METHODS published: 03 November 2021 doi: 10.3389/fpsyt.2021.741170 Methods to Develop an in silico Clinical Trial: Computational Head-to-Head Comparison of Lisdexamfetamine and Methylphenidate Edited by: José Ramón Gutiérrez-Casares 1*† , Javier Quintero 2† , Guillem Jorba 3,4*† , Valentin Junet 3,5 , Michael Noll-Hussong, Vicente Martínez 6 , Tamara Pozo-Rubio 6 , Baldomero Oliva 4 , Xavier Daura 5,7 , Saarland University Hospital, Germany José Manuel Mas 3 and Carmen Montoto 6 Reviewed by: 1 Unidad Ambulatoria de Psiquiatría y Salud Mental de la Infancia, Niñez y Adolescencia, Hospital Perpetuo Socorro, Badajoz, Yoram Vodovotz, Spain, 2 Servicio de Psiquiatría, Hospital Universitario Infanta Leonor, Universidad Complutense, Madrid, Spain, 3 Anaxomics University of Pittsburgh, United States Biotech, Barcelona, Spain, 4 Research Programme on Biomedical Informatics (GRIB), Departament de Ciències Ji-Won Chun, Experimentals i de la Salut, Universitat Pompeu Fabra, Barcelona, Spain, 5 Institute of Biotechnology and Biomedicine, Catholic University of Korea, Universitat Autònoma de Barcelona, Cerdanyola del Vallès, Spain, 6 Medical Department, Takeda Farmacéutica España, South Korea Madrid, Spain, 7 Catalan Institution for Research and Advanced Studies (ICREA), Barcelona, Spain *Correspondence: José Ramón Gutiérrez-Casares jrgutierrezcasares@gmail.com Regulatory agencies encourage computer modeling and simulation to reduce the time Guillem Jorba and cost of clinical trials. Although still not classified in formal guidelines, system guillem.jorba@anaxomics.com biology-based models represent a powerful tool for generating hypotheses with great † These authors have contributed molecular detail. Herein, we have applied a mechanistic head-to-head in silico clinical equally to this work and share first authorship trial (ISCT) between two treatments for attention-deficit/hyperactivity disorder, to wit lisdexamfetamine (LDX) and methylphenidate (MPH). The ISCT was generated through Specialty section: three phases comprising (i) the molecular characterization of drugs and pathologies, (ii) This article was submitted to Computational Psychiatry, the generation of adult and children virtual populations (vPOPs) totaling 2,600 individuals a section of the journal and the creation of physiologically based pharmacokinetic (PBPK) and quantitative Frontiers in Psychiatry systems pharmacology (QSP) models, and (iii) data analysis with artificial intelligence Received: 14 July 2021 methods. The characteristics of our vPOPs were in close agreement with real reference Accepted: 05 October 2021 Published: 03 November 2021 populations extracted from clinical trials, as did our PBPK models with in vivo parameters. Citation: The mechanisms of action of LDX and MPH were obtained from QSP models combining Gutiérrez-Casares JR, Quintero J, PBPK modeling of dosing schemes and systems biology-based modeling technology, Jorba G, Junet V, Martínez V, Pozo-Rubio T, Oliva B, Daura X, i.e., therapeutic performance mapping system. The step-by-step process described Mas JM and Montoto C (2021) here to undertake a head-to-head ISCT would allow obtaining mechanistic conclusions Methods to Develop an in silico that could be extrapolated or used for predictions to a certain extent at the clinical Clinical Trial: Computational Head-to-Head Comparison of level. Altogether, these computational techniques are proven an excellent tool for Lisdexamfetamine and hypothesis-generation and would help reach a personalized medicine. Methylphenidate. Front. Psychiatry 12:741170. Keywords: attention-deficit/hyperactivity disorder, lisdexamfetamine, methylphenidate, mathematical modeling, doi: 10.3389/fpsyt.2021.741170 in silico clinical trial Frontiers in Psychiatry | www.frontiersin.org 1 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate INTRODUCTION 10% (15). Besides, ADHD exhibits an important long-term persistence (15), affecting ∼5% of adults (16–18). Around 30– To reduce clinical trials time and cost and to improve their 50% of children with ADHD continue to manifest symptoms, outcomes’ conclusiveness, regulatory agencies encourage the use inattention in particular, in adulthood (19, 20). Comorbid of computer modeling and simulation (CM&S) approaches to psychiatric disorders are present in up to 67% of ADHD optimize randomized clinical trials (1). CM&S approaches are pediatric-adolescent patients (21) and almost 80% of adults based on the analysis of existing data and experience, including (22). These comorbidities can complicate ADHD diagnosis and real-world data studies, pharmacometrics modeling or, more treatment (23, 24) and include depression, anxiety, bipolar recently, in silico clinical trials (ISCT). Although the concept disorder, binge eating, tics, conduct disorder, personality disorder emerged in the early 2000s (2–4), the term and proper definition and non-alcoholic substance abuse, among others (20, 25). of ISCT was widely established and accepted during the 2010 Recent findings suggest a direct relationship between ADHD decade with the foundation of specific organizations to promote and the development of these comorbidities (24, 26, 27), likely the implementation of these approaches, such as the VPH involving a genetic connection (28), although results on this Institute in 2011 or the Avicenna Alliance, founded by the subject remain controversial. European Commission, to create the research roadmap for ISCT ADHD management comprises pharmacologic and non- (5). In addition to its economic advantages, ISCT allow the pharmacologic treatments. Medications include stimulant exploration of drugs and diseases in many settings, thus, reducing [amphetamines and methylphenidate (MPH)] and non- risks for patients and the use of animal models to test hypotheses. stimulant drugs (atomoxetine, extended-release clonidine, CM&S and artificial intelligence-based approaches are crucial to and guanfacine), with the former being recommended as achieving personalized, preventive, predictive, participative, and first-line treatment (29). Several modifications to improve precise—the so-called 5P—medicine and healthcare (6). the characteristics of amphetamines have been performed, among which the design of the prodrug lisdexamfetamine Systems Biology and MID3 Guidelines (LDX, Vyvanse R in the US and Elvanse R in Europe) and One of the most promising computational tools encompassing the development of extended-release formulations [such as these concepts is systems biology or systems medicine (5, 7–9). the osmotic release oral system (OROS) of MPH, Concerta R During the last 20 years, the US and European medicines agencies or Medikinet R retard]. Although a pediatric clinical trial (FDA and EMA), in collaboration with the pharmaceutical analyzing LDX and MPH is currently ongoing (30, 31), there industry, have been developing the guidelines and good are no explicitly designed head-to-head trials comparing these practices to which these computational approaches should treatments, neither on the pediatric nor adult population. adhere. One of these guidelines is MID3, which describes the We present here the methods of the Therapeutic quantitative framework for predicting and extrapolating models’ Performance Mapping System (TPMS) technology, which conclusions (10, 11). Establishing three categories based on allow the generation of virtual patients and PBPK and systems the relevance of the conclusions, MID3 is meant to guide biology-based models with the purpose of performing ISCTs. industry decision-making (12) or regulatory assessment (13). To demonstrate the applicability of the method, we used Accordingly, models can be classified as (i) “LOW” impact, when as case-study a mechanistic head-to-head ISCT between information obtained from them cannot be directly used to LDX and MPH (Elvanse R vs. Concerta R in the pediatric- make clinical or commercial decisions [e.g., physiologically based adolescent population and Elvanse R vs. Medikinet R retard pharmacokinetic (PBPK)] models; (ii) “MEDIUM” impact, for in the adult population) using a crossover–like design. The models providing helpful information for strategic conditioning objective of this ISCT was to model the efficacies of the of future trial data [e.g., studies to determine optimal dosing, two drugs and compare them in a virtual head-to-head target population, sample size, design of future trials, or setting. Additionally, we describe an approach to measure study of mechanisms of action (MoA) of compounds]; and and compare the output results in terms of efficacy of the (iii) “HIGH” impact, for cases where conclusions support two medications, the molecular mechanisms triggered, and decision-making without the need for additional experimental the response to ADHD management in a diverse population or trial studies (e.g., simulations replacing direct clinical of virtual patients, including patients with the most common trial data in children or oncologic patients that provide psychiatric comorbidities. evidence on efficacy and safety to uphold regulatory submission package and labeling). While pharmacometric models are under evaluation for acceptance as HIGH impact models, systems METHODS biology-based models are still in debate (14). However, they possess an undeniable great potential in providing molecular This methods study details the steps and modeling approaches detail, generating hypotheses, and suggesting specific molecular to carry out the ISCT (Figure 1). Before the study trial solutions to complex pathophysiological problems. (phase I), drugs and pathological conditions were molecularly characterized and reference populations defined. In the Proof-of-Concept: ADHD modeling stage (phase II), a series of virtual populations Attention-deficit/hyperactivity disorder (ADHD) is a complex and PBPK and quantitative systems pharmacology (QSP) ailment with a prevalence in children ranging from 6 to models were generated and embedded in the ISCT as a Frontiers in Psychiatry | www.frontiersin.org 2 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 1 | In silico clinical trial protocol overview. The protocol is divided into three main stages: Phase I, including trial design and information compilation; Phase II, comprising mathematical modeling; and Phase III, consisting of data analysis according to the trial design. ISCT, in silico clinical trial; PBPK, Physiologically based pharmacokinetic; QSP, Quantitative systems biology. means of virtual patient recruitment. At this step, the compared to other clinical trials with equivalent inclusion and models were optimized to reproduce known clinical exclusion criteria. efficacy findings according to the primary outcome of Additionally, standard population distribution data was used the study, i.e., the model-based clinical efficacy-related to fill incomplete demographic parameters. For adults, ESS measure herein proposed, based on modeled protein Round 7 (34) was used, while data from the World Health activity over ADHD molecular definition. Finally, in the Organization (WHO) growth information (35) was retrieved for analysis phase (phase III), the molecular variability among the pediatric-adolescent population. patients was explored by analyzing all ADHD models, patient All virtual patients created had ADHD, and specific branches by patient. for the different comorbidities were also generated, as previously described (36). ADHD and comorbidities definitions were obtained by thorough literature review of current molecular Population Definition—Virtual Patients knowledge on each condition (see Supplementary Methods Two types of virtual populations (vPOPs) were generated: in Supplementary Material 1; Supplementary Tables A, B in adult (>18 years old) and pediatric-adolescent (6–17 years old) Supplementary Material 2). vPOPs. As reference demographic and comorbidity parameters to generate the VPOPs, the following studies were used: Sample Size Calculation NCT00730249 (32) (MPH) and NCT00337285 (33) (LDX) Since data on treated and non-treated patients is not available, for adults; and NCT00763971 study (30) (LDX and MPH) we considered that a number of patients large enough to for the pediatric-adolescent population. These clinical trials discriminate among ADHD patients and healthy individuals presented standard inclusion and exclusion criteria for ADHD would also be large enough to detect efficacy-associated changes evaluation, which were appropriate for the case-study herein for each drug. Therefore, to generate enough patients and ensure proposed and showed homogeneous demographic values when having sufficient statistical power when performing data analyses, Frontiers in Psychiatry | www.frontiersin.org 3 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 2 | Expected percentage of best accuracy as a function of sample size. Dotted blue and discontinuous green lines correspond to the mean % best accuracy reached for each sample size at statistical power 95 and 99%, respectively, assuming a normal distribution of the accuracy variation and estimating the means and the standard deviation for each sample size. The red line shows the 85% Max accuracy level. the sample size approach described below was carried out. repetitions per sample size. Each subset is used to train a linear Because TPMS’ drug efficacy outcomes are based on predicted classifier based on two features extracted by feature selection protein activity (i.e., tSignal, Equation 1—defined in section procedures previously described (36). The accuracy achieved for Systems Biology Maps and Models), this methodology was based each classifier is estimated using k-fold cross-validation (k = 10) on experimental measures that can relate to protein activity (42). Finally, taking as reference the Max accuracy, the percentage variability, particularly gene expression. of max accuracy reached for each subset of samples and total First, gene expression data groups identified as control- samples is calculated using the classifiers obtained for that subset. “healthy” and case- “disease” were retrieved from Gene For the present ADHD study-case, RNAseq records from Expression Omnibus (GEO) experiments (37) and then treated the entry GSE159104 (43) were selected, where two cohorts and normalized using R packages, parameters, and steps defined of patients were already identified and labeled as control by Law et al. (38). Afterwards, a protocol based on the (healthy) and ADHD (case-disease). The variability within the method introduced by Mukherjee et al. (39) and Figueroa et al. genes or proteins involved in the ADHD molecular definition (40) was followed to explore the variation in accuracy and (see Supplementary Methods in Supplementary Material 1; statistical power induced by changes in the sample size. To Supplementary Tables A, B in Supplementary Material 2) was that end, GEO patient-normalized gene expression datasets are evaluated for the 154 samples (78 control, 76 ADHD) included submitted to sampling-without-replacement combined with a in the GEO experiment. After finalizing the abovementioned linear regression classification method (41). The latter allows the procedure, statistical powers of 95 and 99% were used, based on identification of the best classifiers (proteins) to separate control- classification errors (39), and a value of 85% of Max accuracy healthy from case-disease patients, and these classifiers are used was set as minimum valid threshold (Figure 2). Considering a to compute the highest possible accuracy (“Max accuracy”). statistical power of 95%, we deemed 68 samples (34 control and Progressive sampling is then applied to obtain subsets of balanced 34 ADHD) to be enough to achieve the objectives of the analysis samples from both cohorts (case-disease vs. control) in a 1:1 in our simulation. Under these premises, 142 samples (71 control ratio. These subsets are tested for sample sizes ranging from and 71 ADHD) were adequate to reach the target accuracy with eight to the number of the smallest cohort performing 100 99% power (although more RNAseq samples would be required Frontiers in Psychiatry | www.frontiersin.org 4 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 3 | Comorbidities distribution and treatment allocation in the adult virtual population. ADHD, Attention-deficit/hyperactivity disorder; LDX, FIGURE 4 | Comorbidities distribution and treatment allocation in the Lisdexamfetamine; MPH, Methylphenidate; QSP, Quantitative systems biology; pediatric-adolescent virtual population. ADHD, Attention-deficit/hyperactivity vPOP, Virtual population. disorder; LDX, Lisdexamfetamine; MPH, Methylphenidate; QSP, Quantitative systems biology; vPOP, Virtual population. to ensure curve stabilization). Accordingly, at least 100 virtual containing the models of each virtual patient could be cloned; patients were built per each patient group (minimum sample size thus, no wash-out period was needed. of 200 samples per analysis). Intervention Definition According to their population group, the patients included in Patient Distribution the ISCT were treated with different formulations and doses of The two populations, adult and pediatric-adolescent, were LDX and MPH in a two-period crossover-like study design. For segmented into nine arms each (a total of 18 arms) to facilitate all patients, the same initial state was used at each period, hence the simulation and the analysis. One arm accounted for ADHD carryover effect was assumed zero. Dosage, molecular target without any comorbidity, while the eight additional arms profile, and pharmacokinetic information were needed for the contained patients with ADHD and one, or a combination, of QSP modeling herein proposed. comorbid psychiatric conditions. Each of the arms accounting for comorbidities had 100 Dosage patients, while arms related to ADHD alone consisted of 500 Dosage schemes were simulated differently in the pediatric- patients, with the aim of maximizing the number of patients with adolescent and adult populations according to usual clinical different demographical characteristics. Consequently, a total of practices. Adults were treated with LDX (Elvanse R ) 70 mg and 2,600 patients were included in the simulation: 1,300 adults children with LDX 50 mg. Different doses and types of modified (Figure 3) and 1,300 children-adolescents (Figure 4). All of them release systems for MPH were considered in the simulation, were treated sequentially with LDX and MPH using the adequate corresponding to different commercial formulations: (i) for dosing scheme. According to the study’s in silico nature, the files adults, Medikinet R 60 mg with modified-release (also known as Frontiers in Psychiatry | www.frontiersin.org 5 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate TABLE 1 | Identified protein targets for lisdexamfetamine and methylphenidate. Gene name Protein name Effect* Reference of LDX Reference of MPH target target TAAR1 Trace amine-associated receptor 1 1 (49) – SLC18A2 Synaptic vesicular amine transporter (VMAT2) −1 (49, 50) – SLC6A3 Sodium-dependent dopamine transporter (DAT) −1 (50–52) (53–55) SLC6A2 Sodium-dependent noradrenaline transporter (NET) −1 (50, 52) (53–55) SLC6A4 Sodium-dependent serotonin transporter (SERT) −1 (50) – MAOA Amine oxidase (flavin-containing) A −1 (52, 56) – MAOB Amine oxidase (flavin-containing) B −1 (52, 56) – HTR1A 5-hydroxytryptamine receptor 1A 1 – (57, 58) *Effect refers to the drug’s action on the protein, 1 denotes activation of protein function, −1 inhibition of protein function. LDX, lisdexamfetamine; MPH, methylphenidate. Medikinet R XL or Medikinet R Retard), based in multiarticular TABLE 2 | Summary of pharmacometrics information used for PBPK modeling. beads that combine 50% immediate and 50% extended-release (44); and (ii) for the pediatric-adolescent population, Concerta R Drug % Bioavailability Main clearance 36 mg, an osmotic release system (OROS technology) with a (Ref.) organ 22% of the total amount available for immediate release (the Elvanse® 96.4 (59) Kidney (60) remaining 78% corresponding to the osmotically controlled Medikinet® with modified release 30 (61, 62) Kidney (63) extended-release) (45). Concerta® 32 (64) Kidney (63, 65) Molecular Target Profile The molecular target profile identification was performed through a review of official regulatory sources [European drug concentration-related knowledge. Finally, the patient- Medicines Agency—EMA, European Public Assessment specific drug concentration and disease-related data, Report (EPAR)—and Food and Drug Administration—FDA, and protein mapping according to pathophysiological Multidisciplinary and Chemistry reviews and Label], drug- information, are used for generating patient-specific target–dedicated databases [DrugBank (46), STITCH (47), MoA-QSP models of the drugs under study, here MPH SuperTarget (48)] and the scientific literature (the specific and LDX. searches performed can be found in Supplementary Methods in Supplementary Material 1). This information was integrated Virtual Population Modeling into the TPMS technology-based MoA models for each drug. For the construction or recruitment of vPOPs, randomized Table 1 contains the proteins defining the target profile of LDX populational demographic characteristics are generated using and MPH. two types of data sources: (i) original or reference population with demographic characteristics to be mimicked [age, weight, Pharmacokinetics Information height, and/or body mass index (BMI)]; and (ii) standard Bioavailability and drug’s information on main clearance population distributions, retrieved from populational studies. organ were retrieved from published studies and set for the For the present ADHD study-case, the recruitment of each vPOP corresponding PBPK models (Table 2). Moreover, previous was based on the demographical parametric descriptors defined PK studies were used to fit the generated PBPK models, to in section Population Definition—Virtual Patients’ [reference parameterize absorption and drugs’ clearance ratios, and to clinical trials (30, 32, 33), European standard population (34), validate the models. The reference studies used were Krishnan and WHO growth information (35)]. and Zhang (66) for LDX in adults, Boellner et al. (67) for For adult population, an adapted version of the algorithm LDX in children, the EPAR (68) for Medikinet R with modified- proposed by Allen et al. (70) was used to generate the population release, and Maldonado (69) for Concerta R . All three drugs were of individuals virtually recruited in the trial. As a first step, this administered orally and crossed the blood-brain barrier. algorithm generates a multivariate normal distribution (MVND) with the demographic means and standard deviations from Modeling Methodology the original population. The standard population distribution TPMS ISCT is divided into three types of modeling approaches values are used to fill in the potential missing demographic (Figure 1). First, virtual patients are generated containing information. A simulated annealing strategy is then used to demographic information and disease tags. Afterwards, PBPK minimize a cost function by using the patients generated in models are constructed using each patient’s demographic the MVND as starting points (see Supplementary Methods in variables, which are then used to infer inter-patient specific Supplementary Material 1). Frontiers in Psychiatry | www.frontiersin.org 6 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate In the pediatric-adolescent population, a modification of in different organs within the human body. The PBPK the protocol used for adult population was applied to adjust model structure used consists of 14 predefined compartments better the dependence of morphometric measures for ages 0– representing the human body’s main organs and tissues, a 17 years. First, the standard population distribution, taken from simplified version of a previously reported model (72) (Figure 5). the growth information published by the WHO (35), was used Blood acts as the central compartment by interconnecting the rest to create a reference MVND. Then, a sampling strategy based of the system through blood flows, and the whole system can be on a Metropolis-Hastings method (71) was applied to reach the disturbed by administering a drug dose in any of the following original population distributions (see Supplementary Methods in organs or compartments: gut (oral drugs), blood (intravenous Supplementary Material 1). drugs), or skin (subcutaneous drugs). Similarly, clearance of The final distribution values for adult and pediatric-adolescent drugs and compounds is restricted to three compartments: gut, populations were statistically compared (one sample z-test) to liver, and kidneys. The equations associated with blood flow rates the original means and standard deviations; only populations not and organ/tissue volumes are taken from Brochot and Quindroit significantly different from the original population (p-value > (73). These variables depend on cardiac frequency, age, BMI, and 0.05) were accepted and kept for posterior modeling steps. For gender and yield individualized models as described elsewhere both population types, corresponding comorbidity-related tags (74). Here, blood volume was readjusted to fit the volume of were assigned to the patients allocated to each of the 18 