Characteristics of very late-onset schizophrenia-like psychosis as prodromal dementia with Lewy bodies: a cross-sectional study

This study was a retrospective observational study without any intervention, and the information of all participants was anonymized prior to analysis as unlinked data to prevent the identification of personal information. The authors assert that all procedures contributing to this work comply with the ethical standards of the relevant national and institutional committees on human experimentation and with the Helsinki Declaration of 1975, as revised in 2008. The study was approved by the Research Ethical Committee of the Osaka University Hospital (Suita, Japan).

We considered a retrospective cohort of non-demented patients with VLOSLP from the database of the neuropsychology clinic in the Department of Psychiatry at Osaka University Hospital, from January 2014 to March 2021, according to the criteria in the previous studies of VLOSLP [4, 11]. The inclusion criteria were as follows: (1) onset of delusions and/or hallucinations at an age of 60 years or more and (2) existence of delusions and/or hallucinations during the evaluation period. The existence of psychosis (delusions and/or hallucinations) was confirmed by the subitems of delusions and hallucinations in the Neuropsychiatric Inventory (NPI) if there was a reliable family informant or 30 points or more on the Brief Psychiatric Rating Scale (BPRS). The exclusion criteria were as follows: (1) cognitive impairment (Mini-Mental State Examination [MMSE] score < 24); (2) diagnosis of dementia (Clinical Dementia Rating [CDR] ≥ 1); (3) diagnosis of affective disorder; (4) comorbidity of Parkinsonism; (5) abnormal localized findings on magnetic resonance imaging (MRI), such as cerebrovascular disease or brain tumor; and (6) uncontrolled physical disorders that could cause psychosis.

In our neuropsychological clinic, we assessed physical condition, demographic data, and medical history and performed standard neuropsychological examinations. In addition, the patients underwent routine laboratory tests and brain neuroimaging.

General cognition and memory were assessed using MMSE and logical memory in the Wechsler Memory Scale-Revised, respectively. In addition, attention, psychomotor speed, visuospatial cognition, and language were evaluated using the Digit Span, Digit Symbol Substitution Test (DSST), Block Design Test, and Information in Wechsler Adult Intelligence Scale-III (WAIS-III). These assessments were conducted by clinical psychologists and neuropsychiatrists specializing in geriatric psychiatry.

Neuropsychiatric symptoms and cognitive fluctuation were assessed using the NPI-plus [19], which is the original NPI-12 [20] with an additional subitem for cognitive fluctuation, if there was a reliable family informant. If there was no reliable family informant, psychosis was assessed using the BPRS. These assessments were also conducted by neuropsychiatrists specializing in geriatric psychiatry.

During data collection, cases lacking data on MMSE, CDR, and NPI or BPRS were excluded to select patients who met the inclusion criteria. Missing data on WAIS-III was allowed.

We assessed the contents of psychosis according to sub-questions in the delusions and hallucinations section in the NPI. If there were no reliable family informants, the contents of psychosis were judged based on medical history and observation by the geriatric neuropsychiatrists. To evaluate the features of auditory hallucinations in DLB [21], we checked whether auditory hallucinations were accompanied with visual hallucinations (AH+VH) or not (AH–VH).

We collected data from dopamine transporter single-photon emission computed tomography (DAT SPECT) using ¹²³I-labeled N-ω-fluoropropyl-2β-carbomethoxy-3β-[4-iodophenyl] nortropane and MIBG myocardial scintigraphy to evaluate the indicative biomarkers of DLB. We judged the results according to the cutoff values of the specific binding ratio for DAT SPECT [22] and heart-to-mediastinum ratio for MIBG myocardial scintigraphy [23]. Patients who underwent testing for indicative biomarkers of DLB were divided into two groups according to whether the results were positive (VLOSLP+LB) or negative (VLOSLP–LB).

We collected the following MRI data when available. MRI was performed using a l.5T system (Signa Excite HD 12.x, General Electric Medical Systems, Milwaukee). We acquired a three-dimensional volumetric T1-weighted gradient echo sequence to produce a gapless series of thin sagittal sections covering the whole brain. The operating parameters were as follows: field of view = 240 mm; matrix = 256 × 256; 124 × 1.40 mm contiguous sections; repetition time = 12.55 ms; echo time = 4.20 ms; and flip angle = 15°.

