Genetic contribution to disease-course severity and progression in the SUPER-Finland study, a cohort of 10,403 individuals with psychotic disorders

Abstract

Genetic factors contribute to the susceptibility of psychotic disorders, but less is known how they affect psychotic disease-course development. Utilizing polygenic scores (PGSs) in combination with longitudinal healthcare data with decades of follow-up we investigated the contributing genetics to psychotic disease-course severity and diagnostic shifts in the SUPER-Finland study, encompassing 10 403 genotyped individuals with a psychotic disorder. To longitudinally track the study participants' past disease-course severity, we created a psychiatric hospitalization burden metric using the full-coverage and nation-wide Finnish in-hospital registry (data from 1969 and onwards). Using a hierarchical model, ranking the psychotic diagnoses according to clinical severity, we show that high schizophrenia PGS (SZ-PGS) was associated with progression from lower ranked psychotic disorders to schizophrenia (OR = 1.32 [1.23-1.43], p = 1.26e-12). This development manifested already at psychotic illness onset as a higher psychiatric hospitalization burden, the proxy for disease-course severity. In schizophrenia (n = 5 479), both a high SZ-PGS and a low educational attainment PGS (EA-PGS) were associated with increased psychiatric hospitalization burden (p = 1.00e-04 and p = 4.53e-10). The SZ-PGS and the EA-PGS associated with distinct patterns of hospital usage. In individuals with high SZ-PGS, the increased hospitalization burden was composed of longer individual hospital stays, while low EA-PGS associated with shorter but more frequent hospital visits. The negative effect of a low EA-PGS was found to be partly mediated via substance use disorder, a major risk factor for hospitalizations. In conclusion, we show that high SZ-PGS and low EA-PGS both impacted psychotic disease-course development negatively but resulted in different disease-course trajectories.

Fig. 1. Progression to schizophrenia.

a Progression from a lower ranked psychotic diagnosis to schizophrenia. The three panels show individuals who at some point in time had psychotic MDD, BD or SAD as their most severely ranked psychotic diagnosis but later progressed to schizophrenia compared to the individuals who remained at the corresponding lower ranked diagnoses. The results show that a larger proportion of individuals with a high SZ-PGS progressed to schizophrenia (combined HR = 1.23 [1.15-1.31], p = 6.42e-10). [SZ-PGS levels: Low = bottom <20%; Middle=20-60%; High = >80%; ⁽¹⁾ Number of individuals who progressed to schizophrenia; ⁽²⁾ Number of individuals who remained at the specific lower ranked diagnosis.]. b Genetic map of SUPER participants. Polygenic composition of individuals with the four major psychotic diagnoses according to their hierarchical ranking, including the group that progressed to schizophrenia from an initial lower ranked diagnosis (orange). The psychotic and affective dimensions are proxied by the SZ-PGS and MDD-PGS respectively (mean PGS with error bars showing 95% Cl). The progression group displayed a higher SZ-PGS than any of the lower ranked diagnostic groups (p = 0.0178 (SAD), p = 3.06e-16 (BD) and p = 1.95e−07 (psychotic MDD)). [y-axis: SZ-PGS; x-axis: MDD-PGS; *=only individuals who had schizophrenia as their first major psychotic disorder].

Fig. 2. Psychiatric hospitalization burden.

a The psychiatric hospitalization burden for the four major psychotic disorders according to the diagnostic hierarchical model, where the highest ranked psychotic disorder was considered the current lifetime diagnosis. The y-axis shows the proportion hospitalized due to a psychiatric diagnosis in each group and the x-axis shows the age at event. Individuals with schizophrenia had the highest overall burden compared to the lower ranking diagnoses (β = 0.65, p = 3.26e-186). b Associations between endpoints often thought to reflect general functioning and their association to the average yearly psychiatric hospitalization burden. The results show that an increased hospitalization burden was associated with a poor outcome for all endpoints, suggesting that the measurement can serve as a proxy to track disease-severity. Note, the analysis was constrained to individuals with schizophrenia (n = 5479) to not be influence by the cohort’s diagnostic composition. (Parameters have been named to make the direction of effect intuitive; n denotes the number of individuals with non-missing data for the specific endpoint; *Adjusted Paired Associates Learning (PAL) errors, a part of the CANTAB suit to assess cognitive function.).

Fig. 3. Hospital burden for individuals with schizophrenia.

a Heatmap displaying the proportion hospitalized due to a psychiatric diagnosis each year (colored) in relationship to SZ-PGS deciles (y-axis) and age (x-axis). In individuals with schizophrenia a high SZ-PGS was associated with an increased need of psychiatric hospital care (β = 0.067, p = 1.58e-06). b The proportion hospitalized for a primary psychiatric diagnosis in relation to SZ-PGS. The hospitalization data was aligned for the time-point of first recorded psychotic illness. A high SZ-PGS was associated with an increased need of hospital care post psychotic illness onset (β = 0.054, p = 1.00e-04). [SZ-PGS strata: Low = <20% (n = 1 096); Middle = 20–60% (n = 3 287); High = >80% (n = 1 096); total n = 5 479].

Fig. 4. Hospital usage profiles associated with SZ-PGS and EA-PGS.

The upper panels (a, b) display the relationship between the EA-PGS and the SZ-PGS and the total length of stay for the first 15 years post psychotic illness onset. The two lower panels (c + d) show the relationship between the EA-PGS and the SZ-PGS and the median length of each separate hospital visit for the same time-period. The EA-PGS and the SZ-PGS associated with two distinctly different hospitalization profiles. A low EA-PGS associates with a longer total time spent at hospital, but the total time was composed of shorter, more frequent visits, while a high SZ-PRS associated with longer total time spent at hospital and longer individual hospital stays. [PGS strata: Low = <20% (n = 1096); Middle = 20–60% (n = 3287); High = >80% (n = 1096); total n = 5479].