Precision Medicine into Clinical Practice: A Web-Based Tool Enables Real-Time Pharmacogenetic Assessment of Tailored Treatments in Psychiatric Disorders

The management of neuropsychiatric disorders involves different pharmacological treatments. In order to perform efficacious drug treatments, the metabolism of CYP genes can help to foresee potential drug–drug interactions. The NeuroPGx software is an open-source web-based tool for genotype/diplotype/phenotype interpretation for neuropharmacogenomic purposes. The software provides information about: (i) the genotypes of evaluated SNPs (single nucleotide polymorphisms); (ii) the main diplotypes in CYP genes and corresponding metabolization phenotypes; (iii) the list of neuropsychiatric drugs with recommended dosage adjustment (according to CPIC and DPWG guidelines); (iv) the list of possible (rare) diplotypes and corresponding metabolization phenotypes. The combined application of NeuroPGx software to the OpenArray technology results in an easy, quick, and highly automated device ready to be used in routine clinical practice.


Introduction
The advancement in treatment of diseases has allowed a great improvement in life quality and expectancy, with a demographic transition in the age of the population. Usually, the elderly suffers from chronic disorders (i.e., diabetes, hypertension, hypercholesterolemia and depressive disorder) with frequent comorbidities. Clinicians are aware of many potential drug-drug interactions (DDI) that are among the principal causes of adverse drug reactions (ADR). It is estimated that the DDI risk increases with the number of drugs administrated: from 13% in people with two concomitant drugs, to 38% with four concomitant drugs, till 82% with eight or more concomitant drugs [1,2].
The mechanisms of DDI and ADR involve the pharmacokinetic (PK) and pharmacodynamic (PD) pathways that are influenced by individual genomic differences. Many polymorphic variants have been identified to predict the individual response to drugs. A great number of variants have been found in genes coding for cytochromes responsible for drug metabolism and some ADR have been associated with specific HLA-haplotypes (i.e., HLA-B*57:01 in Abacavir ADR) [3]. To support the correct administration of drugs in clinical practice, several pieces of software have been developed so far (i.e., Epocrates, iFacts, Lexi-interact, Medscape, Micromedex) [4]. The genetic typing of polymorphisms associated with a differential cytochrome activity is a laboratory routine. Selected polymorphisms in CYP3A5, CYP2B6, CYP2C9, CYP2C19, and CYP2D6 genes can be genotyped to define alleles and diplotypes for prediction of enzyme activity. An ideal pharmacogenomic panel should investigate the least number of variants for the identification of the greatest number of alleles. Different technologies have been developed to improve the timing and the costs of this typing. In this scenario, genetic laboratories can perform the genetic typing for pharmacogenomic purposes with several different technologies (i.e., Real Time PCR, OpenArray, next-generation sequencing).
To improve the usefulness of the reported association between polymorphisms and drug effectiveness, several international consortia have published pharmacogenomics (PGx) guidelines. These guidelines reported pharmacologic adjustments recommended in patients with increased/reduced enzyme activity. The most widely used guidelines were developed by the Clinical Pharmacogenetics Implementation Consortium (CPIC) [5] and the Dutch Pharmacogenetics Working Group (DPWG) [6]. These guidelines are based on a systematic review of the available literature. In CPIC and DPWG guidelines, haplotype in CYP genes (coding for cytochromes) associated with a major/minor enzyme activity are reported with a specific suggestion for drug administration. From this perspective, the guidelines were developed as useful instruments supporting clinicians in pharmacologic treatment. Unfortunately, the reported recommendations may be dissimilar in different guidelines [7]. Furthermore, sometimes administration suggestions included in drug labels are discordant with what is reported in the guidelines [8].
The treatment of neuropsychiatric disorders often requires multi-drug administration. Furthermore, the high frequency of reduced compliance in neurological patients led to the need for a rapid PGx panel to improve the optimization times of drug treatment.
