Assessing Pharmacokinetic Correlates of Escitalopram-Related Adverse Drug Reactions

Supplemental Digital Content is Available in the Text.

Abstract

Background:

To assess the pharmacokinetic correlates of reported adverse drug reactions (ADRs) under antidepressant treatment with escitalopram (ESC) using a large therapeutic drug monitoring database.

Methods:

A large naturalistic sample of inpatients and outpatients prescribed ESC was analyzed. ADRs were classified using the Udvalg for Kliniske Undersogelser side effect rating scale. We compared ESC-treated patients with (n = 35) and without ADRs (n = 273) using ESC plasma concentrations as the primary outcome. We also compared ADR rates in the 2 groups based on 2 cut-off ESC levels reflecting the recommended upper thresholds of the therapeutic reference range of 80 ng/mL, suggested by the consensus therapeutic drug monitoring guidelines, and 40 ng/mL, based on recent meta-analysis data. The effects of age, sex, smoking, daily ESC dose, plasma concentrations, and concentrations corrected for daily dose were included in a binary logistic regression model to predict ADRs.

Results:

No differences in clinical, demographic, or pharmacokinetic parameters were observed between patients with and without ADRs (P > 0.05). Patients with ESC-related ADRs were more frequently diagnosed with psychotic disorders than those without (25% vs. 7.1%, P = 0.004). None of the variables was associated with ADR risk. Overall, ADR rates were not significantly different in patients above versus below thresholds of ESC concentrations (ESC concentrations >40 [n = 59] vs. ≤40 ng/mL [n = 249] and >80 [n = 8] vs. ≤80 ng/mL [n = 300]; P = 0.56 and P = 1.0, respectively).

Conclusions:

No distinct pharmacokinetic patterns underlying ESC-associated ADRs were observed. Further studies with more specific assessments of ADRs in larger cohorts are required to better identify potential underlying patterns.

INTRODUCTION

Escitalopram (ESC), the active S-(+)-enantiomer of citalopram, is a selective serotonin reuptake inhibitor used as a first-line treatment for major depressive disorder (MDD)¹ and generalized anxiety disorder.² It is the most commonly prescribed antidepressant in the United States.³ Notably intriguing is the exponential surge in the total prescription of SSRIs within the United States, soaring by a staggering 3001% from 1991 to 2018.⁴ ESC undergoes hepatic metabolism mediated by distinct cytochrome P450 isoenzymes, predominantly CYP2C19 and, to a lesser extent, via 2D6 and 3A4.⁵ Moreover, ESC is a P-glycoprotein (P-gp) substrate,⁵ a transport protein that facilitates the elimination of absorbed drugs both in the blood–brain barrier and in the gut. In fact, together with CYP3A4, P-gp acts in tandem in the gut, representing the intestinal first-pass effect.⁶ The half-life of ESC ranges between 27 and 32 hours,⁵ and they are prescribed once daily. Twenty milligrams is the maximum recommended daily dose of ESC. However, the daily dose for older patients (aged >65 years) is limited to 10 mg, depending on the country, because of the risk of corrected QT interval (QTc) prolongation.⁷ Over the past years, the “Consensus Guidelines for Therapeutic Drug Monitoring (TDM) in Neuropsychopharmacology” has provided a therapeutic reference range of 15–80 ng/mL under steady state conditions, with a laboratory alert level at 160 ng/mL.⁵ However, a very recently published meta-analysis of TDM studies suggested a revision of the reference range, adjusting it to a range between 20 and 40 ng/mL,⁸ in which the upper threshold of the reference range was defined by maximum efficacy and not by an increased risk of adverse reactions, as is done for most tricyclic antidepressants.

