Volume 65, Issue 4 p. 558-564
Free Access

Influence of the CYP2D6*4 polymorphism on dose, switching and discontinuation of antidepressants

Monique J. Bijl

Monique J. Bijl

Department of Epidemiology & Biostatistics,

Department of Hospital Pharmacy,

Search for more papers by this author
Loes E. Visser

Loes E. Visser

Department of Epidemiology & Biostatistics,

Department of Hospital Pharmacy,

Search for more papers by this author
Albert Hofman

Albert Hofman

Department of Epidemiology & Biostatistics,

Search for more papers by this author
Arnold G. Vulto

Arnold G. Vulto

Department of Hospital Pharmacy,

Search for more papers by this author
Teun Van Gelder

Teun Van Gelder

Department of Hospital Pharmacy,

Department of Internal Medicine, Erasmus MC, PO Box 2040, 3000 CA, Rotterdam, the Netherlands,

Search for more papers by this author
Bruno H. Ch. Stricker

Corresponding Author

Bruno H. Ch. Stricker

Department of Epidemiology & Biostatistics,

Department of Internal Medicine, Erasmus MC, PO Box 2040, 3000 CA, Rotterdam, the Netherlands,

Drug Safety Unit, Inspectorate for Health Care, PO Box 16119, 2500 BC, The Hague, the Netherlands and

Dr Bruno H.Ch. Stricker, Erasmus MC, Department of Epidemiology & Biostatistics, PO Box 2040, 3000 CA, Rotterdam, the Netherlands.
Tel.: + 31 7 0340 6793;
Fax: + 31 7 0340 6707;
E-mail: [email protected]Search for more papers by this author
Ron H. N. Van Schaik

Ron H. N. Van Schaik

Department of Clinical Chemistry, Erasmus MC, PO Box 2040, 3000 CA, Rotterdam, the Netherlands

Search for more papers by this author
First published: 07 December 2007
Citations: 90

Abstract

What is already known about this subject

• Most antidepressants are metabolized by CYP2D6. The variant allele CYP2D6*4 is the main polymorphism resulting in reduced enzyme activity in Caucasians.

• Reduced enzyme activity potentially leads to increased toxicity of antidepressants, but the relevance of genotyping for clinical practice is unclear. Most clinical studies suffer from small numbers of patients.

What this study adds

• This large population-based cohort study in 1198 elderly Dutch patients examines the influence of the CYP2D6*4 polymorphism on intolerability of antidepressants.

• The risk of switching to another antidepressant in tricyclic antidepressant users is higher in poor metabolizers (PMs), but not in SSRI users. PMs require a lower maintenance dose of antidepressants compared with extensive metabolizers (EMs).

• Antidepressants were initiated in a relatively low dose, with gradual dose increments thereafter, reducing the risk of adverse drug reactions. Therefore, the question remains whether genotyping prior to the start of antidepressant therapy contributes substantially to the optimization of pharmacotherapy.

Aims

To study the effect of CYP2D6*4 on antidepressant dose, switching and discontinuation of therapy.

Methods

The study consisted of all subjects in the Rotterdam Study, who received a first antidepressant prescription between April 1st 1991 and July 1st 2005 and for whom data on CYP2D6 genotype were available. Binary logistic regression was performed to study the association between CYP2D6*4 and switching to any other antidepressant or discontinuation of therapy within 45 days. The difference in mean antidepressant dose was compared between CYP2D6 genotypes using t-tests and repeated measurements analyses.

Results

In users of tricyclic antidepressants (TCAs) the risk of switching to another antidepressant was significantly higher in poor metabolizers (PMs:*4/*4) compared with extensive metabolizers (EMs:*1/*1), with an adjusted OR of 5.77 (95% CI 1.59, 21.03; P = 0.01). In SSRI users there was no significant difference (OR 0.91; 95% CI 0.20, 4.15; P = 0.90). Heterozygous patients did not have an increased risk of switching in both TCA and SSRI users. The mean TCA dose was significantly lower in PMs than in EMs at the third and fourth prescription (difference 0.11 DDD, P = 0.03). In SSRI users the difference in mean dose between PMs and EMs was significant at the third prescription (0.17 DDD; P = 0.02).

