- Email: [email protected]
Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based study
SCHRES-07077; No of Pages 6 Schizophrenia Research xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based study Thomas M. Laursen a,b,c,⁎, Betina B. Trabjerg a,b, Ole Mors a,d, Anders D. Børglum a,d,e, David M. Hougaard f, Manuel Mattheisen a, Sandra M. Meier a,b, Enda M. Byrne h, Preben B. Mortensen a,b, Trine Munk-Olsen a,b, Esben Agerbo a,b,g a
Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark National Centre for Register-Based Research, Aarhus University, Denmark c Mental Health in Primary Care (MEPRICA), Research Unit for General Practice, Department of Public Health, Aarhus University, Aarhus, Denmark d Psychosis Research Unit, Aarhus University Hospital, Risskov, Denmark e Department of Biomedicine and Centre for Integrative Sequencing, iSEQ, Aarhus University, Aarhus, Denmark f Danish Center for Neonatal Screening, Statens Serum Institut, Copenhagen, Denmark g CIRRAU - Centre for Integrated Register-based Research at Aarhus University, Aarhus, Denmark h The University of Queensland, Queensland Brain Institute, Brisbane, Australia b
a r t i c l e
i n f o
Article history: Received 29 August 2016 Received in revised form 30 November 2016 Accepted 2 December 2016 Available online xxxx Keywords: Epidemiology Schizophrenia Mortailty Polygenic risk score
a b s t r a c t Background: It is unknown whether an increased genetic liability to schizophrenia influences the risk of dying early. The aim of the study was to determine whether the genetic predisposition to schizophrenia is associated with the risk of dying early and experience a suicide attempt. Method: Case control study, Denmark. The main measure was the mortality rate ratios (MRR) for deaths and odds ratios (OR) for multiple suicide attempts, associated with one standard deviations increase of the polygenic riskscore for schizophrenia (PRS). Results: We replicated the high mortality MRR = 9.01 (95% CI: 3.56–22.80), and high risk of multiple suicide attempts OR = 33.16 (95% CI: 20.97–52.43) associated with schizophrenia compared to the general population. However, there was no effect of the PRS on mortality MRR = 1.00 (95% CI 0.71–1.40) in the case-control setup or in cases only, MRR = 1.05 (95% CI 0.73–1.51). Similar, no association between the PRS and multiple suicide attempts was found in the adjusted models, but in contrast, family history of mental disorders was associated with both outcomes. Conclusions: A genetic predisposition for schizophrenia, measured by PRS, has little influence on the excess mortality or the risk of suicide attempts. In contrast there is a strong significant effect of family history of mental disorders. Our findings could reflect that the common variants detected by recent PRS only explain a small proportion of risk of schizophrenia, and that future, more powerful PRS instruments may be able to predict excess mortality within this disorder. © 2016 Published by Elsevier B.V.
1. Background Studies have shown that persons with schizophrenia have 2 to 3 fold higher mortality rates compared to the general population (Harris and Barraclough, 1998; Osby et al., 2000; Laursen et al., 2007; Saha et al., 2007), resulting in 15–20 years shorter life expectancy (Laursen, 2011). The excess mortality stem from higher mortality rates from both natural and unnatural causes of death (Laursen et al., 2007). As ⁎ Corresponding author at: National Centre for Register-based Research, Aarhus University, School of Business and Social Sciences, Institute of Economics and Business, Fuglesangs Allé 4, Building K (2631), DK-8210 Aarhus V, Denmark. E-mail address: [email protected] (T.M. Laursen).
early as the start of the 20th century, Odegard (1951) showed an excess mortality in patients with schizophrenia admitted to a Norwegian mental hospital in the period 1916–1941. Since then, different environmentally related explanations of this unacceptable high mortality have been suggested (Thornicroft, 2011; Laursen et al., 2012, 2014b). Firstly, persons with schizophrenia tend to have suboptimal health-related behavior including unhealthy diets, excessive smoking and alcohol use, as well as a sedentary lifestyle, which all are well-known risk factors for early mortality (Juel and Sørensen, 2006; Scheewe et al., 2012). Secondly, antipsychotic drugs may have adverse side effects, e.g. weight gain and serious medical illnesses such as diabetes and sudden cardiovascular death, which have negative effects on the life expectancy in this group of patients (Jerrell et al., 2010; Ribe et al., 2014). Thirdly, the
http://dx.doi.org/10.1016/j.schres.2016.12.001 0920-9964/© 2016 Published by Elsevier B.V.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
2
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
risks of suicide and accidents among patients with schizophrenia are high (Nordentoft et al., 2011). Fourth and lastly, somatic and mental comorbidities, including psychoactive substance use disorders and alcohol use disorders, are common among schizophrenia patients, but diagnosed later and treated insufficiently compared to mentally healthy persons (Laursen and Nordentoft, 2011; Laursen et al., 2011, 2014a). Genetic epidemiological studies of schizophrenia have for decades shown high heritability and family aggregation (McGue et al., 1983; Mortensen et al., 1999). Since 2009 evidence for specific genetic variants has been accumulating (Purcell et al., 2009), and a total of 108 schizophrenia-associated loci have very recently been identified by the Psychiatric Genomics Consortium, confirming that schizophrenia is a highly polygenic disorder. Despite the small effect size of each individual variant, together, the genome-wide-significant loci were estimated to explain 3.4% of the variance in liability, and the cumulative effect of common loci expressed as a polygenic risk score (PRS) was estimated to explain7% of the variance in liability (2014). Furthermore, it has previously been demonstrated that there is a dose-response relationship between the PRS and the risk of schizophrenia, showing that the higher PRS score the higher risk of schizophrenia (Agerbo et al., 2015). Given the strong genetic influence on schizophrenia described above, it seems plausible that the well-described excess mortality in this patient group can also be in part attributed to a genetic vulnerability. We hypothesized that persons with schizophrenia not only have a genetic predisposition towards schizophrenia, but furthermore, that this genetic predisposition may also predict premature death, and thus explain a part of the excess mortality. Ideally, a test of single genes known to cause schizophrenia should be tested against early mortality or suicide attempts. However, as no single genes can accurately be used as biomarkers to diagnose or predict schizophrenia a more broad approach are necessary. We therefore aimed to study the impact of the genetic predisposition, measured by the PRS for schizophrenia for two different outcomes: mortality and suicide attempts. Firstly, our goal was to test to which extent we in our sample could replicate associations of schizophrenia on mortality rates and suicidal behavior. Secondly, we aimed to specifically measure the impact of PRS on the selected outcomes jointly in our cases and controls and separately. Finally, the results of the PRS analysis were compared to the effect of family history of severe mental disorders, which is an alternative and commonly used measure of genetic loading. 2. Method 2.1. Study population, cases and controls We conducted a case control study. People with schizophrenia were defined among all singleton births in Denmark since 1981 with a DNA sample available from the Danish Neonatal Screening Bio-bank (Norgaard-Pedersen and Hougaard, 2007) and an ICD-10 F20 code for schizophrenia between January 1, 1994, and December 31, 2008, N = 1780. We selected a control as a randomly selected person from the CPR register (Pedersen et al., 2006), born in Denmark, with the same gender and the same birthday, not previously having a contact to a psychiatric hospital with a F20 schizophrenia diagnosis (Mors et al., 2011), N = 1768. 2.2. Approval The study did not require informed consent from participants according to Danish legislation (Act of Processing Personal Data), and the study did not involve any contact with study participants. The study was approved by the Danish Data Protection Agency, the Danish Research Ethics Committee, and the Steering Committee for the Danish Neonatal Screening Biobank.