Generally, the pharmacodynamic profile of ESC has a favorable balance among efficacy, effectiveness, and tolerability.^9,10 Common ESC-related adverse drug-induced reactions (adverse drug reactions [ADRs]) include nausea, diarrhea, dizziness,¹¹ blurred vision,¹² and somnolence.¹³ Some dose-dependent patterns have been reported for gastrointestinal ADRs,¹⁴ which may invariably occur during ESC titration.¹⁵ However, commonly reported ADRs after long-term treatment with ESC include weight gain and sexual dysfunction.¹⁵

The assessment of the pharmacokinetic correlates potentially underlying ESC-associated ADRs has not yielded consistent findings. In fact, a previous meta-analysis of TDM studies suggested that the evidence for pharmacokinetic correlates of ESC-associated ADRs is not strong.⁸ For instance, Hart et al^16,17 did not report any relationship between ESC concentration and QTc in a naturalistic sample. The same research group analyzed ADRs in a naturalistic sample of 300 patients with MDD with information on ADRs being available in 50 (out of 300) patients¹⁷; 12 of these 50 patients (24%) reported ESC-associated ADRs with mean drug concentrations of ESC 36.0 ± 33.5 ng/mL and a mean daily dose of 15.6 ± 5.0 mg. The most frequently reported ADR was tension/unrest in 8 of 12 patients. In a large 12-week open-label, partly randomized study, researchers reported elevated ESC levels after 8 weeks in patients with MDD with dry mouth, diarrhea, or drowsiness,¹⁸ and significant associations were detected between diarrhea and the metabolite-to-parent ratio, MPR, dizziness, and metabolite concentration. In an earlier, predominantly female sample, a Taiwanese group reported frequent reports of dry mouth, nausea, and vomiting in carriers of different CYP1A2 genetic polymorphisms, which in turn were associated with high metabolic ratios of S-desmethylcitalopram and S-didesmethylcitalopram, after adjusting for smoking effects after 2 weeks of treatment¹⁹; however, the authors did not assess the role of ESC concentrations separately. A negative association between the severity of sexual dysfunction and ESC concentration was reported in a subgroup of patients who received ESC monotherapy but not in a subgroup of patients who received a combination of ESC and aripiprazole.²⁰ Additionally, in a discontinuation trial of 25 outpatients with MDD, patients who developed discontinuation symptoms had higher daily ESC doses and plasma concentrations than those without MDD.²¹ Finally, in the largest available cohort, Jukic et al investigated switching rates from ESC to other antidepressants as a surrogate for ESC-based treatment failure.²² Thus, treatment failure using this design included both lack of effectiveness and tolerability. Switching from ESC to another antidepressant is approximately 3 times more frequent in CYP2C19 poor metabolizers, with ESC concentrations twice as high as those in CYP2C19 intermediate and extensive metabolizers.²² Accordingly, the authors discussed this pattern in the context of the potentially increased risk of ADRs.

As there is no clear pattern in the relationship between drug concentrations in ESC and ADRs, this study aimed to evaluate the pharmacokinetic correlates of ESC-associated ADRs in a large naturalistic TDM database of inpatients and outpatients under steady-state conditions.

MATERIALS AND METHODS

Patients

The data were analyzed in cooperation with the Department of Psychiatry, Psychotherapy, and Psychosomatics of RWTH Aachen University Hospital, Aachen, Germany, and the Department of Psychiatry and Psychotherapy at the University of Regensburg, Germany. The TDM database with 1109 plasma concentrations of ESC concentrations from adult inpatients and outpatients treated with ESC for different indications was assessed. The data collection took place between 2006 and 2015 as part of the clinical routine in different institutions as part of the AGATE network, “Arbeitsgemeinschaft Arzneimittelsicherheit bei psychischen Erkrankungen” (for details: www.amuep-agate.de), which is a consortium of psychiatric hospitals predominantly in southern Germany performing TDM in clinical routine and pharmacovigilance. The use of TDM is part of routine patient monitoring during ESC-based treatment.²³ Furthermore, TDM was performed when indicated for clinical reasons, such as when ADRs occurred or for other common TDM indications (see Table, Supplemental Digital Content 1, http://links.lww.com/TDM/A718).^24,25 Data registration followed standardized protocols. A retrospective analysis of clinical data was performed in accordance with the 1964 Declaration of Helsinki and its later amendments, and the local regulatory authority of RWTH Aachen University Hospital. For this type of study, a formal patient consent was not required. A group of 541 patients on concomitant medication with possible CYP2D6 inhibitory, CYP3A4, or CYP2C19 inhibitory or inducing properties were excluded (see Table, Supplemental Digital Content 1, http://links.lww.com/TDM/A718).²⁶ Four patients with missing ESC data and 243 patients with incomplete information about ADRs were also excluded from our analysis, resulting in a total of 308 patients. The treating physicians provided detailed narrative reports of (newly occurred) ADRs shortly before blood sampling on the TDM request form. Whether ADRs have occurred must be noted on the request form, and in the case of ADRs, a narrative description is requested. Retrospectively, reported ADRs were classified into 4 major groups (psychic, neurological, autonomic, and other) based on the categories of the Udvalg for Kliniske Undersogelser (UKU) side effect rating scale. The UKU scale was used to assess ADRs.²⁷

