Arranz et al. Translational Psychiatry (2019)9:177 https://doi.org/10.1038/s41398-019-0511-9 Translational Psychiatry ARTICLE Open Access A pharmacogenetic intervention for the improvement of the safety profile of antipsychotic treatments Maria J. Arranz 1,2, Alex Gonzalez-Rodriguez3, Josefina Perez-Blanco2,4, Rafael Penadés2,5, Blanca Gutierrez6, Laura Ibañez1, Barbara Arias2,7, Mercè Brunet8, Jorge Cervilla6, Juliana Salazar9 and Rosa Catalan2,5 Abstract Antipsychotic drugs fail to achieve adequate response in 30–50% of treated patients and about 50% of them develop severe and lasting side effects. Treatment failure results in poorer prognosis with devastating repercussions for the patients, carers and broader society. Our study evaluated the clinical benefits of a pharmacogenetic intervention for the personalisation of antipsychotic treatment. Pharmacogenetic information in key CYP polymorphisms was used to adjust clinical doses in a group of patients who started or switched treatment with antipsychotic drugs (PharmG+, N = 123), and their results were compared with those of a group of patients treated following existing clinical guides (PharmG−, N = 167). There was no evidence of significant differences in side effects between the two arms. Although patients who had their antipsychotic dose adjusted according to CYPs polymorphisms (PharmG+) had a bigger reduction in side effects than those treated as usual (PharmG−), the difference was not statistically significant (p > 0.05 for all comparisons). However, PharmG+ patients treated with CYP2D6 substrates that were carriers of CYP2D6 UMs or 1234567890():,; 1234567890():,; 1234567890():,; 1234567890():,; PMs variants showed a significantly higher improvement in global, psychic and other UKU side effects than PharmG− patients (p = 0.02, p = 0.05 and p = 0.01, respectively). PharmG+ clozapine treated patients with CYP1A2 or CYP2C19 UM and PMs variants also showed higher reductions in UKU scores than PharmG− clozapine patients in general. However, those differences were not statistically significant. Pharmacogenetic interventions may improve the safety of antipsychotic treatments by reducing associated side effects. This intervention may be particularly useful when considering treatment with antipsychotics with one major metabolic pathway, and therefore more susceptible to be affected by functional variants of CYP enzymes. Introduction broader society. The reasons for treatment failure are Antipsychotic drugs are widely used for the treatment unclear and ‘trial and error' naturalistic strategies are of severe mental illness including schizophrenia, bipolar used to select drug type and clinical doses for each disorder and major depression. However, 30–50% of patient. treated patients fail to respond adequately and about Growing evidence indicates that genetic factors play a 50% of them develop severe and lasting side effects. critical role in determining the clinical outcome of anti- Treatment failure results in poorer prognosis with psychotic treatment1,2. Several genetic polymorphisms devastating repercussions for the patients, carers and have been identified in drug metabolic enzymes that influence metabolising rates and clinical outcome. Cyto- chrome P450 enzymes, responsible for the bio- Correspondence: Maria J. Arranz (mjarranz@mutuaterrassa.es) or Rosa Catalan transformation of more than 85% of existing drugs, may (rcatalan@clinic.cat) 1 contain genetic variants that result in poor, intermediate, Fundació Docència i Recerca Mútua Terrassa, Terrassa, Spain 2 Centro de Investigación en Red de Salud Mental, CIBERSAM, Madrid, Spain extensive or ultra-rapid metabolic rates3,4. Individuals Full list of author information is available at the end of the article. © The Author(s) 2019 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/licenses/by/4.0/. Arranz et al. Translational Psychiatry (2019)9:177 Page 2 of 8 with slow metabolising CYP variants are prone to adverse Materials and methods reactions, and require lower doses, whilst patients with Description of sample ultra-rapid metabolising CYP variants require higher Two hundred and ninety patients with a diagnosis of doses to obtain therapeutic response. CYPs are respon- schizophrenia, schizoaffective or delusional disorders sible for the biotransformation of widely used anti- (according to DSM-5) and requiring antipsychotic treat- psychotics such as clozapine (CYP1A2 and CYP2C19), ment completed the study. Patients were recruited in the olanzapine (CYP1A2), risperidone and haloperidol mental health wards of three hospitals (Hospital Clínic (CYP2D6), quetiapine and ziprasidone (CYP3A5) and Hospital de la Santa Creu i Sant Pau, Barcelona; amongst others5,6. The direct