Conclusions

The risk of switching to another antidepressant in TCA users is higher in PMs than in EMs. The maintenance doses of antidepressants were significantly lower in PMs. However, the question whether genotyping prior to the start of antidepressant therapy contributes substantially to the optimization of pharmacotherapy, requires further study.

Introduction

Depression constitutes a major health problem in the elderly. For the treatment of major depression tricyclic antidepressants (TCAs), selective serotonin reuptake inhibitors (SSRIs) and other antidepressants are widely prescribed. However, 50% of patients will not respond adequately to first treatment [1]. This low response rate can be explained by a large interindividual variability in genetic, environmental and pathophysiological factors. In pharmacogenetics, the influence of genes on the pharmacokinetics and pharmacodynamics of drugs is investigated [2].

Most antidepressants are metabolized by cytochrome P450 2D6 (CYP2D6). CYP2D6 accounts for a small percentage of all hepatic CYPs, but is responsible for the metabolism of approximately 25% of all drugs metabolized by CYPs [3]. The CYP2D6 gene is highly polymorphic with more than 60 variant alleles [http://www.cypalleles.ki.se]. Several of these variants encode an inactive protein or no enzyme product (e.g. *3, *4, *5, *6). Subjects with two nonfunctional alleles are classified as ‘poor metabolizers’ (PMs), while carriers of one or two functional alleles (*1, *2) are classified as ‘extensive metabolizers’ (EMs). Approximately 5–10% of the Caucasian population are PMs [4–6]. Subjects with one nonfunctional and one functional allele can also be considered as ‘intermediate metabolizers’ (IMs), although this term also refers to a subject with one nonfunctional allele and one decreased activity allele or two decreased activity alleles (e.g. *10, *41). The clinical impact of the IM phenotype is unclear, probably as a result of diversity in genotypes and may depend on the drug used. ‘Ultra-rapid metabolizers’ (UMs) have >2 functional copies of the CYP2D6 gene and exhibit extremely high enzyme activity. Many genotyping assays determine the duplication of any CYP2D6 gene, including nonfunctional genes, leading to false positive UM assignment. In this way, genotyping will only detect 10–30% of CYP2D6 UMs [7].

CYP2D6*4 is the most common variant allele (allele frequency of 20%) in Caucasians and is the most frequent nonfunctional allele in the PM phenotype; over 75% of the PMs are carriers of this polymorphism [4].

Due to absent CYP2D6-mediated metabolism, PMs have higher plasma concentrations of antidepressants metabolized by CYP2D6 than EMs [8] and are therefore more likely to suffer from dose-dependent adverse drug reactions (ADRs). Especially in patients taking TCAs, PMs may experience cardiotoxicity and other severe ADRs, because TCAs have a narrow therapeutic range. Severe ADRs require dose reductions or discontinuation of antidepressant therapy. In patients receiving a fixed dose of amitriptyline 75 mg twice a day, patients with one nonfunctional CYP2D6 allele had a higher risk of adverse drug reactions than patients with two functional alleles (76.5% vs. 12.1%) [9]. Other studies showed an increased frequency of ADRs in CYP2D6 PMs using antidepressants primarily metabolized by CYP2D6, but these are of limited value due to the small number of patients [10, 11]. In contrast, the CYP2D6 genotype had no effect on paroxetine and mirtazapine discontinuations and adverse events in an 8-week, double-blind randomized study on antidepressant intolerance [12]. Consequently, the actual influence of CYP2D6 polymorphisms on adverse events and clinical outcomes remains unclear.

Therefore, we performed a population-based cohort study to examine the influence of the CYP2D6*4 polymorphism on intolerability of antidepressants by studying dose, frequency of switching to another antidepressant or discontinuation of therapy.

