP; inpatient and outpatient and emergency room; mildly symptomatic based on the BPRS	24.2 (6.3)	62	24/31/0	<5 days	19.7 (39.4)	rsfMRI and morphometry	3T	DUP was negatively correlated with functional connectivity in the default mode network (DFM), salience network (SN), and central executive network (CEN), and surface area in the SN and CEN. DUP was positively correlated with cortical thickness in the DMN and SN.
Niendam	2018	87	SCZ, SAD, or SPD	SCZ, SAD, SPD	19.6 (3.0)	84	NR/NR/61	NR	4.9 (4.5)	fMRI	1.5T	No relationship between DUP and performance or activation in DLPFC
Ren	2013	100	SCZ	First episode; moderately to severely symptomatic based on the PANSS	24.3 (7.5)	41	100/0/0	NA	6.3 (11.0)	Volumetry and rsfMRI	3T	No differences in volumetry or rsfMRI between groups with short (<1 year) or long (>1 year) DUP
Sarpal	2017	83	SCZ, SPD, SAD, PNOS	Mildly to moderately symptomatic based on the BPRS	21.9 (6.0) RD group 21.2 (3.7) NRD group	73	31/NR/NR	15 (59) days	23.8 (18.0)	rsfMRI	3T	Lower functional connectivity was associated with greater DUP between striatum and cortex that overlapped in several regions, including the supramarginal gyrus, middle frontal gyrus, and cingulate gyrus. Positive relationship between increasing DUP and functional connectivity between the ventral striatum/nucleus accumbens and the subcallosal cortex/orbitofrontal cortex, and medial frontal pole
Zhang	2019	60	SCZ	First-episode SCZ; mildly to moderately symptomatic based on the BPRS	25.6 (7.0)	40	60/0/0	NA	8.0 (8.8)	rsfMRI	3T	DUP was associated with decreased functional network connectivity in select sensory-motor subnetworks (left/right upper limb—trunk), but not others (auditory—head/face)

Dx diagnosis, Tx treatment, AP N/F/M antipsychotic naive/free/medicated, FEP first episode of psychosis, SCZ schizophrenia, SAD schizoaffective disorder, SPD schizophreniform disorder, PNOS psychosis NOS, SDUP short duration of untreated psychosis, LDUP long duration of untreated psychosis, DUP duration of untreated psychosis, CT computed tomography, MRI magnetic resonance imaging, fMRI functional magnetic resonance imaging, rsfMRI resting-state functional magnetic resonance imaging, NIRS near-infrared spectroscopy, MRS magnetic resonance spectroscopy, STEAM stimulated echo acquisition mode, PRESS point-resolved spectroscopy, DTIL diffusion tensor imaging, FA fractional anisotropy, EEG electroencephalogram, PET positron emission tomography, T Tesla, NA not applicable, NR not reported, RD responder, NRD non-responder, BPRS Brief Psychiatric Rating Scale, PANSS Positive and Negative Syndrome Scale.

Diffusion tensor/weighted imaging MRI studies

We reviewed eleven studies that used diffusion tensor imaging (DTI) or diffusion-weighted imaging (DWI) to evaluate white matter integrity and its relation to DUP (Table 2). Although four of the eleven studies identified a significant relationship between fractional anisotropy (FA) and DUP, there was little overlap with respect to the brain regions identified. For example, Kraguljac et al. conducted DWI for 66 antipsychotic naive individuals with FEP (both nonaffective and affective) and found that DUP was negatively correlated with whole-brain FA [58]. They also found that there was a mediation effect for FA on treatment outcomes during a 16-week trial of risperidone, suggesting that DUP can affect treatment outcomes by affecting white matter integrity [58]. While these results are intriguing, the lack of replication of the relationship between whole-brain FA and DUP suggests that they must be interpreted with caution. Other studies identified relationships between FA in the genu of the corpus callosum [59] and the left tapetum [60], but not whole-brain FA. Notably, of the studies with significant findings, the relationship between FA and DUP was inversely correlated in two studies and positively correlated in the other two [59, 61].

In contrast, seven DTI/DWI studies identified no relationship between FA and DUP [54, 62–67] including the largest of the DTI/DWI studies which included 153 antipsychotic naive participants with schizophrenia [54]. The absence of a statistically significant relationship between FA and DUP in the majority of studies (including those with the largest sample size) and the lack of a clear direction of effect or convergent brain region suggests that positive findings in the literature between FA and DUP should be interpreted with caution. Several studies examined mean diffusivity (MD) and its relationship with DUP though the results were inconclusive. Mandl et al. found an increase in MD in the genu with increasing DUP [59] while Filippi et al. found decreased MD with longer DUP across a variety of brain regions [61]. Given the opposite direction of effect and a lack of relationship between MD and DUP found in the largest study done by Kraguljac et al. [58], the relationship between MD and DUP remains unclear.