The vast majority of genetic analyses are performed through DNA sequencing. This approach can be carried out by sequencing a subset of genes specifically included in a custom panel (target resequencing) or by analyzing the entire exome. The cost reduction at the moment allows the routine use of the exome analysis and the standardization of the analyses, avoiding the development and validation of specific panels. However, the amount of data generated makes the interpretation phase challenging and virtual filters have been proposed that limit the interpretation to the genes of interest only. Alongside the sequencing platforms, a useful approach in many contexts is the genotyping, the genetic analysis limited to single specific polymorphisms. There are many massive genotyping platforms on the market capable of analyzing over a million polymorphisms. These approaches have been used successfully for GWAS, but their use in genetic diagnostics is limited. We believe that the genotyping platform could be an excellent tool in some areas of genetic diagnostics such as pharmacogenetics.
In this paper, we aimed at describing an integrated NeuroPGx system for the rapid evaluation of samples for neuropsychiatric-pharmacogenomic purposes. The automated genotyping (NeuroPGx panel) is followed by software data analysis and drug administration suggestion (NeuroPGx software), in order to provide useful information in about five hours from the sample collection ( Figure 1). OpenArray™ technology is an easy and quick automated system for high-throughput genotyping. In particular, OpenArray is a nanofluidic real-time PCR system that utilizes a stainless steel, microscopic-scale plate with 384 wells, where TaqMan Assays are spotted according to customer specifications. The plate design allows for the genotyping of several samples in a unique experiment. In fact, the system can process one to four QuantStudio 12K Flex OpenArray plates simultaneously, in order to genotype over 1700 samples in a day.
Similar to what happens in genetic sequencing analyzes, we applied the concept of virtual interpretation panels to genotyping.
The TaqMan ® OpenArray ® Pharmacogenomics (PGx) Panel analyzes a large number of polymorphisms, but the interpretation is limited only to those of interest for neuropharmacogenomic purposes to reduce the analysis and reporting times. In particular, out of the 60 polymorphisms in 14 genes, 39 SNPs in 5 genes were selected for their known role in the metabolism of neurological drugs. Table 1 summarizes polymorphic variants selected for neuropharmacogenomic (NeuroPGx panel) OpenArray™ panel.
In order to reduce the time for analysis, a user-friendly software for haplotype definition has been created. The NeuroPGx software is an open-source facility for genotype/diplotype/phenotype interpretation for neuroPGx purposes. It has been developed for the OpenArray™ NeuroPGx panel, but it can elaborate data from each genotyping platform.  OpenArray™ technology is an easy and quick automated system for high-throughput genotyping. In particular, OpenArray is a nanofluidic real-time PCR system that utilizes a stainless steel, microscopic-scale plate with 384 wells, where TaqMan Assays are spotted according to customer specifications. The plate design allows for the genotyping of several samples in a unique experiment. In fact, the system can process one to four QuantStudio 12K Flex OpenArray plates simultaneously, in order to genotype over 1700 samples in a day.
Similar to what happens in genetic sequencing analyzes, we applied the concept of virtual interpretation panels to genotyping.
The TaqMan ® OpenArray ® Pharmacogenomics (PGx) Panel analyzes a large number of polymorphisms, but the interpretation is limited only to those of interest for neuropharmacogenomic purposes to reduce the analysis and reporting times. In particular, out of the 60 polymorphisms in 14 genes, 39 SNPs in 5 genes were selected for their known role in the metabolism of neurological drugs. Table 1 summarizes polymorphic variants selected for neuropharmacogenomic (NeuroPGx panel) OpenArray™ panel.
In order to reduce the time for analysis, a user-friendly software for haplotype definition has been created. The NeuroPGx software is an open-source facility for genotype/diplotype/phenotype interpretation for neuroPGx purposes. It has been developed for the OpenArray™ NeuroPGx panel, but it can elaborate data from each genotyping platform.