2.3. Calculation of the PRS for schizophrenia DNA was extracted from the dried blood samples stored in the Danish neonatal bio-bank (Norgaard-Pedersen and Hougaard, 2007), whole-genome amplified (in triplicate using the Qiagen REPLI-g mini kit and the 3 separate reactions were pooled), and genotyped with Illumina Human 610-Quad BeadChip array (Borglum et al., 2014) and Illumina's HumanCoreExome beadchip (Illumina, San Diego, CA, USA) (Meier et al., 2015). The PRS for schizophrenia with p-value threshold 0.05 and normalized to the sample was calculated using the SNP information from the Psychiatric Genomics Consortium, (discovery sample of 34,600 cases and 45,968 control individuals, excluding the Danish data). We also calculated the PRS using threshold 1.00. More detailed description of the cases has been published previously (Nature, 2014; Agerbo et al., 2015; Borglum et al., 2014). We used the score as a continuous variable in the analysis, but we also tabulated the 10% percentiles in the first descriptive table. 2.4. Definition of outcome variables: mortality and suicide attempts The outcome measures, suicide attempts and mortality, were defined from the Danish National Registers. The suicide attempts measure was constructed from the National patients register (Lynge et al., 2011) and the Psychiatric Central Register (Mors et al., 2011): We chose all contacts, where the reason for the contact was suicide attempts. Furthermore, all ICD10 F diagnosis with comorbid diagnosis code T36T50, T52-T60, S51, S55, S59, S61, S65, S69 or any hospital contact with diagnosis code T39, T42, T43, T58, X60-X84. For further description of the definition of suicide attempts, see Christiansen (Christiansen et al., 2015). We have truncated the number of suicide attempts to 0, 1 or 2+ attempts. Mortality was defined as the date of death stated in the death certificates in the Cause of Death Register (Helweg-Larsen, 2011). 2.5. Definition of covariates All analysis had a basic adjustment for age, gender and calendar year. Analyses including the PRS were also adjusted for ancestry using the first 10 principal components estimated from genome-wide SNP genotypes (Price et al., 2006). Furthermore, we examined the impact of any mental disorders in the family. We used an overall yes/no contact (in or out patient) to a psychiatric hospital for any mental disorder in the mother or the father of the proband. Somatic comorbidity was defined using the Charlson Comorbidity Index (Charlson et al., 1987). The Index includes 19 chronic diseases to which a weight from one to six to each disorder is assigned according to the severity of the disease. The score of the Index is the sum of all weights. In the present study we have truncated the index to include the scores 0, 1, 2+. 2.6. Statistical analysis Mortality rates were analyzed in a Cox regression model with time since first schizophrenia diagnosis in cases and time since matching in controls as the underlying time scale. Number of suicide attempts was analyzed in a multinomial logistic regression model with outcome consisting of 0, 1, or 2+ suicide attempts. The polygenic score was analyzed as a continuous variable with one standard deviation increase as the measurement unit. In Table 1 the score is also shown in deciles with the lowest decile as the reference. For each outcome the effect of polygenic score risk was evaluated in the case-control setting and in cases and controls only. Because of limited power due to the fact that only 5 controls died during follow-up, the association between the polygenic score and mortality was not evaluated in controls. The two outcomes, mortality and suicide attempts, were analyzed in two models:
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx Table 1 Sample characteristics: Number of cases and controls and crude Odds Ratios for schizophrenia. Schizophrenia (n = 1780)
Non-schizophrenia (n = 1768)
Crude ORa (95% CI)
Death Yes No
44 (2.5) 1736 (97.5)
5 (0.3) 1763 (99.7)
9.01 (3.56; 22.80)b 1
Suicide attempts 2+ 1 0
399 (22.4) 257 (14.4) 1124 (63.2)
20 (1.1) 42 (2.4) 1706 (96.5)
33.16 (20.97; 52.43) 9.72 (6.95; 13.61) 1
Polygenic score Continuous measured
–
–
1.41 (1.31; 1.51)
Polygenic score 9 8 7 6 5 4 3 2 1 0
238 (13.4) 225 (12.6) 189 (10.6) 173 (9.7) 179 (10.1) 164 (9.2) 158 (8.9) 178 (10.0) 150 (8.4) 126 (7.1)
116 (6.6) 130 (7.4) 166 (9.4) 182 (10.3) 176 (10.0) 191 (10.8) 197 (11.1) 177 (10.0) 205 (11.6) 228 (12.9)
3.73 (2.73; 5.09) 3.13 (2.31; 4.26) 2.06 (1.52; 2.78) 1.72 (1.27; 2.33) 1.84 (1.36; 2.49) 1.55 (1.15; 2.10) 1.45 (1.07; 1.96) 1.82 (1.34; 2.46) 1.32 (0.98; 1.79) 1
Family history of mental disorder Yes 536 (30.1) No 1244 (69.9)
232 (13.1) 1536 (86.9)
2.86 (2.41; 3.39) 1
Year of birth 1996–2000 1991–1995 1986–1990 1981–1985
4 (0.2) 141 (7.9) 827 (46.5) 808 (45.4)
4 (0.2) 141 (8.0) 831 (47.0) 792 (44.8)
0.99(0.21; 4.68)c 0.98 (0.60; 1.61)c 0.98 (0.75; 1.27)c 1
Gender M F
994 (55.8) 786 (44.2)
990 (56.0) 778 (44.0)
0.99 (0.87; 1.13)c 1
OR = odds ratio, CI = confidence interval, Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a The crude OR are only adjusted for age and time at matching and sex (not C1–C10). b Death will obviously not occur before schizophrenia but the OR can be interpreted as the OR of death by schizophrenia status. c The dataset we uses is from a matched case-control study and the matching on date of birth and gender makes these estimates close to 1. d Measurement unit = 1 standard deviation.
1. Basic adjustment where the covariates PRS, schizophrenia, and family history were adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. 2. Mutually adjusted, where the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
3
Table 2 Mortality rate ratios: polygenic score and family history.
Polygenic score Schizophrenia Family history
Yes vs. no Yes vs. no
Basic adjustmenta MRR (95% CI)
Mutually adjustedb MRR (95% CI)
1.21 (0.86; 1.69) 8.76 (3.46; 22.18) 2.61 (1.47; 4.63)
1.00 (0.71; 1.40) 7.76 (3.02; 19.91) 1.83 (1.02; 3.27)
Time from the date when schizophrenia cases were matched to controls was used as the underlying time axis in the Cox Proportional Hazards Model MMR = mortality rate ratios, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a Basic adjustment is adjusting for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. b Mutually adjusted: the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
in a crude OR of 3.73 (95% CI: 2.73–5.09) in the highest decile, compared to the lowest decile. The crude OR for family history of mental disorders was 2.86 (95% CI: 2.41–3.39), Table 1.
3.2. Mortality The polygenic score was not associated with mortality in the basic adjusted model; MRR = 1.21 (95% CI 0.86–1.69), when measured as a continuous variable. In the mutually adjusted model the MRR was 1.00 (95% CI 0.71–1.40), Table 2. In contrast, a family history of any mental disorder was significantly associated in both the basic adjusted and mutually adjusted model. In the case-only setting, the polygenic score had no effect on mortality, while the effect of family history was 2.03 (95% CI 1.12–3.71), Table 3. In the control only setup there were only 5 people who had died and thus not enough cases to analyze.
3.3. Suicide attempts The PRS was significantly associated with two or more suicide attempts 1.18 (95% CI 1.04–1.34), but the corresponding mutually adjusted effect was not significant, OR 0.97 (95% CI 0.85–1.12). An analogous pattern was seen for 1 suicide attempt, Table 4. Among persons with schizophrenia there was no effect of the PRS, while a small effect of family history was found (although not significant for 1 suicide attempt), Table 4. In controls, family history had a highly significant effect with an OR of almost three associated with one suicide attempt and an OR of more than three for 2 or more suicide attempts, Table 5. Little effect of the PRS was found in this outcome.
3. Results 3.1. Confirmation of previous results The sample comprised 1780 cases with schizophrenia and 1768 age and gender matched controls. We found a crude Odds Ratio (OR), i.e. adjusted only for age, gender, and calendar time at matching time, of 9.01 (95% CI: 3.56–22.80) for death in the cases with schizophrenia compared to the control group. Two or more suicide attempts was associated with a crude OR of 33.16 (95% CI: 20.97–52.43) while one suicide attempt was associated with a crude OR of 9.72 (95% CI: 6.95–13.61), Table 1. The crude OR for the polygenic score was 1.41 (95% CI: 1.31–1.51) measured as a continuous variable with one standard deviation as the measure unit. When the PRS was divided into deciles an increasing OR from the lower deciles towards the higher deciles was found, ending
Table 3 Mortality rate ratios, cases: polygenic score and familyhistory.