ISCT distribution of each compound for optimized modeling. arms (Figure 4). Parameters related to the anatomy and physiology of each Demographic parameters were used to obtain accurate and specific patient’s human body were used to mathematically individualized PBPK models of the drugs, while comorbidity describe drugs’ internal flow, i.e., drugs’ absorption, distribution, data, once translated into molecular information, influenced the metabolism, and excretion (ADME) processes (75). The drug’s patients’ corresponding QSP models. absorption and clearance constant parameters were calculated by fitting the general model to existing real pharmacokinetics Systems Biology Maps and Models data points for d-amphetamine (d-Amph, active compound TPMS technology (36) generates mathematical models for LDX) and MPH (Table 2) (66–69). For other ADHD drugs that use known biological, medical, and pharmacological (see Supplementary Methods in Supplementary Material 1), information as training data (see Supplementary Table C in pharmacokinetics data used can be found in Supplementary Supplementary Material 2) to simulate the behavior of drugs Table D in Supplementary Material 2. Regarding MPH, and the pathophysiology of diseases in terms of changes in as extended-release capsules are not easily simulated, an protein activity. This methodology uses supervised machine approximation using repeated administration of lower doses was learning methods based on a human protein functional used. This strategy had already been described for the two MPH network to infer information at the clinical and protein extended-release formulations used here, and resulted in similar levels. Here, TPMS was used to build the mathematical concentration dynamics: (i) for Medikinet R with modified- models to simulate the behavior of LDX and MPH over release, considered to have an equivalent MPH bioavailability to ADHD by modeling the changes in proteins’ activity defining Ritalin R (56), a twice-a-day administration was simulated with the disease. While generating TPMS models, molecular half the dosage for each simulated administration, and (ii) for information relating to psychiatric comorbidities was Concerta R , three administrations were simulated with one-third added to denote the different neurophysiological ADHD of the original dose for each administration (76). patient types. The whole PBPK compartments model is implemented in The resulting models allowed the extraction of several protein MATLABTM , and differential equations describing the kinetics of activity measures. Therefore, the model-derived parameter the compounds and the fitting procedures are integrated by using tSignal (Equation 1) (36), which ranges between 1 and −1, SimBiology Toolkit (77). applied to the molecular definition of clinical conditions (in this case, ADHD molecular definition, as detailed in Supplementary Quantitative Systems Pharmacology Table B in Supplementary Material 2) permitted access to Models—Quantitative Mechanism of Action clinically relevant information at a model-patient level. A QSP model enclosing PBPK model outputs and TPMS model maps was generated for each patient of the vPOPs. 1 Xn tSignal = − vi yi (1) QSP models are generated following the TPMS methodology n i=1 previously described (36) but incorporating drug concentration Where n is the number of proteins defining the protein set; data at different timepoints in addition to molecular inputs, vi are the protein signs (active or inactive) according to each which add patient-specific quantitative data. To this end, a disease/comorbidity definitions; and yi are the resulting modeled set of drug concentration timepoints in the target tissue— signal values achieved by each protein “i” after stimulating the brain in this study—can be associated with the modulation of model with the corresponding drug. the drug’s target proteins. Additionally, by applying the EC50 equation definition and using clinical efficacy observations, the Physiologically Based Pharmacokinetic Models drug’s effect on the disease-characterized proteins in the target A PBPK model per virtual patient was built to describe tissue can also be calculated (see Supplementary Methods in the relationship between drug doses and drug concentration Supplementary Material 1). Accordingly, the resulting MPH Frontiers in Psychiatry | www.frontiersin.org 7 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 5 | Schematic representation of the multi-compartment model for physiologically based pharmacokinetic modeling. and LDX drug’s target modulation-efficacy relationships were As previously described (36), after modeling a drug MoA on each used as extra parameters in the TPMS training set, resulting in patient, a protein activity value in the range (−1, 1) was obtained. the final QSP models. The latter had the same output format as These values can be either analyzed individually or combined in the systems biology MoA models previously described (36), but protein functional groups to evaluate biological concepts, such as included quantitative information related to drug concentration. diseases or comorbidities. Hence, these models were used to answer additional questions related to individual differences among patients or treatment Efficacy Outcome comparisons. At least 50 mathematical solutions per patient were As for any clinical trial, in which the primary outcome is usually computed during the QSP modeling to account for intra-patient related to the drug’s efficacy, our primary case-study goal was variability, with accuracies >85% with respect to the TPMS to identify and compare both drug’s efficacies. Accordingly, training set (36). a selection and conversion methodology were defined to select the protein set within the ADHD definition that best explained a chosen efficacy metric, and we transformed the Efficacy Outcomes and Measures protein activities of that set into a model-derived measure that Definition and Optimization correlated with an actual clinical measure. The clinical variable Molecular Measures used here was the ADHD Rating Scale IV (ADHD-RS IV, Due to the systems-biology–based nature of the virtual patients’ change from baseline). Three steps were followed to convert resulting models, all measures were centered on protein activity. TPMS-model protein activities into ADHD-RS IV values: (i) Frontiers in Psychiatry | www.frontiersin.org 8 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate select a model-derived activity measure (i.e., tSignal) that could of response model tSignal for each Xi parameter variation can be be used as a proxy for efficacy; (ii) carry out ADHD molecular expressed as: characterization, which consisted on a curated review of the scientific literature available in the PubMed database to identify dTPMS d(tSignal) = (3) proteins functionally involved in ADHD (see Supplementary d (Xi ) d (Xi ) Methods in Supplementary Material 1; and Supplementary The tSignal difference compared to the original model was Tables A, B in Supplementary Material 2; and (iii) optimize by computed for all values in the range tested, and the mean for trimming the ADHD molecular definition using real clinical trial each protein was calculated and evaluated as a percentage with efficacy observations (using ADHD-RS IV). In the third step, respect to the maximal possible tSignal variation, set as 2 [(−1, 1) a series of eligible ADHD clinical trials meeting our inclusion difference] minus the original tSignal. criteria and measuring ADHD-RS IV in relevant drugs (Table 3; An unsupervised clustering strategy was applied to obtain Supplementary Methods in Supplementary Material 1) groups of two to seven clusters of MoAs to evaluate the were compiled. The reported ADHD-RS IV values were molecular variability of the generated models. The two (adults then used for ADHD molecular definition refinement and children-adolescents) complete sets of 1,000 QSP ADHD through Pearson’s correlation (Supplementary Methods in patient mechanistic models (500 for LDX and 500 for MPH) Supplementary Material 1); the final ADHD definitions used were evaluated separately, taking into account the final activation for outcome measurement are displayed in Supplementary Table values of the ADHD protein effectors modulated by both E in Supplementary Material 2. drugs. Clusters were obtained using K means algorithm (93). Model-derived ADHD outcome measures were optimized The clustering analysis was performed using all features separately for adults’ and pediatric-adolescent’s clinical trials to (effector proteins) and principal component analysis (PCA) reduce noise on the molecular definition. dimensionality reduction with five dimensions (94). Four quality Data Analysis indicators were used to select the optimal number of clusters: For the analysis of the population demographic and PBPK Hopkins statistics (95) to measure the cluster tendency of a data parameters, descriptive statistics were used (mean and standard set; Silhouette index (96) to weigh the cohesion of the clusters and deviation, frequency tables, or pie charts), and appropriate Jaccard Bootstrap Index (97) to gauge the similarity and diversity parametric and non-parametric tests applied. The p-value was of sample sets. Clusters were also filtered by heavily unbalanced taken as a measure of the significance of the fitting to the groups, according to the Silhouette index ratio (96). Classification reference population. analysis, as described elsewhere (36), were applied to molecularly The data was analyzed employing MATLABTM functions describe the identified clusters. and Python or R packages to compare means and/or standard Ethics deviation between data distributions. Analyses with <30 samples Only aggregated patient data from published clinical trials were were treated with non-parametric tests, while comparisons used in the current project (30, 32, 33, 78–90, 98). Aggregated involving more than 30 samples were performed assuming a patient data prevents individual patients’ identification and, thus, normal distribution and treated with parametric tests; in all avoids the need for approval from an ethics committee or cases, the applied test was reported. The statistical significance institutional review board. level was set at p < 0.05. False discovery rate (FDR) was used to control type I errors by applying the Benjamini- Computational Availability Hochberg (91) multi-test correction method, whenever relevant. All simulations described in this project were executed in All analyses were performed according to the described the Anaxomics’ cloud computing, which integrates more than analytical strategy. 800 computational threads in machines with 64 Gigabytes The accuracies of systems biology and QSP models were of RAM. Software, databases, and tools are the property of calculated for each solution within each individual model Anaxomics Biotech. and expressed as the percentage of compliance of all drug- pathophysiology relationships included in the training set (36). RESULTS To evaluate the sensitivity of systems biology models, a local sensitivity analysis based in the SOBOL methodology (92) was Demographic Characteristics performed to explore whether the variation in the protein activity The characteristics of adult and pediatric-adolescent vPOPs (−1, 1) of the proteins in the models influenced the MoA models of ADHD patients were generated from the proportions of response of the two drugs (ADHD, as defined in Supplementary demographic characteristics reported in the corresponding Tables A, B in Supplementary Material 2). According to the clinical trials. Additionally, eight subpopulations with different SOBOL terminology, TPMS models could be redefined as: comorbidities (depression, anxiety, bipolar disorder, tics, and binge eating disorder) were created using the same method for tSignal(ADHD) = TPMS (X) for X = {X1 , X2 , X3 , . . . , Xn } (2) both populations to evaluate the impact of comorbidities on the drugs’ efficacy. The characteristics of our modeled vPOPs can be Where Xi corresponds to each one of the parameters (here found in Figure 6A for adults and Figure 6B for the pediatric- protein nodes activity) used in the models. Then, the variation adolescent population. The characteristics of the adult vPOP Frontiers in Psychiatry | www.frontiersin.org 9 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate TABLE 3 | List of clinical trials used for attention-deficit/hyperactivity disorder model-derived efficacy measure optimization. Clinical trial Title References number/PMID Adult clinical trials PMID: 17137560 Efficacy and safety of dexmethylphenidate extended-release capsules in adults with attention-deficit/hyperactivity