Quantification of ESC

Blood samples were collected immediately before drug administration (trough concentration) at steady state (>5 elimination half-lives under the same drug dose). ESC concentrations were determined using HPLC with ultraviolet detection (HPLC/UV). The method was validated according to Deutsche Industrie Norm 32,645, as described in the guidelines of the Society of Toxicology and Forensic Chemistry, in accordance with ISO 5725 (International Organization for Standardization),²⁸ FDA (US Food and Drug Administration) guidance,²⁹ and ICH (International Conference on Harmonization) requirements.³⁰ The limit of quantification was 6 ng/mL in accordance with a signal-to-noise ratio of 3:1.³¹ This laboratory conducts regular internal quality controls and participates in external quality assessment schemes using INSTAND (www.instand-ev.de; Düsseldorf, Germany).

Statistical Analysis

The study design included a comparison of 2 study groups: patients reporting ESC-related ADRs (ADRs; n = 35) and those without ADRs (n = 273). Continuous variables are presented as medians and interquartile ranges (IQRs), and categorical variables are presented as numbers and percentages. The Shapiro–Wilk test was used to assess normality, and Levene test was used to assess the homogeneity of variance. Comparisons were performed using the Mann–Whitney U test. Categorical variables were analyzed using Pearson χ² test or Fisher exact test, as appropriate. Our primary outcome was ESC plasma concentration at trough levels. A binary logistic regression model was used to assess independent predictors associated with ADRs. The following prespecified variables were included in the multivariable models: age, sex, body mass index (BMI), smoking status, ESC daily doses, and ESC plasma concentrations. Finally, we compared the frequencies of ADRs in the 2 groups based on 2 different cut-off levels of ESC plasma concentrations: (1) 40 ng/mL, which is the upper threshold of a more recently suggested therapeutic reference range⁸ and (2) 80 ng/mL, which is the upper threshold of the therapeutic reference range suggested by the AGNP in the last consensus update.⁵ Differences were considered statistically significant at a two-sided P-value <0.05. All analyses were performed using the SPSS for Windows 25.0 (IBM, Armonk, NY) software.

RESULTS

The demographic, clinical, and relevant pharmacokinetic data for both groups are summarized in Table 1. There were no differences in age, sex, percentage of older patients, BMI, smoking status, and co-medication with antipsychotics, other antidepressants, mood stabilizers, or benzodiazepines (P > 0.05) between patients with and without reported ADRs. The daily ESC doses, plasma concentrations, and concentration-to-dose ratios (C/D ratios) did not differ between patients with and without ESC-associated ADRs (P > 0.05). Psychotic disorders were more frequent in patients with ADRs than those in patients without ADRs (25% vs. 7.1%, odds ratio [OR] 4.4; 95% confidence interval [CI], 1.5–12.9, P = 0.004). There were no differences in the other ICD-10 diagnostic categories between the 2 groups (P > 0.05). The most frequently reported UKU types were other ADRs in 14 patients (40%), followed by autonomic ADRs in 12 (34.3%), psychic in 10 (28.6%), and neurological ADRs in 5 patients (14.3%; Table 2). In the multivariate analysis, none of the prespecified variables were associated with ADR risk (Table 3).

TABLE 1.