Polygenic score Family history
Yes vs. no
Schizophrenia group MRR (95% CI)
Non schizophrenia group MRR (95% CI)
1.05 (0.73;1.51) 2.03 (1.12;3.71)
– –
Note. Time from the date when schizophrenia cases were matched to controls was used as the underlying time axis in the Cox Proportional Hazards Model MMR = mortality rate ratios, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. The estimates are adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components and the model includes both the polygenic score and the family history.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
4
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
Table 4 Polygenic score and odds ratios of suicide attempts. Basic adjustmenta
Polygenic score Schizophrenia Family history
Yes vs. no Yes vs. no
Mutually adjustedb
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
1.17 (1.01; 1.35) 9.86(7.03; 13.82) 2.03 (1.56; 2.64)
1.18 (1.04; 1.34) 33.49 (21.16; 53.00) 2.13 (1.69; 2.68)
0.98 (0.84; 1.14) 9.43 (6.70; 13.27) 1.39 (1.06; 1.83)
0.97 (0.85; 1.12) 32.17 (20.28; 51.03) 1.37 (1.07; 1.75)
Note. Multinomial logistic model using Suicide attempts = 0 as reference category. OR = odds ratio, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a Basic adjustment is adjusting for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. b Mutually adjusted: the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
3.4. Sensitivity analysis on all SNPs We also performed a sensitivity analysis where we calculated the PRS using threshold 1.00. The results were almost identical to the main results; see Appendix Tables A1 and A2. 4. Discussion We replicated the very high mortality rates found in persons with schizophrenia, with an almost 10 times higher mortality rate compared to the controls. Furthermore, we replicated the very high ORs of N30 associated with two or more suicide attempts in persons with schizophrenia. The genetic component in people with schizophrenia or in controls, measured by the PRS, was associated with suicide attempts in the basic adjusted model, but not in the mutually adjusted model. There was no significant association with mortality in either model. In contrast, we found that family history of mental disorders was associated with both higher mortality and higher risk of suicide attempts. 4.1. Strengths and limitations Our sample comprised 1780 persons with schizophrenia and 1768 matched controls. This limited sample size could affect our ability to detect the influence of the PRS. However, we could easily reproduce the known results of excess mortality and larger risks of suicide in persons with schizophrenia. In the same way, we could show that family history of mental disorders affected the risk of the two outcomes, suggesting that it was not a lack of power that prompted our results. The PRS is a relatively new approach to measure the genetic architecture of schizophrenia. The lack of a positive association between the PRS score and the excess mortality in this study could therefore be attributable to the fact that the PRS score is not yet a sufficient tool to measure genetic risk to schizophrenia. The measure may be improved in the future as the sample sizes used in GWAS to estimate the individual SNP effects become larger and thus the estimates become more accurate. However, at the moment the polygenic score is the best candidate for a single measure that simultaneously takes many variants into account (Kendler, 2016). There have been suggestions of a locus/loci in the XY homologous regions (Crow, 2013; De Lisi et al., 1994) and there is a major interaction with the epigenetic mechanism of X inactivation here (Crow, 2015). Thus our study cannot preclude that a solution lies in “hidden heritability” rather than hitherto unsuspected family interactions. Paternal age and maternal grand-father age at the birth of the proband (Crow, 2012) are consistent with an epigenetic explanation which may suggest that sex chromosomal influence could be a scope for further research. We have an unbiased and incident sample of all people diagnosed with schizophrenia in Denmark as well as population based controls drawn at random from the entire population. This ensured that not only very severe cases of schizophrenia were enrolled in the analysis, avoiding a bias towards larger effects of the covariates. Likewise, not
only very healthy people were drawn as controls ensuring true representation of an entire population (Meier et al., 2015). Because of limited power, we adjusted for parental history of any mental disorder, and we only adjust for the genetics of schizophrenia using the PRS score. Thus e.g. genetic variants associated with depression or anxiety could be passed down to the children, affecting mortality and then be captured by the general family history of mental disorders. 4.2. Interpretation The hypothesis that the genetic factors contributing to the etiology of schizophrenia may also be risk factors for reduced life expectancy (Laursen et al., 2014b) was not supported in the study. Our study suggests that there only is a weak influence, if any, of the PRS on excess mortality found among people with schizophrenia or controls. The elucidation of the genetic architecture of schizophrenia has been a key issue in schizophrenia research for many years. GWAS studies have had the problem that the lack of significant findings could be due to a lack of statistical power, owing to the large number of statistical tests combined with small effect sizes of individual genetic variants. In contrast, the PRS score does not have equivalent challenges with multiple testing. With a very large discovery sample the precision of the PRS improves, but, although it is a very promising measure of genetic liability to schizophrenia, it only captures a small fraction of the overall risk as well as the genetic risk of schizophrenia (Agerbo et al., 2012, 2015). It is important to note that the risk score is only generated using results from studies that focus on genetic variants that are common in the population (minor allele frequency N 1%). They do not fully capture rare genetic variants that potentially contribute substantially to schizophrenia risk. It has been reported that the PRS score was higher for persons prescribed Clozapine (taken as an indication of treatment resistance) in a smaller sample with 434 Clozapine users and 370 non-users (Frank et al., 2015) showing that it is possible to detect differences in smaller samples. We replicated that family history of mental disorders is associated with higher mortality in all models; a finding that often is ascribed to shared genetic factors and environmental factors. As we made an adjustment of the genetic risk factor of schizophrenia, our results could suggest that the environmental factors of the family history of mental disorders could be more influential than the genetic factors of schizophrenia. However, it should be noted that the PRS in its current form may not be an optimal indicator of the genetic risk, and thus the assessment of the balance between genes and environment should be made with caution. 5. Conclusion Genetic predisposition for schizophrenia, as measured by the most recent (and most informative) PRS, was not associated with excess mortality or risk of suicide attempts in those with schizophrenia. In contrast, these outcomes were significantly associated with family history of mental disorders.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
5
Table 5 Odds ratios of suicide attempts, cases and controls: polygenic score and family history. Schizophrenia group
Polygenic score Family history
Yes vs. no
Non schizophrenia group
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
0.94 (0.80; 1.11) 1.25 (0.93; 1.68)
0.95 (0.82; 1.10) 1.29 (1.00; 1.67)
1.22 (0.82; 1.83) 2.69 (1.34; 5.40)
1.29 (0.72; 2.31) 3.16 (1.18; 8.45)
Multinomial logistic model using suicide attempts = 0 as reference category. OR = odds ratio, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. The estimates are adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components and the model includes both the polygenic score and the family history.
Our findings could reflect that the common variants detected by recent PRS only explain a small proportion of risk of schizophrenia, and thus it is not surprising that nested outcomes related to mortality within schizophrenia are not associated with PRS. Thus, our findings should be considered as inconclusive, and do not exclude the possibility that future, more powerful PRS instruments may be able to predict mortality within this disorder. Regardless of these issues, it is important to remain mindful that patients with schizophrenia have an increased risk of premature death, and that health service providers should focus on interventions related to modifiable risk factors such as smoking, sedentary life lifestyle and comorbid physical disorders. Role of funding source The study was supported by an unrestricted grant from the Lundbeck Foundation (R155-2014-1724). The funding bodies played no role in the design of the study, preparation of the manuscript, or decision to publish.
Author contributions Thomas Munk Laursen, Esben Agerbo: Designed the study and wrote the first draft of the manuscript. Betina Trabjerg, Ole Mors, Anders D. Børglum, David M. Hougaard, Manuel Mattheisen, Sandra Meier, Enda Byrne, Preben B. Mortensen, and Trine Munk-Olsen: Assisted with editing and preparation of the final manuscript.
Conflicts of interest None.
Acknowledgement The manuscript has been read and approved by all named authors, and all authors have approved the order of authorship. There are no other individuals who satisfied the criteria for authorship but are not listed.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.schres.2016.12.001. References Agerbo, E., Mortensen, P.B., Wiuf, C., Pedersen, M.S., McGrath, J., Hollegaard, M.V., Nørgaard-Pedersen, B., Hougaard, D.M., Mors, O., Pedersen, C.B., 2012. Modelling the contribution of family history and variation in single nucleotide polymorphisms to risk of schizophrenia: a Danish national birth cohort-based study. Schizophr Res. 134 (2-3), 246–252. Agerbo, E., Sullivan, P.F., Vilhjalmsson, B.J., Pedersen, C.B., Mors, O., Borglum, A.D., Hougaard, D.M., Hollegaard, M.V., Meier, S., Mattheisen, M., Ripke, S., Wray, N.R., Mortensen, P.B., 2015. Polygenic risk score, parental socioeconomic status, family history of psychiatric disorders, and the risk for schizophrenia: a Danish populationbased study and meta-analysis. JAMA Psychiatry. Biological insights from 108 schizophrenia-associated genetic lociNature 511, 421–427. Borglum, A.D., Demontis, D., Grove, J., Pallesen, J., Hollegaard, M.V., Pedersen, C.B., Hedemand, A., Mattheisen, M., Uitterlinden, A., Nyegaard, M., Orntoft, T., Wiuf, C., Didriksen, M., Nordentoft, M., Nothen, M.M., Rietschel, M., Ophoff, R.A., Cichon, S., Yolken, R.H., Hougaard, D.M., Mortensen, P.B., Mors, O., 2014. Genome-wide study of association and interaction with maternal cytomegalovirus infection suggests new schizophrenia loci. Mol. Psychiatry 19, 325–333. Charlson, M.E., Pompei, P., Ales, K.L., MacKenzie, C.R., 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis. 40, 373–383.