Methods

Setting

The Rotterdam Study is a prospective population-based cohort study that investigates the incidence and risk factors of cardiovascular, neurodegenerative, locomotor and ophthalmologic diseases in the elderly. From 1990 to 1993, all inhabitants of Ommoord, a district of the city of Rotterdam in the Netherlands, aged 55 years or over, were invited to participate. The rationale and design of this study have been described elsewhere [13]. The Medical Ethics Committee of the Erasmus Medical Center approved the study and written informed consent was obtained from all participants. The cohort encompasses 7983 individuals who were all interviewed and investigated at baseline. Since the start of the study, follow-up examinations were conducted periodically. In addition, the total cohort is continuously monitored for major morbidity and mortality through linkage with the records of the patient's general practioner. More than 99% of the participants have their drug prescriptions filled at seven regional pharmacies, which are all fully computerized. Complete data on drug use are available as of January 1, 1991. The pharmacy data include the Anatomical Therapeutical Chemical (ATC)-code, the dispensing date, the total amount of drug units per prescription, the prescribed daily number of units, and product name of the drugs.

Cohort definition

The study cohort consisted of all subjects in the Rotterdam Study, who received a first prescription of an antidepressant between April 1st 1991 and July 1st 2005, and for whom there were data on CYP2D6 genotype available (n = 1198). The start date of April 1, 1991 was chosen to exclude patients who were treated with antidepressants in the preceding 3 months. Subjects were followed until one of the outcomes, death or the end of study period on July 1st 2005, whichever came first.

Outcome definition

In this study three types of outcomes were used: switching, discontinuation and dose. Switching was defined as a switch to any other antidepressant, irrespective of class, within 45 days after the start of the first prescription. A switch within 45 days was assumed to occur due to intolerance of the drug, since the efficacy of an antidepressant can only be assessed after at least 6 weeks of therapy. Secondly, we looked at discontinuation of antidepressant therapy within 45 days. Discontinuation of therapy was defined as no further prescriptions for that particular drug after the initial 45 days. Thirdly, the influence of the CYP2D6 genotype on the mean antidepressant dose was analyzed. To compare doses of different antidepressants between genotypes the prescribed daily dose (PDD) was divided by the defined daily dose (DDD), according to the World Health Organization. For each prescription the mean PDD/DDD ratio was calculated for the tricyclic antidepressants (TCAs) amitriptyline, nortriptyline, imipramine, clomipramine, opipramol, doxepin, dosulepin, maprotiline and for the selective serotonin reuptake inhibitors (SSRIs) fluoxetine, paroxetine, fluvoxamine, citalopram, escitalopram and sertraline.

Covariates

The following baseline covariates were considered as potential confounders: age, gender, body mass index (BMI) and renal function. BMI was calculated by dividing weight (kg) by height (m2). Renal function was estimated using the formula of Cockroft & Gault [14]. In addition, the starting dose of an antidepressant was considered a determinant affecting the risk of switching and discontinuation.

Genotyping

At the baseline examination of the Rotterdam Study, blood was taken from which DNA was isolated. The CYP2D6*4 (1846G > A) genotyping was done using Taqman allelic discrimination assays on the ABI Prism 9700 HT Sequence detection system. Primers and probes were designed by Applied Biosystems by their Assay-by-Design service. Polymerase chain reactions (PCR) were performed in a reaction volume of 2.0 μl, containing assay-specific primers, allele-specific Taqman MGB probes, Abgene Absolute QPCR Rox Mix and genomic DNA (1 ng). The thermal profile consisted of an initial denaturation step at 95°C for 15 min, followed by 40 cycles of denaturation at 92°C for 15 s and annealing and extension at 60°C for 1 min. Genotypes were scored by measuring allele-specific fluorescence using the SDS 2.2.2 software for allelic discrimination (Applied Biosystems).