Structural imaging studies and DUP

We reviewed 49 MRI and 2 CT studies that examined structural measures including volume and thickness across a variety of brain regions and their relationship to DUP (Table 3). There was significant heterogeneity in study design among the studies reviewed. Some studies focused on one brain region while others tested for associations across many regions. We thus separated the results of these studies into whole brain, cortical, cerebellar, ventricular, subcortical, and white matter subgroups (Table 4).

Table 4.

Summary of structural imaging studies associations.

Structure/region	Studies with significant association	Studies with no association	Percent of studies with significant association (%)
Whole-brain volume	1	9	10
Cortical thickness	0	15	0
Frontal lobe	3	8	27.3
Orbitofrontal cortex	0	5	0
Temporal lobe	4	7	36.4
Temporal pole	0	1	0
Planum temporale	1	0	100
Fusiform gyrus	3	0	100
Parietal lobe	1	5	16.7
Occipital lobe	2	4	33.3
Insular cortex	0	4	0
Ventricles	1	10	9.1
Cerebellum	1	6	14.3
Amygdala	0	3	0
Caudate	1	9	10
Hippocampus	5	7	41.7
Nucleus accumbens	0	1	0
Putamen	0	5	0
Thalamus	1	4	20
White matter volume	0	11	0
Corpus callosum	0	4	0

Whole brain, cortical, cerebellar, and ventricular studies

With respect to the relationship between whole-brain volume and DUP, a study by Malla et al. including 80 individuals with FEP found that individuals with a shorter DUP (<18 weeks) had larger total brain volumes than those with a longer DUP (>18 weeks) [68]. However, nine other studies found no relationship between whole-brain volume and DUP [69–77]. Moreover, none of the 15 studies that tested for an association between total cortical thickness and DUP identified a significant relationship [68–71, 73–75, 77–84]. These results suggest that DUP is unlikely to contribute substantially to whole-brain volume loss or cortical thinning.

Studies of the four main cortical lobes were more likely to identify significant associations, though the results remain mixed with inconsistent findings at the subregional level. Several studies examining the effects of DUP on the frontal lobe identified a significant relationship. Guo et al. found an inverse relationship between the right superior frontal gyrus and DUP in 57 antipsychotic naive individuals with schizophrenia [85]. Malla et al. identified an inverse relationship in three frontal lobe areas including the bilateral medial frontal gyri, bilateral straight gyri, and the left subgyral region [86]. Lastly, a CT scan study by Madsen et al. identified a relationship between greater sulcal enlargement and DUP in 21 individuals with first-episode schizophrenia [87]. In contrast to these findings, five studies specifically examining the orbitofrontal cortex [55, 84, 88–90] and eight studies of the frontal cortex did not identify a relationship between these regions and DUP [70, 71, 74, 78, 84, 91–93].

Two volumetric studies of the relationship between DUP and the temporal lobe found generalized gray matter reductions [82, 94] while two others found an inverse relationship between DUP and the superior temporal gyrus [31, 85]. A single study examining the effects of DUP on the planum temporale also identified an inverse relationship [93]. However, seven studies of the temporal lobe found no relationship between temporal lobe volumes and DUP [71, 73, 74, 78, 81, 84, 91]. Four studies specifically examining the insular cortex found no relationship with DUP [55, 95–97]. Interestingly, all three studies that specifically examined the fusiform gyrus identified a significant inverse relationship between volumetric measures of the left fusiform gyrus and DUP [82, 85, 98]. Studies of the occipital lobe were also mixed with one finding an association between DUP and left occipital lobe volume [82] and another identifying an inverse relationship between the lingual gyrus and DUP [98], but four studies found no relationship [71, 74, 78, 84]. Only one study [86] identified an association between parietal lobe structural measures and DUP while five found no association [55, 71, 74, 78, 84]. One study focused on the cerebellum [98] detected an inverse association between the declive of the cerebellar vermis and DUP while six other studies of the cerebellum found no relationship [31, 69, 70, 74, 80, 99]. Lastly, ten studies that examined the relationship between ventricular size and DUP found no relationship between the two [69–74, 77, 80, 81, 100].