Christiansen, E., Agerbo, E., Larsen, K.J., Bilenberg, N., Stenager, E., 2015. Youth, suicide attempts and low level of education: a Danish historical register-based cohort study of the outcome of suicide attempt. Int. J. Soc. Psychiatry 61, 802–810. Crow, T.J., 2012. Grandparental and parental age effects suggest epigenetic transmission on both X and Y chromosomes. Schizophr. Res. 139, 267–268. Crow, T.J., 2013. The XY gene hypothesis of psychosis: origins and current status. Am. J. Med. Genet. B Neuropsychiatr. Genet. 162B, 800–824. Crow, T.J., 2015. Is psychosis a disorder of XY epigenetics? EBioMedicine 2, 794–795. De Lisi, L.E., Friedrich, U., Wahlstrom, J., Boccio-Smith, A., Forsman, A., Eklund, K., Crow, T.J., 1994. Schizophrenia and sex chromosome anomalies. Schizophr. Bull. 20, 495–505. Frank, J., Lang, M., Witt, S.H., Strohmaier, J., Rujescu, D., Cichon, S., Degenhardt, F., Nothen, M.M., Collier, D.A., Ripke, S., Naber, D., Rietschel, M., 2015. Identification of increased genetic risk scores for schizophrenia in treatment-resistant patients. Mol. Psychiatry 20, 150–151. Harris, E.C., Barraclough, B., 1998. Excess mortality of mental disorder. Br. J. Psychiatry 173, 11–53. Helweg-Larsen, K., 2011. The Danish register of causes of death. Scand. J. Public Health 39, 26–29. Jerrell, J.M., McIntyre, R.S., Tripathi, A., 2010. Incidence and costs of cardiometabolic conditions in patients with schizophrenia treated with antipsychotic medications. Clin. Schizophr. Relat. Psychoses 4, 161–168. Juel, K., Sørensen, J., 2006. Risikofaktorer og folkesundhed i Danmark. Copenhagen. Ref Type: Conference Proceeding. Kendler, K.S., 2016. The schizophrenia polygenic risk score: to what does it predispose in adolescence? JAMA Psychiatry 73, 193–194. Laursen, T.M., 2011. Life expectancy among persons with schizophrenia or bipolar affective disorder. Schizophr. Res. 131, 101–104. Laursen, T.M., Nordentoft, M., 2011. Heart disease treatment and mortality in schizophrenia and bipolar disorder - changes in the Danish population between 1994 and 2006. J. Psychiatr. Res. 45, 29–35. Laursen, T.M., Munk-Olsen, T., Nordentoft, M., Mortensen, P.B., 2007. Increased mortality among patients admitted with major psychiatric disorders: a register-based study comparing mortality in unipolar depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia. J. Clin. Psychiatry 68, 899–907. Laursen, T.M., Munk-Olsen, T., Gasse, C., 2011. Chronic somatic comorbidity and excess mortality due to natural causes in persons with schizophrenia or bipolar affective disorder. PLoS One 6, e24597. Laursen, T.M., Munk-Olsen, T., Vestergaard, M., 2012. Life expectancy and cardiovascular mortality in persons with schizophrenia. Curr. Opin. Psychiatry 25, 83–88. Laursen, T.M., Mortensen, P.B., Mac Cabe, J.H., Cohen, D., Gasse, C., 2014a. Cardiovascular drug use and mortality in patients with schizophrenia or bipolar disorder: a Danish population-based study. Psychol. Med. 44, 1625–1637. Laursen, T.M., Nordentoft, M., Mortensen, P.B., 2014b. Excess early mortality in schizophrenia. Annu. Rev. Clin. Psychol. 10, 425–448. Lynge, E., Sandegaard, J.L., Rebolj, M., 2011. The Danish National Patient Register. Scand. J. Public Health 39, 30–33. McGue, M., Gottesman, I.I., Rao, D.C., 1983. The transmission of schizophrenia under a multifactorial threshold model. Am. J. Hum. Genet. 35, 1161–1178. Meier, S.M., Agerbo, E., Maier, R., Pedersen, C.B., Lang, M., Grove, J., Hollegaard, M.V., Demontis, D., Trabjerg, B.B., Hjorthoj, C., Ripke, S., Degenhardt, F., Nothen, M.M., Rujescu, D., Maier, W., Werge, T., Mors, O., Hougaard, D.M., Borglum, A.D., Wray, N.R., Rietschel, M., Nordentoft, M., Mortensen, P.B., Mattheisen, M., 2015. High loading of polygenic risk in cases with chronic schizophrenia. Mol. Psychiatry. Mors, O., Perto, G.P., Mortensen, P.B., 2011. The Danish psychiatric central research register. Scand. J. Public Health 39, 54–57. Mortensen, P.B., Pedersen, C.B., Westergaard, T., Wohlfahrt, J., Ewald, H., Mors, O., Andersen, P.K., Melbye, M., 1999. Effects of family history and place and season of birth on the risk of schizophrenia. N. Engl. J. Med. 340, 603–608. Nordentoft, M., Mortensen, P.B., Pedersen, C.B., 2011. Absolute risk of suicide after first hospital contact in mental disorder. Arch. Gen. Psychiatry 68, 1058–1064. Norgaard-Pedersen, B., Hougaard, D.M., 2007. Storage policies and use of the Danish newborn screening biobank. J. Inherit. Metab. Dis. 30, 530–536. Odegard, O., 1951. Mortality in Norwegian mental hospitals 1926–1941. Acta Genet. Stat. Med. 2, 141–173. Osby, U., Correia, N., Brandt, L., Ekbom, A., Sparen, P., 2000. Mortality and causes of death in schizophrenia in Stockholm county, Sweden. Schizophr. Res. 45, 21–28. Pedersen, C.B., Gotzsche, H., Moller, J.O., Mortensen, P.B., 2006. The Danish civil registration system. A cohort of eight million persons. Dan. Med. Bull. 53, 441–449. Price, A.L., Patterson, N.J., Plenge, R.M., Weinblatt, M.E., Shadick, N.A., Reich, D., 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38, 904–909.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
6
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
Purcell, S.M., Wray, N.R., Stone, J.L., Visscher, P.M., O'Donovan, M.C., Sullivan, P.F., Sklar, P., 2009. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460, 748–752. Ribe, A.R., Laursen, T.M., Sandbaek, A., Charles, M., Nordentoft, M., Vestergaard, M., 2014. Long-term mortality of persons with severe mental illness and diabetes: a population-based cohort study in Denmark. Psychol. Med. 44, 3097–3107. Saha, S., Chant, D., McGrath, J., 2007. A systematic review of mortality in schizophrenia: is the differential mortality gap worsening over time? Arch. Gen. Psychiatry 64, 1123–1131.