disorder (78) PMID: 20576091 Randomized, double-blind, placebo-controlled, crossover study of the efficacy and safety of lisdexamfetamine dimesylate in (79) adults with attention-deficit/hyperactivity disorder: novel findings using a simulated adult workplace environment design NCT00337285 A long-term, open-label, and single-arm study of NRP104 30, 50, or 70 mg per day in adults with attention deficit hyperactivity (33) disorder (ADHD) NCT01270555 Efficacy of bupropion SR for attention deficit hyperactivity disorder (ADHD) in adults with recent past or current substance use (80) disorders NCT01259492 A 40-week, randomized, double-blind, placebo-controlled, multicenter efficacy and safety study of methylphenidate HCl (81) extended release in the treatment of adult patients with childhood-onset ADHD NCT02141113 Double-blind, randomized, placebo-controlled, single- center, dose optimization study evaluating efficacy and safety of (82) guanfacine hydrochloride in combination with Psychostimulants in adults aged 18–65 years with a diagnosis of ADHD NCT02604407 A phase 3, randomized, double-blind, multicenter, placebo-controlled, forced-dose titration, safety and efficacy study of SHP465 (83) in adults aged 18–55 years with attention-deficit/hyperactivity disorder (ADHD) Pediatric-adolescent clinical trials NCT00507065 A phase III, randomized, multicenter, double-blind, parallel-group, placebo-controlled safety and efficacy study of ADDERALL XR (84) with an open label extension, in the treatment of adolescents aged 13–17 with ADHD PMID: 17577466 Efficacy and tolerability of lisdexamfetamine dimesylate (NRP-104) in children with attention-deficit/hyperactivity disorder: a phase (85) III, multicenter, randomized, double-blind, forced-dose, parallel-group study NCT00447278 A study comparing the effect of atomoxetine vs. other standard care therapy on the long term functioning in (86) attention-deficit/hyperactivity disorder (ADHD) children and adolescents (ADHD LIFE) NCT00393042 Sleep and tolerability of extended release dexmethylphenidate vs. mixed amphetamine salts: a double blind, placebo controlled (87) study (SAT STUDY) PMID: 21241954 Clonidine extended-release tablets for pediatric patients with attention-deficit/hyperactivity disorder (88) NCT00763971 A phase III, randomized, double-blind, multicentre, parallel-group, placebo- and active-controlled, dose-optimization safety and (30) efficacy study of lisdexamfetamine dimesylate (LDX) in children and adolescents aged 6–17 with attention-deficit/hyperactivity disorder (ADHD) NCT01244490 A phase 3, randomized, double-blind, multicentre, parallel-group, placebo- and active-reference, dose-optimization efficacy and (89) safety study of extended-release guanfacine hydrochloride in children and adolescents aged 6–17 years with attention-deficit/hyperactivity disorder NCT01328756 A phase 4, open-label, multicentre, safety study of lisdexamfetamine dimesylate in children and adolescents with (90) attention-deficit/hyperactivity disorder (ADHD) showed no significant differences with real reference populations Physiologically Based Pharmacokinetic extracted from clinical trials (Table 4). The same was true for the Models pediatric-adolescent vPOP (Table 5). PBPK models simulating the available drug concentration in blood over time were obtained for LDX and MPH and for the two studied populations. Drug concentration models were fitted Local Sensitivity Analysis of Systems to real data resulting in similar blood drug concentration levels Biology-Based Models for a standard adult (male, 40 years old, 175 cm, 70 kg) and TPMS-derived MoA models were subjected to sensitivity child (male, 8 years old, 30 kg, 130 cm) (Figure 7). PBPK model SOBOL analysis to evaluate whether the variation of molecular simulations complied with the observed in vivo curves, even for parameters would affect the models’ response and to identify the case of MPH in children and adults, where approximating key molecules. The sensitivity evaluation was carried out for repeated administration of lower doses was required to model the a range of values (−1, 1) for each protein. Although these modified-release formulations. models have about 5000 parameters, less than a third of them showed a real impact (difference >15%) on the output, Efficacy Outcomes and Measures which was less notorious in MPH (max difference ∼17%) Definition and Optimization than in LDX (max difference ∼32%) (Supplementary Table F After the process of optimizing by trimming, 83 proteins in Supplementary Material 2). Interestingly, from the 30 most (out of 86) were included in the pediatric-adolescent ADHD sensitive proteins, some were shared between both mechanisms definition (ρ = −0.81) and 66 proteins (out of 86) in the (namely, NFKB1, PRKCA, PRKCZ, TRAF6, and PRKCB). adult ADHD definition (ρ = −0.79). The resulting molecular Frontiers in Psychiatry | www.frontiersin.org 10 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 6 | Demographic characteristics (sex, age, BMI, height, and weight) of (A) the adult virtual population (N = 500) and (B) the pediatric-adolescent virtual population (N = 500). definitions found after optimizing the model-derived efficacy Quantitative Systems Pharmacology measures for each conditions’ clinical efficacy can be found Models in the Virtual Populations in Supplementary Table G in Supplementary Material 2. The The MoA of LDX and MPH in our populations of interest, subsequent regression lines, as well as the different study inferred from QSP models, were obtained by combining PBPK points used, are represented in Supplementary Figures A, B in models of the dosing schemes of these drugs and TPMS Supplementary Material 1. technology, which modeled the MoA of both drugs in ADHD. Frontiers in Psychiatry | www.frontiersin.org 11 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate TABLE 4 | Demographic characteristics of the adult virtual population and the Clusters were represented using the two main components reference population. of PCA (Figure 8), which explained 66.7 and 12.4% of the Virtual population Reference population p-valuea observed variability in adults, and 61.4 and 19.4% in children- (N = 500) (N = 511) adolescents, respectively. The five most relevant proteins in the PC1 (eigenvector 1) of each population were – IL4, AKT3, Sex (% females) 41.6 41.6 NA NTRK2, IL5, and NTF3 for adults and – CRY1, AKT3, CRY2, Age (years) 36.98 ± 10.31 36.58 ± 10.10 0.37 AKT1, and AKT2 for children-adolescents. Height (cm) 172.03 ± 9.98 171.7 ± 9.4 0.43 We also found that clustering was associated to differences in Weight (kg) 77.74 ± 16.95 78.75 ± 17.20 0.19 treatment efficacy in adults (ANOVA p-value = 2.515e-08) and BMI (kg/m2 ) 26.51 ± 6.55 NAb NAb children-adolescents (ANOVA p-value = 1.194e-09). In adults, cluster 4 showed the highest mean tSignal (p-value = 2.263e- Figures are mean ± standard deviation unless otherwise stated. BMI, Body mass index; NA, not applicable. 09), while cluster 2 was the one presenting the lowest (p-value a Calculated with the unpaired two-tailed T Student’s test. = 6.835e-04). In children-adolescents, the tSignal of cluster b For demographic data not provided in the reference clinical trial, European mean values 1 was significantly higher (p-value = 2.752e-05) and that of were used. cluster 3 was significantly lower (p-value = 7.397e-04) than the rest (Student’s T-test). To further characterize the clusters, we TABLE 5 | Demographic characteristics of the pediatric-adolescent virtual performed an ANOVA analysis to identify potential differences population and the reference population. on the demographic characteristics within the clusters. In the overall analysis, only weight was significantly different (p-value Virtual population Reference population p-valuea < 0.05) in adults (ANOVA p-value = 0.019), and no differences (N = 500) (N = 111) were found in children-adolescents. When comparing each Sex (% females) 20.0 20.0 NA cluster against the rest in adults, we found that BMI, weight, and Age (years) 11.11 ± 2.73 10.90 ± 2.80 0.09 gender ratio were significantly lower in cluster 4, while weight Height (cm) 147 ± 15.92 NAb NAb was slightly higher in cluster 2. In children- adolescents, BMI Weight (kg) 42.41 ± 12.70 43.60 ± 15.10 0.08 and weight were significantly lower in cluster 1, while weight was BMI (kg/m2 ) 19.1 ± 2.7 19.1 ± 3.4 0.81 slightly higher in cluster 3 (Table 7). Figures are mean ± standard deviation unless otherwise stated. BMI, Body mass index; NA, not applicable. a Calculated with the unpaired two-tailed T Student’s test. DISCUSSION b For demographic data not provided in the reference clinical trial, European mean values were used. Herein, the technology to create populations of virtual patients and the subsequent ISCT is described in the case-study of LDX and MPH head-to-head comparison in the context of ADHD The simulation analyzed the whole available data on pathologies, treatment. Adult and pediatric-adolescent vPOPs were obtained, drugs, and the population. The mean accuracy values obtained and PBPK and QSP models were generated successfully to in mechanistic models for ADHD virtual patients were: 91.63% provide the basis for identifying mechanistic differences between (adults treated with LDX), 91.71% (adults treated with MPH), the two drugs, patient cohort differences, inter- and intra-patient 91.68% (children-adolescents treated with LDX), and 91.69% response variability. (children-adolescents treated with MPH). Thus, for each patient, Preliminary evaluation of the models revealed some insights activation/inhibition patterns of all proteins associated with the on the factors affecting MoA-related treatment efficacy. The MoA of LDX and MPH were obtained. Drugs’ efficacy on ADHD sensitivity analysis of systems biology MoA models provided a measured over each virtual patient was exclusively estimated list of common proteins that might affect both drugs’ efficacy: using the above mentioned tSignal formula (Equation 1), which proteins involved in the NF-κB signaling pathway (NFKB1 summed up the activity values of ADHD effector proteins. The and TRAF6) and PKC (alpha, beta, and zeta types). The tSignal formula was applied to the list of ADHD effector proteins pleiotropic nature of these proteins and their involvement in optimized to fit clinical observations and provided high accuracy several signaling processes could explain their potential impact QSP models for the whole set of 1,300 patients comprising adults in the sensitivity of mechanistic models. However, a detailed and children. evaluation of each drug mechanistic model should provide The ADHD population was subjected to clustering analysis further knowledge on the key proteins involved. to explore molecular variability within the LDX and MPH QSP model clustering analysis indicated the presence of mechanistic models. The optimal number of clusters for adults several response patterns, not clearly defined by drug treatment. was four different clusters, whereas three main clusters were The protein activity-based unsupervised clustering was somehow identified for children, according to Hopkins statistics (0.82 and associated to response level, and PCA analysis revealed some 0.89, respectively), Silhouette index (0.31 and 0.33, respectively), relevant proteins that could be exerting this effect, including: and Jaccard Bootstrap index (0.52 and 0.57, respectively). These dopamine signaling-related AKT proteins (99), neurotrophins results reflected drug-independent patient intrinsic variability related to neural viability and dopamine regulation in ADHD since they clustered in a non–drug-dependent manner (Table 6). (100–103), circadian rhythm proteins related to ADHD and Frontiers in Psychiatry | www.frontiersin.org 12 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 7 | Blood d-Amph and MPH concentration comparison between real datapoints and the curve resulting from the PBPK model. (A) Generated for a standard adult patient after a single 70 mg dose of LDX, real datapoints obtained from Krishnan et al. (66); (B) generated for a standard adult patient after two 10 mg doses of MPH every 4 h, real datapoints obtained from BfArM (68); (C) generated for a standard pediatric patient after a single 50 mg dose of LDX, real datapoints obtained from Boellner et al. (67); and (D) model generated for a standard pediatric patient after three 5 mg doses of MPH every 4 h, real datapoints obtained from Maldonado et al. (69). d-Amph, d-Amphetamine; LDX, Lisdexamfetamine; MPH, Methylphenidate; PBPK, Physiologically based pharmacokinetic. comorbidities–associated sleep disturbances (104–106), and evaluated pharmacokinetic