Differences in Demographic and Clinical Characteristics of Patients With and Without ADRs

	Total (N = 308)	ADRs (n = 35)	No ADRs (n = 273)	P
Age, yr, median (IQR)*	43.0 (31.0–54.0)	43.0 (33.0–54.0)	43.0 (31.0–54.0)	0.76
Sex, male, n (%)	115 (37.3)	13 (37.1)	102 (37.4)	0.98
Aged >65 yrs, n (%)	34 (11.0)	5 (14.3)	29 (10.6)	0.56
Body mass index (kg/m2), median (IQR)†	24.3 (22.0–28.2)	24.1 (23.3–28.4)	24.3 (22.0–28.2)	0.75
Smokers, n (%)‡	106 (35.5)	13 (38.2)	93 (35.1)	0.72
ESC daily dose (mg/d), median (IQR)	20.0 (10.0–20.0)	20.0 (10.0–20.0)	20.0 (10.0–20.0)	0.77
ESC plasma concentration (ng/mL), median (IQR)	20.0 (10.0–34.0)	19.0 (11.0–38.0)	20.0 (10.0–34.0)	0.85
C/D ratio ([ng/mL]/[mg/d]), median (IQR)	1.10 (0.68–1.87)	1.30 (0.75–1.90)	1.07 (0.67–1.87)	0.62
Co-medication with, n (%)
Antipsychotics	100 (32.5)	14 (40.0)	86 (31.5)	0.30
Other antidepressants	91 (29.5)	10 (28.6)	81 (29.7)	0.90
Mood stabilizers	1 (0.4)	0 (0.0)	1 (0.4)	1.00
Benzodiazepines or Z-substances	42 (13.6)	3 (8.6)	39 (14.3)	0.40
Diagnoses according to ICD-10, n (%)§
Affective disorders (F3x)	157 (75.5)	15 (62.5)	142 (77.2)	0.12
Neurotic, stress-related, somatoform disorders (F4x)	47 (22.6)	3 (12.5)	44 (23.9)	0.30
Psychotic disorders (F2x)	19 (9.1)	6 (25.0)	13 (7.1)	0.004
Other disorders (F0x, F1x, F5x, F6x, and F7x)	38 (18.3)	6 (25.0)	32 (17.4)	0.36

Bold number indicates statistically significant differences between patient groups with versus without ADRs.

^*Data were missing for 1 patient (0.3%).

^†Data were missing for 40 patients (13.0%).

^‡Data were missing for 9 patients (2.9%).

^§Data were missing for 101 patients (32.5%).

ADR, adverse drug reaction; C/D ratio, concenration to dose ratio; ESC, escitalopram; ICD, International Classification of Diseases; IQR, interquartile range.

TABLE 2.

UKU Side Effects Rating Subscales

UKU Subscales	n (%)
Psychic	10 (28.6)
Neurologic	5 (14.3)
Autonomic	12 (34.3)
Other	14 (40.0)

TABLE 3.

Odds Ratios (95% Confidence Intervals) of Escitalopram-Related Adverse Drug Reactions

	Total ADRs
Age (yr)	1.00 (0.97–1.03)
Sex (male)	1.18 (0.54–2.57)
BMI (kg/m2)	1.00 (0.93–1.07)
Smoker status	1.27 (0.58–2.77)
ESC daily dose (mg/d)	0.98 (0.91–1.05)
ESC plasma concentration (ng/mL)	1.02 (0.97–1.08)

ADRs, adverse drug reactions; BMI, body mass index; ESC, escitalopram.

Data were missing for 48 (15.6%) patients.

By stratifying patients' groups into a “high” ESC plasma concentration group with drug concentrations >40 ng/mL (n = 59) and a “low” ESC concentrations group with drug concentrations ≤40 ng/mL (n = 249), no significant difference was found between the groups in frequency of reported ADRs (13.6% vs. 10.8%, OR 1.29; 95% CI, 0.6–3.0, P = 0.56). Similarly, there was no difference in specific UKU types between the 2 comparison groups (2-tailed Fisher exact test, P > 0.05, Table 4).

TABLE 4.