Scheewe, T.W., Takken, T., Kahn, R.S., Cahn, W., Backx, F.J., 2012. Effects of exercise therapy on cardiorespiratory fitness in patients with schizophrenia. Med. Sci. Sports Exerc. 44, 1834–1842. Thornicroft, G., 2011. Physical health disparities and mental illness: the scandal of premature mortality. Br. J. Psychiatry 199, 441–442.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
Contents lists available at ScienceDirect
Schizophrenia Research journal homepage: www.elsevier.com/locate/schres
Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based study Thomas M. Laursen a,b,c,⁎, Betina B. Trabjerg a,b, Ole Mors a,d, Anders D. Børglum a,d,e, David M. Hougaard f, Manuel Mattheisen a, Sandra M. Meier a,b, Enda M. Byrne h, Preben B. Mortensen a,b, Trine Munk-Olsen a,b, Esben Agerbo a,b,g a
Lundbeck Foundation Initiative for Integrative Psychiatric Research, iPSYCH, Denmark National Centre for Register-Based Research, Aarhus University, Denmark c Mental Health in Primary Care (MEPRICA), Research Unit for General Practice, Department of Public Health, Aarhus University, Aarhus, Denmark d Psychosis Research Unit, Aarhus University Hospital, Risskov, Denmark e Department of Biomedicine and Centre for Integrative Sequencing, iSEQ, Aarhus University, Aarhus, Denmark f Danish Center for Neonatal Screening, Statens Serum Institut, Copenhagen, Denmark g CIRRAU - Centre for Integrated Register-based Research at Aarhus University, Aarhus, Denmark h The University of Queensland, Queensland Brain Institute, Brisbane, Australia b
a r t i c l e
i n f o
Article history: Received 29 August 2016 Received in revised form 30 November 2016 Accepted 2 December 2016 Available online xxxx Keywords: Epidemiology Schizophrenia Mortailty Polygenic risk score
a b s t r a c t Background: It is unknown whether an increased genetic liability to schizophrenia influences the risk of dying early. The aim of the study was to determine whether the genetic predisposition to schizophrenia is associated with the risk of dying early and experience a suicide attempt. Method: Case control study, Denmark. The main measure was the mortality rate ratios (MRR) for deaths and odds ratios (OR) for multiple suicide attempts, associated with one standard deviations increase of the polygenic riskscore for schizophrenia (PRS). Results: We replicated the high mortality MRR = 9.01 (95% CI: 3.56–22.80), and high risk of multiple suicide attempts OR = 33.16 (95% CI: 20.97–52.43) associated with schizophrenia compared to the general population. However, there was no effect of the PRS on mortality MRR = 1.00 (95% CI 0.71–1.40) in the case-control setup or in cases only, MRR = 1.05 (95% CI 0.73–1.51). Similar, no association between the PRS and multiple suicide attempts was found in the adjusted models, but in contrast, family history of mental disorders was associated with both outcomes. Conclusions: A genetic predisposition for schizophrenia, measured by PRS, has little influence on the excess mortality or the risk of suicide attempts. In contrast there is a strong significant effect of family history of mental disorders. Our findings could reflect that the common variants detected by recent PRS only explain a small proportion of risk of schizophrenia, and that future, more powerful PRS instruments may be able to predict excess mortality within this disorder. © 2016 Published by Elsevier B.V.
1. Background Studies have shown that persons with schizophrenia have 2 to 3 fold higher mortality rates compared to the general population (Harris and Barraclough, 1998; Osby et al., 2000; Laursen et al., 2007; Saha et al., 2007), resulting in 15–20 years shorter life expectancy (Laursen, 2011). The excess mortality stem from higher mortality rates from both natural and unnatural causes of death (Laursen et al., 2007). As ⁎ Corresponding author at: National Centre for Register-based Research, Aarhus University, School of Business and Social Sciences, Institute of Economics and Business, Fuglesangs Allé 4, Building K (2631), DK-8210 Aarhus V, Denmark. E-mail address: [email protected] (T.M. Laursen).
early as the start of the 20th century, Odegard (1951) showed an excess mortality in patients with schizophrenia admitted to a Norwegian mental hospital in the period 1916–1941. Since then, different environmentally related explanations of this unacceptable high mortality have been suggested (Thornicroft, 2011; Laursen et al., 2012, 2014b). Firstly, persons with schizophrenia tend to have suboptimal health-related behavior including unhealthy diets, excessive smoking and alcohol use, as well as a sedentary lifestyle, which all are well-known risk factors for early mortality (Juel and Sørensen, 2006; Scheewe et al., 2012). Secondly, antipsychotic drugs may have adverse side effects, e.g. weight gain and serious medical illnesses such as diabetes and sudden cardiovascular death, which have negative effects on the life expectancy in this group of patients (Jerrell et al., 2010; Ribe et al., 2014). Thirdly, the
http://dx.doi.org/10.1016/j.schres.2016.12.001 0920-9964/© 2016 Published by Elsevier B.V.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
2
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
risks of suicide and accidents among patients with schizophrenia are high (Nordentoft et al., 2011). Fourth and lastly, somatic and mental comorbidities, including psychoactive substance use disorders and alcohol use disorders, are common among schizophrenia patients, but diagnosed later and treated insufficiently compared to mentally healthy persons (Laursen and Nordentoft, 2011; Laursen et al., 2011, 2014a). Genetic epidemiological studies of schizophrenia have for decades shown high heritability and family aggregation (McGue et al., 1983; Mortensen et al., 1999). Since 2009 evidence for specific genetic variants has been accumulating (Purcell et al., 2009), and a total of 108 schizophrenia-associated loci have very recently been identified by the Psychiatric Genomics Consortium, confirming that schizophrenia is a highly polygenic disorder. Despite the small effect size of each individual variant, together, the genome-wide-significant loci were estimated to explain 3.4% of the variance in liability, and the cumulative effect of common loci expressed as a polygenic risk score (PRS) was estimated to explain7% of the variance in liability (2014). Furthermore, it has previously been demonstrated that there is a dose-response relationship between the PRS and the risk of schizophrenia, showing that the higher PRS score the higher risk of schizophrenia (Agerbo et al., 2015). Given the strong genetic influence on schizophrenia described above, it seems plausible that the well-described excess mortality in this patient group can also be in part attributed to a genetic vulnerability. We hypothesized that persons with schizophrenia not only have a genetic predisposition towards schizophrenia, but furthermore, that this genetic predisposition may also predict premature death, and thus explain a part of the excess mortality. Ideally, a test of single genes known to cause schizophrenia should be tested against early mortality or suicide attempts. However, as no single genes can accurately be used as biomarkers to diagnose or predict schizophrenia a more broad approach are necessary. We therefore aimed to study the impact of the genetic predisposition, measured by the PRS for schizophrenia for two different outcomes: mortality and suicide attempts. Firstly, our goal was to test to which extent we in our sample could replicate associations of schizophrenia on mortality rates and suicidal behavior. Secondly, we aimed to specifically measure the impact of PRS on the selected outcomes jointly in our cases and controls and separately. Finally, the results of the PRS analysis were compared to the effect of family history of severe mental disorders, which is an alternative and commonly used measure of genetic loading. 2. Method 2.1. Study population, cases and controls We conducted a case control study. People with schizophrenia were defined among all singleton births in Denmark since 1981 with a DNA sample available from the Danish Neonatal Screening Bio-bank (Norgaard-Pedersen and Hougaard, 2007) and an ICD-10 F20 code for schizophrenia between January 1, 1994, and December 31, 2008, N = 1780. We selected a control as a randomly selected person from the CPR register (Pedersen et al., 2006), born in Denmark, with the same gender and the same birthday, not previously having a contact to a psychiatric hospital with a F20 schizophrenia diagnosis (Mors et al., 2011), N = 1768. 2.2. Approval The study did not require informed consent from participants according to Danish legislation (Act of Processing Personal Data), and the study did not involve any contact with study participants. The study was approved by the Danish Data Protection Agency, the Danish Research Ethics Committee, and the Steering Committee for the Danish Neonatal Screening Biobank.