parameters in a short time. However, cytokines related to neuroinflammation and Th2 response and they were limited in the number of virtual patients that could be ADHD (107–110). Regarding clinical characteristics, a possible generated accurately. Therefore, considering that ADHD is not as correlation was found between lower weight, female gender, varying in brief periods and that such pharmacokinetic detail was and lower BMI, with a higher tSignal or better efficacy in unnecessary, a higher number of patients could be generated in adult population. Similar results were found in the children- our study. On the other hand, a multi-compartment model with adolescent population, were higher tSignals were found in the a large virtual population size has been published on trauma- group with lower weight and BMI. In this sense, previous reports induced critical illness that showed how the molecular and had already suggested a relationship between drug efficacy and cellular events taken as a whole could manifest heterogeneously BMI (111). on individuals (116). These results were in agreement with ours, which showed different clusters of patients that could correspond Related Work to different response profiles to a certain point, independent from Virtual populations have been generated in the past to assist in drug treatment. solving complex medical issues. The FDA has accepted a type Virtual populations combined with PBPK modeling have been 1 diabetes simulator to replace animal testing in pre-clinical used successfully to predict the pharmacokinetic profile of a trials (112). Besides, in silico cloning of data from individual drug and evaluate potential drug-drug interactions for a specific type 1 diabetes patients to improve algorithms for closed-loop ethnicity (117). In addition, a PBPK model combined with insulin delivery systems has been reported in 12 and 47 virtual systems-biology techniques has been reported and validated as an patients in studies that aimed to tackle the challenging problem efficient tool for assessing risk exposure to certain volatile organic of inter- and intra-subject variability (113, 114). Likewise, a compounds (118). Furthermore, multi-compartment QSP has virtual population of 50 individuals has been generated to test in been used to model immunotherapies in breast cancer (119). silico drug cardiotoxicity and account for inter-subject variability When associated with pharmacokinetics and pharmacodynamics in clinical studies with toxicological endpoints (115). Such data, it has been reported in an in silico virtual clinical trial to approaches were warranted considering the high variability of the analyze predictive biomarkers in certain breast cancers (120). Frontiers in Psychiatry | www.frontiersin.org 13 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate TABLE 6 | Distribution of LDX and MPH mechanistic models in the generated allowing the exploration of the complete clinical and molecular clusters. landscape of each patient. Furthermore, our ISCT design had a Drug Cluster 1 Cluster 2 Cluster 3 Cluster 4 large enough sample size and considered pools of mathematical (Red) (Green) (Blue) (Purple) solutions for each patient—instead of a single mathematical model per patient—which ensured that the simulations were Adult models clustering robust and appropriate for data analysis. LDX 122 123 150 105 However, our study presented some limitations. Firstly, MPH 98 154 106 142 our models depended on the current knowledge of human physiology, particularly on the drugs and disease under Drug Cluster 1 Cluster 2 Cluster 3 (Red) (Green) (Blue) investigation, as well as protein interactions and pathways described and involved in the MoA. Therefore, our models Pediatric-adolescent models clustering could have been susceptible to missing data, errors, and bias, LDX 230 179 91 and some aspects could have been overlooked. For instance, MPH 215 118 167 unknown targets or yet undescribed pathophysiological ADHD processes might play a role in the MoA of the evaluated LDX, Lisdexamfetamine; MPH, Methylphenidate. drugs. ADHD and its associated comorbid psychiatric disorders present a high genetic and signaling overlap (126, 127), which could act as confounding factors at the clinical and molecular Hence, PBPK and QSP models have been established as powerful levels. Accordingly, the molecular characterizations used for computational tools for in silico simulations. modeling could be biased; prospective data could expand the Finally, only a few in silico head-to-head trials have been knowledge on these diseases and, therefore, improve our model- published. A recent study compared two insulin therapies for derived conclusions. type 1 diabetes treatment by using the abovementioned FDA- Secondly, our approach considered only the impact of approved simulator and pharmacokinetics models to compare demographic characteristics on the PBPK modeling (i.e., two designs, crossover and parallel (121). The parallel design was drugs’ absorption, distribution, metabolism, and excretion). justified because it would likely be preferred in a real setting for However, other consequences of these characteristics at practical reasons, which is not necessarily true in the case of the ADHD pathophysiology level were not considered, our study on ADHD. Another head-to-head mechanistic study because of the absence of (i) clear molecular information comparing two lung cancer treatments has been reported by our to include in the ADHD definitions for each patient profile, group, whereby a similar approach to the one here described was and (ii) reliable sources of information to properly model undertaken (122). However, our previous study did not require these characteristics at the molecular level. This limitation generating virtual populations nor used PBPK or QSP models to could prevent the modeling and detection of relevant reach its conclusions. results regarding these characteristics, such as age-related These examples of application of in silico modeling neurodevelopment (128, 129), differences between children, approaches in different therapeutic areas bear witness to an adolescent and adults (130, 131) or the potential role of increasing tendency to use newly available high performance sex-dependent differences (132–134). Future data on large computing technologies in the field of biomedicine. The sets of patients, or specific research on the impact of those use of these technologies will help advancing toward characteristics on ADHD, might allow to improve our models the implementation of precision medicine pipelines and and derived conclusions. personalizing the healthcare provided to patients. Thirdly, all mathematical models are subjected to the limitation of not being able to fit 100% the training data Strengths and Limitations information. In our approach, while we obtained a pediatric TPMS models are constructed considering the whole human virtual population with demographic values non-significantly protein network and a wide range of drug-pathology different from the reference clinical trial population, the obtained relationships, not only limited to ADHD or psychiatric p-values for age and weight were close to the significance ailments, which, in part, attenuates the potential bias on threshold. These parameters proved to be more difficult to fit in information regarding drugs or disease of interest. As defined pediatric than in adult virtual populations. While clinical trials by Jorba et al. (36), only MoA models with accuracies above only report average weight and age, general pediatric population 85% against the training set were used to ensure good quality weight distributions obtained from growth information (35) and general extrapolation of results. This systems biology-based are age-dependent. Accordingly, setting a higher threshold of methodology has been reported to be successful, with results significance (p > 0.1 or even p > 0.2) during the randomization validated by in vitro and/or in vivo models (123–125). procedure might ensure obtaining a fitter population, especially Only limited by computational power, ISCT allows enrolling a regarding the pediatric case. In this specific scenario, as the large number of patients with several neurophysiological ADHD case-study objective for the generation of the ISCT was a subtypes, which can be difficult, costly, and even not feasible in head-to-head between LDX and MPH using the exact same a conventional clinical trial setting. Virtual patients generated population, this bias was not expected to significantly affect in our study were defined by the drugs’ molecular mechanisms, the results. TPMS-based models are not an exception either Frontiers in Psychiatry | www.frontiersin.org 14 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate FIGURE 8 | PCA representation, based on the modulation of ADHD effectors, of LDX and MPH mechanisms of action in (A) ADHD adult patients and (B) ADHD children-adolescent patients. ADHD, Attention-deficit/hyperactivity disorder; LDX, Lisdexamfetamine; MPH, Methylphenidate; PCA, Principal component analysis. Frontiers in Psychiatry | www.frontiersin.org 15 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate TABLE 7 | Results of the comparison analysis between demographic characteristics within the clusters. ID Age, years Height, cm Weight, kg BMI, kg/m2 Gender, M:F ratio Adults 1 36.49 ± 9.96 172.19 ± 10.62 78.79 ± 17.67 26.85 ± 6.82 0.63 (0.141) (0.424) (0.797) (0.297) (0.387) 2 37.16 ± 10.73 172.02 ± 9.94 79.58 ± 16.95 27.02 ± 6.04 0.60 (0.454) (0.739) (0.979) (0.034) (0.131) 3 36.20 ± 9.75 171.80 ± 9.92 77.33 ± 15.73 26.53 ± 6.67 0.59 (0.826) (0.158) (0.659) (0.651) (0.953) 4 38.04 ± 10.73 172.16 ± 10.01 75.17 ± 17.23 25.62 ± 6.67 0.52 (0.016) (0.064) (0.820) (0.006) (0.014) Children-adolescents 1 10.95 ± 2.82 145.96 ± 16.13 41.42 ± 12.60 18.93 ± 2.69 0.81 (0.634) (0.095) (0.064) (0.026) (0.031) 2 11.22 ± 2.57 147.44 ± 14.53 42.70 ± 12.00 19.25 ± 2.71 0.76 (0.067) (0.431) (0.568) (0.646) (0.400) 3 11.27 ± 2.75 148.29 ± 16.96 43.80 ± 13.51 19.36 ± 2.68 0.83 (0.170) (0.283) (0.131) (0.041) (0.118) Figures are mean ± standard deviation (and p-values*). *T-student-test: each cluster vs. rest of clusters. ANOVA, Analysis of variance; BMI, Body mass index; ID, Cluster ID; M:F ratio, Male to female ratio. p-values in bold are those considered significant (p < 0.05). (36). Each virtual patient was constructed with at least 50 associated to the outcome measure used would be of clinical solutions, and a population sample size larger than the nature; ADHD symptom scales are based on questionnaires to minimum calculated was used to dampen this effect. TPMS the patient or the physician, that comprise several aspects of a models present an inherent variability, rendering them useful complex psychiatric disease. These clinical measures might not to explore molecular variability within human physiology (36); be as directly associated to molecular or biologically measurable through an adequate management of the model’s variability and factors (such as blood pressure when studying hypertension). considering an appropriate sample size, the best solutions could To minimize both technical and clinical limitations of the be obtained. outcome measure used, we selected the ADHD-RS IV scale Finally, our study’s primary outcome was generated with as this was the scale with the largest amount of clinical trial information from literature on the drugs used for ADHD information for different mechanisms of action, so the model treatment and their measured clinical effect. The values used efficacy measures could be properly optimized to fit clinical data. for the training process were the average values reported in Our approach tried to compile the largest amount of available those publications (Table 3). However, a great dispersion was information around patients, disease, and treatments at the observed. For instance, while the mean ADHD-RS IV value molecular and clinical level and provided benchmarks to validate associated with amphetamine was −18.1, the authors report a the different steps of the study. Nonetheless, a corroboration range of response between −4.68 and −31.52, representing a of the herein described procedure to infer new actual clinical 74% deviation from the mean, clearly higher than the dispersion results with independent (existent or new) experiments is values generated with our models (Supplementary Figure A in called for. Supplementary Material 1). The dispersion identified in clinical In our ISCT, most of these sources of error could translate trials was probably due to demographic and metabolic differences into an error in the estimation of