Differences in Demographic and Clinical Characteristics Between Patients With Escitalopram Plasma Concentrations Above Versus Below 40 ng/mL and Above Versus Below 80 ng/mL Respectively

	ESC Level >40 ng/mL (n = 59)	ESC Level ≤40 ng/mL (n = 249)	P	ESC Level >80 ng/mL (n = 8)	ESC Level ≤80 ng/mL (n = 300)	P
Age, yr, median (IQR)*	44.0 (33.0–56.0)	42.0 (31.0–53.0)	0.27	56.0 (40.5–68.3)	43.0 (31.0–54.0)	0.042
Sex, male, n (%)	20 (33.9)	95 (38.2)	0.54	4 (50.0)	111 (37.0)	0.48
BMI (kg/m2), median (IQR)†	23.5 (21.9–26.7)	24.5 (22.1–28.7)	0.12	23.5 (22.3–28.0)	24.3 (22.0–28.3)	0.79
Smokers, n (%)‡	21 (36.2)	85 (35.3)	0.89	4 (50.0)	102 (35.1)	0.46
ADRs, n (%)	8 (13.6)	27 (10.8)	0.56	1 (12.5)	34 (11.3)	1.0
UKU side effects rating scale, n (%)
Psychic	2 (3.4)	8 (3.2)	1.0	NA	NA	NA
Neurologic	0 (0.0)	5 (2.0)	0.59	NA	NA	NA
Autonomic	4 (6.8)	8 (3.2)	0.25	NA	NA	NA
Other	5 (8.5)	9 (3.6)	0.11	NA	NA	NA

In a small number of patients there were reports of more than 1 ADRs.

^*Data were missing for 1 (0.3%) patient.

^†Data were missing for 40 (13.0%) patients.

^‡Data were missing for 9 (2.9%) patients.

ADR, adverse drug reaction; BMI, body mass index; ESC Level, escitalopram plasma concentration; IQR, interquartile range; NA, not available; UKU, Udvalg for Kliniske Undersogelser.

When comparing reported ADRs in patients with ESC concentrations >80 ng/mL (n = 8) versus with ESC concentrations ≤80 ng/mL (n = 300), there were no differences between groups regarding the frequency of reported ADRs (12.5% vs. 11.3%, OR 1.1; 95% CI, 0.1–9.4, Fisher exact test, 2-tailed, P = 1.0, Table 4).

DISCUSSION

Our findings do not suggest any clear pharmacokinetic correlates for ESC-associated ADRs, despite previous reports on the dose-dependent patterns of ESC-associated ADRs.^32,33 In fact, neither ESC concentrations nor C/D levels differed between patients with and without ADRs. Reported ADR rates neither differed between patients with “high” ESC concentrations (>40 ng/mL) versus patients with “low” ESC concentrations (≤40 ng/mL) with 40 ng/mL as the upper threshold of a recently suggested therapeutic reference range for ESC of 20–40 ng/mL by Eichentopf et al⁸ nor between patients with ESC concentrations >80 ng/mL versus patients ≤80 ng/mL as the upper threshold of the current therapeutic reference range suggested by Hiemke et al.⁵ Previous reports have suggested elevated ESC levels in patients with dry mouth, diarrhea, or drowsiness,¹⁸ whereas no pharmacokinetic differences have been reported for other ADRs such as QTc prolongation.¹⁶ Thus, the heterogeneity of the assessed ADRs and their underlying mechanisms may hamper the detection of clear pharmacokinetic patterns directly involved in ESC-associated ADRs. The most reported UKU type (40%) in our retrospective analysis was other ADRs. This UKU category included symptoms of sexual dysfunction, weight changes, different skin symptoms (rash, pruritus, and photosensitivity), and several rare ADRs, such as changes in menstrual bleeding, galactorrhea, and gynecomastia. Some of these reactions may be underpinned by a so-called “serotonin overactivity,”³⁴ whereas some emerging evidence suggested more weight gain in CYP2C19 poor/intermediate metabolizers—where drug exposure may be increased—treated with the parent compound of ESC, citalopram compared with normal metabolizers.³⁵