Selection of Drugs and CYP Variants
Forty-seven neuropsychiatric drugs were selected for their metabolic association with CYP enzymes. One non-neurological drug (clopidogrel) has been added to the list of selected drugs due to its large use in clinical practice. The selected drugs and CYP genes associated with a different drug metabolism are summarized in Table 2. From all CYP genes, CYP3A5, CYP2B6, CYP2C9, CYP2C19, and CYP2D6 were selected for their large representation in guidelines. Thirty-nine SNPs were selected in order to define haplotypes associated with differential enzyme activity at CYP3A5, CYP2B6, CYP2C9, CYP2C19, and CYP2D6. The selected SNPs and allele frequencies across populations [5] are summarized in Table 3.

DNA Purification and Quantification
Genomic DNA was extracted from 400 µL of peripheral blood using MagPurix Blood DNA Extraction Kit and MagPurix Automatic Extraction System (Resnova, Genzano di Roma, Rome, Italy) according to the manufacturer's instructions. The concentration and quality of the extracted DNA was checked by DeNovix Spectrophotometer (Resnova).

OpenArray™ Technology
The OpenArray™ Technology (Thermo Fisher Scientific, Waltham, MA, USA) is a high-throughput real-time PCR genotyping method that allows a rapid screening of several TaqMan assays in many samples. This real-time method involves the use of an array composed of 3072 through-holes running on QuantStudio 12K Flex Real Time PCR System (Thermo Fisher Scientific) with OpenArray™ block. The OpenArray™ system is composed of a specific plate (OpenArray™ plate) divided into 48 subarrays. The plate customized for this work consists of 60 probes pre-spotted in each of these 48 wells [62].
For each sample, 125-150 ng of extracted DNA and 3µL of 2X TaqMan OpenArray™ Genotyping Master Mix were manually loaded into 384-well plates according to manufacturer's instructions (Thermo Fisher Scientific). Negative control was obtained by adding 3µL of pure distilled water to the Master Mix. The QuantStudio 12K Flex OpenArray™ AccuFill System transfers the previously generated mix to TaqMan OpenArray™ plate. The amplification is performed by QuantStudio 12K Flex Real Time PCR System (Thermo Fisher Scientific) instrument and the results have been analyzed by the Taqman Genotyper Software v1.3 and verified on Genotype app on Thermo Fisher Cloud (Thermo Fisher Scientific). For each SNP, the call rate was calculated through the Genotype app on Thermo Fisher Cloud.

Statistical Analysis
At the end of the analytical assay, the TaqMan Genotyper Software indicates the percentage of successful genotyping, reported as call rate. The call rate is defined as the fraction of SNPs that were assigned a genotype call by the software compared to the total number of SNPs typed. The successful call rate is defined as more than 90%.

NeuroPGx Software Designing
Based on CPIC allele definition, an automated algorithm (NeuroPGx software) was created for the identification of diplotypes compatible with genotypes at selected SNPs. In particular, a user-friendly system was developed the genotype-diplotype definition for each CYP included in the neuroPGx panel. All possible diplotypes were evaluated for their enzymatic phenotype [9] and the frequency in the reference population [5].
The NeuroPGx software takes subjects' genotyping on five core genes (CYP2B6, CYP2C19, CYP2C9, CYP2D6, CYP3A5) as input and calculates all of the possible haplotype combinations, which are more than 5 million. As only a set of diplotypes are assignable to each subject based on their genotype, NeuroPGx outputs a table containing the possible diplotypes and highlights the most probable one based on population frequencies. Moreover, it reports genetic variation-associated drug metabolism profiles (Normal Metabolizer, Poor Metabolizer, Intermediate Metabolizer, Rapid Metabolizer, Ultrarapid Metabolizer). Finally, this information highlights CPIC and DWPG guidelines about drug use for the associated metabolism profiles [5][6][7].