the principal variable, between patient cohorts and how the principal variable was evaluated by the Pearson correlation coefficient (Supplementary measured. This effect appeared in all analyzed drugs, and we Figures A, B in Supplementary Material 1). Interestingly, the estimated an average dispersion of 57% for all of them. The Pearson correlation coefficients obtained after lineal regression dispersion in the efficacy measured from the clinical trials cannot adjustment for the adult and pediatric-adolescent populations be mathematically treated without accessing patient data, which were high given the large dispersion shown in values from is not available; at this point, we had to resort to a naïve pooled clinical practice for the same drugs (Supplementary Table E in approach, risking its associated limitations. In such cases, the Supplementary Material 2). best approach to obtain a drug efficacy value is to compute the mean of the values reported by different authors. Selection bias CONCLUSION can also induce errors when using external data. To attenuate its effect, clinical trials assessing a wide range of drugs were The methods here illustrated described the step-by-step process used in our study. On the other hand, another limitation for creating a virtual population of patients treated with two Frontiers in Psychiatry | www.frontiersin.org 16 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate drugs for ADHD management, LDX and MPH, with the aim of JRG-C, JQ, BO, XD, and JM supervised and validated it. GJ and designing an ISCT for their comparison head-to-head. Our study VJ aided in visualization tasks. All authors reviewed and edited provided adult and pediatric-adolescent vPOPs and generated its final version. QSP models to infer, after analysis, the MoA of these two drugs. This theoretical model, and its use for a head-to-head analysis, would allow obtaining conclusions classified as MEDIUM impact FUNDING according to MID3 guidelines. Although experimental and This study was funded by Takeda. Public funders provided clinical assays are warranted to validate or refute these potential support for some of the authors’ salaries: GJ has received results before translation into clinical practice, the mechanistic- funding from the European Union’s Horizon 2020 research driven modeling techniques used here should be accepted as and innovation program under the Marie Skłodowska-Curie hypothesis-generation solid tools with a remarkable ability to Grant Agreement No. 765912. VJ is part of a project (COSMIC; provide molecular detail. Besides, from a scientific evidence www.cosmic-h2020.eu) that has received funding from the point of view, complementing meta-analyses with theoretical European Union’s Horizon 2020 research and innovation models, such as the ones here presented, can palliate the program under the Marie Skłodowska-Curie Grant Agreement lack of costly, though necessary, head-to-head clinical trials. No. 765158. Altogether, in silico techniques can contribute to advancing the understanding of diseases’ pathophysiology and the molecular MoA of available therapies, with the ultimate goal of reaching ACKNOWLEDGMENTS personalized medicine. The authors would like to thank Juan Manuel García Illarramendi (Anaxomics Biotech, Barcelona) for his assistance on the RNAseq DATA AVAILABILITY STATEMENT data to protein transformation for sample size calculation; Mireia The original contributions presented in the study are included Coma, Cristina Segú-Vergés, and Helena Bartra (Anaxomics in the article/Supplementary Material, further inquiries can be Biotech, Barcelona) for compiling data; Judith Farrés (Anaxomics directed to the corresponding authors. Biotech, Barcelona) for supervising the methodology; and Matías Rey-Carrizo (BCN Medical Writing) for providing editorial support. AUTHOR CONTRIBUTIONS JRG-C, JQ, VM, TP-R, CM, and JM conceived the study. JRG-C, SUPPLEMENTARY MATERIAL JQ, GJ, and VJ performed the investigation and drafted the first version of the manuscript. GJ, VJ, BO, XD, and JM developed the The Supplementary Material for this article can be found methodology of the study. GJ, VJ, and JM undertook software online at: https://www.frontiersin.org/articles/10.3389/fpsyt. related tasks. VM, TP-R, and CM managed the project and 2021.741170/full#supplementary-material REFERENCES 8. Wolkenhauer O, Auffray C, Brass O, Clairambault J, Deutsch A, Drasdo D, et al. Enabling multiscale modeling in systems medicine. Genome Med. 1. Holford N, Ma SC, Ploeger BA. Clinical trial simulation: a review. Clin (2014) 6:21. doi: 10.1186/gm538 Pharmacol Ther. (2010) 88:166–82. doi: 10.1038/clpt.2010.114 9. Wang RS, Maron BA, Loscalzo J. Systems medicine: evolution of systems 2. Gomeni R, D’Angeli C, Bye A. In silico prediction of optimal in vivo delivery biology from bench to bedside. Wiley Interdiscip Rev Syst Biol Med. (2015) properties using convolution-based model and clinical trial simulation. 7:141–61. doi: 10.1002/wsbm.1297 Pharm Res. (2002) 19:99–103. doi: 10.1023/A:1013667718695 10. Visser SAG, Norton JD, Marshall S, O’Kelly M. Common best practice 3. Clermont G, Bartels J, Kumar R, Constantine G, Vodovotz Y, Chow C. In in modeling and simulation across quantitative disciplines: a comparison silico design of clinical trials: a method coming of age. Crit Care Med. (2004) of independently emerging proposals. Stat Biopharm Res. (2018) 10:72– 32:2061–70. doi: 10.1097/01.CCM.0000142394.28791.C3 5. doi: 10.1080/19466315.2017.1385520 4. Girard P, Cucherat M, Guez D, Boissel J-P, Cucherat M, Durrleman S, et al. 11. Manolis E, Rohou S, Hemmings R, Salmonson T, Karlsson M, Milligan Simulation des essais cliniques dans le développement des médicaments. PA. the role of modeling and simulation in development and registration Therapies. (2004) 59:287–95. doi: 10.2515/therapie:2004056 of medicinal products: output from the EFPIA/EMA modeling and 5. Viceconti M, Henney A, Morley-Fletcher E. In silico clinical trials: how simulation workshop. CPT Pharmacometrics Syst Pharmacol. (2013) computer simulation will transform the biomedical industry. Int J Clin Trials. 2:e31. doi: 10.1038/psp.2013.7 (2016) 3:37–46. doi: 10.18203/2349-3259.ijct20161408 12. Marshall S, Hemmings R, Josephson F, Karlsson M, Posch M, Steimer 6. Blobel B, Ruotsalainen P, Brochhausen M, Oemig F, Uribe GA. Autonomous J-L. Modeling and simulation to optimize the design and analysis of systems and artificial intelligence in healthcare transformation to 5P confirmatory trials, characterize risk-benefit, and support label claims. medicine - ethical challenges. Stud Heal Technol Inf. (2020) 270:1089– CPT Pharmacometrics Syst Pharmacol. (2013) 2:e27. doi: 10.1038/psp. 93. doi: 10.3233/SHTI200330 2013.4 7. Bousquet J, Jorgensen C, Dauzat M, Cesario A, Camuzat T, 13. European Medicines Agency. Role of Modelling and Simulation in Regulatory Bourret R, et al. Systems medicine approaches for the definition Decision Making in Europe. (2011). Available online at: https://www. of complex phenotypes in chronic diseases and ageing. From ema.europa.eu/en/documents/presentation/presentation-role-modelling- concept to implementation and policies. Curr Pharm Des. (2014) simulation-regulatory-decision-making-europe_en.pdf (accessed July, 20:5928–44. doi: 10.2174/1381612820666140314115505 2020). Frontiers in Psychiatry | www.frontiersin.org 17 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate 14. Nguyen TH, Mouksassi MS, Holford N, Al-Huniti N, Freedman I, Hooker ADHD: a randomized placebo-controlled, multi-centre study with AC, et al. Model evaluation of continuous data pharmacometric models: extended-release methylphenidate. World J Biol Psychyatry. (2012) metrics and graphics. CPT Pharmacometrics Syst Pharmacol. (2017) 6:87– 13:48–59. doi: 10.3109/15622975.2010.540257 109. doi: 10.1002/psp4.12161 33. Ginsberg L, Katic A, Adeyi B, Dirks B, Babcock T, Lasser R, et al. Long- 15. Adler LA, Faraone SV, Spencer TJ, Berglund P, Alperin S, Kessler RC. term treatment outcomes with lisdexamfetamine dimesylate for adults with The structure of adult ADHD. Int J Methods Psychiatr Res. (2017) attention-deficit/hyperactivity disorder stratified by baseline severity. Curr 26:e1555. doi: 10.1002/mpr.1555 Med Res Opin. (2011) 27:1097–107. doi: 10.1185/03007995.2011.567256 16. Kessler RC, Adler L, Barkley R, Biederman J, Conners CK, Demler O, et al. 34. The European Social Survey. ESS Round 7: European Social Survey Round 7 The prevalence and correlates of adult ADHD in the United States: results Data (2014). Data file edition 2.2. NSD - Norwegian Centre for Research Data, from the national comorbidity survey replication. Am J Psychiatry. (2006) Norway - Data Archive and distributor of ESS data for ESS ERIC. (2018). 163:716–23. doi: 10.1176/ajp.2006.163.4.716 Available online at: https://www.europeansocialsurvey.org/data/download. 17. Polanczyk GV, Willcutt EG, Salum GA, Kieling C, Rohde LA, html?r=7 (accessed June, 2020). ADHD. prevalence estimates across three decades: an updated 35. de Onis M, Onyango AW, Borghi E, Siyam A, Nishida C, systematic review and meta-regression analysis. Int J Epidemiol. (2014) Siekmann J. Development of a WHO growth reference for school- 43:434–42. doi: 10.1093/ije/dyt261 aged children and adolescents. Bull World Health Organ. (2007) 18. Xu G, Strathearn L, Liu B, Yang B, Bao W. Twenty-year trends 85:660–7. doi: 10.2471/BLT.07.043497 in diagnosed attention-deficit/hyperactivity disorder among US 36. Jorba G, Aguirre-Plans J, Junet V, Segu-Verges C, Ruiz JL, Pujol A, children and adolescents, 1997-2016. JAMA Netw Open. (2018) et al. In-silico simulated prototype-patients using TPMS technology 1:e181471. doi: 10.1001/jamanetworkopen.2018.1471 to study a potential adverse effect of sacubitril and valsartan. 19. Achenbach TM, Howell CT, McConaughy SH, Stanger C. Six-year PLoS ONE. (2020) 15:e0228926. doi: 10.1371/journal.pone.02 predictors of problems in a national sample: IV. Young adult signs 28926 of disturbance. J Am Acad Child Adolesc Psychiatry. (1998) 37:718– 37. Barrett T, Wilhite SE, Ledoux P, Evangelista C, Kim IF, Tomashevsky M, et al. 27. doi: 10.1097/00004583-199807000-00011 NCBI GEO: archive for functional genomics data sets–update. Nucleic Acids 20. Wilens TE, Faraone S V, Biederman J. Attention-deficit/hyperactivity Res. (2013) 41:D991–5. doi: 10.1093/nar/gks1193 disorder in adults. JAMA. (2004) 292:619–23. doi: 10.1001/jama.292.5.619 38. Law CW, Alhamdoosh M, Su S, Dong X, Tian L, Smyth GK, et al. RNA- 21. Larson K, Russ SA, Kahn RS, Halfon N. Patterns of comorbidity, functioning, seq analysis is easy as 1-2-3 with limma, Glimma and edgeR. F1000Research. and service use for US children with ADHD, 2007. Pediatrics. (2011) (2018) 5:1408. doi: 10.12688/f1000research.9005.3 127:462–70. doi: 10.1542/peds.2010-0165 39. Mukherjee S, Tamayo P, Rogers S, Rifkin R, Engle A, Campbell C, et al. 22. Katzman MA, Bilkey TS, Chokka PR, Fallu A, Klassen LJ. Adult ADHD and Estimating dataset size requirements for classifying DNA microarray data. comorbid disorders: clinical implications of a dimensional approach. BMC J Comput Biol. (2003) 10:119–42. doi: 10.1089/106652703321825928 Psychiatry. (2017) 17:302. doi: 10.1186/s12888-017-1463-3 40. Figueroa RL, Zeng-Treitler Q, Kandula S, Ngo LH. Predicting sample size 23. McGough JJ, Barkley RA. Diagnostic controversies in adult required for classification performance. BMC Med Inf Decis Mak. (2012) attention deficit hyperactivity disorder. Am J Psychiatry. (2004) 12:8. doi: 10.1186/1472-6947-12-8 161:1948–56. doi: 10.1176/appi.ajp.161.11.1948 41. Fukunaga K. Chapter 1 introduction. In: Fukunaga K, editor. Introduction 24. Ohnishi T, Kobayashi H, Yajima T, Koyama T, Noguchi K. Psychiatric to Statistical Pattern Recognition. 2nd ed. Boston: Academic Press. (1990). comorbidities in adult attention-deficit/hyperactivity disorder: prevalence p. 1–10. and patterns in the routine clinical setting. Innov Clin Neurosci. 42. Kohavi R. A study of cross-validation and bootstrap for accuracy estimation (2019) 16:11–6. and model selection. In: International Joint Conference on Artificial 25. Mannuzza S, Klein RG, Bessler A, Malloy P, LaPadula M. Adult psychiatric Intelligence (IJCAI), p. 1137–43. Available online at: https://dl.acm.org/ status of hyperactive boys grown up. Am J Psychiatry. (1998) 155:493– (accessed September, 2020). 