Safety and tolerability of duloxetine in elderly patients... : International Clinical Psychopharmacology (2024)

Introduction

Major depressive disorder (MDD) is a public health problem in the elderly, and frequently represents a challenge for the treating physician (^{Shanmugham et al., 2005; Fiske et al., 2009; Alexopoulos, 2011}). If left untreated, MDD in the elderly can lead to poor quality of life, problems with social and physical functioning, worsening of chronic medical conditions, and increased morbidity and mortality from suicide and other causes (^{Unutzer, 2007; Djernes et al., 2011}).

Key clinical considerations when treating depression in the elderly include the presence of comorbid anxiety and other medical comorbidities, adverse events, potential pharmaco*kinetic and pharmacodynamic changes, as well as the spectrum of depression symptoms for each patient (^{Barkin et al., 2000; Beyer, 2007; Unutzer, 2007; Mutasingwa et al., 2011}). Additional factors such as polypharmacy, patient attitudes, family/caregiver support, stigma, substance abuse, and cost of treatment have been associated with adherence to medical treatment (^{Zivin and Kales, 2008}). Despite these challenges, there are effective and well-tolerated treatments for late-life depression, including antidepressants, psychotherapy, and electroconvulsive therapy (^{Shanmugham et al., 2005}).

In primary care, antidepressant medications are the most commonly used treatments for MDD in the elderly, and antidepressant monotherapy is preferred to minimize side effects and drug interactions (^{Unutzer, 2007}). A number of different antidepressants have been approved to treat depression in elderly patients; however, the risks and side effect profiles, as well as the chemical structure, metabolism, and pharmaco*kinetics of these medications vary (^{Baumann, 1998; Unutzer, 2007; Dolder et al., 2010}). Second-generation antidepressants, including selective serotonin reuptake inhibitors (SSRIs) and serotonin-norepinephrine reuptake inhibitors (SNRIs), are most often used as the first-line antidepressant treatment option in the elderly, primarily because these medications have an improved safety and tolerability profile than first-generation tricyclic antidepressants (^{Unutzer, 2007; Alexopoulos, 2011}). In addition, second-generation antidepressants have shown efficacy overall, albeit modest, in double-blind, placebo-controlled studies (^{Nelson et al., 2008}). The SNRIs, in particular, may be useful for elderly patients with depression who experience concomitant pain (^{Unutzer, 2007}). Nevertheless, second-generation antidepressants are potentially problematic in that they carry a significant risk of adverse events, such as increased bleeding, hyponatremia, increases in blood pressure and pulse, orthostatic hypotension or serotonin syndrome as well as side effects such as nausea, sedation, insomnia, dry mouth, and sexual dysfunction (^{Glassman, 1998; Mottram et al., 2006; Unutzer, 2007; Nelson et al., 2008; Dolder et al., 2010; Mutasingwa et al., 2011}). These safety issues and side effects may limit the use of these antidepressants, and may adversely affect long-term treatment adherence.

Duloxetine hydrochloride is an antidepressant medication that inhibits both serotonin and norepinephrine reuptake, and previous studies suggest that it may be suitable for the treatment of MDD in elderly patients. In a long-term, open-label study, using data from a subset of patients 65 years of age and older, duloxetine induced significant improvements in both patient-rated and clinician-rated depression, and the safety profile was consistent with increasing concentrations of serotonin and norepinephrine (^{Wohlreich et al., 2004}). ^{Raskin et al. (2007)} have reported that duloxetine improved cognitive, emotional, and some painful symptoms associated with MDD and showed significantly superior antidepressant efficacy to placebo in elderly patients with MDD. Furthermore, in a meta-analysis of eight double-blind, randomized, placebo-controlled, and active comparator trials of patients at least 18 years of age, duloxetine appeared to have a cardiovascular profile comparable to other medications considered to be first-line treatment options for depression (^{Thase et al., 2005}). A review of 42 duloxetine placebo-controlled trials in various disease states including MDD, diabetic peripheral neuropathic pain, fibromyalgia, generalized anxiety disorder, and lower urinary tract disorders found a very low occurrence of cardiovascular adverse events during duloxetine treatment, including no significant differences by age in vital signs or ECGs (^{Wernicke et al., 2007}). Furthermore, at supratherapeutic doses in healthy volunteers, duloxetine did not prolong cardiac intervals (^{Zhang et al., 2007}).

