Determination of Patient Adherence for Duloxetine in Urine

Abstract

Duloxetine, known by its brand name, Cymbalta^TM, is a selective serotonin and norepinephrine reuptake inhibitor used to treat major depressive disorders. Determination of patient compliance for duloxetine is typically determined through medication possession ratio (MPR) or plasma concentrations. The purpose of this paper was to characterize normal urinary duloxetine concentrations in patients prescribed duloxetine to monitor patient adherence. Patient data collected from routine screens for duloxetine concentrations in urine were included in this study. Inclusion criteria consisted of patients who were prescribed duloxetine and (i) tested positive for duloxetine, (ii) tested negative for illicit substances and (iii) included creatinine, age and duloxetine dose administered. Of the 5,592 patient urines screened, 2,004 of the results fit into the inclusion criteria. Positive urine concentrations of duloxetine ranged from 50 to 2,722 ng/mL. Duloxetine urine concentrations were normalized to creatinine and dose further characterized by sex, age, body mass index (BMI) and dose in milligrams. Sample distribution included urines collected from 1,487 females and 517 males. The age range of the specimen donors was between 15 and 90 years old with an average age of 52. BMI levels ranged from 13.9 (underweight) to 88.1 (obese), with the average BMI being 33.5. The most common dose of duloxetine prescribed was a daily, oral dose of 60 mg. Analysis of the normalized, transformed creatinine concentrations showed that there was a significant statistical difference (P < 0.05) in the urinary duloxetine concentrations by sex and by dose (mg). Female patients further showed a statistical difference in urinary duloxetine concentration in age groups 18–64 and 64 and older. By characterizing urinary duloxetine concentrations in patients prescribed the medication, normalized distributions of data ranges have been established. These data ranges for urinary duloxetine concentrations can be used to determine patient compliance with duloxetine in routine, clinical samples.

Introduction

Duloxetine is a potent selective serotonin and norepinephrine inhibitor (SSRI/SNRI) sold under the brand name of Cymbalta^TM. Duloxetine was approved for the use in the treatment of major depressive disorders (MDDs) but is also used to treat generalized anxiety disorder, diabetic peripheral neuropathic pain, fibromyalgia and chronic musculoskeletal pain (1). Duloxetine safety and efficacy has been extensively studied. While there are case reports that detail fatal and non-fatal overdoses of duloxetine, there were only two reported deaths in association with duloxetine in 2019 and 2020, respectively (2–10). Duloxetine overdoses are usually presented in association with other drugs of abuse (7, 9–11).

The pharmacokinetics and pharmacodynamics of duloxetine have been extensively studied in humans. Enteric-coated duloxetine in a 60-mg dose has an average bioavailability of 50% (range 30–80%) (12). It is highly protein bound and is extensively metabolized by cytochrome p450 enzymes CYP2D6 and CYP2A1 (12, 13). The clearance kinetics for a 60-mg dose is 101 L/h, and the majority of duloxetine is excreted as glucuronide and or sulfate conjugates in urine (12). Age, sex and dosage have been reported as having the greatest effect on duloxetine absorption, distribution, metabolism and excretion (12, 14, 15). Women tend to have a higher systemic duloxetine concentration in blood due to lower CYP2A1 activity, and women >65 years are reported as having lower clearance kinetics with duloxetine (14, 15). Despite these observed differences, they have no bearing on clinical outcomes and do not influence recommendations for duloxetine dosing (12, 15).

Patients undergoing treatment for MDD are typically prescribed 60-mg enteric-coated capsules or higher, but the range of possible doses is 20–120 mg. Duloxetine is regarded as having a good response in the treatment of MDD over its predecessors, tricyclic antidepressants (16). Despite this, patient adherence to medical treatments in patients diagnosed with MDD and related disorders is a cause for concern (17–19). Patient adherence is defined as the extent to which a patient takes their medication as prescribed (20). Patients who do not adhere to their dosing recommendations in the treatment of MDD often relapse in their disease state in comparison with patients who adhere to their dosing regimen (18). Patient adherence is typically measured through medication possession ratio (MPR). MPR is calculated from the sum of a patient’s pill count over the number of days for a given refill period. A patient is considered adherent to their treatment plan if the MPR is greater than or equal to a ratio of 0.8. For duloxetine patients, adherence is ∼34–39% (18, 19).

There are many factors that can contribute to low patient adherence: lack of therapeutic response, patient-related factors, disease-related factors, medication-related factors, physician-related factors and psychological factors (18). Lack of therapeutic response can be determined through therapeutic drug monitoring (16), but therapeutic drug monitoring for duloxetine is reported through plasma duloxetine concentrations. Patient adherence through MPR requires access to patient prescriptions or reliance on patients accurately reporting this pill consumption. Some studies only have access to pharmacy records, further complicating matters in determining if a patient is adhering to their treatment plan (17, 18).