correlation between pre- Hospital Universitario, Granada) and were mostly out- sence of slow metabolising CYP variants and patients. Patients were randomly selected for PI (PharmG antipsychotic-induced adverse reactions is well docu- + arm) or treatment as usual (PharmG− arm). Rando- mented, although their influence on the level of efficacy in misation was conducted using a web programme (www. less clear7,8. It has been suggested that the adjustment of randomiser.org). Pharmacogenetic arm (PharmG+): 123 clinical doses according to the CYP genetic profile (pre- patients were genotyped using a pharmacogenetic test at sence of functional polymorphisms) of patients has the the beginning of their treatment with a new antipsychotic potential to significantly improve the efficacy (15%) and and clinical doses were adjusted accordingly when safety (25% reduction in side effects) of pharmacological required (PharmG+). Naturalistic arm (PharmG−): 167 treatments9. patients were treated as usual following standard clinical There are several commercially available pharmacoge- practices. This sample has a statistical power > 95% to netic tests that interrogate CYP functional polymorph- detect genetic associations with a medium effect size (f = isms, which can be potentially useful for the 0.25, α error = 0.05, calculated with GPower version personalisation of antipsychotic drugs10. However, their 3.0.10). use in clinical settings is minimal. Recent studies have proven the benefits of using pharmacogenetic information Response assessment for the personalisation of treatment in psychiatry. CYPs, The positive and negative syndrome scale for schizo- UGTs and multidrug resistance 1 (MDR1) genotyping for phrenia (PANSS19) and the UKU-side effect rating scale antidepressant dose adjustment improves remission11. (UKU-SERS20) scores were obtained in all patients at the The use of commercial tests interrogating key CYP beginning and after 12 weeks to assess the efficacy and polymorphisms and serotonin genes resulted in improved safety of the treatment. PANSS and UKU values were response to antidepressants12–15. A recent study in elderly scored by trained psychiatrists who were blind to the patients treated with antidepressants and antipsychotics patients' pharmacogenetic arm. Within each participating according to their pharmacogenetic profile had a sig- hospital, the same psychiatrist evaluated the pre and post nificant decrease in hospitalisations and emergency treatment patient status for both arms. Adherence was department visits16. Pharmacogenetic-guided prescription confirmed by plasma levels at week 12 of treatment. of antidepressants also resulted in improved adherence Table 1 summarises the clinical and demographic data of and reduced pharmacy costs17,18. However, most studies the study samples. All participants gave informed consent performed to date have focused on antidepressant medi- to the study, which was approved by the local ethics cations. It is not yet sufficiently established the clinical committees. and economic benefits of introducing pharmacogenetic interventions (PIs) for the improvement of antipsychotic Biological samples treatment. In addition, rapid testing protocols that deliver Whole blood samples were collected from all partici- clear recommendations on antipsychotic choice and dose pants at the time of recruitment. DNA was obtained using are required to increase the implementation of pharma- commercial kits (QIAmp DNA mini Kit, Qiagen) within cogenetics in clinical practice. hours of blood extraction and was immediately sent for The aim of this study was to evaluate the clinical ben- pharmacogenetic testing (PharmaG+ patients) or stored efits of a PI for the personalisation of antipsychotic at −80 °C until required. treatment. Pharmacogenetic information in key CYP polymorphisms was used to adjust clinical doses in a Pharmacogenetic intervention group of patients who started or switched treatment with A commercial pharmacogenetic test (Brainchip, antipsychotic drugs, and their results were compared with Brainco, Bilbao, Spain) was used to characterise common those of a group of patients treated following existing and functional polymorphisms in CYP1A2, CYP2D6, clinical guides. We obtained encouraging results indicat- CYP2C19 and CYP3A5 enzymes, the main metabolic ing that pharmacogenetic-guided antipsychotic prescrip- pathways of currently available antipsychotics. Table 2 tion may increase the safety of antipsychotic treatments. summarises the list of key CYP polymorphisms genotyped Arranz et al. Translational Psychiatry (2019)9:177 Page 3 of 8 Table 1 Clinical and demographic data of the study Table 2 Key CYP polymorphisms genotyped in the samples samples Study