NeuroPGx Software Application
NeuroPGx is written in R programming language (version 4.0.5) and the published interactive app is a dashboard made with Shiny (version 1.6.0) [63,64]. The application can be run locally or on a server. Running NeuroPGx requires installing the latest version of R and Rstudio [64]. NeuroPGx is freely available on GitHub (https://github.com/Andreater/ NeuroPGx, accessed on 20 August 2021) and it's released under the AGPLv3 license.
Accessing GitHub (https://github.com/Andreater/NeuroPGx, accessed on accessed on 20 August 2021) users can download the software and the instruction files, which are reported on the readme file of the repository. Extensive installation instructions are reported both on the readme and in the package homepage on GitHub. Users can find several examples of genotyping tables in the samples folder of the software (NeuroPGx/data/samples). The genotyping table is the input required to run NeuroPGx and it should contain all genotypes at selected SNPs (Table 3) for one or more samples. The input file can be prepared in an Excel spreadsheet with 4 columns: Sample, Gene, rsID, Genotype. Samples' ID should be reported in the Sample column while Gene Symbols and dbSNP IDs should fill the Gene and rsID columns. The Genotype column should be filled with genotype information for each sample and a "/" should be used as a separator. Deletions in a SNP can be coded as -/-or A/-, while more complex configurations, such as CTT/CTT, can be easily reported (Table S1). NeuroPGx also accepts .tsv and .csv files. If the genotyping table lacks some SNP or gene, the NeuroPGx will evaluate possible haplotype combinations based on the available data. When the app starts, it needs the genotyping table to be inputted by clicking on the "browse" button at the top left of the page (Figure 2). The elaboration starts and takes about 5-10 min to finish and show the output ( Figure S1). The software page is divided into 6 panels. The "How it works" panel contains information about how to use the app. The "Samples" panel reports an interactive view of the input table, while the "Assigned diplotypes" panel shows an interactive view of the output table. The "Plots" panel shows plots for both the phenotype and the EHR, divided into two different tabs. The last two panels, namely "Suggested drug w/o interaction" and "Suggested drug with interaction" summarize therapy adjustments based on CPIC and DWPG guidelines, as well as literature-retrieved indications. These suggestions are based on subject metabolic profiles without the interaction between genotypes and metabolites or with their interaction, respectively. be run locally or on a server. Running NeuroPGx requires installing the latest version of R and Rstudio [64]. NeuroPGx is freely available on GitHub (https://github.com/Andreater/NeuroPGx, accessed on 20 August 2021) and it's released under the AGPLv3 license.
Accessing GitHub (https://github.com/Andreater/NeuroPGx, accessed on accessed on 20 August 2021) users can download the software and the instruction files, which are reported on the readme file of the repository. Extensive installation instructions are reported both on the readme and in the package homepage on GitHub. Users can find several examples of genotyping tables in the samples folder of the software (Neu-roPGx/data/samples). The genotyping table is the input required to run NeuroPGx and it should contain all genotypes at selected SNPs (Table 3) for one or more samples. The input file can be prepared in an Excel spreadsheet with 4 columns: Sample, Gene, rsID, Genotype. Samples' ID should be reported in the Sample column while Gene Symbols and dbSNP IDs should fill the Gene and rsID columns. The Genotype column should be filled with genotype information for each sample and a "/" should be used as a separator. Deletions in a SNP can be coded as -/-or A/-, while more complex configurations, such as CTT/CTT, can be easily reported (Table S1). NeuroPGx also accepts .tsv and .csv files. If the genotyping table lacks some SNP or gene, the NeuroPGx will evaluate possible haplotype combinations based on the available data. When the app starts, it needs the genotyping table to be inputted by clicking on the "browse" button at the top left of the page ( Figure 2). The elaboration starts and takes about 5-10 min to finish and show the output ( Figure S1). The software page is divided into 6 panels. The "How it works" panel contains information about how to use the app. The "Samples" panel reports an interactive view of the input table, while the "Assigned diplotypes" panel shows an interactive view of the output table. The "Plots" panel shows plots for both the phenotype and the EHR, divided into two different tabs. The last two panels, namely "Suggested drug w/o interaction" and "Suggested drug with interaction" summarize therapy adjustments based on CPIC and DWPG guidelines, as well as literature-retrieved indications. These suggestions are based on subject metabolic profiles without the interaction between genotypes and metabolites or with their interaction, respectively. .

Results
A NeuroPGx system for the rapid pharmacogenomics evaluation of samples was developed and tested. Genotypes of SNPs involved in neuropsychiatric and neurological drug metabolisms (NeuroPGx Panel) were tested on 100 samples through the OpenArray platform, revealing an average call rate of 94.1%. The Taqman Genotyper Software v1.3 on local computer devices and Genotype app on Thermo Fisher Cloud permit the analyzing of genotyping results. Thermo Fisher Cloud is an easy-to-use platform designed to support analyses and interpretation of instrumental results. In particular, Thermo Fisher Cloud gives figurative report of the OpenArray experiments, improving the interpretation and the quality evaluation time. The average call rate and allelic discrimination calls can be visualized in a condensed image that permits the evaluation of the quality of the experiment.
The development and application of NeuroPGx software permits the automated identification of all possible diplotypes compatible with genotypes at each CYP gene included in the virtual NeuroPGx Panel. Based on population characteristics, the software selects the most likely diplotype-phenotype (Table 3 and Figure 2). Otherwise, all possible diplotype-phenotype combinations are identified and listed in the final report. Therefore, the neuroPGx software report provides information about the: (i) genotypes at evaluated SNPs; (ii) main diplotypes at CYP genes and corresponding metabolization phenotypes; (iii) list of neuropsychiatric drugs with recommended dosage adjustment; (iv) list of possible (rare) diplotypes and corresponding metabolization phenotypes.
The application of NeuroPGx software to genotypic data allows one to obtain complete results for a single sample, for a pool of samples, and for samples with incomplete genotype information (i.e., if one gene and/or one SNP is missing). A complete README and "How to use" section on GitHub (https://github.com/Andreater/NeuroPGx, accessed on 20 August 2021) helps users to install and use the software. The software permits the analysis of samples even if the genotypic data are incomplete; detailed examples of how to build the input file (for one sample, for pool of samples, and if one gene/SNP is missing) are included on GitHub (https://github.com/Andreater/NeuroPGx/tree/main/data/ samples, accessed on 20 August 2021).