The purpose of this study is to characterize urinary duloxetine concentrations to determine normal, therapeutic ranges for duloxetine adherence. Urine is a clean biological matrix that requires noninvasive collection techniques compared to blood. Establishing normal, therapeutic urinary duloxetine concentrations would allow prescribers to easily determine patient adherence without bloodwork or relying on patient reports. In this study, urinary duloxetine concentrations from routine urine drug screenings were characterized by various patient factors to establish normal, therapeutic urinary duloxetine concentrations.

Methods

Analytical methods

Reference standards were purchased from Cerilliant (Round Rock, TX, USA). All solvents, including methanol (Optima grade), formic acid (88%), acetonitrile (Optima grade), ammonium acetate (Optima grade) and isopropanol (Optima grade), were purchased from VWR (Radnor, PA, USA). Drug-free human urine was acquired from UTAK Laboratories (Valencia, CA, USA).

Preparation of standards

The solid standard was initially dissolved into a stock solution of 1 mg/mL in methanol prior to use. Standards were prepared by diluting 100 µg/mL stocks into combined methanolic working stocks that were then further diluted into normal, drug-free human urine to reach the appropriate calibrator (50, 100, 250, 500, 1,000, 2,500 and 5,000 ng/mL) and quality control (250 ng/mL) concentration. An internal standard stock solution was produced by diluting duloxetine-d₃ (100 μg/mL) into 0.1% formic acid to reach a concentration of 250 ng/mL. The reporting cutoff for duloxetine was 50 ng/mL.

Sample preparation

To 50 μL of the sample (i.e., patient sample, calibrator or quality control), 450 μL of the internal standard solution was added to the sample vial or 96-well plate position. Vials or plates were then capped or sealed and vortexed for 10 s. Plates and vials were placed in the autosampler of the liquid chromatograph with tandem mass spectrometer (LC–MS-MS) for analysis.

LC–MS-MS conditions

Samples were analyzed by LC–MS-MS on an Agilent system comprised of a 1290 series liquid chromatography pump system with a micro-well plate autosampler connected to a 6460 triple quadrupole mass spectrometer. Chromatographic separation was performed using a Kinetex^® Phenyl-Hexyl, 2.1 × 50 mm, 2.6 μm column (Phenomenex, Torrance, CA, USA, Part No. 00B-4495-AN). Mobile phase A was 0.1% formic acid in 10 mM ammonium formate, and mobile phase B was 0.1% formic acid in 50:50 (v/v) in acetonitrile:methanol. The following gradient conditions were used: initial mobile phase conditions were 20% mobile phase B, which increased to 50% at 2.1 min. The gradient was increased to 95% mobile phase B at 2.11 min and was then returned to 20% mobile phase B at 2.3 min. The total run time was 2.4 min. The injection volume was 2 μL. The acquisition method was run in dynamic multiple reaction monitoring, and the analyte transitions were as followed: duloxetine, m/z 298 > 44 and m/z 298 > 154; duloxetine-d₃, m/z 301 > 47 and m/z 301 > 157. The validation of this method followed College of American Pathologists (CAP) and Clinical Laboratory Improvements Amendments (CLIA) guidelines. The validation protocol was described in detail by Enders and McIntire (21).

Patient criteria

All specimens that were used in this analysis were de-identified. Ameritox is accredited by the CAP and abides by CAP, CLIA and Health Insurance Portability and Accountability Act requirements. The use of the results of the de-identified samples for this study was approved as qualified for exemption by the Virginia Commonwealth University Institutional Review Board by the US Code of Federal Regulations (45 CFR 46.102).

The de-identified patient urine sample results used in this study were collected at single time points over a 16-month period from routine urine drug screens for duloxetine. The specimens were generally received and analyzed within 72 h of the collection as part of a therapeutic mental health medication monitoring programs, with sample collection primarily taken within 24 h of the last dose. In addition to duloxetine, the urine samples were screened for fluoxetine, mirtazapine, norfluoxetine, paroxetine and trazodone. Sample validity tests were also performed and included creatinine, specific gravity, nitrite, chromate and pH.