arm Totals Gene Alleles genotyped Known function PharmG+ PharmG− Total CYP1A2 *1 Normal (N = 123) (N = 167) (N = 290) *1F Higher inducibility CYP2C19 *1 Normal Gender *2 No activity Male 63 (51.2%) 94 (56.3%) 157 (54.1%) *3 No activity Female 60 (48.8%) 73 (43.7%) 133 (45.9%) *4 No activity Age *8 Decreased activity Mean (SD) (years) 46.11 ± 13.7 48.68 ± 13.46 CYP2D6 *1 Normal Diagnosis *2 Normal Schizophrenia 86% 69% 76% *3 No activity Schizoaffective 5% 4% 4% *4 No activity Delusional disorder 9% 27% 20% *5 No activity Treatment *6 No activity Drug *9 Decreased activity Clozapine 43 (35%) 88 (52.7%) 131 (45.2%) *10 Decreased activity Risperidone 16 (13%) 20 (12%) 36 (12.4%) *35 Normal Olanzapine 25 (20.3%) 14 (8.4%) 39 (13.4%) *41 Decreased Paliperidone 16 (13%) 22 (13%) 38 (13.1%) *2×N Increased activity Aripiprazole 7 (5.7%) 13 (7.8%) 20 (6.9%) *4×N Increased activity Quetiapine 11 (8.9%) 5 (3%) 16 (5.5%) CYP3A5 *1 Normal Ziprasidone 1 (0.8%) 2 (1.2%) 3 (1%) *3 Reduced activity Trifluoperazine 1 (0.8%) 1 (0.6%) 2 (0.7%) Haloperidol 1 (0.8%) 1 (0.6%) 2 (0.7%) Asenapine 1 (0.8%) 1 (0.6%) 2 (0.7%) Dose adjustment protocol Pimozide 1 (0.8%) 0 1 (0.3%) Patients were recruited when starting treatment with an Dose (olanzapine equivalent) antipsychotic or when a change in their antipsychotic Average dose (SD) 11.57 ± 7.28 10.65 ± 5.70 treatment was required. All patients (PharmG+ and PANSS basal scores PharmG−) were informed of the study and were blind to Mean (SD) 99.23 ± 18.08 91.55 ± 17.55 the PI. No pharmacogenetic information was provided to the patients, nor were they told if changes in their med- UKU basal scores ication had been required as a result. At the time of the Mean (SD) 9.15 ± 5.98 6.83 ± 5.40 study, there were no clinical guidelines on the level of dose adjustment required for patients with CYP func- tional polymorphisms. The clinical recommendations that included the most frequently detected in Caucasian (dose adjustments) used in the study were decided by the populations21. A DNA sample from PharmG+ patients clinical teams at the beginning of the project and are was obtained at the time of initiating or changing anti- summarised in Table 3. Doses were adjusted according to psychotic treatment and sent to Brainco for character- main metabolic pathways:5,6 clozapine doses were adjus- isation with the Brainchip array. Results were returned ted according to the genotypic variants observed in within 1–2 weeks of extraction to the responsible clin- CYP1A2 and CYP2C19; olanzapine doses were adjusted ician, and dose adjustments, if necessary, were performed according to CYP1A2 polymorphisms; risperidone, aripi- according to the study protocol. Patients were maintained prazole, haloperidol, pimozide and trifluoperazine doses on antipsychotic monotherapy during the duration of the were adjusted according to CYP2D6 polymorphisms; study. finally, quetiapine and ziprasidone doses were adjusted Arranz et al. Translational Psychiatry (2019)9:177 Page 4 of 8 Table 3 Recommended adjustment of clinical doses for patients undergoing pharmacogenetic intervention (PharmG+) Antipsychotic Standard CYP1A2 CYP2D6 CYP2C19 CYP3A5 clinical dose PM IM UM PM IM UM PM IM PM IM UM Clozapine 150–900 mg/day <25–50% NC >25–30% <25–30% NC Olanzapine 7.5–30 mg/day <25–50% NC >25–30% Risperidone 3–12 mg/day <25–30% Gradual >25–30% Haloperidol 10–20 mg/day <50% <10–20% >25–30% Aripiprazole 10–30 mg/day <25–50% Gradual >25–30% Quetiapine 300–750 mg/day <25–50% <10–20% >25–30% Ziprasidone 80–160 mg/day <5–50% <10–20% >25–30% Gradual: 5–10% reduction every 3 days (up to 30% reduction) depending on side effects according to CYP3A5 genetic variants. Paliperidone and between arms (see Table 1). To minimise these differ- asenapine doses were not adjusted unless they were given ences, the PANSS or UKU score differences (basal scores in combination with a second antipsychotic, in which case − scores at 12 weeks of treatment) were considered as the the second antipsychotic was used as a guide for adjust- dependent variables. In addition, treatment (antipsychotic ment. According to this protocol, N = 55 PharmG+ drug) was one of the covariates considered in the patients required dose adjustments of the antipsychotic regression model. drug received. Despite adjustments, clinical doses No significant differences in variation in PANSS scores remained within recommended ranges in both arms. were observed between PharmG+ (26.81 ± 1.3 score reduction after 12 weeks, on average) and PharmG− Genotypic characterisation of samples patients (26.6 ± 1 score reduction after 12 weeks, on All patients (PharmG+ and PharmG−, N = 290) were average) (see Table 4). Linear model regression con- genotyped for the key CYP