Machine Learning-Based Prediction of Escitalopram and Sertraline Side Effects With Pharmacokinetic Data in Children and Adolescents.

Selective serotonin reuptake inhibitors (SSRI) are the first-line pharmacologic treatment for anxiety and depressive disorders in children and adolescents. Many patients experience side effects that are difficult to predict, are associated with significant morbidity, and can lead to treatment discontinuation. Variation in SSRI pharmacokinetics could explain differences in treatment outcomes, but this is often overlooked as a contributing factor to SSRI tolerability. This study evaluated data from 288 escitalopram-treated and 255 sertraline-treated patients ≤ 18 years old to develop machine learning models to predict side effects using electronic health record data and Bayesian estimated pharmacokinetic parameters. Trained on a combined cohort of escitalopram- and sertraline-treated patients, a penalized logistic regression model achieved an area under the receiver operating characteristic curve (AUROC) of 0.77 (95% confidence interval (CI): 0.66-0.88), with 0.69 sensitivity (95% CI: 0.54-0.86), and 0.82 specificity (95% CI: 0.72-0.87). Medication exposure, clearance, and time since the last dose increase were among the top features. Individual escitalopram and sertraline models yielded an AUROC of 0.73 (95% CI: 0.65-0.81) and 0.64 (95% CI: 0.55-0.73), respectively. Post hoc analysis showed sertraline-treated patients with activation side effects had slower clearance (P = 0.01), which attenuated after accounting for age (P = 0.055). These findings raise the possibility that a machine learning approach leveraging pharmacokinetic data can predict escitalopram- and sertraline-related side effects. Clinicians may consider differences in medication pharmacokinetics, especially during dose titration and as opposed to relying on dose, when managing side effects. With further validation, application of this model to predict side effects may enhance SSRI precision dosing strategies in youth.

Influence of CYP2D6 Metabolizer Status on Risperidone and Paliperidone Tolerability in Children and Adolescents.

Background: Risperidone and, to a lesser extent, paliperidone are metabolized by CYP2D6; however, there are limited data related to variation in CYP2D6 phenotypes and the tolerability of these medications in children and adolescents. Furthermore, the impact of CYP2D6 on the association of risperidone and paliperidone with hyperprolactinemia in youth is not well understood. Methods: A retrospective chart review was performed in psychiatrically hospitalized children and adolescents prescribed risperidone (n = 263, age = 3-18 years, mean age = 13 ± 3 years, 49% female) or paliperidone (n = 124, age = 5-18 years, mean age = 15 ± 2 years, 44% female) who had CYP2D6 genotyping performed as part of routine care. CYP2D6 phenotypes were determined based on Clinical Pharmacogenetics Implementation Consortium guidelines and CYP2D6 inhibitors causing phenoconversion. Adverse effects were obtained from a review of the electronic health record, and patients were selected, in part, to enrich non-normal metabolizers. Results: Among risperidone-treated patients, 45% experienced an adverse effect, whereas 36% of paliperidone-treated patients experienced adverse effects. Discontinuation of risperidone due to lack of efficacy was more frequent in the CYP2D6 normal metabolizers and ultrarapid metabolizers compared with intermediate metabolizers (IMs) and phenoconverted poor metabolizers (pPMs) (54.5% vs. 32.7%, p < 0.001). Discontinuation due to weight gain was more common among risperidone- than paliperidone-treated patients (17% vs. 7%, p = 0.011). Among those taking paliperidone, CYP2D6 was associated with discontinuation due to side effects (p = 0.008), and youth with slower CYP2D6 metabolism (i.e., pPMs and IMs) were more likely to discontinue. Hyperprolactinemia was found in 10% of paliperidone-treated patients and 5% of risperidone-treated patients, and slower CYP2D6 metabolizers required higher risperidone doses to cause hyperprolactinemia (p = 0.011). Conclusions: CYP2D6 phenotype is associated with discontinuation of risperidone due to lack of efficacy and the dose of risperidone that induced hyperprolactinemia, as well as discontinuation of paliperidone due to adverse effects. Future studies should evaluate exposure-response and toxicity relationships in risperidone- and paliperidone-treated youth.

Escitalopram and Sertraline Population Pharmacokinetic Analysis in Pediatric Patients.