The current study examined the safety and tolerability of duloxetine hydrochloride in a large cohort of elderly patients with MDD, 65 years of age and older, to help inform clinicians providing individualized care to this patient population. To obtain the cohort, individual patient data were pooled from two 60 mg/day, double-blind, randomized, placebo-controlled, duloxetine trials that exclusively enrolled elderly patients.

Materials and methods

Study I (http://www.clinicaltrials.gov identifier: NCT00062673; ^{Raskin et al., 2007}) was a multicenter, double-blind, placebo-controlled, phase 4 study of 311 elderly outpatients with recurrent MDD carried out in the USA and Puerto Rico. After a 1-week screening phase and a 1-week double-blind, placebo lead-in phase to blind the start of treatment, patients were randomized (2 : 1) to duloxetine 60 mg/day or placebo for 8 weeks of acute treatment.

Study II (http://www.clinicaltrials.gov identifier: NCT00406848; ^{Robinson et al., 2012}) was a multicenter, randomized, placebo-controlled, double-blind, phase 4 study of 370 elderly outpatients with MDD carried out in the USA, France, Mexico, and Puerto Rico. After a 3- to 30-day screening phase and a 1-week double-blind placebo lead-in phase, patients were randomized (2 : 1) to duloxetine or placebo for 12 weeks of acute double-blind treatment. Patients receiving duloxetine were administered 30 mg/day for 1 week, followed by forced titration to 60 mg/day for the remaining 11 weeks. Patients were instructed to take food with the study drug. Beginning with the placebo lead-in, patients requiring a dosage decrease because of safety or tolerability, or an increase because of lack of efficacy, were discontinued from the study. Following the acute phase, patients entered a double-blind continuation phase for an additional 12 weeks.

Patient population

Patients in both studies were male or female outpatients at least 65 years of age or older. The Mini-International Neuropsychiatric Interview (^{Sheehan et al., 1998}) was used to document the presence of MDD, as defined by the Diagnostic and Statistical Manual of Mental Disorders, 4th ed. (DSM-IV) (^{American Psychiatric Association, 1994}), in study I and the DSM-IV, 4th ed. text revision (DSM-IV-TR) (^{American Psychiatric Association, 2000}), in study II.

Patients had to fulfill specific inclusion criteria for MDD in each study. Patients in study I had to have a 17-Item Hamilton Depression Rating Scale (HAMD₁₇) (^{Hamilton, 1960, 1967}) total score of at least 18, whereas patients in study II had to have a Montgomery–Åsberg Depression Rating Scale (^{Montgomery and Asberg, 1979}) total score of at least 20. Patients in both studies had to have a Mini-Mental State Examination (^{Folstein et al., 1975}) score of at least 20, and a previous episode of depression. Key exclusion criteria were current primary axis I diagnosis, other than MDD; previous diagnosis of psychotic disorder; organic mental disorder or mental retardation diagnosis; and serious or unstable medical illness (^{Raskin et al., 2007; Robinson et al., 2012}); however, specific phobias, and nicotine and caffeine dependence or abuse were not exclusionary (^{Raskin et al., 2007; Robinson et al., 2012}).

Safety assessments

Safety measures recorded at every visit included spontaneously reported adverse events, weight, and supine and standing (for orthostatic change) blood pressure and pulse. Treatment-emergent adverse events (TEAEs) were defined as events that first occurred or worsened postrandomization during the acute treatment phase, as compared with the maximum severity at or before randomization. All TEAEs were reported according to the Medical Dictionary for Regulatory Activities (MedDRA), version 13. Treatment-emergent falls were not pooled in the current study because of differences in the methodology in collecting fall events in the two studies. For TEAE fall results from study I, see ^{Raskin et al. (2008)} and for TEAE fall results for study II see the clinical trial registry (NCT00406848).