functional polymorphisms sidering PI (PharmG+ or PharmG− arm), age, gender, included in the pharmacogenetic array using iPlex® Gold dose (olanzapine equivalent) and drug as covariates con- chemistry and the MassARRAY platform (CEGEN-PRB2- firmed this observation (model χ2 = 6.72, df = 5, p = 0.24, ISCIII, University of Santiago de Compostela, Spain) and PI Wald coefficient = 0.04, p = 0.84). Similarly, no sig- TaqMan probes and technology for the CNVs (University nificant differences in PANSS scores improvement were of Granada). High-throughput genotyping results coin- observed between PharmG+ and PharmG− patients cided with those obtained with the commercial array in treated with clozapine (29.36 ± 2.3 vs 26 ± 1.4, respec- overlapping samples and polymorphisms. tively; model χ2 = 8.54, df = 4, p = 0.08, Wald coefficient for PI = 1.32, p = 0.25). Analyses within the subgroup of Statistical analyses patients treated with drugs mainly metabolised by the Linear regression models to assess the influence of the enzyme CYP2D6 (risperidone, haloperidol, aripiprazole, PI were calculated considering level of response and side pimozide and trifluoperazine) also showed no differences effects improvement as dependent variables. The PI with PharmG+ and PharmG− patients showing similar (study arm PharmG+ or PharmG−) was included as a variations in PANSS scores (24.33 ± 3 vs 26.47 ± 2). This predictor variable and drug, gender, dose (olanzapine difference was not statistically significant (χ2 = 2.90, df = equivalent), age of patients and PI as additional covari- 5, p = 0.72 and Wald = 0.34, p = 0.56 for model and PI, ables. All analyses were performed using the statistic respectively). When analysing patients with functional package SPSS (version 22, IBM). variants in CYP1A2, CYP2D6 and/or CYP2C19 (N = 155), no differences were observed between PharmG+ and Results PharmG− patients in PANSS improvement, nor was it Tables 4 and 5 summarise the results of the statistical observed in the subgroups of patients treated with clo- analyses. The CYP minor allele frequencies observed zapine (p > 0.05 for all comparisons, see Table 5). How- agreed with those described in European populations21. ever, there was a significant difference when analysing Despite randomisation during the recruitment, differ- PANSS score improvement between PharmG+ patients ences were observed in the distribution of treatments (p (26.6 ± 3.5) and PharmG− patients (28.63 ± 3.7) treated = 0.01) and in the severity of symptoms (p = 0.002) with CYP2D6 substrates. This significant difference was Arranz et al. Translational Psychiatry (2019)9:177 Page 5 of 8 Table 4 Summary of statistical analyses in all samples Clinical variable All patients Clozapine treated patients Patients treated with (N = 290) (N = 131) CYP2D6 substrates (N = 61) PharmG+ PharmG− PharmG+ PharmG− PharmG+ PharmG− (N = 123) (N = 167) (N = 43) (N = 88) (N = 26) (N = 35) Difference in PANSS average 26.81 + 1.3 26.60 + 1 29.36 + 2.3 26.0 + 1.4 24.33 + 3.0 26.47 + 2.0 Model χ2 (p value) 6.72 (0.24) 8.54 (0.08) 2.90 (0.72) PI Wald (p value) 0.04 (0.84) 1.32 (0.25) 0.34 (0.56) Difference in UKU scores 3.62 ± 0.66 3 ± 0.56 3.81 ± 1.10 2.67 ± 1.03 4.04 ± 1.34 2.69 ± 1.11 Model χ2 (p value) 10.19 (0.07) 5.67 (0.23) 2.79 (0.73) PI Wald (p value) 0.40 (0.53) 0.002 (0.97) 0.38 (0.54) Difference in psychic SE 2.02 ± 0.36 1.26 ± 0.30 1.84 ± 0.63 1.38 ± 0.52 1.60 ± 0.72 0.53 ± 0.53 Model χ2 (p value) 10.38 (0.07) 3.26 (0.52) 7.71 (0.17) PI Wald (p value) 1.53 (0.22) 0.03 (0.86) 0.64 (0.42) Difference in neurological SE 0.87 ± 0.20 0.94 ± 0.25 0.98 ± 0.32 0.54 ± 0.51 1.04 ± 0.48 1.25 ± 0.56 Model χ2 (p value) 12.49 (0.03) 12.65 (0.01) 2.01 (0.85) PI Wald (p value) <0.01 (0.99) 0.01 (0.91) 0.08 (0.78) Difference autonomic SE 0.32 ± 0.26 0.17 ± 0.12 0.16 ± 0.51 0.37 ± 0.37 0.71 ± 0.45 0.28 ± 0.16 Model χ2 (p value) 4.09 (0.54) 4.42 (0.35) 10.03 (0.07) PI Wald (p value) 0.35 (0.56) 0.47 (0.49) 0.93 (0.33) Difference in other side 0.54 ± 0.24 0.52 ± 0.23 0.93 ± 0.30 0.38 ± 0.41 0.56 ± 0.30 0.44 ± 0.29 effects Model χ2 (p value) 9.67 (0.09) 2.53 (0.64) 2.45 (0.78) PI Wald (p value) 0.003 (0.95) 1.32 (0.25) 0.07 (0.79) PI pharmacogenetic intervention, SE side effects mainly caused by patients' gender (Wald = 6.25, p = 0.01) side effects (1.08 vs −0.3, Wald = 6.19, p = 0.01) in the and dose (Wald = 6.25, p = 0.01) and not by the PI (Wald subgroup of patients treated with CYP2D6 substrates (see = 1.03, p = 0.31). Table 5). Analyses of improvement in side effects after 12 weeks of treatment (UKU basal scores − UKU scores after 12 weeks) Discussion revealed a slightly higher improvement (bigger difference in Despite growing evidence showing the influence of UKU score reduction) in