2.3. Calculation of the PRS for schizophrenia DNA was extracted from the dried blood samples stored in the Danish neonatal bio-bank (Norgaard-Pedersen and Hougaard, 2007), whole-genome amplified (in triplicate using the Qiagen REPLI-g mini kit and the 3 separate reactions were pooled), and genotyped with Illumina Human 610-Quad BeadChip array (Borglum et al., 2014) and Illumina's HumanCoreExome beadchip (Illumina, San Diego, CA, USA) (Meier et al., 2015). The PRS for schizophrenia with p-value threshold 0.05 and normalized to the sample was calculated using the SNP information from the Psychiatric Genomics Consortium, (discovery sample of 34,600 cases and 45,968 control individuals, excluding the Danish data). We also calculated the PRS using threshold 1.00. More detailed description of the cases has been published previously (Nature, 2014; Agerbo et al., 2015; Borglum et al., 2014). We used the score as a continuous variable in the analysis, but we also tabulated the 10% percentiles in the first descriptive table. 2.4. Definition of outcome variables: mortality and suicide attempts The outcome measures, suicide attempts and mortality, were defined from the Danish National Registers. The suicide attempts measure was constructed from the National patients register (Lynge et al., 2011) and the Psychiatric Central Register (Mors et al., 2011): We chose all contacts, where the reason for the contact was suicide attempts. Furthermore, all ICD10 F diagnosis with comorbid diagnosis code T36T50, T52-T60, S51, S55, S59, S61, S65, S69 or any hospital contact with diagnosis code T39, T42, T43, T58, X60-X84. For further description of the definition of suicide attempts, see Christiansen (Christiansen et al., 2015). We have truncated the number of suicide attempts to 0, 1 or 2+ attempts. Mortality was defined as the date of death stated in the death certificates in the Cause of Death Register (Helweg-Larsen, 2011). 2.5. Definition of covariates All analysis had a basic adjustment for age, gender and calendar year. Analyses including the PRS were also adjusted for ancestry using the first 10 principal components estimated from genome-wide SNP genotypes (Price et al., 2006). Furthermore, we examined the impact of any mental disorders in the family. We used an overall yes/no contact (in or out patient) to a psychiatric hospital for any mental disorder in the mother or the father of the proband. Somatic comorbidity was defined using the Charlson Comorbidity Index (Charlson et al., 1987). The Index includes 19 chronic diseases to which a weight from one to six to each disorder is assigned according to the severity of the disease. The score of the Index is the sum of all weights. In the present study we have truncated the index to include the scores 0, 1, 2+. 2.6. Statistical analysis Mortality rates were analyzed in a Cox regression model with time since first schizophrenia diagnosis in cases and time since matching in controls as the underlying time scale. Number of suicide attempts was analyzed in a multinomial logistic regression model with outcome consisting of 0, 1, or 2+ suicide attempts. The polygenic score was analyzed as a continuous variable with one standard deviation increase as the measurement unit. In Table 1 the score is also shown in deciles with the lowest decile as the reference. For each outcome the effect of polygenic score risk was evaluated in the case-control setting and in cases and controls only. Because of limited power due to the fact that only 5 controls died during follow-up, the association between the polygenic score and mortality was not evaluated in controls. The two outcomes, mortality and suicide attempts, were analyzed in two models:
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx Table 1 Sample characteristics: Number of cases and controls and crude Odds Ratios for schizophrenia. Schizophrenia (n = 1780)
Non-schizophrenia (n = 1768)
Crude ORa (95% CI)
Death Yes No
44 (2.5) 1736 (97.5)
5 (0.3) 1763 (99.7)
9.01 (3.56; 22.80)b 1
Suicide attempts 2+ 1 0
399 (22.4) 257 (14.4) 1124 (63.2)
20 (1.1) 42 (2.4) 1706 (96.5)
33.16 (20.97; 52.43) 9.72 (6.95; 13.61) 1
Polygenic score Continuous measured
–
–
1.41 (1.31; 1.51)
Polygenic score 9 8 7 6 5 4 3 2 1 0
238 (13.4) 225 (12.6) 189 (10.6) 173 (9.7) 179 (10.1) 164 (9.2) 158 (8.9) 178 (10.0) 150 (8.4) 126 (7.1)
116 (6.6) 130 (7.4) 166 (9.4) 182 (10.3) 176 (10.0) 191 (10.8) 197 (11.1) 177 (10.0) 205 (11.6) 228 (12.9)
3.73 (2.73; 5.09) 3.13 (2.31; 4.26) 2.06 (1.52; 2.78) 1.72 (1.27; 2.33) 1.84 (1.36; 2.49) 1.55 (1.15; 2.10) 1.45 (1.07; 1.96) 1.82 (1.34; 2.46) 1.32 (0.98; 1.79) 1
Family history of mental disorder Yes 536 (30.1) No 1244 (69.9)
232 (13.1) 1536 (86.9)
2.86 (2.41; 3.39) 1
Year of birth 1996–2000 1991–1995 1986–1990 1981–1985
4 (0.2) 141 (7.9) 827 (46.5) 808 (45.4)
4 (0.2) 141 (8.0) 831 (47.0) 792 (44.8)
0.99(0.21; 4.68)c 0.98 (0.60; 1.61)c 0.98 (0.75; 1.27)c 1
Gender M F
994 (55.8) 786 (44.2)
990 (56.0) 778 (44.0)
0.99 (0.87; 1.13)c 1
OR = odds ratio, CI = confidence interval, Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a The crude OR are only adjusted for age and time at matching and sex (not C1–C10). b Death will obviously not occur before schizophrenia but the OR can be interpreted as the OR of death by schizophrenia status. c The dataset we uses is from a matched case-control study and the matching on date of birth and gender makes these estimates close to 1. d Measurement unit = 1 standard deviation.
1. Basic adjustment where the covariates PRS, schizophrenia, and family history were adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. 2. Mutually adjusted, where the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
3
Table 2 Mortality rate ratios: polygenic score and family history.
Polygenic score Schizophrenia Family history
Yes vs. no Yes vs. no
Basic adjustmenta MRR (95% CI)
Mutually adjustedb MRR (95% CI)
1.21 (0.86; 1.69) 8.76 (3.46; 22.18) 2.61 (1.47; 4.63)
1.00 (0.71; 1.40) 7.76 (3.02; 19.91) 1.83 (1.02; 3.27)
Time from the date when schizophrenia cases were matched to controls was used as the underlying time axis in the Cox Proportional Hazards Model MMR = mortality rate ratios, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a Basic adjustment is adjusting for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. b Mutually adjusted: the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
in a crude OR of 3.73 (95% CI: 2.73–5.09) in the highest decile, compared to the lowest decile. The crude OR for family history of mental disorders was 2.86 (95% CI: 2.41–3.39), Table 1.
3.2. Mortality The polygenic score was not associated with mortality in the basic adjusted model; MRR = 1.21 (95% CI 0.86–1.69), when measured as a continuous variable. In the mutually adjusted model the MRR was 1.00 (95% CI 0.71–1.40), Table 2. In contrast, a family history of any mental disorder was significantly associated in both the basic adjusted and mutually adjusted model. In the case-only setting, the polygenic score had no effect on mortality, while the effect of family history was 2.03 (95% CI 1.12–3.71), Table 3. In the control only setup there were only 5 people who had died and thus not enough cases to analyze.
3.3. Suicide attempts The PRS was significantly associated with two or more suicide attempts 1.18 (95% CI 1.04–1.34), but the corresponding mutually adjusted effect was not significant, OR 0.97 (95% CI 0.85–1.12). An analogous pattern was seen for 1 suicide attempt, Table 4. Among persons with schizophrenia there was no effect of the PRS, while a small effect of family history was found (although not significant for 1 suicide attempt), Table 4. In controls, family history had a highly significant effect with an OR of almost three associated with one suicide attempt and an OR of more than three for 2 or more suicide attempts, Table 5. Little effect of the PRS was found in this outcome.
3. Results 3.1. Confirmation of previous results The sample comprised 1780 cases with schizophrenia and 1768 age and gender matched controls. We found a crude Odds Ratio (OR), i.e. adjusted only for age, gender, and calendar time at matching time, of 9.01 (95% CI: 3.56–22.80) for death in the cases with schizophrenia compared to the control group. Two or more suicide attempts was associated with a crude OR of 33.16 (95% CI: 20.97–52.43) while one suicide attempt was associated with a crude OR of 9.72 (95% CI: 6.95–13.61), Table 1. The crude OR for the polygenic score was 1.41 (95% CI: 1.31–1.51) measured as a continuous variable with one standard deviation as the measure unit. When the PRS was divided into deciles an increasing OR from the lower deciles towards the higher deciles was found, ending
Table 3 Mortality rate ratios, cases: polygenic score and familyhistory.
Polygenic score Family history
Yes vs. no
Schizophrenia group MRR (95% CI)
Non schizophrenia group MRR (95% CI)
1.05 (0.73;1.51) 2.03 (1.12;3.71)
– –
Note. Time from the date when schizophrenia cases were matched to controls was used as the underlying time axis in the Cox Proportional Hazards Model MMR = mortality rate ratios, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. The estimates are adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components and the model includes both the polygenic score and the family history.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
4
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
Table 4 Polygenic score and odds ratios of suicide attempts. Basic adjustmenta
Polygenic score Schizophrenia Family history
Yes vs. no Yes vs. no
Mutually adjustedb
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
1.17 (1.01; 1.35) 9.86(7.03; 13.82) 2.03 (1.56; 2.64)
1.18 (1.04; 1.34) 33.49 (21.16; 53.00) 2.13 (1.69; 2.68)
0.98 (0.84; 1.14) 9.43 (6.70; 13.27) 1.39 (1.06; 1.83)
0.97 (0.85; 1.12) 32.17 (20.28; 51.03) 1.37 (1.07; 1.75)
Note. Multinomial logistic model using Suicide attempts = 0 as reference category. OR = odds ratio, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. a Basic adjustment is adjusting for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components. b Mutually adjusted: the covariates had the same basic adjustment, but, were mutually adjusted for the variables in the table, as well as adjusted for the Charlson Comorbidity score.