It may also be difficult to compare our findings with those of other reports that have studied the role of metabolic ratios or metabolite(s) in ESC.^18,19 This was because ESC metabolite assessments were not available for our TDM dataset. Other challenges in comparing our results with previous findings may lie in differences in the study samples. Patients in our group had various psychiatric diagnoses, whereas previous studies invariably assessed cohorts of patients with MDD.^16,18,19 Interestingly, rates of ESC-associated ADRs were more frequent in patients with versus without psychotic disorders (25% vs 7.1%). In this context, it could be speculated (and assuming that patients with psychotic disorders were also receiving antipsychotic treatment) that the interactions of ESC with other medications may explain this pattern. However, we excluded patients receiving medications with known strong inducing or inhibiting properties. However, if pharmacokinetic drug–drug interactions did occur, they may have affected secondary metabolic pathways of ESC, for example, CYP2D6 activity, and thus may not be reflected in changes in the concentrations of the parent compound or the main metabolite.³⁶ This may explain the lack of distinct pharmacokinetic patterns underlying ADRs, despite their more frequent reporting in patients in whom combinations of ESC with other psychotropic agents, such as antipsychotics, are presumed. Alternatively, we may consider the role of pharmacodynamic drug–drug interactions, which cannot be assessed using TDM data. Indeed, previous reports have suggested an increased risk of extrapyramidal side effects in patients who concomitantly received SSRIs.³⁷ There are also reports of extrapyramidal side effect risk even in patients who received selective serotonin reuptake inhibitor monotherapy.³⁸ The main hypothesis considers the indirect antagonizing effect of serotonin release on dopamine,³⁹ which leads to a disinhibitory effect on acetylcholine.⁴⁰ However, evidence for this interaction is still lacking and its clinical relevance remains controversial.^40,41 Additionally, in our large naturalistic sample, the frequency of neurological ADRs was the lowest (14.3%) compared with the other 3 ADRs subgroups at UKU.

Our findings, together with previous efforts to define the therapeutic reference range of ESC,⁸ support the view that the upper threshold level is best defined by maximal efficacy and not by decreased tolerability.

Interestingly, none of the potential confounders (age, sex, BMI, smoking status, daily dose, plasma concentration, or concentration-to-dose ratio) was associated with ADR risk.

Limitations

This study has several limitations that must be considered. The main limitation is the retrospective design of our study, which may imply that patient information might have been less precise compared with a prospective design. Furthermore, data on potentially relevant variables, such as clinical response, antidepressant treatment duration, comorbidities, and liver and renal function (which might have affected ESC clearance¹⁷), were not available. Although the reported ADRs were described in the context of ESC, as the majority of patients were not receiving ESC monotherapy, it is not possible to exclude a causal (or contributing) role for co-medications. Another shortcoming is the pooling of patients with different types of ADRs; using more sensitive assessment methods for ADRs may have provided different insights. Clinicians were instructed to draw plasma concentrations at the trough, potentially leading to a short lag between the report and ADR onset. When analyzing the UKU subscales separately, considerably smaller subgroups were formed, which might have been underpowered to detect relevant differences. Because of its low power, we also avoided analyses regarding the role of co-medications potentially underlying the higher ADR rate in patients with psychotic disorders compared with that in other diagnostic groups. In the future, we encourage researchers to apply standardized follow-up protocols for data collection to capture a dynamic overview of the pharmacokinetic pathways related to ADRs in common clinical scenarios. Additionally, the lack of data on ESC metabolites(s) did not allow us to gain valuable insights into the role of ESC metabolites, which had previously received some attention, yielding interesting findings.^18,19 Finally, a critical aspect refers to the unraveling of ESC-related ADRs from several mood symptoms; as we did not have clinical response data for the included patients, we could consider this information in our analysis.

CONCLUSIONS

Our retrospective analysis revealed no direct association between ESC concentrations and ESC-related ADRs. Testing different upper thresholds of the reference range for ESC did not detect a value above which the ADRs' risk substantially increased. We believe that the upper threshold reflects the ESC concentration required for maximal efficacy. Furthermore, we did not detect any clinical or demographic characteristics of patients at risk for ESC-related ADRs other than psychotic disorders, implying diagnosis-dependent tolerability patterns, potentially due to drug–drug interactions. Further studies with larger sample sizes and more focused outcomes may help identify the pharmacokinetic patterns involved in some ESC-related ADRs.