BACKGROUND AND OBJECTIVE: Escitalopram and sertraline are commonly prescribed for anxiety and depressive disorders in children and adolescents. The pharmacokinetics (PK) of these medications have been evaluated in adults and demonstrate extensive variability, but studies in pediatric patients are limited. Therefore, we performed a population PK analysis for escitalopram and sertraline in children and adolescents to characterize the effects of demographic, clinical, and pharmacogenetic factors on drug exposure. METHODS: A PK dataset was generated by extracting data from the electronic health record and opportunistic sampling of escitalopram- and sertraline-treated psychiatrically hospitalized pediatric patients aged 5-18 years. A population PK analysis of escitalopram and sertraline was performed using NONMEM. Concentration-time profiles were simulated using MwPharm++ to evaluate how covariates included in the final models influence medication exposure and compared to adult therapeutic reference ranges. RESULTS: The final escitalopram cohort consisted of 315 samples from 288 patients, and the sertraline cohort consisted of 265 samples from 255 patients. A one-compartment model with a proportional residual error model best described the data for both medications. For escitalopram, CYP2C19 phenotype and concomitant CYP2C19 inhibitors affected apparent clearance (CL/F), and normalizing CL/F and apparent volume of distribution (V/F) to body surface area (BSA) improved estimations. The final escitalopram model estimated CL/F and V/F at 14.2 L/h/1.73 m2 and 428 L/1.73 m2, respectively. For sertraline, CYP2C19 phenotype and concomitant CYP2C19 inhibitors influenced CL/F, and empirical allometric scaling of patient body weight on CL/F and V/F was significant. The final sertraline model estimated CL/F and V/F at 124 L/h/70 kg and 4320 L/70 kg, respectively. Normalized trough concentrations (Ctrough) for CYP2C19 poor metabolizers taking escitalopram were 3.98-fold higher compared to normal metabolizers (151.1 ng/mL vs 38.0 ng/mL, p < 0.0001), and normalized Ctrough for CYP2C19 poor metabolizers taking sertraline were 3.23-fold higher compared to normal, rapid, and ultrarapid metabolizers combined (121.7 ng/mL vs 37.68 ng/mL, p < 0.0001). Escitalopram- and sertraline-treated poor metabolizers may benefit from a dose reduction of 50-75% and 25-50%, respectively, to normalize exposure to other phenotypes. CONCLUSION: To our knowledge, this is the largest population PK analysis of escitalopram and sertraline in pediatric patients. Significant PK variability for both medications was observed and was largely explained by CYP2C19 phenotype. Slower CYP2C19 metabolizers taking escitalopram or sertraline may benefit from dose reductions given increased exposure.

The Impact of Marijuana on Antidepressant Treatment in Adolescents: Clinical and Pharmacologic Considerations.

The neuropharmacology of marijuana, including its effects on selective serotonin reuptake inhibitor (SSRI)/antidepressant metabolism and the subsequent response and tolerability in youth, has received limited attention. We sought to (1) review clinically relevant pharmacokinetic (PK) and pharmacodynamic (PD) interactions between cannabinoids and selected SSRIs, (2) use PK models to examine the impact of cannabinoids on SSRI exposure (area under curve (AUC)) and maximum concentration (CMAX) in adolescents, and (3) examine the frequency of adverse events reported when SSRIs and cannabinoids are used concomitantly. Cannabinoid metabolism, interactions with SSRIs, impact on relevant PK/PD pathways and known drug-drug interactions were reviewed. Then, the impact of tetrahydrocannabinol (THC) and cannabidiol (CBD) on exposure (AUC24) and CMAX for escitalopram and sertraline was modeled using pediatric PK data. Using data from the Food and Drug Administration Adverse Events Reporting System (FAERS), the relationship between CBD and CYP2C19-metabolized SSRIs and side effects was examined. Cannabis and CBD inhibit cytochrome activity, alter serotonergic transmission, and modulate SSRI response. In PK models, CBD and/or THC increases sertraline and es/citalopram concentrations in adolescents, and coadministration of CBD and CYP2C19-metabolized SSRIs increases the risk of cough, diarrhea, dizziness, and fatigue. Given the significant SSRI-cannabinoid interactions, clinicians should discuss THC and CBD use in youth prescribed SSRIs and be aware of the impact of initiating, stopping, or decreasing cannabinoid use as this may significantly affect es/citalopram and sertraline exposure.