3.4. Sensitivity analysis on all SNPs We also performed a sensitivity analysis where we calculated the PRS using threshold 1.00. The results were almost identical to the main results; see Appendix Tables A1 and A2. 4. Discussion We replicated the very high mortality rates found in persons with schizophrenia, with an almost 10 times higher mortality rate compared to the controls. Furthermore, we replicated the very high ORs of N30 associated with two or more suicide attempts in persons with schizophrenia. The genetic component in people with schizophrenia or in controls, measured by the PRS, was associated with suicide attempts in the basic adjusted model, but not in the mutually adjusted model. There was no significant association with mortality in either model. In contrast, we found that family history of mental disorders was associated with both higher mortality and higher risk of suicide attempts. 4.1. Strengths and limitations Our sample comprised 1780 persons with schizophrenia and 1768 matched controls. This limited sample size could affect our ability to detect the influence of the PRS. However, we could easily reproduce the known results of excess mortality and larger risks of suicide in persons with schizophrenia. In the same way, we could show that family history of mental disorders affected the risk of the two outcomes, suggesting that it was not a lack of power that prompted our results. The PRS is a relatively new approach to measure the genetic architecture of schizophrenia. The lack of a positive association between the PRS score and the excess mortality in this study could therefore be attributable to the fact that the PRS score is not yet a sufficient tool to measure genetic risk to schizophrenia. The measure may be improved in the future as the sample sizes used in GWAS to estimate the individual SNP effects become larger and thus the estimates become more accurate. However, at the moment the polygenic score is the best candidate for a single measure that simultaneously takes many variants into account (Kendler, 2016). There have been suggestions of a locus/loci in the XY homologous regions (Crow, 2013; De Lisi et al., 1994) and there is a major interaction with the epigenetic mechanism of X inactivation here (Crow, 2015). Thus our study cannot preclude that a solution lies in “hidden heritability” rather than hitherto unsuspected family interactions. Paternal age and maternal grand-father age at the birth of the proband (Crow, 2012) are consistent with an epigenetic explanation which may suggest that sex chromosomal influence could be a scope for further research. We have an unbiased and incident sample of all people diagnosed with schizophrenia in Denmark as well as population based controls drawn at random from the entire population. This ensured that not only very severe cases of schizophrenia were enrolled in the analysis, avoiding a bias towards larger effects of the covariates. Likewise, not
only very healthy people were drawn as controls ensuring true representation of an entire population (Meier et al., 2015). Because of limited power, we adjusted for parental history of any mental disorder, and we only adjust for the genetics of schizophrenia using the PRS score. Thus e.g. genetic variants associated with depression or anxiety could be passed down to the children, affecting mortality and then be captured by the general family history of mental disorders. 4.2. Interpretation The hypothesis that the genetic factors contributing to the etiology of schizophrenia may also be risk factors for reduced life expectancy (Laursen et al., 2014b) was not supported in the study. Our study suggests that there only is a weak influence, if any, of the PRS on excess mortality found among people with schizophrenia or controls. The elucidation of the genetic architecture of schizophrenia has been a key issue in schizophrenia research for many years. GWAS studies have had the problem that the lack of significant findings could be due to a lack of statistical power, owing to the large number of statistical tests combined with small effect sizes of individual genetic variants. In contrast, the PRS score does not have equivalent challenges with multiple testing. With a very large discovery sample the precision of the PRS improves, but, although it is a very promising measure of genetic liability to schizophrenia, it only captures a small fraction of the overall risk as well as the genetic risk of schizophrenia (Agerbo et al., 2012, 2015). It is important to note that the risk score is only generated using results from studies that focus on genetic variants that are common in the population (minor allele frequency N 1%). They do not fully capture rare genetic variants that potentially contribute substantially to schizophrenia risk. It has been reported that the PRS score was higher for persons prescribed Clozapine (taken as an indication of treatment resistance) in a smaller sample with 434 Clozapine users and 370 non-users (Frank et al., 2015) showing that it is possible to detect differences in smaller samples. We replicated that family history of mental disorders is associated with higher mortality in all models; a finding that often is ascribed to shared genetic factors and environmental factors. As we made an adjustment of the genetic risk factor of schizophrenia, our results could suggest that the environmental factors of the family history of mental disorders could be more influential than the genetic factors of schizophrenia. However, it should be noted that the PRS in its current form may not be an optimal indicator of the genetic risk, and thus the assessment of the balance between genes and environment should be made with caution. 5. Conclusion Genetic predisposition for schizophrenia, as measured by the most recent (and most informative) PRS, was not associated with excess mortality or risk of suicide attempts in those with schizophrenia. In contrast, these outcomes were significantly associated with family history of mental disorders.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
5
Table 5 Odds ratios of suicide attempts, cases and controls: polygenic score and family history. Schizophrenia group
Polygenic score Family history
Yes vs. no
Non schizophrenia group
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
Suicide attempts = 1 OR (95% CI)
Suicide attempts = 2+ OR (95% CI)
0.94 (0.80; 1.11) 1.25 (0.93; 1.68)
0.95 (0.82; 1.10) 1.29 (1.00; 1.67)
1.22 (0.82; 1.83) 2.69 (1.34; 5.40)
1.29 (0.72; 2.31) 3.16 (1.18; 8.45)
Multinomial logistic model using suicide attempts = 0 as reference category. OR = odds ratio, CI = confidence interval. Polygenic score = polygenic risk profile score for schizophrenia with p-value threshold 0.05 and normalized to the sample. The estimates are adjusted for age at the matching time in one-year groups, year at the matching time in one-year groups, sex and the first 10 genomic principal components and the model includes both the polygenic score and the family history.
Our findings could reflect that the common variants detected by recent PRS only explain a small proportion of risk of schizophrenia, and thus it is not surprising that nested outcomes related to mortality within schizophrenia are not associated with PRS. Thus, our findings should be considered as inconclusive, and do not exclude the possibility that future, more powerful PRS instruments may be able to predict mortality within this disorder. Regardless of these issues, it is important to remain mindful that patients with schizophrenia have an increased risk of premature death, and that health service providers should focus on interventions related to modifiable risk factors such as smoking, sedentary life lifestyle and comorbid physical disorders. Role of funding source The study was supported by an unrestricted grant from the Lundbeck Foundation (R155-2014-1724). The funding bodies played no role in the design of the study, preparation of the manuscript, or decision to publish.
Author contributions Thomas Munk Laursen, Esben Agerbo: Designed the study and wrote the first draft of the manuscript. Betina Trabjerg, Ole Mors, Anders D. Børglum, David M. Hougaard, Manuel Mattheisen, Sandra Meier, Enda Byrne, Preben B. Mortensen, and Trine Munk-Olsen: Assisted with editing and preparation of the final manuscript.
Conflicts of interest None.
Acknowledgement The manuscript has been read and approved by all named authors, and all authors have approved the order of authorship. There are no other individuals who satisfied the criteria for authorship but are not listed.
Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.schres.2016.12.001. References Agerbo, E., Mortensen, P.B., Wiuf, C., Pedersen, M.S., McGrath, J., Hollegaard, M.V., Nørgaard-Pedersen, B., Hougaard, D.M., Mors, O., Pedersen, C.B., 2012. Modelling the contribution of family history and variation in single nucleotide polymorphisms to risk of schizophrenia: a Danish national birth cohort-based study. Schizophr Res. 134 (2-3), 246–252. Agerbo, E., Sullivan, P.F., Vilhjalmsson, B.J., Pedersen, C.B., Mors, O., Borglum, A.D., Hougaard, D.M., Hollegaard, M.V., Meier, S., Mattheisen, M., Ripke, S., Wray, N.R., Mortensen, P.B., 2015. Polygenic risk score, parental socioeconomic status, family history of psychiatric disorders, and the risk for schizophrenia: a Danish populationbased study and meta-analysis. JAMA Psychiatry. Biological insights from 108 schizophrenia-associated genetic lociNature 511, 421–427. Borglum, A.D., Demontis, D., Grove, J., Pallesen, J., Hollegaard, M.V., Pedersen, C.B., Hedemand, A., Mattheisen, M., Uitterlinden, A., Nyegaard, M., Orntoft, T., Wiuf, C., Didriksen, M., Nordentoft, M., Nothen, M.M., Rietschel, M., Ophoff, R.A., Cichon, S., Yolken, R.H., Hougaard, D.M., Mortensen, P.B., Mors, O., 2014. Genome-wide study of association and interaction with maternal cytomegalovirus infection suggests new schizophrenia loci. Mol. Psychiatry 19, 325–333. Charlson, M.E., Pompei, P., Ales, K.L., MacKenzie, C.R., 1987. A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J. Chronic Dis. 40, 373–383.