Subcortical regions and white matter studies

There were no significant associations between DUP and several subcortical regions including the amygdala [77, 80, 101], putamen [71, 72, 80, 102, 103], and nucleus accumbens [102]. Although there was one study by Crespo-Facorro et al. of 76 individuals in their first episode of psychosis that identified an inverse relationship between DUP and caudate size [30], there were nine studies that did not find a similar relationship [44, 70–72, 80, 102–105]. Another study by the same group [106] identified an inverse association between the thalamic volume and DUP, this was only significant in the regression analysis and not in the correlational analysis. Moreover, four other studies of the thalamus, including two by the same group of researchers, did not identify a significant relationship between the thalamus and DUP [30, 55, 70, 71]. Studies of the hippocampus were more mixed in their findings. Two studies [85, 98] identified an inverse relationship between DUP and the right parahippocampal gyrus and an additional two studies identified an inverse relationship between DUP and the left [40] and right hippocampus [83], respectively. A study by Goff et al. did not identify a relationship between DUP and left or right hippocampal volumetric integrity at baseline, but did observe a left-sided annualized reduction in hippocampal volumetric integrity correlated with DUP in the 8-week follow-up scan [107]. Seven other studies examining the relationship between DUP and the hippocampus found no association [70, 72, 76, 77, 80, 108, 109]. Lastly, structural imaging studies of white matter and its relation to DUP have been unrevealing to date. Four studies of the corpus callosum [62, 110–112] and eleven studies of generalized white matter volume revealed no association with DUP [69–71, 74, 75, 80, 82, 83, 86, 113, 114].

DISCUSSION

In this manuscript, we built upon previous work examining imaging correlates of DUP [32] by reviewing 83 studies of neurobiologic correlates of DUP including structural and functional imaging, EEG, and neurobiochemical measures. Overall, 27 out of the total 83 studies (32.5%) reported a significant association with DUP. Here, we review what the field can learn from these studies and provide recommendations for future studies of the neurobiological basis of DUP.

Overall, studies were of medium quality with an average total quality score of 5.08 out of 10 (Table 5). Potential sources of bias were the lack of reporting of non-participation rates (0% of studies fully met these criteria), lack of adjustment for potential confounds (15.6% of studies), and lack of clear definitions and details of ascertainment for both participant’s primary diagnosis (18.1% of studies) and DUP/DUI (24.1% of studies). The most notable limitation of the studies we reviewed was their sample size (mean = 58.8, median = 51). Recent research has shown that the effect size of imaging differences in psychological traits tends to be small and that brain-wide behavioral phenotypic associations become more reliable around 2000 participants [115]. Moreover, smaller studies (n < 200) may suffer from the “underpowered correlation paradox” [115] wherein small studies are only powered to detect the largest effects/correlations, which are often spurious due to sampling variability. This problem becomes exacerbated by publication bias which then leads to improper power calculations for future studies based on inflated effect sizes [115]. While we appreciate that the cost to conduct such large-scale studies for any individual group is prohibitive, we advocate for the approach taken by large-scale consortia such as the Enhancing Neuro Imaging Genetics through Meta-Analysis consortium [116] and the Adolescent Brain Cognitive Development study [117]. Indeed, through these efforts, we are beginning to identify robust statistical associations in schizophrenia [118] that point toward high confidence brain differences in schizophrenia for further study.

Table 5.

Assessment of study quality and risk of bias.