The results included duloxetine urine concentrations as well as specifications about the patient such as sex, age, height, weight and prescribed duloxetine dosage (mg). The inclusion criteria for the samples were as follows:

1. Only patients who were prescribed duloxetine and tested positive for duloxetine were included.

2. Patients who were positive for illicit drugs or had invalid specimens were excluded.

3. Patients who failed sample validity tests (e.g., pH, creatinine and specific gravity) were excluded.

4. Patients who did not have age or dosage (mg) associated with their duloxetine concentrations were excluded.

5. Duplicate patient samples were excluded.

Urinary duloxetine results were broken down by sex, age, body mass index (BMI) and duloxetine dose (mg). Age classifications were broken down as <18, 18–64 and >64 years. BMI ranges were classified as <18.5 (underweight), 18.5–24.9 (normal weight), 25–29.9 (overweight) and > 30 (obesity).

Statistical analyses

Statistical testing was carried out with JMP version 14 software. Descriptive statistics for the normalized and transformed patient samples were expressed as mean and SD. Urinary duloxetine concentrations were normalized to creatinine (mg) and dose (mg) with the following equation:

(1)

The distribution of normalized, urinary duloxetine concentrations (ng/mL duloxetine/mg creatinine/mg duloxetine dose) was heavily right-skewed, requiring logarithmic transformation of the data to achieve normality, as shown in the following equation:

(2)

Normality was determined by reviewing histograms for Gaussian distributions and Q–Q plots. Normalized and transformed samples that showed a normal distribution were further assessed with the Brown–Forsythe test for equal variances. Acceptable results for equal variances were determined with P-values >0.05 (P > 0.05). The means of sample sets that had equal variances were further assessed with a pooled t-test or an analysis of variance (ANOVA) for statistical differences. Samples were considered statistically different if the P-value was <0.05 (P < 0.05). If a sample set had a statistically different result, the means were further compared with a Tukey’s Honest Significant Difference (HSD) test to compare means between groups. In cases where a sample was not normally distributed, the non-parametric chi-squared or Kruskil–Wallis test was used for statistical analyses.

Results

Of the 5,592 patient urines screened, 2,004 of the patient urines fit into the inclusion criteria. Sample distribution included urines collected from 1,487 females and 517 males. The prescribed dose of duloxetine was 20–120 mg taken orally, once or twice daily. The average and most frequent dose was 60 mg, taken orally, once daily (N = 1,593). Positive urine concentrations of urinary duloxetine ranged from 50 to 2,722 ng/mL. The average age range of specimen donors was between 15 and 90 years old, with an average age of 52. The BMI ranged from 13.9 to 88.1, with an average BMI of 33.5 (N = 1,855).

Table I summarizes the distribution of the normalized, transformed patient urinary duloxetine concentrations. Statistical analysis of patient sample results was performed by normalizing the data to creatinine and dose, as described in Equation (1). A log transformation was then performed to create a Gaussian distribution, as described in Equation (2) (Figure 1). Histograms and Q–Q plots showed no departures from normality. Using the Brown–Forsythe test, all sample groups were determined to have equal variances (P > 0.05). Therefore, pooled t-tests and ANOVA were used for all statistical test models. Results of the statistical tests and interpretations are summarized in Tables II and III. Analysis of the duloxetine concentrations showed that there was a statistically significant difference (P < 0.05) in urinary duloxetine concentrations between sexes (P < 0.0001). There were only two patients under the age of 18 for both sexes. Because of the low sample volume (N < 3), normalized, transformed urinary duloxetine concentrations were not subjected to statistical analysis. Female patients aged 18 and older, however, showed a statistical difference (P = 0.0239) in urinary duloxetine concentrations in age groups 18–64 and 64 and older (Table II). For both sexes, there was a statistically significant difference in normalized, transformed urinary duloxetine concentrations by prescribed duloxetine dose (P < 0.0001), with the exception of 90 vs 120 mg doses (P = 0.8240). Female patients showed a statistical difference at all dosing concentrations, with the exception of 90 vs 120 mg (P = 0.9290). Male patients showed a statistical difference at all dosing concentrations, with the exception of 20 vs 30 mg (P = 0.5995), 60 vs 90 mg (P = 0.0626) and 90 vs 120 mg (P = 0.9096).

Table I.