The individual MR images were segmented using “New segment” implemented in Statistical Parametric Mapping (SPM) version 12 (Welcome Department of Cognitive Neurology, London, UK) running under MATLAB R2018b (MathWorks Inc., Sherborn, MA) and transformed into a standard stereotaxic anatomical space with the brain template from the Montreal Neurological Institute (voxel size: 1.5 mm × 1.5 mm × 1.5 mm) using diffeomorphic anatomic registration through exponentiated lie algebra (DARTEL) flow created in the DARTEL process.

N-isopropyl-p-[¹²³I] iodoamphetamine (¹²³I-IMP) SPECT was conducted using an integrated SPECT/CT system (Symbia T-6; Siemens Healthcare, Erlangen, Germany) to evaluate cerebral blood flow. Studies were conducted in a resting state with eyes closed and ears unplugged. A 167 MBq dose of ¹²³I-IMP was injected into the antecubital vein. The SPECT scan was initiated 15 min after the ¹²³I-IMP injection, and data acquisition was performed for 30 min in dynamic mode with gamma cameras rotating over a 360° range in 4° angular steps (90 views). For image reconstruction, the scatter component in the projection data was estimated using the triple-energy window method. Scatter-subtracted tomographic data were applied for the filtered back-projection with Chang’s attenuation correction method with a 0.15 cm⁻¹ attenuation coefficient.

To evaluate what kind of patients with VLOSLP were examined for indicative biomarkers of DLB, the characteristics were compared between patients who had been tested and those who had not. The characteristics of VLOSLP+LB and VLOSLP-LB were also compared. For the aforementioned comparisons, we used the Mann–Whitney U test for continuous and ordinal variables and the chi-squared test for nominal variables. These analyses were performed using SPSS for Mac version 25.0 (IBM Corp., Armonk, NY, USA). The level of statistical significance was set at a two-tailed p < 0.05.

Differences in regional gray matter volume were examined using a two-sample t-test model in SPM between VLOSLP+LB and VLOSLP–LB. The specific effect of each parameter was tested using [1] or [− 1] t-contrast with additional zeros for the remaining nuisance covariates. Age, gender, and intracranial volume were used as nuisance covariates. The statistical threshold was set to an uncorrected p < 0.01, and the extent threshold was set at more than 300 voxels.

To compare cerebral blood flow over the entire brain between VLOSLP+LB and VLOSLP–LB, three-dimensional stereotactic surface projection (3D-SSP) images were firstly created using the Neurological Statistical Image Analysis Software (NEUROSTAT) [24]. Image analysis was performed using iSSP version 3.5 on FALCON version 6.1.0.0 (Nihon Medi-Physics Co., Ltd., Japan). In this process, the spatial distribution of abnormal cerebral blood flow was calculated for each SPECT data using the age-gender-matched normal control database of Osaka University Hospital. The spatial distribution of abnormal cerebral blood flow was compared between VLOSLP+LB and VLOSLP–LB by calculating two-sample t-statistic values (converted to Z) at each voxel using iSSP3.5_2tZ (Nihon Medi-Physics Co., Ltd., Japan). The statistical Z-threshold was set at 1.

A total of 16 men and 48 women with VLOSLP were included (Fig. 1). The mean (SD) age and MMSE scores were 78.5 (6.4) and 26.4 (1.8), respectively (Table 1). Two patients had no reliable family caregiver; therefore, their psychotic symptoms were assessed using the BPRS (scores were 45 and 30). Forty-nine patients (76.6%) had delusions, 35 (54.7%) had hallucinations, and 20 (31.3%) had both. The most frequently observed delusions were persecution delusions (40.6%), followed by theft delusions (39.1%), and phantom-border delusions (31.3%). The most frequently observed hallucinations were auditory hallucinations (35.9%), followed by visual hallucinations (28.1%).