Subjects were defined as PM if they were homozygous for the *4 allele. In case of heterozygosity the subjects were defined as IMs. When the *4 allele was absent, subjects were classified as EMs. Our method was unable to discriminate between *4/*4 (PMs) and *4/*5 (PMs) individuals, since the gene deletion *5 will result in no PCR product. Likewise, *1/*5 individuals will be genotyped as *1/*1. However, this does not affect the distribution of PMs and non-poor metabolizers.

Statistical analysis

Genotype frequency was tested for deviations from Hardy–Weinberg equilibrium by using a χ2-test. Binary logistic regression analysis was used to analyze the association between CYP2D6 genotype and switching and the association between genotype and discontinuation of therapy. Confounders were defined as covariates associated with the outcome at a P value of 0.1 in the univariate analysis and if they changed the point estimate by 10% or more in the multivariate model in addition to age and gender. Analyses were carried out in all antidepressant users, and separately for the different antidepressants classes (TCAs and SSRIs).

To compare the mean PDD/DDD ratio between EMs, IMs and PMs independent sample t-tests were used at consecutive prescriptions. In addition, repeated measurements analysis was performed on the mean PDD/DDD ratio between EMs, IMs and PMs in series of consecutive prescriptions to adjust for dependency of observations within each patient and to adjust for potential confounders. Logistic regression analysis and t-tests were performed with SPSS for Windows, version 11.0. Repeated measurement analysis was performed with SAS, version 8.2, using the Proc Mixed program.

Results

The baseline characteristics of the study population are shown in Table 1. The mean age was approximately 69 years and 68% were women. Eight hundred and seven patients used TCAs, 833 used SSRIs and 213 patients used other antidepressants at any time during the study period. Of the TCAs amitriptyline was the most frequently used drug (68.3%), paroxetine was the most frequently used drug in the SSRI-group (46.8%) and mirtazapine was the most frequently used drug in the group of other antidepressants (34.3%). In Table 2 the frequencies of antidepressants used and extent of CYP2D6 metabolism are given.

Table 1.
Baseline characteristics of the study population
Variable Number of patients (%)
n = 1198
Age, average (SD) 69.4 (8.2) years
Gender
 Male 378 (31.6)
 Female 820 (68.4)
BMI (SD) n = 1155 26.5 (3.9) kg m−2
Serum creatinine (SD) n = 921 82.2 (20.1) µmol l−1
CYP2D6 genotype*
*1/*1 777 (64.9)
*1/*4 341 (28.5)
*4/*4 80 (6.7)
  • * Hardy–Weinberg equilibrium; χ 2  = 23.285 (P = 0.00001).
Table 2.
Frequencies and types of antidepressants used during the study period (1991–2005) and the extent of CYP2D6 metabolism for each antidepressant
Antidepressant Number of patients (%)
n = 1198
Extent of
CYP2D6 metabolism *
TCA use
 Amitriptyline 551 (45.9) ++
 Maprotiline 99 (8.2) +++
 Clomipramine 79 (6.6) ++
 Nortriptyline 35 (2.9) +++
 Imipramine 29 (2.4) ++
 Dosulepin 8 (0.7)
 Doxepin 4 (0.3) ++
 Opipramol 2 (0.2)
SSRI use
 Paroxetine 390 (32.5) +++
 Fluvoxamine 161 (13.4) ++
 Fluoxetine 139 (11.6) +++
 Sertraline 87 (7.3)
 Citalopram 55 (4.6) –/+
 Escitalopram 1 (0.1)
Other
 Mirtazapine 73 (6.1) ++
 Mianserine 70 (5.8) ++
 Venlafaxine 41 (3.4) ++
 Trazodon 29 (2.4)
 Moclobemide 9 (0.8)
 Nefazodon 1 (0.1)
  • Some patients used more than one antidepressant during the study period.
  • * Extent of CYP2D6 metabolism [20, 22]. + minor metabolism route; ++ partly metabolized by CYP2D6; +++ major metabolism route; – no CYP2D6 metabolism.