8. doi: 10.1176/ajp.155.4.493 43. McCaffrey TA, St. Laurent G, III, Shtokalo D, Antonets D, Vyatkin Y, Jones 26. Howard AL, Kennedy TM, Mitchell JT, Sibley MH, Hinshaw SP, Arnold D, et al. Biomarker discovery in attention deficit hyperactivity disorder: LE, et al. Early substance use in the pathway from childhood attention- RNA sequencing of whole blood in discordant twin and case-controlled deficit/hyperactivity disorder (ADHD) to young adult substance use: cohorts. BMC Med Genomics. (2020) 13:160. doi: 10.1186/s12920-020- evidence of statistical mediation and substance specificity. Psychol Addict 00808-8 Behav. (2020) 34:281–92. doi: 10.1037/adb0000542 44. Connor DF, Steingard RJ. New formulations of stimulants for attention- 27. Biederman J. Impact of comorbidity in adults with attention- deficit hyperactivity disorder: therapeutic potential. CNS Drugs. (2004) deficit/hyperactivity disorder. J Clin Psychiatry. (2004) 65:3–7. 18:1011–30. doi: 10.2165/00023210-200418140-00005 28. Kittel-Schneider S, Reif A. Adulte Aufmerksamkeitsdefizit- 45. Döpfner M, Ose C, Fischer R, Ammer R, Scherag A. Comparison of the /Hyperaktivitätsstörung und Komorbidität: neue Befunde zu efficacy of two different modified release methylphenidate preparations epidemiologischen und genetischen Faktoren [Adult attention for children and adolescents with attention-deficit/hyperactivity disorder deficit hyperactivity disorder and comorbidity: new findings in a natural setting: comparison of the efficacy of Medikinet R retard on epidemiological and genetic factors]. Nervenarzt. (2020) and Concerta R -A rand. J Child Adolesc Psychopharmacol. (2011) 21:445– 91:575–82. doi: 10.1007/s00115-020-00900-5 54. doi: 10.1089/cap.2010.0082 29. Cortese S. Pharmacologic treatment of attention deficit-hyperactivity 46. Wishart DS, Feunang YD, Guo AC, Lo EJ, Marcu A, Grant JR, et al. disorder. N Engl J Med. (2020) 383:1050–6. doi: 10.1056/NEJMra1917069 DrugBank 5.0: a major update to the DrugBank database for 2018. Nucleic 30. Coghill D, Banaschewski T, Lecendreux M, Soutullo C, Johnson Acids Res. (2018) 46:D1074–82. doi: 10.1093/nar/gkx1037 M, Zuddas A, et al. European, randomized, phase 3 study of 47. Szklarczyk D, Santos A, von Mering C, Jensen LJ, Bork P, Kuhn M, et al. 5: lisdexamfetamine dimesylate in children and adolescents with attention- augmenting protein-chemical interaction networks with tissue and affinity deficit/hyperactivity disorder. Eur Neuropsychopharmacol. (2013) data. Nucleic Acids Res. (2016) 44:D380–4. doi: 10.1093/nar/gkv1277 23:1208–18. doi: 10.1016/j.euroneuro.2012.11.012 48. Hecker N, Ahmed J, von Eichborn J, Dunkel M, Macha K, Eckert A, et al. 31. Coghill DR, Banaschewski T, Lecendreux M, Zuddas A, Dittmann RW, SuperTarget goes quantitative: update on drug-target interactions. Nucleic Otero IH, et al. Efficacy of lisdexamfetamine dimesylate throughout the Acids Res. (2012) 40:D1113–7. doi: 10.1093/nar/gkr912 day in children and adolescents with attention-deficit/hyperactivity disorder: 49. Ekstrand E, Murphy HM, Wideman CH. The effects of the prodrug Vyvanse results from a randomized, controlled trial. Eur Child Adolesc Psychiatry. on spatial working memory and adiposity in rats. Pharmacol Biochem Behav. (2014) 23:61–8. doi: 10.1007/s00787-013-0421-y (2019) 186:172765. doi: 10.1016/j.pbb.2019.172765 32. Retz W, Rösler M, Ose C, Scherag A, Alm B, Philipsen A, 50. Hutson PH, Pennick M, Secker R. Preclinical pharmacokinetics, et al. Multiscale assessment of treatment efficacy in adults with pharmacology and toxicology of lisdexamfetamine: a novel Frontiers in Psychiatry | www.frontiersin.org 18 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate d-amphetamine pro-drug. Neuropharmacology. (2014) 87:41– a single-dose, randomized, open-label, crossover. Clin Ther. (2010) 50. doi: 10.1016/j.neuropharm.2014.02.014 32:252–64. doi: 10.1016/j.clinthera.2010.02.011 51. Strajhar P, Vizeli P, Patt M, Dolder PC, Kratschmar DV, Liechti ME, et al. 68. Bundesinstitut für Arzeimittel und Medizinprodukte. Decentralised Effects of lisdexamfetamine on plasma steroid concentrations compared Procedure. Public Assessment Report. Medikinet 5, 10, 20 mg. Medikinet with d-amphetamine in healthy subjects: a randomized, double-blind, retard 5, 10, 20, 30, 40 mg. Methylphenidate hydrochloride. Available online placebo-controlled study. J Steroid Biochem Mol Biol. (2019) 186:212– at: http://mri.cts-mrp.eu/download/DE_H_2223_003_PAR.pdf (June, 25. doi: 10.1016/j.jsbmb.2018.10.016 2020). 52. Cheney MD, Chen MH, Zhang D, Phillips JG, Loffredo MJ, D’Amico AV. 69. Maldonado R. Comparison of the pharmacokinetics and clinical efficacy of Greatest percentage of involved core length and the risk of death from new extended-release formulations of methylphenidate. Expert Opin Drug prostate cancer in men with highest Gleason score ≥ 7. Clin Genitourin Metab Toxicol. (2013) 9:1001–14. doi: 10.1517/17425255.2013.786041 Cancer. (2014) 12:234–40. doi: 10.1016/j.clgc.2014.01.006 70. Allen RJ, Rieger TR, Musante CJ. Efficient generation and selection of 53. Rajala AZ, Populin LC, Jenison RL. Methylphenidate affects task-switching virtual populations in quantitative systems pharmacology models. CPT and neural signaling in non-human primates. Psychopharmacol. (2020) Pharmacometrics Syst Pharmacol. (2016) 5:140–6. doi: 10.1002/psp4.12063 237:1533–43. doi: 10.1007/s00213-020-05478-z 71. Chib S, Greenberg E. Understanding the Metropolis-Hastings 54. Zhang CL, Feng ZJ, Liu Y, Ji XH, Peng JY, Zhang XH, et al. Methylphenidate algorithm. Am Stat. (1995) 49:327–35. doi: 10.1080/00031305.1995. enhances NMDA-receptor response in medial prefrontal cortex via sigma-1 10476177 receptor: a novel mechanism for methylphenidate action. PLoS ONE. (2012) 72. Peters SA. Evaluation of a generic physiologically based pharmacokinetic 7:e51910. doi: 10.1371/journal.pone.0051910 model for lineshape analysis. Clin Pharmacokinet. (2008) 55. Miller EM, Quintero JE, Pomerleau F, Huettl P, Gerhardt GA, Glaser PEA. 47:261–75. doi: 10.2165/00003088-200847040-00004 Chronic methylphenidate alters tonic and phasic glutamate signaling in the 73. Brochot C, Quindroit P. Modelling the Fate of Chemicals in Humans Using frontal cortex of a freely-moving rat model of ADHD. Neurochem Res. (2019) a Lifetime Physiologically Based Pharmacokinetic (PBPK) Model in MERLIN 44:89–101. doi: 10.1007/s11064-018-2483-1 Expo. Cham: Springer International Publishing (2018). p. 215–57. 56. Dew RE, Kollins SH. Lisdexamfetamine dimesylate: a new option in 74. Ciffroy P, Alfonso B, Altenpohl A, Banjac Z, Bierkens J, Brochot C, et al. stimulant treatment for ADHD. Expert Opin Pharmacother. (2010) 11:2907– Modelling the exposure to chemicals for risk assessment: a comprehensive 13. doi: 10.1517/14656566.2010.531009 library of multimedia and PBPK models for integration, prediction, 57. Markowitz JS, DeVane CL, Pestreich LK, Patrick KS, Muniz R. A uncertainty and sensitivity analysis - the MERLIN-Expo tool. Sci Total Env. comprehensive in vitro screening of d-, l-, and dl-threo-methylphenidate: (2016) 568:770–84. doi: 10.1016/j.scitotenv.2016.03.191 an exploratory study. J Child Adolesc Psychopharmacol. (2006) 16:687– 75. Rostami-Hodjegan A. Physiologically based pharmacokinetics joined 98. doi: 10.1089/cap.2006.16.687 with in vitro-in vivo extrapolation of ADME: a marriage under the 58. Markowitz JS, DeVane CL, Ramamoorthy S, Zhu HJ. The psychostimulant arch of systems pharmacology. Clin Pharmacol Ther. (2012) 92:50– d-threo-(R,R)-methylphenidate binds as an agonist to the 5HT(1A) receptor. 61. doi: 10.1038/clpt.2012.65 Pharmazie. (2009) 64:123–5. doi: 10.1691/ph.2009.8691 76. Guze SB. Diagnostic and statistical manual of mental disorders, 4th ed. 59. Läkemedelsverket/Swedish Medical Products Agency. Public Assessment (DSM-IV). Am J Psychiatry. (1995) 152:1228. doi: 10.1176/ajp.152.8.1228 ReportScientific discussion. Elvanse (lisdexamfetamine dimesylate) 77. The MathWorks Inc. MATLAB and SimBiology Toolbox Release 5.7. Natick, SE/H/1839/01-03. (2013). Available online at: https://docetp.mpa.se/ MA: The MathWorks Inc. (2017). LMF/Elvanse capsule, hard ENG PAR_09001bee80cde265.pdf (accessed 78. Spencer TJ, Wilens TE, Biederman J, Weisler RH, Read SC, Pratt R. June, 2020). Efficacy and safety of mixed amphetamine salts extended release 60. Najib J, Wimer D, Zeng J, Lam KW, Romanyak N, Paige Morgan (Adderall XR) in the management of attention-deficit/hyperactivity E, et al. Review of lisdexamfetamine dimesylate in adults with disorder in adolescent patients: a 4-week, randomized, double- attention-deficit/hyperactivity disorder. J Cent Nerv Syst Dis. (2017) blind, placebo-controlled, parallel-group study. Clin Ther. (2006) 9:1179573517728090. doi: 10.1177/1179573517728090 28:266–79. doi: 10.1016/j.clinthera.2006.02.011 61. Novartis Pharmaceutical Canada Inc. Product Monograph. Ritalin R 79. Wigal T, Brams M, Gasior M, Gao J, Squires L, Giblin J. Randomized, (Methylphenidate Hydrochloride) 10 mg and 20 mg Tablets Novartis double-blind, placebo-controlled, crossover study of the efficacy and safety Standard. (2017). Available online at: https://s3-us-west-2.amazonaws. of lisdexamfetamine dimesylate in adults with attention-deficit/hyperactivity com/drugbank/cite_this/attachments/files/000/004/483/original/Health_ disorder: novel findings using a simulated adult workplace environment Canada_-_Ritalin.PDF?1556050208 (accessed October, 2020). design. Behav Brain Funct. (2010) 6:34. doi: 10.1186/1744-90 62. Electronic Medicines Compendium (eMC). Medikinet 20 mg Tablets. 81-6-34 (2013). Available online at: https://www.medicines.org.uk/emc/product/ 80. ClinicalTrials.gov. Efficacy of Bupropion for Attention Deficit Hyperactivity 8236/smpc#gref (accessed September, 2020). Disorder (ADHD) in Adults (NCT01270555). (2013). Available online 63. Electronic Medicines Compendium (eMC). Medikinet 5 mg Tablets. (2013). at: https://clinicaltrials.gov/ct2/show/NCT01270555 (accessed September, Available online at: https://www.medicines.org.uk/emc/product/328/smpc# 2020). PRECLINICAL_SAFETY (accessed September, 2020). 81. Huss M, Ginsberg Y, Tvedten T, Arngrim T, Philipsen A, Carter K, 64. Chan YP, Swanson JM, Soldin SS, Thiessen JJ, Macleod SM, Logan W. et al. Methylphenidate hydrochloride modified-release in adults with Methylphenidate hydrochloride given with or before breakfast: II. Effects attention deficit hyperactivity disorder: a randomized double-blind placebo- on plasma concentration of methylphenidate and ritalinic acid. Pediatrics. controlled trial. Adv Ther. (2014) 31:44–65. doi: 10.1007/s12325-013- (1983) 72:56–9. 0085-5 65. (eMC) EMC. Concerta XL 36 mg Prolonged-Release Tablets. (2012). Available 82. Butterfield ME, Saal J, Young B, Young JL. Supplementary guanfacine online at: https://www.medicines.org.uk/emc/product/8236/smpc#gref hydrochloride as a treatment of attention deficit hyperactivity disorder (accessed July, 2020). in adults: a double blind, placebo-controlled study. Psychiatry Res. (2016) 66. Krishnan S, Zhang Y. Relative bioavailability of lisdexamfetamine 70-mg 236:136–41. doi: 10.1016/j.psychres.2015.12.017 capsules in fasted and fed healthy adult volunteers and in solution: a single- 83. Weisler RH, Greenbaum M, Arnold V, Yu M, Yan B, Jaffee M, et al. Efficacy dose, crossover pharmacokinetic study. J Clin Pharmacol. (2008) 48:293– and safety of SHP465 mixed amphetamine salts in the treatment of attention- 302. doi: 10.1177/0091270007310381 deficit/hyperactivity disorder in adults: results of a randomized, double- 67. Boellner SW, Stark JG, Krishnan S, Zhang Y. Pharmacokinetics blind, placebo-controlled, forced-dose clinical study. CNS Drugs. (2017) of lisdexamfetamine dimesylate and its active metabolite, d- 31:685–97. doi: 10.1007/s40263-017-0455-7 amphetamine, with increasing oral doses of lisdexamfetamine 84. Spencer TJ, Adler LA, McGough JJ, Muniz R, Jiang H, Pestreich L. Efficacy dimesylate in children with attention-deficit/hyperactivity disorder: and safety of dexmethylphenidate extended-release capsules in adults with Frontiers in Psychiatry | www.frontiersin.org 19 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate attention-deficit/hyperactivity disorder. Biol Psychiatry. (2007) 61:1380– 103. Park S, Kim BN, Kim JW, Shin MS, Cho SC, Kim JH, et al. 7. doi: 10.1016/j.biopsych.2006.07.032 Neurotrophin 3 genotype and emotional adverse effects of osmotic- 85. Biederman J, Krishnan S, Zhang Y, McGough JJ, Findling RL. Efficacy release oral system methylphenidate (OROS-MPH) in children with and tolerability of lisdexamfetamine dimesylate (NRP-104) in children attention-deficit/ hyperactivity disorder. J Psychopharmacol. (2014) 28:220– with attention-deficit/hyperactivity disorder: a phase III, multicenter, 6. doi: 10.1177/0269881113480989 randomized, double-blind, forced-dose, parallel-group study. Clin Ther. 104. Accardo JA, Marcus CL, Leonard MB, Shults J, Meltzer LJ, Elia J. Associations (2007) 29:450–63. doi: 10.1016/S0149-2918(07)80083-X between