Summary of Urinary Duloxetine Concentrations in Male and Female Patients^a

		Female		Male
			Log[Norm. Duloxetine]		Log[Norm. Duloxetine]
		N	(average ± SD)	N	(average ± SD)
Age (years)	<18	2	4.54 ± 1.23	2	4.28 ± 1.24
	18–64	1,325	4.75 ± 1.12	473	4.35 ± 1.01
	>64	160	4.96 ± 1.04	42	4.57 ± 0.96
BMI	<18.5	12	4.64 ± 0.74	6	4.36 ± 0.52
	18.5–24.9	175	4.64 ± 1.11	55	4.20 ± 1.09
	25–29.9	337	4.81 ± 1.07	135	4.48 ± 0.94
	>29.9	847	4.75 ± 1.03	288	4.33 ± 1.02
Dose (mg)	20	27	3.18 ± 1.09	9	3.04 ± 0.79
	30	196	3.92 ± 0.98	63	3.52 ± 0.97
	60	1,176	4.87 ± 0.96	417	4.47 ± 0.92
	90	62	5.47 ± 0.98	19	5.05 ± 0.81
	120	26	5.65 ± 0.91	8	5.37 ± 0.86

^aConcentrations are normalized to creatinine and dose and logarithmically transformed (log[Norm. Duloxetine]).

Figure 1.

Histogram and box and whisker plots for (A) all patient urinary duloxetine concentration, normalized to creatinine and dose (ng/mL urinary duloxetine/mg creatinine/mg duloxetine dose) and (B) the normalized duloxetine after logarithmic transformation. Box and whisker plots represent the median/mean value, second and third quartiles, and minimum and maximum values. Logarithmically transformed data created a Gaussian distribution, from which normal patient urinary duloxetine values could be established.

Table III.

Post Hoc, Tukey’s HSD Analysis of Patient’s Urinary Duloxetine Concentrations by Dose (ng/mL Urinary Duloxetine/mg Creatinine/mg Duloxetine Dose)^a

Comparison group	Both sexes P-value	Female P-value	Male P-value
20 vs 30	0.0010b	0.0023b	0.5995
20 vs 60	<0.0001b	<0.0001b	<0.0001b
20 vs 90	<0.0001b	<0.0001b	<0.0001b
20 vs 120	<0.0001b	<0.0001b	<0.0001b
30 vs 60	<0.0001b	<0.0001b	<0.0001b
30 vs 90	<0.0001b	<0.0001b	<0.0001b
30 vs 120	<0.0001b	<0.0001b	<0.0001b
60 vs 90	<0.0001b	<0.0001b	0.0626
60 vs 120	<0.0001b	<0.001b	0.0330b
90 vs 120	0.8240	0.9290	0.9096

^aConcentrations are normalized to creatinine and dose and logarithmically transformed.

^bSignificant difference between the two doses was concluded if the P-value was <0.05.

Table II.

Statistical Tests and Results of Patient’s Urinary Duloxetine Concentrations (ng/mL Urinary Duloxetine/mg Creatinine/mg Duloxetine Dose)^a

	Category	Test	P-value
	By Sex	Pooled t-test	<0.0001b
	By dose	ANOVA	<0.0001b
Female	By age	Pooled t-test	0.0239b
	By BMI	ANOVA	0.3211
	By dose (mg)	ANOVA	<0.0001b,c
Male	By age	Pooled t-test	0.1640
	By BMI	ANOVA	0.2948
	By dose (mg)	ANOVA	<0.0001b,c

^aConcentrations are normalized to creatinine and dose and logarithmically transformed (Log[Norm. Duloxetine]).

^bSignificant difference was concluded if the P-value was <0.05.

^cIn cases of significant differences, the multiple comparison test, Tukey’s HSD test, was conducted to determine between which groups the statistical difference was occurring.

Discussion

The purpose of this project was to characterize normal urinary duloxetine concentrations in patients prescribed duloxetine to monitor patient adherence. Traditional methods currently use MPR or duloxetine plasma levels to monitor patient adherence and therapeutic drug monitoring (16–19). The use of urine as a routine, noninvasive method for monitoring duloxetine levels in patients can be used alongside MPR data to help determine patient compliance. To our knowledge, there are no published studies that use urinary duloxetine concentrations to assist in monitoring patient compliance.

Duloxetine pharmacokinetics has been extensively studied in humans. Pharmacokinetic profiles of duloxetine have shown that increasing dose reduces clearance kinetics, resulting in an increase in duloxetine plasma concentration in humans (12). This was reflected in the presented data that showed an increase in normalized, transformed urinary duloxetine concentrations with increasing doses (Table I). Statistical analyses showed that there was a significant difference (P < 0.05) in urinary duloxetine concentration between all dosage levels (Figure 3). Overlap between concentration ranges was observed. This is anticipated for two reasons: the first is that individual patients will have biological variations in metabolism and excretion of duloxetine. The second is that, despite normalizing to creatinine, the samples collected for this study were done at random, single time points that did not take into account factors such as time since the last dose, last urination before next dose, etc. The male population in this study had no statistical difference in normalized, transformed urinary duloxetine concentrations at 90 and 120 mg, and at 60 and 90 mg (Table III). It is possible that the increasing duloxetine concentration plateaus earlier in males than in females. Similar to other research studies, body weight has no significant effect on the pharmacokinetics of duloxetine in patients. This was reflected in urine as there was no statistical difference (P > 0.05) in urinary duloxetine concentrations when classifying them by BMI.