Thirty-four patients underwent testing for at least one of the indicative biomarkers of DLB, of which 17 underwent DAT SPECT and 22 underwent MIBG myocardial scintigraphy. Table 1 shows the differences between patients who underwent DLB biomarkers and those who did not. The scores of delusions (p = 0.002), hallucinations (p = 0.003), and cognitive fluctuation (p = 0.003) in the NPI-plus were significantly higher, and the scores for agitation/aggression (p = 0.027) were significantly lower in those who underwent indicative biomarkers than in those who did not. Delusions of persecution (53.9% vs. 26.7%, p = 0.033), phantom-border delusions (44.1% vs. 16.7%, p = 0.018), and visual hallucinations (44.1% vs. 10.0%, p = 0.002) were more prevalent in patients who underwent testing for indicative biomarkers than in those who did not. There were no differences in age, gender, or cognitive battery between the groups.

Among 34 patients who underwent testing for indicative biomarkers of DLB, 17 patients underwent DAT SPECT and eight showed positive results, while 22 underwent MIBG myocardial scintigraphy and six showed positive results. Five patients underwent testing for both indicative biomarkers; three were positive in both, one was negative in both, and one was positive in MIBG myocardial scintigraphy but negative in DAT SPECT. Therefore, 11 patients (32.4%) showed positive results for the indicative biomarkers (VLOSLP+LB).

The standardized score of DSST (mean [SD] score = 6.9 [3.1] vs. 10.0 [2.7], p = 0.005) and the score of delusions in the NPI-plus (mean [SD] score = 2.4 [2.3] vs. 6.9 [4.3], p = 0.005) were significantly lower in the VLOSLP+LB group than in the VLOSLP–LB group (Table 2). There were no differences in gender, age, and other cognitive battery between the two groups. Visual hallucinations were significantly more prevalent in the VLOSLP+LB group than in the VLOSLP–LB group (81.8% vs. 26.1%, p = 0.002), although the prevalence of delusions and other hallucinations between the groups was not different. Auditory hallucinations were prevalent in both groups (43.5% in VLOSLP–LB, and 45.5% in VLOSLP+LB). Among patients with auditory hallucinations, AH–VH tended to be more prevalent in VLOSLP–LB (7 out of 10) than in VLOSLP+LB patients (1 out of 5) (Fig. 2).

Ten patients in the VLOSLP+LB group and 14 in VLOSLP–LB group underwent 3D T1 MRI. Gray matter volume was not significantly different in any region between the VLOSLP+LB and VLOSLP–LB groups.

Eleven patients in the VLOSLP+LB group and 16 in the VLOSLP–LB group underwent ¹²³I IMP-SPECT. Cerebral blood flow was lower in VLOSLP+LB patients as compared to VLOSLP–LB patients in the posterior regions, including the occipital lobe (Fig. 3). Cerebral blood flow was higher in the bilateral precentral gyrus, postcentral gyrus, cingulate, thalamus, brainstem, and right temporal lobe in the VLOSLP+LB group than in the VLOSLP–LB group.

This study has some limitations. It should be noted that no biomarker directly assessed the presence of Lewy pathology. Although the utility of both imaging in distinguishing DLB from AD is well-established with high sensitivity and specificity [34, 35], the utility for the prodromal stage of DLB is not very high [16, 17]. A previous neuropathological study reported that some patients with late-onset schizophrenia and delusional disorders, which overlap with VLOSLP, had corticobasal degeneration [6], which may have caused the reduced uptake on DAT SPECT [36, 37]. In addition, neurodegenerations other than LBD, such as argyrophilic grain disease and primary age-related tauopathy, are also reportedly associated with VLOSLP [5, 6]. Some case reports have shown AD pathology in patients with VLOSLP [38, 39]. We should consider the rate of false positives and false negatives of indicative biomarkers and the influence of neurodegeneration other than LBD in interpreting the results.

There are other limitations in addition to the selection bias in conducting neuroimaging tests, the characteristics of biomarkers, and the influence of other neurodegenerations as described above. First, the sample size was too small to perform multivariate analysis. Second, this was a retrospective, cross-sectional, single-center study. Third, we did not control confounders of psychosis such as medications and sensory impairment. Finally, we did not conduct statistical correction for multiple comparisons. We considered this study to be at the stage of exploring the characteristics of patients suggestive of prodromal DLB in VLOSLP. Therefore, we focused on avoiding the risk of β-error rather than α-error in this small-sample-size study. For this reason, we decided that it would be preferable to present the results without correction for multiple comparisons. In fact, many of the items that were judged to be significantly different and some of the items that were judged to be not significantly different showed medium-to-large effect sizes. Further longitudinal prospective studies with larger sample sizes are required.