Discussion and Conclusions
Here, we report the first NeuroPGx system ready for application in clinical practice. The exact interfacing between genotyping technologies and NeuroPGx software allows one to obtain results in a very short time. In particular, the application of virtual NeuroPGx Panel throughout OpenArray™ System to our new NeuroPGx software enables the full analysis (from sample to report) in about 4 h (Figure 1). It is well known that the time of analysis is often crucial in clinical practice and timely therapy of neuropsychiatric disorders can make all the difference. For this reason, the combination of a high-throughput and rapid technology such as OpenArray and the design of an automated interpretation system can improve the effectiveness of pharmacological therapy.
The NeuroPGx panel involves 39 SNPs for cytochrome enzymatic phenotypes' estimation and prevision of drug response. SNPs were selected from CPIC and DPWG guidelines for pharmacogenomic characterization of patients. These SNPs were tested on 100 samples, revealing an average call rate of 94.1%. The OpenArray technology and the automated interpretation of results through Thermo Fisher Cloud reduce the analysis and interpretation times. In particular, the figurative report of the experiment permits a rapid evaluation of run quality. The evaluation of OpenArray technology confirms that it can easily detect genotypes at multiple loci, allowing one to obtain neuroPGx information in a short time. Compared with classical real-time PCR or with sequencing approaches, the OpenArray technology presents several advantages: it is quite fully automated, it is less time-consuming, and it is easy to analyze. Furthermore, the implementation of data in the Thermo Fisher Cloud facilitates the protection of experimental data from accidental loss. Moreover, the high rate of automated steps in sample processing permits the diffusion of this technology even in laboratories without a high expertise in molecular genetics.
Although the processing optimization requires many samples (almost 16), the cost and time are really attractive. In particular, the NeuroPGx panel applied to OpenArray requires less than 4 h for analysis and about 20 min for interpretation.
The NeuroPGx software was designed to provide an automated interpretation of genotyping data that usually require a long time and are subject to human error. The NeuroPGx software is an open-source platform that provides a complete interpretation of genotyping data according to population distribution of compatible CYP alleles. In order to give an accurate and useful interpretation of genotypes, a full evaluation of administration details was performed on neuropsychiatric drugs ( Table 2). As well as those of authorities in other countries [8], labels approved by the Food and Drug Administration (FDA), European Medicines Agency (EMA), and Italian Medicines Agency (AIFA) sometimes reported administration suggestions that are discordant with guideline recommendations (Table 4). Thus, the evaluation of cytochrome genotypes and metabolizing phenotypes in clinical practice should take into consideration suggestions of dosage adjustments in order to quickly identify the right dosage for the patient. For this reason, the PGx report provided by NeuroPGx software includes not only genetic data, but also administration recommendations according to the main international guidelines (CPIC and DPWG). Considering the well-known differences among suggestions in guidelines [7], the PGx report highlights all possible recommendations, in order to give complete information useful for clinicians ( Figure S2).  To our knowledge, this is the first software that provides complete information from the genotype to the drug administration. Traditionally, interpretation tools help in defining metabolization profiles of selected CYP genes, without any reference to guidelines. The NeuroPGx software is designed to become a useful device in clinical practice; in fact, it is easy to use, it is applicable to data from all genotyping platforms, and it provides updated administration suggestions for all listed drugs.
In conclusion, the combined application of NeuroPGx software to the OpenArray technology results in an easy, quick, and highly automated device ready to be used in clinical practice. In particular, the quickness of this combination will help neurologists and psychiatrists in pharmacological therapy, giving important genetic information in about 4 h. Considering time and costs of the NeuroPGx panel on the OpenArray platform, it is expected that analysis will allow one to obtain useful pharmacogenomic results (with a full genotype-phenotype and drugs report) in less than one working day. Taking into account the high frequency of reduced compliance and the necessity for multi-drug administration in the treatment of neuropsychiatric disorders, a rapid NeuroPGx system will permit the improvement of drug effectiveness. From this perspective, the validation of one-day pharmacogenomic test will promote its diffusion in clinical practice, supporting the precise administration of drugs. Furthermore, the NeuroPGx software by itself can be used to quickly define CYP metabolization profiles and guideline recommendations. Designed to meet the needs of psychiatrists, it will be extended to other pharmaceuticals fulfilling the needs of several specialties.
Funding: This work is also funded by the Italian Ministry of Health (5XMille-2018) and the National Research Council CNR (project "A multifactorial intervention for successful aging").

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Informed consent was obtained from all subjects involved in the study.

Data Availability Statement:
The data generated in the present study are included within the manuscript and its supplementary file.

Conflicts of Interest:
The authors declare no conflict of interest.