Christiansen, E., Agerbo, E., Larsen, K.J., Bilenberg, N., Stenager, E., 2015. Youth, suicide attempts and low level of education: a Danish historical register-based cohort study of the outcome of suicide attempt. Int. J. Soc. Psychiatry 61, 802–810. Crow, T.J., 2012. Grandparental and parental age effects suggest epigenetic transmission on both X and Y chromosomes. Schizophr. Res. 139, 267–268. Crow, T.J., 2013. The XY gene hypothesis of psychosis: origins and current status. Am. J. Med. Genet. B Neuropsychiatr. Genet. 162B, 800–824. Crow, T.J., 2015. Is psychosis a disorder of XY epigenetics? EBioMedicine 2, 794–795. De Lisi, L.E., Friedrich, U., Wahlstrom, J., Boccio-Smith, A., Forsman, A., Eklund, K., Crow, T.J., 1994. Schizophrenia and sex chromosome anomalies. Schizophr. Bull. 20, 495–505. Frank, J., Lang, M., Witt, S.H., Strohmaier, J., Rujescu, D., Cichon, S., Degenhardt, F., Nothen, M.M., Collier, D.A., Ripke, S., Naber, D., Rietschel, M., 2015. Identification of increased genetic risk scores for schizophrenia in treatment-resistant patients. Mol. Psychiatry 20, 150–151. Harris, E.C., Barraclough, B., 1998. Excess mortality of mental disorder. Br. J. Psychiatry 173, 11–53. Helweg-Larsen, K., 2011. The Danish register of causes of death. Scand. J. Public Health 39, 26–29. Jerrell, J.M., McIntyre, R.S., Tripathi, A., 2010. Incidence and costs of cardiometabolic conditions in patients with schizophrenia treated with antipsychotic medications. Clin. Schizophr. Relat. Psychoses 4, 161–168. Juel, K., Sørensen, J., 2006. Risikofaktorer og folkesundhed i Danmark. Copenhagen. Ref Type: Conference Proceeding. Kendler, K.S., 2016. The schizophrenia polygenic risk score: to what does it predispose in adolescence? JAMA Psychiatry 73, 193–194. Laursen, T.M., 2011. Life expectancy among persons with schizophrenia or bipolar affective disorder. Schizophr. Res. 131, 101–104. Laursen, T.M., Nordentoft, M., 2011. Heart disease treatment and mortality in schizophrenia and bipolar disorder - changes in the Danish population between 1994 and 2006. J. Psychiatr. Res. 45, 29–35. Laursen, T.M., Munk-Olsen, T., Nordentoft, M., Mortensen, P.B., 2007. Increased mortality among patients admitted with major psychiatric disorders: a register-based study comparing mortality in unipolar depressive disorder, bipolar affective disorder, schizoaffective disorder, and schizophrenia. J. Clin. Psychiatry 68, 899–907. Laursen, T.M., Munk-Olsen, T., Gasse, C., 2011. Chronic somatic comorbidity and excess mortality due to natural causes in persons with schizophrenia or bipolar affective disorder. PLoS One 6, e24597. Laursen, T.M., Munk-Olsen, T., Vestergaard, M., 2012. Life expectancy and cardiovascular mortality in persons with schizophrenia. Curr. Opin. Psychiatry 25, 83–88. Laursen, T.M., Mortensen, P.B., Mac Cabe, J.H., Cohen, D., Gasse, C., 2014a. Cardiovascular drug use and mortality in patients with schizophrenia or bipolar disorder: a Danish population-based study. Psychol. Med. 44, 1625–1637. Laursen, T.M., Nordentoft, M., Mortensen, P.B., 2014b. Excess early mortality in schizophrenia. Annu. Rev. Clin. Psychol. 10, 425–448. Lynge, E., Sandegaard, J.L., Rebolj, M., 2011. The Danish National Patient Register. Scand. J. Public Health 39, 30–33. McGue, M., Gottesman, I.I., Rao, D.C., 1983. The transmission of schizophrenia under a multifactorial threshold model. Am. J. Hum. Genet. 35, 1161–1178. Meier, S.M., Agerbo, E., Maier, R., Pedersen, C.B., Lang, M., Grove, J., Hollegaard, M.V., Demontis, D., Trabjerg, B.B., Hjorthoj, C., Ripke, S., Degenhardt, F., Nothen, M.M., Rujescu, D., Maier, W., Werge, T., Mors, O., Hougaard, D.M., Borglum, A.D., Wray, N.R., Rietschel, M., Nordentoft, M., Mortensen, P.B., Mattheisen, M., 2015. High loading of polygenic risk in cases with chronic schizophrenia. Mol. Psychiatry. Mors, O., Perto, G.P., Mortensen, P.B., 2011. The Danish psychiatric central research register. Scand. J. Public Health 39, 54–57. Mortensen, P.B., Pedersen, C.B., Westergaard, T., Wohlfahrt, J., Ewald, H., Mors, O., Andersen, P.K., Melbye, M., 1999. Effects of family history and place and season of birth on the risk of schizophrenia. N. Engl. J. Med. 340, 603–608. Nordentoft, M., Mortensen, P.B., Pedersen, C.B., 2011. Absolute risk of suicide after first hospital contact in mental disorder. Arch. Gen. Psychiatry 68, 1058–1064. Norgaard-Pedersen, B., Hougaard, D.M., 2007. Storage policies and use of the Danish newborn screening biobank. J. Inherit. Metab. Dis. 30, 530–536. Odegard, O., 1951. Mortality in Norwegian mental hospitals 1926–1941. Acta Genet. Stat. Med. 2, 141–173. Osby, U., Correia, N., Brandt, L., Ekbom, A., Sparen, P., 2000. Mortality and causes of death in schizophrenia in Stockholm county, Sweden. Schizophr. Res. 45, 21–28. Pedersen, C.B., Gotzsche, H., Moller, J.O., Mortensen, P.B., 2006. The Danish civil registration system. A cohort of eight million persons. Dan. Med. Bull. 53, 441–449. Price, A.L., Patterson, N.J., Plenge, R.M., Weinblatt, M.E., Shadick, N.A., Reich, D., 2006. Principal components analysis corrects for stratification in genome-wide association studies. Nat. Genet. 38, 904–909.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001
6
T.M. Laursen et al. / Schizophrenia Research xxx (2016) xxx–xxx
Purcell, S.M., Wray, N.R., Stone, J.L., Visscher, P.M., O'Donovan, M.C., Sullivan, P.F., Sklar, P., 2009. Common polygenic variation contributes to risk of schizophrenia and bipolar disorder. Nature 460, 748–752. Ribe, A.R., Laursen, T.M., Sandbaek, A., Charles, M., Nordentoft, M., Vestergaard, M., 2014. Long-term mortality of persons with severe mental illness and diabetes: a population-based cohort study in Denmark. Psychol. Med. 44, 3097–3107. Saha, S., Chant, D., McGrath, J., 2007. A systematic review of mortality in schizophrenia: is the differential mortality gap worsening over time? Arch. Gen. Psychiatry 64, 1123–1131.
Scheewe, T.W., Takken, T., Kahn, R.S., Cahn, W., Backx, F.J., 2012. Effects of exercise therapy on cardiorespiratory fitness in patients with schizophrenia. Med. Sci. Sports Exerc. 44, 1834–1842. Thornicroft, G., 2011. Physical health disparities and mental illness: the scandal of premature mortality. Br. J. Psychiatry 199, 441–442.
Please cite this article as: Laursen, T.M., et al., Association of the polygenic risk score for schizophrenia with mortality and suicidal behavior - A Danish population-based stu, Schizophr. Res. (2016), http://dx.doi.org/10.1016/j.schres.2016.12.001