Genotype distributions of CYP2D6*4 are given in Table 1. The allele frequency of the CYP2D6*4 polymorphism in our population was 20.8%. We identified 777 patients (64.9%) with the wild type genotype (EMs), 341 patients (28.5%) were heterozygous (IMs), and 80 patients (6.7%) were homozygous for the *4 allele (PMs). Genotype frequencies significantly deviated from the Hardy–Weinberg equilibrium.

Only approximately 4% of the antidepressant users switched to another antidepressant within 45 days. Table 3 shows the association between CYP2D6 genotype and switching. Overall PMs had a higher risk of switching to another antidepressant than EMs, although this difference was not statistically significant. In TCA users the risk of switching was significantly higher in PMs compared with EMs (OR = 5.77; 95% CI 1.59, 21.03; P = 0.01). This effect was not seen in SSRI users. An increased risk of discontinuation of therapy was seen in PMs compared with EMs (OR = 1.45; 95% CI 0.91, 2.32; P = 0.12), but this difference was not statistically significant (Table 4). This association was somewhat more explicit in TCA users (OR = 1.62; 95% CI 0.84, 3.12; P = 0.15) than in SSRI users (OR = 1.20; 95% CI 0.56, 2.57; P = 0.65).

Table 3.
Association between CYP2D6 genotype and switching to any other antidepressant <45 days
CYP2D6 genotype Number of switchers ORadj* switching (95% CI) P value
All antidepressants
*1/*1 33 ref
*1/*4 13 0.90 (0.47, 1.73) 0.74
*4/*4 6 1.80 (0.72, 4.48) 0.20
TCAs
*1/*1 11 ref
*1/*4 4 0.84 (0.26, 2.70) 0.77
*4/*4 4 5.77 (1.59, 21.03) 0.01
SSRIs
*1/*1 19 ref
*1/*4 9 1.09 (0.48, 2.49) 0.84
*4/*4 2 0.91 (0.20, 4.15) 0.90
  • * Adjusted for age and gender.
Table 4.
Association between CYP2D6 genotype and discontinuation of antidepressive therapy <45 days
CYP2D6 genotype Number of discontinuations ORadj* discontinuation (95% CI) P value
All antidepressants
*1/*1 285 ref
*1/*4 135 1.13 (0.87, 1.47) 0.36
*4/*4 36 1.45 (0.91, 2.32) 0.12
TCAs
*1/*1 144 ref
*1/*4 74 1.33 (0.92, 1.90) 0.13
*4/*4 19 1.62 (0.84, 3.12) 0.15
SSRIs
*1/*1 118 ref
*1/*4 51 0.96 (0.63, 1.45) 0.83
*4/*4 13 1.20 (0.56, 2.57) 0.65
  • * Adjusted for age and gender.

The change in mean dose (PDD/DDD ratio) over time for TCAs is given in Figure 1. The average TCA starting dose was 0.36 DDD. At the first prescription the mean doses did not significantly differ between CYP2D6 genotypes. For all genotypes the prescribed daily dose increased over the first four prescriptions although this increase was smaller in PMs and IMs. At the third and fourth prescription the mean dose was significantly lower in PMs than in EMs (difference 0.11 DDD, P = 0.03). Additionally, we analyzed the difference in TCA doses between genotypes with a repeated measurements analysis. In PMs the adjusted difference in PDD/DDD ratio was 0.05 DDD compared with EMs (95% CI −0.07, 0.16, P = 0.46). The mean dose of the IMs was lower compared with EMs (difference 0.02 DDD, 95% CI −0.03, 0.08, P = 0.43). In the repeated measurements analyses we adjusted for age, gender and starting dose. Adjusting for renal function and BMI did not significantly affect the results.