Figure 3.

Box and whisker plots for all normalized, transformed patient urinary duloxetine values (log[Norm. Duloxetine]). Median/mean, second and third quartiles, and minimum and maximum values for each group, including sample size (N), are reported. Post hoc analysis with Tukey’s HSD showed a statistically significant difference (P < 0.05) in all duloxetine values, with the exception of 90 vs 120 mg (P = 0.8240).

The normalized, transformed urinary duloxetine concentrations are relative to plasma duloxetine concentrations. Pharmacokinetic studies show that duloxetine is primarily metabolized through enzymes CYP2A1 and CYP2D6 with CYP2A1 being the leading enzyme for the metabolism of duloxetine (12). CYP2A1 activity is reported as being lower in women than in men, resulting in higher duloxetine plasma concentrations in women. Women had an average normalized, transformed duloxetine value of 4.74 ± 1.05, compared with men that had an average normalized, transformed duloxetine value of 4.38 ± 1.01 (Figure 2). While there was some overlap in the distribution of normalized, transformed urinary concentration, as seen in the SD of the box and whisker plots (Figure 2), there was a statistically significant difference (P < 0.05) in the normalized, transformed urinary duloxetine concentration between sexes (Table II).

Figure 2.

Box and whisker plots of male and female normalized, transformed urinary duloxetine values (log[Norm. Duloxetine]). Median/mean, second and third quartiles, and minimum and maximum values for each group, including sample size (N), are reported. Male and females had a statistically significant difference (P < 0.05) in mean, normalized and transformed duloxetine values.

The statistical difference of urinary duloxetine concentrations in women above the age of 64 is consistent with the decreased clearance kinetics of duloxetine in women in this age range (15). Pharmacokinetic data from previous research have indicated a decrease in clearance concentrations due to a change in clearance kinetics for women above the age of 65 (14, 15). This change in clearance kinetics was reflected in the urinary duloxetine concentrations of women above the age of 64. Women over the age of 64 had an average, normalized and transformed duloxetine value of 4.96 ± 1.04, and women between the ages of 18 and 64 had an average, normalized and transformed urinary duloxetine value of 4.75 ± 1.12. Similar to the differences in sex, there was overlap between the two group ranges (Table I), but there was a statistically significant difference (P < 0.05) in the age ranges (Table II). The reduction in clearance for the older population is believed to result from aging factors such as slower hepatic metabolism and CYP enzyme activity (15). While studies have noted the differences in sex and age in regard to duloxetine plasma concentrations, they do not have any bearing on the patient dosing recommendations. The reflection of these data trends in urine may help to further establish criteria for patient adherence and explain the differences in sex and age populations.

While urinary duloxetine concentrations can be useful to determine patient adherence, there are some limitations. Individual patients have many variables (e.g., race, enzyme activity, hepatic clearance and renal function) that can create wide variations in individual duloxetine pharmacokinetics and clinical outcomes. This data set provides >1,000 patient samples to create a good foundation for establishing normal ranges of patients who are adhering to their treatment plan. However, physicians should be aware of an individual’s urinary duloxetine profile to prevent misinterpretations of data. Furthermore, traditional limitations of urine results should be applied when assessing the data. Drug presence in urine is not correlated with therapeutic concentrations in plasma and only indicates recent exposure and not long-term exposure (22). As these were single time point, random urine specimens, the extent of full pharmacokinetic profiles cannot be determined from this study. Despite these limitations, single time point urinary duloxetine concentrations can be used to help assess current patient adherence.

Conclusions

By characterizing urinary duloxetine concentrations in patients prescribed the medication, normalized, transformed distributions of data ranges have been established. Patients positive for urinary duloxetine had normalized, transformed values between 1.53 and 7.53. The normalized, transformed data are significantly different depending on sex, age and dosing amount. With MPR, these data ranges for urinary duloxetine concentrations can be used to determine patient compliance with duloxetine in routine, clinical samples. By ensuring that patients are adhering to their treatment plan, they have less risk of relapse into their disease state and can have improved medical outcomes.

Funding

This work was supported in part by the National Institutes of Health (P30 DA033934).

Data availability

The data underlying this article were provided by Ameritox, LLC under permission. Data will be shared on request to the corresponding author with the permission of Ameritox.