psychiatric comorbidities and sleep disturbances in children with 86. ClinicalTrials.gov. A Study Comparing the Effect of Atomoxetine Versus attention-deficit/hyperactivity disorder. J Dev Behav Pediatr. (2012) 33:97– Other Standard Care Therapy on the Long Term Functioning in Attention- 105. doi: 10.1097/DBP.0b013e31823f6853 Deficit/Hyperactivity Disorder (ADHD) Children and Adolescents (ADHD 105. Emre Onat O, Ece Kars M, Gül S, Bilguvar K, Wu Y, Özhan A, et al. Human LIFE) (NCT00447278). (2010). Available online at: https://clinicaltrials.gov/ CRY1 variants associate with attention deficit/hyperactivity disorder. J Clin ct2/show/NCT00447278 (accessed September, 2020). Invest. (2020) 130:3885–900. doi: 10.1172/JCI135500 87. Stein MA, Waldman ID, Charney E, Aryal S, Sable C, Gruber 106. Coogan AN, Schenk M, Palm D, Uzoni A, Grube J, Tsang R, et al. Dose effects and comparative effectiveness of extended AH, et al. Impact of adult attention deficit hyperactivity release dexmethylphenidate and mixed amphetamine salts. J Child disorder and medication status on sleep/wake behavior and Adolesc Psychopharmacol. (2011) 21:581–8. doi: 10.1089/cap.20 molecular circadian rhythms. Neuropsychopharmacology. (2019) 11.0018 44:1198–206. doi: 10.1038/s41386-019-0327-6 88. Jain R, Segal S, Kollins SH, Khayrallah M. Clonidine extended- 107. Leffa DT, Torres ILS, Rohde LA A review on the role of inflammation release tablets for pediatric patients with attention-deficit/hyperactivity in attention-deficit/hyperactivity disorder. Neuroimmunomodulation. (2018) disorder. J Am Acad Child Adolesc Psychiatry. (2011) 50:171– 25:328–33. doi: 10.1159/000489635 9. doi: 10.1016/j.jaac.2010.11.005 108. Verlaet AA, Noriega DB, Hermans N, Savelkoul HF. Nutrition, 89. Hervas A, Huss M, Johnson M, McNicholas F, van Stralen J, Sreckovic immunological mechanisms and dietary immunomodulation in ADHD. Eur S, et al. Efficacy and safety of extended-release guanfacine hydrochloride Child Adolesc Psychiatry. (2014) 23:519–29. doi: 10.1007/s00787-014-0522-2 in children and adolescents with attention-deficit/hyperactivity disorder: a 109. Verlaet AAJ, Breynaert A, Ceulemans B, De Bruyne T, Fransen E, randomized, controlled, phase III trial. Eur Neuropsychopharmacol. (2014) Pieters L, et al. Oxidative stress and immune aberrancies in attention- 24:1861–72. doi: 10.1016/j.euroneuro.2014.09.014 deficit/hyperactivity disorder (ADHD): a case-control comparison. Eur Child 90. Coghill DR, Banaschewski T, Nagy P, Otero IH, Soutullo C, Yan B, Adolesc Psychiatry. (2019) 28:719–29. doi: 10.1007/s00787-018-1239-4 et al. Long-term safety and efficacy of lisdexamfetamine dimesylate in 110. Jackson-Cowan LD, Cole EF, Arbiser JL, Silverberg JI, Lawley LP. TH2 children and adolescents with ADHD: a phase IV, 2-year, open-label sensitization in the skin-gut-brain axis: how early-life Th2-mediated study in Europe. CNS Drugs. (2017) 31:625–38. doi: 10.1007/s40263-017- inflammation may negatively perpetuate developmental and psychologic 0443-y abnormalities. Pediatr Dermatol. (2021). doi: 10.1111/pde.14657. [Epub 91. Benjamini Y, Hochberg Y. Controlling the false discovery rate - a practical ahead of print]. and powerful approach to multiple testing. J R Stat Soc B. (1995) 57:289– 111. Faraone SV, Spencer TJ, Kollins SH, Glatt SJ, Goodman D. Dose 300. doi: 10.1111/j.2517-6161.1995.tb02031.x response effects of lisdexamfetamine dimesylate treatment in adults 92. Zhang XY, Trame MN, Lesko LJ, Schmidt S. Sobol sensitivity analysis: a tool with ADHD: an exploratory study. J Atten Disord. (2012) 16:118– to guide the development and evaluation of systems pharmacology models. 27. doi: 10.1177/1087054711403716 CPT Pharmacometrics Syst Pharmacol. (2015) 4:69–79. doi: 10.1002/psp4.6 112. Visentin R, Dalla Man C, Kovatchev B, Cobelli C. The university 93. Vassilvitskii S, Arthur D. K-means++: The Advantages of Careful Seeding. of virginia/padova type 1 diabetes simulator matches the in Proceedings of the Eighteenth Annual ACM-SIAM Symposium on glucose traces of a clinical trial. Diabetes Technol Ther. (2014) Discrete Algorithms. Available online at: http://ilpubs.stanford.edu:8090/778/ 16:428–34. doi: 10.1089/dia.2013.0377 (accessed June, 2020). 113. Haidar A, Wilinska ME, Graveston JA, Hovorka R. Stochastic 94. Jolliffe IT. Principal Component Analysis. Berlin: Springer (2002). virtual population of subjects with type 1 diabetes for the 95. Banerjee A, Davë RN. Validating clusters using the Hopkins statistic. In: assessment of closed-loop glucose controllers. IEEE Trans IEEE International Conference on Fuzzy Systems. Budapest: IEEE (2004). p. Biomed Eng. (2013) 60:3524–33. doi: 10.1109/TBME.2013.22 149–53. 72736 96. Rousseeuw PJ. Silhouettes: a graphical aid to the interpretation and 114. Visentin R, Man CD, Cobelli C. One-day bayesian cloning of type 1 validation of cluster analysis. J Comput Appl Math. (1987) 20:53– diabetes subjects: toward a single-day UVA/Padova type 1 diabetes simulator. 65. doi: 10.1016/0377-0427(87)90125-7 IEEE Trans Biomed Eng. (2016) 63:2416–24. doi: 10.1109/TBME.2016. 97. Chung NC, Miasojedow B, Startek M, Gambin A. Jaccard/Tanimoto 2535241 similarity test and estimation methods for biological presence-absence data. 115. Polak S, Fijorek K, Glinka A, Wisniowska B, Mendyk A. Virtual BMC Bioinformatics. (2019) 20:644. doi: 10.1186/s12859-019-3118-5 population generator for human cardiomyocytes parameters: in silico 98. Coghill DR, Banaschewski T, Bliss C, Robertson B, Zuddas A. Cognitive drug cardiotoxicity assessment. Toxicol Mech Methods. (2012) 22:31– function of children and adolescents with attention-deficit/hyperactivity 40. doi: 10.3109/15376516.2011.585477 disorder in a 2-year open-label study of lisdexamfetamine dimesylate. CNS 116. Brown D, Namas RA, Almahmoud K, Zaaqoq A, Sarkar J, Barclay Drugs. (2018) 32:85–95. doi: 10.1007/s40263-017-0487-z DA, et al. Trauma in silico: individual-specific mathematical 99. Kitagishi Y, Minami A, Nakanishi A, Ogura Y, Matsuda S. Neuron membrane models and virtual clinical populations. Sci Transl Med. (2015) trafficking and protein kinases involved in autism and ADHD. Int J Mol Sci. 7:285ra61. doi: 10.1126/scitranslmed.aaa3636 (2015) 16:3095–115. doi: 10.3390/ijms16023095 117. Kim Y, Hatley O, Rhee S, Yi S, Lee HA, Yoon S, et al. Development 100. Liu DY, Shen XM, Yuan FF, Guo OY, Zhong Y, Chen JG, et al. The Physiology of a Korean-specific virtual population for physiologically based of BDNF and its relationship with ADHD. Mol Neurobiol. (2015) 52:1467– pharmacokinetic modelling and simulation. Biopharm Drug Dispos. 76. doi: 10.1007/s12035-014-8956-6 (2019) 40:135–50. doi: 10.1002/bdd.2178 101. Banaschewski T, Becker K, Scherag S, Franke B, Coghill D. Molecular 118. Ruiz P, Emond C, McLanahan ED, Joshi-Barr S, Mumtaz M. Exploring genetics of attention-deficit/hyperactivity disorder: an overview. Eur Child mechanistic toxicity of mixtures using PBPK modeling and computational Adolesc Psychiatry. (2010) 19:237–57. doi: 10.1007/s00787-010-0090-z systems biology. Toxicol Sci. (2020) 174:38–50. doi: 10.1093/toxsci/kfz243 102. Bilgiç A, Toker A, Işik Ü, Kilinç I. Serum brain-derived neurotrophic factor, 119. Wang H, Milberg O, Bartelink IH, Vicini P, Wang B, Narwal R, et al. glial-derived neurotrophic factor, nerve growth factor, and neurotrophin- In silico simulation of a clinical trial with anti-CTLA-4 and anti-PD-L1 3 levels in children with attention-deficit/hyperactivity disorder. Eur Child immunotherapies in metastatic breast cancer using a systems pharmacology Adolesc Psychiatry. (2017) 26:355–63. doi: 10.1007/s00787-016-0898-2 model. R Soc Open Sci. (2019) 6:190366. doi: 10.1098/rsos.190366 Frontiers in Psychiatry | www.frontiersin.org 20 November 2021 | Volume 12 | Article 741170 Gutiérrez-Casares et al. ADHD ISCT: Lisdexamfetamine vs. Methylphenidate 120. Wang H, Sové RJ, Jafarnejad M, Rahmeh S, Jaffee EM, Stearns V, 130. Zalsman G, Shilton T. Adult ADHD: a new disease? Int J Psychiatry Clin et al. Conducting a virtual clinical trial in HER2-negative breast cancer Pract. (2016) 20:70–6. doi: 10.3109/13651501.2016.1149197 using a quantitative systems pharmacology model with an epigenetic 131. Franke B, Michelini G, Asherson P, Banaschewski T, Bilbow A, Buitelaar modulator and immune checkpoint inhibitors. Front Bioeng Biotechnol. JK, et al. Live fast, die young? A review on the developmental trajectories (2020) 8:141. doi: 10.3389/fbioe.2020.00141 of ADHD across the lifespan. Eur Neuropsychopharmacol. (2018) 28:1059– 121. Schiavon M, Visentin R, Giegerich C, Sieber J, Dalla Man C, Cobelli C, 88. doi: 10.1016/j.euroneuro.2018.08.001 et al. In silico head-to-head comparison of insulin glargine 300 U/mL and 132. Quinn PO. Treating adolescent girls and women with ADHD: gender- insulin degludec 100 U/mL in type 1 diabetes. Diabetes Technol Ther. (2020) specific issues. J Clin Psychol. (2005) 61:579–87. doi: 10.1002/jclp.20121 22:553–61. doi: 10.1089/dia.2020.0027 133. Loke H, Harley V, Lee J. Biological factors underlying sex differences in 122. Carcereny E. Fernández-nistal A, López A, Montoto C, Naves A, Segú- neurological disorders: focus on SRY. Int J Biochem Cell Biol. (2015) 65:139– Vergés C, et al. Head to head evaluation of second generation ALK inhibitors 50. doi: 10.1016/j.biocel.2015.05.024 brigatinib and alectinib as first-line treatment for ALK + NSCLC using 134. Franceschini A, Fattore L. Gender-specific approach in an in silico systems biology-based approach. Oncotarget. (2021) 12:316– psychiatric diseases: because sex matters. Eur J Pharmacol. (2021) 32. doi: 10.18632/oncotarget.27875 896:173895. doi: 10.1016/j.ejphar.2021.173895 123. Romeo-Guitart D, Forés J, Herrando-Grabulosa M, Valls R, Leiva- Rodríguez T, Galea E, et al. Neuroprotective drug for nerve Conflict of Interest: JRG-C has served as speaker for Takeda and Shire and has trauma revealed using artificial intelligence. Sci Rep. (2018) received research funding from Shire. JQ has served as speaker and/or on scientific 8:1879. doi: 10.1038/s41598-018-19767-3 advisory boards for Takeda, Janssen, and Rubio. GJ, VJ, and JM are full-time 124. Iborra-Egea O, Santiago-Vacas E, Yurista SR, Lupón J, Packer M, employees at Anaxomics Biotech. VM, TP-R, and CM are full-time employees Heymans S, et al. Unraveling the molecular mechanism of action of at Takeda. empagliflozin in heart failure with reduced ejection fraction with or without diabetes. JACC Basic Transl Sci. (2019) 4:831–40. doi: 10.1016/j.jacbts.2019. The remaining authors declare that the research was conducted in the absence of 07.010 any commercial or financial relationships that could be construed as a potential 125. Loren V, Garcia-Jaraquemada A, Naves JE, Carmona X, Manosa M, Aransay conflict of interest. AM, et al. ANP32E, a protein involved in steroid-refractoriness in ulcerative colitis, identified by a systems biology approach. J Crohns Colitis. (2019) Publisher’s Note: All claims expressed in this article are solely those of the authors 13:351–61. doi: 10.1093/ecco-jcc/jjy171 and do not necessarily represent those of their affiliated organizations, or those of 126. Faraone SV, Mick E. Molecular genetics of attention deficit the publisher, the editors and the reviewers. Any product that may be evaluated in hyperactivity disorder. Psychiatr Clin North Am. (2010) 33:159– this article, or claim that may be made by its manufacturer, is not guaranteed or 80. doi: 10.1016/j.psc.2009.12.004 127. Faraone SV, Larsson H. Genetics of attention deficit hyperactivity disorder. endorsed by the publisher. Mol Psychiatry. (2019) 24:562–75. doi: 10.1038/s41380-018-0070-0 128. Sonuga-Barke EJS, Halperin JM. Developmental phenotypes and causal Copyright © 2021 Gutiérrez-Casares, Quintero, Jorba, Junet, Martínez, Pozo-Rubio, pathways in attention deficithyperactivity disorder: potential targets Oliva, Daura, Mas and Montoto. This is an open-access article distributed under the for early intervention? J Child Psychol Psychiatry. (2010) 51:368– terms of the Creative Commons Attribution License (CC BY). The use, distribution 89. doi: 10.1111/j.1469-7610.2009.02195.x or reproduction in other forums is permitted, provided the original author(s) and 129. Dark C, Homman-Ludiye J, Bryson-Richardson RJ. The role of the copyright owner(s) are credited and that the original publication in this journal ADHD associated genes in neurodevelopment. Dev Biol. (2018) is cited, in accordance with accepted academic practice. No use, distribution or 438:69–83. doi: 10.1016/j.ydbio.2018.03.023 reproduction is permitted which does not comply with these terms. Frontiers in Psychiatry | www.frontiersin.org 21 November 2021 | Volume 12 | Article 741170