Details are in the caption following the image


Change in mean tricyclic antidepressant dose over time per genotype (*1/*1, (); *1/*4, (inline image); *4/*4, (inline image)). For each prescription the mean prescribed daily dose (PDD)/defined daily dose (DDD) ratio was calculated. ⊥ 95% confidence interval of the mean PDD/DDD ratio of CYP2D6 PMs. *P value < 0.05

For SSRIs the effect of CYP2D6 genotype on mean dose (PDD/DDD ratio) over time is given in Figure 2. The average starting dose was 0.84 DDD. The starting doses of EM, IM or PM patients did not significantly differ. In SSRI users the difference in mean dose between PMs and EMs was significant at the third prescription (difference 0.17 DDD; P = 0.02), but not significant for the following prescriptions. The curve of the IMs overlapped the EM mean dose to a large extent. With repeated measurements analysis the adjusted difference was 0.08 DDD (95% CI −0.02, 0.20; P = 0.11) between PMs and EMs and 0.13 DDD (95% CI −0.01, 0.24; P = 0.04) between PMs and IMs.

Details are in the caption following the image


Change in mean SSRI dose over time per genotype (*1/*1, (); *1/*4, (inline image); *4/*4, (inline image)). For each prescription the mean prescribed daily dose (PDD)/defined daily dose (DDD) ratio was calculated. ⊥ 95% confidence interval of the mean PDD/DDD ratio of CYP2D6 PMs. *P value < 0.05

Discussion

This study demonstrated that in users of TCA the risk of switching to any other antidepressant within 45 days is significantly higher in PMs than in EMs. These findings are in concordance with Mulder et al.[15] who found a HR of 3.50 (95% CI 1.52, 8.10) for switching to another drug in the same therapeutic class in PMs vs. EMs. In their study switching could be seen as an overall expression of unsatisfactory response to treatment including ineffectiveness and adverse drug reactions. We studied switching within 45 days of antidepressant use, since after a period of 6 weeks ineffectiveness could be the reason for switching instead of adverse drug reactions. The increased risk of switching to another antidepressant was not seen in SSRI users. Selection of SSRIs primarily metabolized by CYP2D6 (fluoxetine, paroxetine, fluvoxamine) did not alter these results. In contrast to TCAs, for which a narrow therapeutic range exists, no clear relationship between clinical effect and plasma concentration has been found for SSRIs. ADRs seemed not to be associated with high plasma concentrations of SSRIs [16]. Therefore, higher plasma concentrations of SSRIs in PMs would not lead to an increased frequency of switching. In IMs the risk of switching to any other antidepressant was not increased. Unlike PMs, in whom CYP2D6 enzyme activity is absent, patients heterozygous for the *4 allele have decreased enzyme activity. Plasma concentrations of antidepressants in these IMs would be slightly higher compared with EMs, but apparently did not lead to more switching. The clinical relevance of this genotype was less important than PMs. The low frequency of switching in our study (∼4%) may be the result of a carefully chosen low starting dose by general practitioners and psychiatrists diminishing the occurrence of ADRs or adjusting the initial dose. In our study a large proportion of amitriptyline users started on a low dose, probably also because this drug is not only used as antidepressant, but is also prescribed for the treatment of neuropathic pain [17]. As information on the exact indication was not available we have included all amitriptyline users in the study. We have repeated the analysis after exclusion of low dose amitriptyline users (<25 mg amitriptyline), but this did not alter our results. Due to the low frequency of switching in our cohort genotyping patients prior to the start of antidepressive therapy could only prevent a few patients from switching.

Discontinuation of antidepressant therapy was also studied, but was not associated with CYP2D6 genotype. Over 30% of our patients stopped their antidepressive medication within 45 days. Reasons for discontinuation of initial therapy are ADRs, non-compliance, lack of improvement and patient's belief about depression and antidepressants, with non-compliance being the most frequent reason [18].

After titrating the dose of an antidepressant to an optimal level of effectiveness with minimum ADRs, the mean antidepressant dose was significantly lower in PMs than in EMs. The absolute dose difference between PMs and EMs was smaller than 0.10 DDD, corresponding to 2–15 mg depending on the drug. This small difference in mean dose has limited impact on clinical outcome.