PharmG+ than in PharmG− genetic factors on antipsychotic treatment efficacy, phar- patients in general, especially in the UKU psychic subscales macogenetic information is rarely used in clinical settings (Table 4). However, the contribution of the PI to these for the personalisation of treatment. Several studies have improvements was not statistically significant (p > 0.05 for shown that most clinicians and patients are in favour of all comparisons). Analyses of patients with functional var- PIs22–24. However, the scarce evidence on the clinical and iants in CYP1A2, CYP2D6 or CYP2C19 revealed clearer economic benefits of PIs in psychiatry hamper their improvements in the profile of side effects in the PharmG+ implementation in clinical practice. In this study, we patients in comparison to PharmG− patients. Psychic side investigated the clinical benefits of using genetic infor- effects were further reduced in PharmG+ patients (2.40 ± mation in CYP metabolising enzymes for the adjustment 0.41) in comparison to PharmG− patients (1.21 ± 0.37; of clinical doses of a variety of commonly used anti- χ2 = 10.90, p = 0.05 and Wald = 3.30, p = 0.07 for model psychotics and compared the results with those of and PI, respectively). The PI was significantly associated to patients treated as usual. improvements in total (6.92 improvement in PharmG+ vs Several commercial pharmacogenetic tests are available 1.27 in PharmG−, Wald = 5.56, p = 0.02), psychic (2.50 vs for use in psychiatry25. Most of them interrogate key func- 0.53 improvement, Wald = 3.96, p = 0.05) and other UKU tional polymorphisms in drug metabolising enzymes. Arranz et al. Translational Psychiatry (2019)9:177 Page 6 of 8 Table 5 Summary of statistical analyses in individuals with functional variants in relevant CYPs Clinical variable All patients Clozapine treated patients Patients treated with (N = 155) (N = 68) CYP2D6 substrates (N = 41) PharmG+ PharmG− PharmG+ PharmG− PharmG+ PharmG− (N = 79) (N = 76) (N = 30) (N = 38) (N = 21) (N = 20) Difference in PANSS average 26.88 + 1.4 25.92 + 1.3 30.10 + 3 24.74 + 2 26.58 + 3.5 28.63 + 3.7 Model χ2 (p value) 1.68 (0.89) 5.19 (0.27) 12.88 (0.03) PI Wald (p value) 0.16 (0.69) 1.78 (0.18) 1.03 (0.31) Difference in UKU scores 4.20 ± 0.76 3.01 ± 0.66 4.24 ± 1.39 3.08 ± 1.25 6.92 ± 2.09 1.27 ± 1.78 Model χ2 (p value) 8.80 (0.12) 3 (0.56) 8.20 (0.15) PI Wald (p value) 1.23 (0.27) 0.02 (0.89) 5.56 (0.02) Difference in psychic SE 2.40 ± 0.41 1.21 ± 0.37 2.14 ± 0.82 1.42 ± 0.70 2.50 ± 1.14 0.53 ± 0.77 Model χ2 (p value) 10.90 (0.05) 1.20 (0.88) 10.83 (0.06) PI Wald (p value) 3.30 (0.07) 0.14 (0.71) 3.96 (0.05) Difference in neurological SE 0.99 ± 0.24 1.08 ± 0.26 1.24 ± 0.40 0.67 ± 0.67 1.67 ± 0.64 0.53 ± 0.76 Model χ2 (p value) 11.27 (0.05) 15.91 (0.003) 3.63 (0.60) PI Wald (p value) 0.01 (0.91) 0.03 (0.87) 1.67 (0.20) Difference autonomic SE 0.31 ± 0.29 0.20 ± 0.13 0.07 ± 0.64 0.75 ± 0.41 1.33 ± 0.91 0.13 ± 0.13 Model χ2 (p value) 3.04 (0.69) 2.14 (0.71) 5.13 (0.40) PI Wald (p value) 0.22 (0.64) 0.66 (0.42) 1.29 (0.26) Difference in other side 0.59 ± 0.28 0.45 ± 0.29 0.86 ± 0.39 0.25 ± 0.74 1.08 ± 0.57 −0.3 ± 0.41 effects Model χ2 (p value) 7.39 (0.19) 6.20 (0.19) 7.99 (0.16) PI Wald (p value) 0.10 (0.75) 0.08 (0.78) 6.19 (0.01) PI pharmacogenetic intervention, SE side effects However, several of these tests include other polymorphisms pharmacokinetics, but also by pharmacodynamic inter- of unclear clinical value. We used a pharmacogenetic test actions with the many receptors targeted by currently that gives detailed information on CYPs functional poly- available antipsychotic drugs. Therefore, the possible morphisms. CYPs genotypes were used for dose adjust- influence of altered metabolism on drug efficacy may be ments, whilst the additional polymorphisms genotyped by diluted. Previous genetic association studies have failed to the test were not considered for the intervention. This find a clear association between CYPs functional poly- alternative information should not be considered until morphisms and antipsychotic treatment response8. confirmation of their relevance is obtained, as it may hamper Nevertheless, strong evidence supports the influence of the clinical value and implementation of pharmacogenetic CYPs variants on the development of drug-induced tests26. In the case of antipsychotics, the most robust results adverse reactions7,22,28–30. Interestingly, we found that are those that associate CYP functional polymorphisms with patients who had their antipsychotic dose adjusted adverse reactions7, whereas other polymorphisms in according to key polymorphisms (PharmG+) had a bigger dynamic genes have been also proven to be useful for the reduction in side effects than those