Laboratory tests (hematology, clinical chemistry, and urinalysis) were carried out at screening and at the last visit of the acute treatment phase of each study. Values for abnormal laboratory tests were based on Covance reference ranges (Covance Central Laboratory Services, Indianapolis, Indiana, USA). A treatment-emergent high chemistry value was defined as a change from a value less than or equal to the high limit in the baseline assessments to a value greater than the high limit at the relevant postbaseline visit(s). A treatment-emergent low chemistry value was defined as a change from a value greater than or equal to the low limit in the baseline assessments to a value less than the low limit at the relevant postbaseline visit(s).

Abnormal weight gain was defined as at least a 7% increase from baseline. Abnormal weight loss was defined as at least a 7% decrease from baseline. High supine systolic blood pressure was defined as a value of at least 140 mmHg with an increase of at least 10 mmHg from baseline. High supine diastolic blood pressure was defined as a value of at least 90 mmHg with an increase of at least 10 mmHg from baseline. High supine pulse was defined as a value of at least 100 bpm with an increase of at least 10 from baseline. Treatment-emergent orthostatic hypotension was defined as standing systolic blood pressure at least 20 mmHg less than the supine systolic blood pressure or as standing diastolic blood pressure at least 10 mmHg less than the supine diastolic blood pressure at any postbaseline interval, where a patient did not fulfill the criteria at any visit during the baseline period.

All ECGs were recorded once before random assignment and at the last visit of the acute treatment phase for study I. For study II, ECGs were recorded before random assignment and at the last visit of the 24-week study, but they were not recorded at the end of 12 weeks of acute treatment. Because of nonoverlapping postbaseline ECG assessments, pooling was not possible. For ECG results from study I, see ^{Raskin et al. (2008)} and for ECG results for study II see the clinical trial registry (NCT00406848).

Statistical methods

Individual data from all randomized patients were pooled for the analyses presented in this article, unless otherwise specified. Pooled safety analyses included data from the 8-week acute treatment phase for study I and the 12-week acute treatment phase for study II. The term ‘significant’ for treatment comparisons indicates statistical significance (two sided P≤0.05). No adjustments were made for multiple comparisons. Unless otherwise specified, baseline refers to the last nonmissing prerandomization observation and endpoint refers to the last nonmissing observation during the acute treatment phase.

Categorical variables such as baseline patient demographics and characteristics (e.g. sex and ethnicity); rates of overall discontinuation and discontinuation because of adverse events; postrandomization rates of TEAEs (preferred terms in MedDRA, version 13); and rates of abnormal changes in vital signs, weight, and laboratory analytes were summarized and compared between treatments using a Cochran–Mantel–Haenszel test for a general association controlling for study. For the predefined postrandomization abnormalities, only patients with a normal baseline value were included in the analysis.

Continuous variables such as baseline patient characteristics and rank-transformed changes in laboratory analytes were evaluated using a fixed-effects analysis of variance model that included terms for treatment and study. Changes from the baseline to the endpoint in vital signs and weight were evaluated using an analysis of covariance (ANCOVA) model that included the terms treatment, study, and baseline. For early dropout patients, the endpoint was imputed using the last observation carried forward approach.

Age categories (<75 vs. ≥75 years) were utilized to determine patient strata for subgroup analyses, which assessed and compared duloxetine treatment effects between patient subgroups. For continuous variables, such as changes in vital sign, subgroup analyses were carried out by adding the relevant subgroup and treatment×subgroup interaction terms to the ANCOVA model described above. For categorical variables, such as TEAE and discontinuation because of adverse events, incidence rates were summarized by treatment within each patient subgroup, and the Breslow–Day test was carried out for assessment of hom*ogeneity of treatment odds ratios across patient subgroups. The Breslow–Day test and interaction term in the ANCOVA model were tested at a two sided 0.10 significance level.