Author	Year	Sample size	Nonparticipation rate	Clinical definitions, ascertainment, and reporting of primary diagnosis	Clinical definitions, ascertainment, and reporting of DUP/DUI	Adjustment for confounding and/or bias	Methodology and reporting of neurobiological data	Total quality score
Bangalore	2009	82	−	+	+	+	+	8
Behere	2009	14	−	*	−	−	+	3
Boonstra	2011	57	−	*	+	*	+	6
Briend	2020	54	*	*	+	+	+	8
Bryll	2020	40	−	*	*	−	+	4
Cahn	2002	20	−	*	+	*	+	6
Chou	2014	62	−	+	+	+	+	8
Chou	2015	28	−	+	+	−	+	7
Collinson	2014	113	−	+	*	−	+	5
Crespo-Facorro	2007b	61	*	+	+	*	+	8
Crespo-Facorro	2007a	76	*	+	+	*	+	8
Crespo-Facorro	2009	142	*	+	+	+	+	9
Crespo-Facorro	2010	142	*	+	+	−	+	7
Crespo-Facorro	2011	142	*	+	+	*	+	8
Duggal	2005	30	−	*	*	−	+	4
Ebdrup	2010	38	−	*	*	−	+	4
Emsley	2017	23	*	*	*		+	5
Fannon	2000	37	−	*	*	−	+	4
Filippi	2014	43	−	*	+	−	*	4
Galinska	2009	30	−	*	+	−	+	5
Goff	2018	57	*	+	+	*	+	8
Gunduz	2002	51	−	*	*	−	+	4
Guo	2012	20	−	*	*	−	*	3
Guo	2013	57	*	*	+	+	+	8
Guo	2014b	49	−	*	*	−	*	3
Guo	2014a	49	−	*	*	−	*	3
Gutierrez-Galve	2010	37	*	*	*	*	*	5
Hafizi	2017	19	*	*	*	*	+	6
Haukvik	2016	79	*	*	*	−	*	4
Hietala	2003	14	−	*	−	*	*	3
Ho	2003	102	*	*	*	+	*	6
Ho	2005	105	−	*	*	+	*	5
Hoff	2000	50	−	*	+	−	+	5
Ichimiya	2001	20	−	*	*	−	+	4
Jayakumar	2005	18	−	*	*	−	+	4
Jayakumar	2006	12	−	*	+	−	+	5
John	2008	23	−	+	*	*	+	6
Joyal	2003	18	−	*	*	−	+	4
Keshavan	1998a	25	−	*	+	+	+	7
Keshavan	1998b	25	−	*	+	*	+	6
Keshavan	2002	31	−	*	+	−	+	5
Kraguljac	2020	66	*	*	*	*	+	6
Lappin	2006	81	*	*	*	*	+	6
Lee	2018	95	−	*	*	−	*	3
Liu	2013	17	−	*	*	*	*	4
Liu	2021	153	−	*	*	+	+	6
Luck	2010	32	−	*	*	−	+	4
Luck	2011	44	−	*	*	−	+	4
Madsen	1999	21	−	*	*	−	*	3
Malla	2002	114	−	*	*	−	*	3
Malla	2011	80	*	*	*	*	*	5
Manchanda	2005	122	−	−	*	−	*	2
Manchanda	2008	117	*	−	*	*	*	5
Manchanda	2014	103	*	−	−	−	*	2
Mandl	2013	16	*	+	*	*	+	7
Manivannan	2019	37	−	+	*	*	+	6
Maximo	2020	55	*	+	*	+	+	8
Niendam	2018	87	−	*	*	*	+	5
Nishimura	2014	73	−	*	−	−	+	3
Penttila	2010	61	*	−	*	−	*	3
Rapp	2017	23	−	+	*	*	+	6
Ren	2013	100	*	*	*	+	*	6
Roiz-Santianez	2010b	118	*	*	*	*	+	6
Roiz-Santianez	2010a	80	*	*	*	−	*	4
Roiz-Santianez	2011	141	*	*	*	*	+	6
Sarpal	2017	83	−	*	*	*	+	5
Shirayama	2010	19	−	*	*	−	+	4
Smith	2012	58	*	*	*	*	+	6
Takahashi	2007	38	*	*	*	+	+	7
Takayanagi	2010	42	−	*	*	*	+	5
Tauscher-Wisniewski	2005	37	−	*	*	*	+	5
Theberge	2004	19	−	*	*	*	*	4
Venkatasubramanian	2003	15	−	*	*	−	+	4
Venkatasubramanian	2008	51	−	*	*	*	+	5
Venkatasubramanian	2010	66	−	*	*	−	+	4
Wang	2005	18	−	*	*	−	+	4
Watts	2020	27	−	*	*	+	+	6
Whitworth	1998	41	−	*	*	−	*	3
Xiao	2015	128	−	*	*	−	+	4
Yao	2014	64	−	*	*	−	+	4
Zanetti	2008	129	*	*	*	−	+	5
Zeng	2021	56	−	*	*	−	+	4
Zhang	2019	60	−	*	*	*	+	5

− Criteria not met

^*Criteria partially met

+ Criteria met; 0 points for −; 1 point for *; 2 points for +.