treated as usual improvement of antidepressant medication27. (PharmG−). This finding was not statistically significant Our results showed that a PI consisting in adjustments when investigating all the participants in the study. of clinical doses in patients with CYP alterations did not However, when looking at the subgroup of patients who have a significant influence on the level of efficacy of had functional variants in the CYPs responsible for the antipsychotic treatments (p > 0.05 in all comparisons). metabolism of their prescribed antipsychotic, the results This was somehow expected. Given the complex phar- were clearer. Our PI was particularly useful in the macodynamic profile on antipsychotic drugs, treatment reduction of psychic symptoms (p = 0.07) in patients with success may be influenced not only by drug CYPs functional variants. The subgroup of patients Arranz et al. Translational Psychiatry (2019)9:177 Page 7 of 8 treated with CYP2D6 substrates (risperidone, haloperidol, (antipsychotic treatment and basal symptom severity) aripiprazole, pimozide and trifluoperazine) that were were included as covariates in the analyses. Sample size carriers of CYP2D6 UMs or PMs variants clearly benefited was dimed sufficient for the study aims. However, the from the intervention: model regression analyses showed limited number of patients carrying CYP functional that the PI resulted in a significant improvement in global, polymorphisms did not allow the investigation of func- psychic and other UKU side effects (p = 0.02, p = 0.05 and tional groups (UMs and PMs separately). In addition, the p = 0.01, respectively). PharmG+ clozapine treated variety of treatments used in the study hampered the patients also showed higher reductions in UKU scores investigation of individual drugs. Nevertheless, we per- than PharmG− clozapine patients in general. However, formed analyses on the subgroups of patients carrying those differences were not statistically significant, even CYPs functional variants (either UMs or PMs), and when considering patients with CYP1A2 or CYP2C19 UM investigated the two largest antipsychotic groups within and PMs variants (p > 0.05 in all comparisons). These the study (clozapine and CYP2D6 substrates). Finally, a results suggest that a PI would be particularly useful when CYP2D6*17 polymorphism with reduced activity was not considering treatment with antipsychotic drugs with one investigated in our study as it was not included in the main CYP pathway, where the influence of UMs and PMs array used for the PI. However, the reported frequency of variants should be more evident. this allele in the Spanish population is very low (0.0093) Previous studies on the benefits of PIs in psychiatry have and should not greatly influence the results34. focused mainly on antidepressant treatments. Several studies In summary, although we found no evidence of greater have proven the clinical value of a commercial array com- efficacy, a PI may improve the safety profile of anti- bining genetic information in metabolic enzymes (CYP1A2, psychotic treatments in patients presenting poor or ultra- CYP2D6 and CYP2C19) and antidepressant targets (ser- rapid CYP variants. The PI described in our study may be otonin transporter, SLC6A4 and serotonin receptor 2A, particularly useful when considering treatment with HTR2A). Patients whose antidepressant drug and dose were antipsychotics with one major metabolic pathway, and selected according to the array polymorphisms showed therefore more susceptible to be affected by functional better symptom improvement than patients treated as variants of CYP metabolising enzymes. usual12,13. Similar results were found in a recent study in Spanish patients, also using a commercial kit that included CYPs, multidrug resistance 1 (MDR1 o ABCB1) and other Acknowledgements We are extremely grateful to allparticipants. We thank the Spanish Ministry of target variants1. A study using the same commercial kit in a Economy, Industry, and the European Funds for Regional Development (Fondo retrospective study on psychiatric patients receiving a variety Europeo de Desarrollo Regional, FEDER) for their support. We are indebted to of psychotropic drugs observed that those patients who had the IDIBAPS Biobank, integrated in the Spanish National Biobank Network, for samples and data procurement. This research was funded by grants from the followed the test recommendations showed better treatment Institute Carlos III (FIS PI11/02006; FIS PI16/01029). A.G.