The need for a refined understanding of CYP2D6 in second-generation antipsychotic outcomes in children and adolescents.

Tweetable abstract High-quality studies examining the influence of CYP2D6 on the exposure and tolerability of antipsychotics in youth are needed to mitigate the limitations of prior studies.

The IRF5-TNPO3 association with systemic lupus erythematosus has two components that other autoimmune disorders variably share.

Exploiting genotyping, DNA sequencing, imputation and trans-ancestral mapping, we used Bayesian and frequentist approaches to model the IRF5-TNPO3 locus association, now implicated in two immunotherapies and seven autoimmune diseases. Specifically, in systemic lupus erythematosus (SLE), we resolved separate associations in the IRF5 promoter (all ancestries) and with an extended European haplotype. We captured 3230 IRF5-TNPO3 high-quality, common variants across 5 ethnicities in 8395 SLE cases and 7367 controls. The genetic effect from the IRF5 promoter can be explained by any one of four variants in 5.7 kb (P-valuemeta = 6 × 10(-49); OR = 1.38-1.97). The second genetic effect spanned an 85.5-kb, 24-variant haplotype that included the genes IRF5 and TNPO3 (P-valuesEU = 10(-27)-10(-32), OR = 1.7-1.81). Many variants at the IRF5 locus with previously assigned biological function are not members of either final credible set of potential causal variants identified herein. In addition to the known biologically functional variants, we demonstrated that the risk allele of rs4728142, a variant in the promoter among the lowest frequentist probability and highest Bayesian posterior probability, was correlated with IRF5 expression and differentially binds the transcription factor ZBTB3. Our analytical strategy provides a novel framework for future studies aimed at dissecting etiological genetic effects. Finally, both SLE elements of the statistical model appear to operate in Sjögren's syndrome and systemic sclerosis whereas only the IRF5-TNPO3 gene-spanning haplotype is associated with primary biliary cirrhosis, demonstrating the nuance of similarity and difference in autoimmune disease risk mechanisms at IRF5-TNPO3.

Lupus risk variants in the PXK locus alter B-cell receptor internalization.

Genome wide association studies have identified variants in PXK that confer risk for humoral autoimmune diseases, including systemic lupus erythematosus (SLE or lupus), rheumatoid arthritis and more recently systemic sclerosis. While PXK is involved in trafficking of epidermal growth factor Receptor (EGFR) in COS-7 cells, mechanisms linking PXK to lupus pathophysiology have remained undefined. In an effort to uncover the mechanism at this locus that increases lupus-risk, we undertook a fine-mapping analysis in a large multi-ancestral study of lupus patients and controls. We define a large (257kb) common haplotype marking a single causal variant that confers lupus risk detected only in European ancestral populations and spans the promoter through the 3' UTR of PXK. The strongest association was found at rs6445972 with P < 4.62 × 10(-10), OR 0.81 (0.75-0.86). Using stepwise logistic regression analysis, we demonstrate that one signal drives the genetic association in the region. Bayesian analysis confirms our results, identifying a 95% credible set consisting of 172 variants spanning 202 kb. Functionally, we found that PXK operates on the B-cell antigen receptor (BCR); we confirmed that PXK influenced the rate of BCR internalization. Furthermore, we demonstrate that individuals carrying the risk haplotype exhibited a decreased rate of BCR internalization, a process known to impact B cell survival and cell fate. Taken together, these data define a new candidate mechanism for the genetic association of variants around PXK with lupus risk and highlight the regulation of intracellular trafficking as a genetically regulated pathway mediating human autoimmunity.

Genetic susceptibility to lupus: the biological basis of genetic risk found in B cell signaling pathways.

Over 50 genetic variants have been statistically associated with the development of SLE (or lupus). Each genetic association is a key component of a pathway to lupus pathogenesis, the majority of which requires further mechanistic studies to understand the functional changes to cellular physiology. Whereas their use in clinical practice has yet to be established, these genes guide efforts to develop more specific therapeutic approaches. The BCR signaling pathways are rich in lupus susceptibility genes and may well provide novel opportunities for the understanding and clinical treatment of this complex disease.