The variety of participant diagnoses, medication and substance use status, study designs, imaging techniques, and statistical approaches allowed us to consider numerous variables while synthesizing the results. With respect to participant diagnosis, studies ranged from using broadly defined first episode of psychosis including affective psychosis to schizophrenia. Some studies included participants with co-occurring substance use disorders including alcohol and cannabis use disorders, both of which have inconsistently been shown to affect brain structure and function [119, 120], and many others did not. In addition, studies examined both medicated and unmedicated participants, and often included the duration of treatment in their analyses. This is an important variable to consider given that antipsychotic medications have been correlated with cortical thinning [121, 122] and the fact that those with shorter DUPs had more time on medication. The studies included in this review spanned a wide range of DUP, with some studies having a mean DUP as short as 1.4 months [87] and some nearly 4 years [44]. Relatedly, while some structural and functional studies examined one or several regions, others were brain-wide association studies looking at dozens of regions.

We will highlight several potential takeaways and areas of interest for future investigation in well-powered studies. First, while there is clear clinical evidence that a longer DUP is associated with worse functional outcomes [4, 5], the data reviewed do not suggest that there is a large effect of DUP on brain-wide cortical thickness or white matter volume as all fifteen studies of total cortical thickness and eleven studies of white matter volume were negative. Similar conclusions can likely be drawn for whole-brain volume and ventricular volume where only one study out of ten and eleven total studies were positive, respectively. While regional cortical studies were mostly mixed or negative, interestingly, half the studies examining temporal regions (i.e., including planum temporale, fusiform gyrus, hippocampus) and all three studies of fusiform gyrus found a significant association, highlighting a region for potential future study with larger samples. The functional and molecular imaging studies reviewed were generally equivocal to negative.

There were several limitations of the studies included in this review. As many studies likely did not predefine their regions of interest before conducting their research, significant results may have been identified through post-hoc analyses which are subject to increased type I error [123]. In addition, the majority of studies did not account for multiple comparisons, and those that reported significant findings noted that the results would not remain significant when accounting for multiple comparisons. Further, two studies in particular identified more significant associations with DUP and discrete brain regions than any other with Bangalore et al. [98] identifying eight and Malla et al. [86] identifying 14 significant associations. Notably, these two studies accounted for the majority of brain region-specific associations with DUP in a prior review [32]. Lastly, the studies reviewed differed in the equipment used, motion artifact correction, and image processing techniques, further exacerbating the issues posed by the clinical heterogeneity of the studies. Unfortunately, we were unable to address this heterogeneity either through meta-analysis or other statistical techniques to combine the data as nearly half of the studies did not report parameter estimates for their associations and few of the studies made their data publicly available. Moreover, given the significant heterogeneity of the studies, creating smaller homogenous subgroups for deeper analysis would significantly limit our statistical power.

These results provide evidence against the notion of psychosis as structurally or functionally neurotoxic on a global scale [29, 124] and suggest that specific regions of the brain, such as temporal regions, may be more vulnerable to the effects of DUP. It is also possible that current methodologies lack the resolution needed to more accurately examine the effects of DUP on the brain. For example, it has been suggested that atrophy in hippocampal circuits may be limited to neuropil [125]. Synaptic dysfunction has long been proposed as a central mechanism in the development of schizophrenia [25, 126]. Post-mortem studies have demonstrated reduced synaptic spine density in schizophrenia [127, 128]. Newer methodologies, such as MR scanners with stronger magnets, and PET imaging with newer ligands capable of measuring subcellular structures (e.g., the PET ligand [¹¹C]UCB-J [129]), may be better able to capture these limited neuropathologic processes.

Here, we suggest additional procedural recommendations that could further ensure rigor and scientific value in future work on the neurobiologic effects of DUP. For example, the studies reviewed typically enroll any individual in their first episode of psychosis without separating individuals based on their severity or clinical course, potentially missing relevant subtypes of schizophrenia. Indeed, there is existing evidence that patients with very poor outcome psychosis can exhibit progressive changes on neuroimaging [130] and that there are clearly different trajectories in the first episode of psychosis with some patients having a stable and others a more deteriorative course of schizophrenia [131]. Future studies that aim to stratify individuals by their clinical course based on pre and post-morbid cognitive testing or illness severity may be better able to detect structural and functional brain imaging changes related to DUP. Longitudinal studies that examine neurobiology over six months or longer, and in particular those that incorporate treatment components, may help the field to better understand effects over time and in particular which patients may respond better to antipsychotic medications or other treatments. In sum, future studies of the potential effects of DUP on brain structure and function should be well powered, conducted in a systematic fashion with careful attention paid to potential confounders and methodogical issues, include longitudinal and preferably treatment components, consider stratifying participants based on their course of illness and disease severity, and incorporate newer imaging methodologies that are more sensitive to detect subtle abnormalities that may be associated with DUP.