-R. was partially funded response31. A recent study in which clinicians were provided by a private grant from the Jové family. Genotyping was partially conducted by pharmacogenetic information based on CYP2D6 and the CEGEN-PRB2-ISCIII node, which is supported by grant PT13/001, ISCIII- SGEFI/FEDER. CYP2C19 variants to adjust antipsychotic and anti- depressant treatments in 80 psychiatric patients suggested a favourable opinion on the outcome of the intervention, Author details although no control group was considered in the study32. 1 Fundació Docència i Recerca Mútua Terrassa, Terrassa, Spain. 2Centro de Interestingly, Herbid et al. showed in a previous study that Investigación en Red de Salud Mental, CIBERSAM, Madrid, Spain. 3Department the adjustment of clinical doses according to CYP2D6 and of Mental Health, Parc Taulí University Hospital Sabadell, Barcelona, Spain. 4 Department of Psychiatry, Hospital de la Santa Creu i Sant Pau, Barcelona, CYP2C19 polymorphisms resulted in a reduction of treat- Spain. 5Barcelona Clinic Schizophrenia Unit (BCSU), Neurosciences Institute, ment costs in schizophrenia patients showing poor or ultra- Hospital Clinic of Barcelona, University of Barcelona, IDIBAPS, Barcelona, Spain. rapid metabolism33. However, no previous study has focused 6 Department of Psychiatry, University of Granada, Granada, Spain. 7Department of Biologia Evolutiva, Ecologia i Ciències Ambientals, Facultat de Biologia, on the prospective evaluation of clinical improvement when Institut de Biomedicina de la Universitat de Barcelona (IBUB), Barcelona, Spain. using pharmacogenetics for the adjustment of antipsychotic 8 Pharmacology and Toxicology Unit, Department of Biochemistry and treatments. To our knowledge, our findings are the first Molecular Genetics, Hospital Clinic, Barcelona, Spain. 9Genetics Department, Hospital de la Santa Creu i Sant Pau, Barcelona, Spain evidence of the clinical benefits obtained by a pharmaco- genetic selection of antipsychotic clinical doses in compar- ison to naturalistic practices. Conflict of interest Our study has several limitations. Despite randomisa- M.J.A. consulted for Brainco, the company commercialising the tion during recruitment, antipsychotic treatments and pharmacogenetic test used in the study. However, the company closed in severity of symptoms were not evenly distributed in the 2014 and did not influence in any way the results of this study. R.C. has received honoraria or travel grants from Janssen, Pfizer, Ferrer, Adamed and study arms (PharmG+ and PharmG−). To minimise the Lundbeck-Otsuka. A.G.-R. has received honoraria and/or travel grants from effect of these differences on results, both factors Janssen, Pfizer, Ferrer and Lundbeck-Otsuka. Arranz et al. Translational Psychiatry (2019)9:177 Page 8 of 8 Publisher’s note 16. Brixner, D. et al. The effect of pharmacogenetic profiling with a clinical deci- Springer Nature remains neutral with regard to jurisdictional claims in sion support tool on healthcare resource utilization and estimated costs in the published maps and institutional affiliations. elderly exposed to polypharmacy. J. Med. Econ. 19, 213–228 (2016). 17. Winner, J. G. et al. Combinatorial pharmacogenomic guidance for psychiatric Received: 15 January 2019 Revised: 9 May 2019 Accepted: 31 May 2019 medications reduces overall pharmacy costs in a 1 year prospective evalua- tion. Curr. Med. Res. Opin. 31, 1633–1643 (2015). 18. Fagerness, J. et al. Pharmacogenetic-guided psychiatric intervention asso- ciated with increased adherence and cost savings. Am. J. Manag. Care 20, e146–e156 (2014). References 19. Kay, S. R., Fiszbein, A. & Opler, L. A. The positive and negative syndrome scale 1. Arranz M., Perez-Blanco J. & Arias B. Pharmacogenetics of the efficacy of (PANSS) for schizophrenia. Schizophr. Bull. 13, 261–276 (1987). antipsychotic drugs in schizophrenia. in Genetic Influences on Response to Drug 20. Lingjaerde, O., Ahlfors, U. G., Bech, P., Dencker, S. J. & Elgen, K. The UKU side Treatment for Major Psychiatric Disorders (eds Rybakowski J. & Serretti A.) effect rating scale. A new comprehensive rating scale for psychotropic drugs (Springer International Publishing, Switzerland, 2016). and a cross-sectional study of side effects in neuroleptic-treated patients. Acta 2. Pouget, J. G., Shams, T. A., Tiwari, A. K. & Müller, D. J. Pharmacogenetics and Psychiatr. Scand. Suppl. 334, 1–100 (1987). outcome with antipsychotic drugs. Dialogues Clin. Neurosci. 16, 555–566 21. Zhou, Y., Ingelman-Sundberg, M. & Lauschke, V. M. Worldwide distribution of (2014). cytochrome P450 alleles: a meta-analysis of population-scale sequencing 3. Bertilsson, L. Metabolism of antidepressant and neuroleptic drugs by cyto- projects. Clin. Pharm. Ther. 102, 688–700 (2017). chrome p450s: clinical and interethnic aspects. Clin. Pharm. Ther. 82, 606–609 22. Müller, D. J., Kekin, I., Kao, A. C. & Brandl, E. J. Towards the implementation of (2007). CYP2D6 and CYP2C19 genotypes in clinical practice: update and report from a 4. Zanger, U. M., Turpeinen, M., Klein, K. & Schwab, M. Functional pharmaco- pharmacogenetic service clinic. Int. Rev. Psychiatry 25, 554–571 (2013). genetics/genomics of human cytochromes P450 involved in drug bio- 23. Lanktree, M. B. et al. Positive perception of pharmacogenetic testing for transformation. Anal. Bioanal. Chem. 392, 1093–1108 (2008). psychotropic medications. Hum. Psychopharmacol. 