In TCA users with genotype *1/*4 (IMs) the optimum dose lay between EMs and PMs, while the mean SSRI dose of heterozygous patients overlapped with EMs. This difference could be explained by the amount of drug metabolized by CYP2D6. CYP2D6 is not the only cytochrome P450 enzyme involved in the metabolism of antidepressants. CYP2C19 plays a role in the demethylation of amitriptyline, imipramine, clomipramine, sertraline and citalopram, CYP1A2 and CYP3A4 contribute to a lesser extent [19, 20]. However, CYP2D6 exerts a higher influence on antidepressants metabolism than CYP2C19 [21]. Most tricyclic antidepressants are (partly) metabolized by CYP2D6, whereas some SSRIs are not metabolized by CYP2D6 (citalopram, escitalopram and sertraline). Our results indicate that the metabolism of TCAs depends more on CYP2D6 than that of SSRIs. Due to the low number of prescriptions for the newer SSRIs (non CYP2D6 substrates) we did not separate the SSRI group into substrates and non-substrates in our analysis.

Some potential limitations of our cohort study should be considered. Selection bias was unlikely because all antidepressant users were identified in a population based cohort study and prescribing doctors were not aware of CYP2D6 status of their patients. Missing blood samples and difficulties with genotyping (due to the suboptimal quality of long-term storage of DNA or a homozygous *5 subject) were probably not related to CYP2D6 genotype. The frequency of the CYP2D6*4 allele in our study (20.8%) was in accordance with the literature [4], but interestingly no Hardy–Weinberg equilibrium was observed. The assay was validated by DNA sequencing, and thus it seems not to be responsible for this discrepancy. However, our genotyping assay discriminated between the presence and absence of the CYP2D6*4 allele, but was unable to distinguish *4/*5 from *4/*4 individuals, who were classified as CYP2D6*4/*4. This led to an overestimation of *4/*4 individuals in the Hardy–Weinberg equilibrium. However, this does not affect phenotype classification since both *4/*4 and *4/*5 are PMs.

In this study we only determined CYP2D6*4, because this polymorphism is the most cost-effective in this large number of subjects and cost-effectiveness is a consideration of increasing importance in healthcare. High throughput assays are very expensive; moreover determination of CYP2D6*4 in our population should predict >75% of PMs [4]. Taking into account other, less frequent genetic variants (*3, *5, *6, *7, *8, *10, *41), it can be calculated that we missed approximately 209 IMs (17%), 10 PMs (0.8%) and 24 UMs (2%), who were all included in the EM group. Nevertheless, we believe that these misclassifications would have led to an underestimation of our association.

Information bias is not likely, since data on genotype and prescription data were collected prospectively without prior knowledge of the study hypothesis. We assessed potential confounding factors such as age, gender, BMI, renal function and starting dose in the multivariate analyses, but no association was found between CYP2D6 and BMI, renal function and starting antidepressant dose. Elderly subjects frequently use multiple drugs, due to comorbidity. Potentially, this could lead to ADRs due to drug–drug interactions. However, we assumed that comedication of CYP2D6 inhibitors did not confound our results, because comedication is prescribed independent of CYP2D6 genotype.

Our observational study demonstrated that CYP2D6 PM genotype is associated with an increased risk of switching to another antidepressant within the first 6 weeks of TCA pharmacotherapy and showed that PMs required a lower maintenance dose compared with EMs. Our data showed that starting doses of antidepressants prescribed to the elderly general population are carefully low and are titrated to the optimum dose, reducing the risk of adverse drug reactions. Therefore, although this study demonstrated that the CYP2D6 polymorphism is associated with antidepressant use the question remains whether genotyping prior to the start of antidepressant therapy contributes substantially to the optimization of pharmacotherapy.

Competing interests: None declared.