29, 287–291 (2014). 5. Arranz, M. J. & de Leon, J. Pharmacogenetics and pharmacogenomics of 24. Walden, L. M. et al. Physicians’ opinions following pharmacogenetic testing for schizophrenia: a review of last decade of research. Mol. Psychiatry 12, 707–747 psychotropic medication. Psychiatry Res. 229, 913–918 (2015). (2007). 25. Bousman, C. A., Jaksa, P. & Pantelis, C. Systematic evaluation of commercial 6. Ravyn, D., Ravyn, V., Lowney, R. & Nasrallah, H. A. CYP450 pharmacogenetic pharmacogenetic testing in psychiatry: a focus on CYP2D6 and CYP2C19 allele treatment strategies for antipsychotics: a review of the evidence. Schizophr. coverage and results reporting. Pharm. Genom. 27, 387–393 (2017). Res. 149, 1–14 (2013). 26. de Leon, J. & Spina, E. What is needed to incorporate clinical pharmacogenetic 7. Fleeman, N. et al. The clinical effectiveness and cost-effectiveness of testing for tests into the practice of psychopharmacotherapy? Expert Rev. Clin. Pharmacol. cytochrome P450 polymorphisms in patients with schizophrenia treated with 9, 351–354 (2016). antipsychotics: a systematic review and economic evaluation. Health Technol. 27. Fabbri, C. et al. Consensus paper of the WFSBP Task Force on Genetics: Assess. 14, 1–157 (2010). genetics, epigenetics and gene expression markers of major depressive dis- 8. Fleeman, N. et al. Cytochrome P450 testing for prescribing antipsychotics in order and antidepressant response. World J. Biol. Psychiatry 18, 5–28 (2017). adults with schizophrenia: systematic review and meta-analyses. Pharm. J. 11, 28. Lee, H. J. & Kang, S. G. Genetics of tardive dyskinesia. Int. Rev. Neurobiol. 98, 1–14 (2011). 231–264 (2011). 9. Ingelman-Sundberg, M. Pharmacogenetics of cytochrome P450 and its 29. Zackrisson, A. L., Lindblom, B. & Ahlner, J. High frequency of occurrence of applications in drug therapy: the past, present and future. Trends Pharm. Sci. CYP2D6 gene duplication/multiduplication indicating ultrarapid metabolism 25, 193–200 (2004). among suicide cases. Clin. Pharm. Ther. 88, 354–359 (2010). 10. Bousman, C. A. & Hopwood, M. Commercial pharmacogenetic-based deci- 30. Chaudhry, S. R. et al. Pharmacogenetic prediction of individual variability in sion-support tools in psychiatry. Lancet Psychiatry 3, 585–590 (2016). drug response based on CYP2D6, CYP2C9 and CYP2C19 genetic poly- 11. Singh, A. B. Improved antidepressant remission in major depression via a morphisms. Curr. Drug Metab. 15, 711–718 (2014). pharmacokinetic pathway polygene pharmacogenetic report. Clin. Psycho- 31. Espadaler, J., Tuson, M., Lopez-Ibor, J. M., Lopez-Ibor, F. & Lopez-Ibor, M. I. pharmacol. Neurosci. 13, 150–156 (2015). Pharmacogenetic testing for the guidance of psychiatric treatment: a multi- 12. Altar, C. A. et al. Clinical validity of cytochrome P450 metabolism and serotonin center retrospective analysis. CNS Spectr. 22, 315–324 (2017). gene variants in psychiatric pharmacotherapy. Int. Rev. Psychiatry 25, 509–533 32. Walden L. M. et al. Genetic testing for CYP2D6 and CYP2C19 suggests (2013). improved outcome for antidepressant and antipsychotic medication. Psy- 13. Hall-Flavin, D. K. et al. Using a pharmacogenomic algorithm to guide the chiatry Res. pii: S0165-1781(17)31330-6. https://doi.org/10.1016/j.psychres.2018. treatment of depression. Transl. Psychiatry 2, e172 (2012). 02.055. (2018). [Epub ahead of print]. 14. Pérez, V. et al. Efficacy of prospective pharmacogenetic testing in the treat- 33. Herbild, L., Andersen, S. E., Werge, T., Rasmussen, H. B. & Jürgens, G. Does ment of major depressive disorder: results of a randomized, double-blind pharmacogenetic testing for CYP450 2D6 and 2C19 among patients with clinical trial. BMC Psychiatry 17, 250 (2017). diagnoses within the schizophrenic spectrum reduce treatment costs? Basic 15. Rosenblat, J. D., Lee, Y. & McIntyre, R. S. Does pharmacogenomic testing Clin. Pharm. Toxicol. 113, 266–272 (2013). improve clinical outcomes for major depressive disorder? A systematic review 34. Naranjo, M. E. et al. High frequency of CYP2D6 ultrarapid metabolizers in Spain: of clinical trials and cost-effectiveness studies. J. Clin. Psychiatry 78, 720–729 controversy about their misclassification in worldwide population studies. (2017). Pharm. J. 16, 485–490 (2016).