All statistical analyses were carried out using SAS, version 9.1 (SAS Institute Inc., Cary, North Carolina, USA).

Results and discussion

Patients

The safety population included 681 patients (duloxetine, N=456; placebo, N=225) who were randomized to treatment. Of these, 341 (74.8%) duloxetine-treated patients and 158 (70.2%) placebo-treated patients completed the acute treatment phases.

Patient demographics and other baseline characteristics are presented in Table 1. Of the 681 patients randomized to treatment, 66.8% (N=455) were less than 75 years of age and 33.2% (N=226) were at least 75 years of age. There were no statistically significant differences between the treatment groups in any of the patient demographic variables, including ethnicity, age, sex, and weight. Patients in both treatment groups were predominantly women (duloxetine, 63.2%; placebo, 58.2%), white (duloxetine, 78.7%; placebo, 77.3%), and of comparable mean age (duloxetine, 72.77 years; placebo, 73.17 years). The mean severity of depression was mild and there were no statistically significant differences between the treatment groups in the mean HAMD₁₇ Total score and Maier subscale score.

Discontinuations

Overall

The overall discontinuation rates for any reason did not differ significantly between the duloxetine and the placebo groups in the total patients (25.2 vs. 29.8%; P=0.198). Similarly, the incidence of discontinuation because of any adverse event was not significantly different between duloxetine and placebo in the total patients (10.7 vs. 7.1%; P=0.13). Further, the incidence of any individual adverse event leading to discontinuation was not significantly different between the treatment groups (P>0.05; data not shown).

<75 and ≥75 years of age

Discontinuation rates for any reason did not differ significantly between the duloxetine and the placebo groups in the less than 75-year (19.9 vs. 26.4%; P=0.117) or at least 75-year (36.2 vs. 36.4%; P=0.954) subgroups and no significant age subgroup heterogeneity was observed in the effect of duloxetine (P=0.335), as determined by the Breslow–Day test. In addition, the incidence of discontinuation because of any adverse event was not significantly different between duloxetine and placebo in the less than 75-year (7.2 vs. 6.1%; P=0.667) and at least 75-year (18.1 vs. 9.1%; P=0.075) subgroups. Although the incidence of discontinuations because of any adverse event was higher in the at least 75-year subgroup than in the less than 75-year subgroup, there was no significant heterogeneity (P=0.307) in the effect of duloxetine between age subgroups, as determined by the Breslow–Day test.

Serious adverse events

Overall

There were no deaths during either study. During the acute treatment phases, serious adverse events (SAEs) occurred in nine (2.0%) duloxetine-treated patients and in six (2.7%) placebo-treated patients. Chest pain, arrhythmia, coronary artery occlusion, hip fracture, intentional overdose, and intracranial aneurysm occurred in duloxetine-treated patients. Chest pain, bladder transitional cell carcinoma, chronic obstructive pulmonary disease, hyperglycemia, and influenza occurred in placebo-treated patients. Of all the SAEs, only chest pain occurred in more than one duloxetine-treated patient (n=2). Only one SAE was judged by the investigator to be related to the study drug: hip fracture in a duloxetine-treated patient resulting from a fall. The patient with this SAE had a history of osteoporosis, was also receiving antihypertensive medication, and was noted to be hypotensive at the study visit before the fall, when her antihypertensive medication was decreased to a minimum dose.

Treatment-emergent adverse events

Overall

The TEAEs that were at least 5% for duloxetine and twice the rate of placebo are presented in Table 2, and are also stratified by age subgroup. Overall, 70.0% of patients reported one or more TEAEs during acute treatment, and the percentage of patients reporting at least 1 TEAE was significantly higher for duloxetine-treated patients (73.7%) than for placebo-treated patients (62.7%; P=0.003). Of these events, the overall incidence of dry mouth, constipation, nausea, diarrhea, and dizziness occurred significantly more frequently for duloxetine compared with placebo.

<75 and ≥75 years of age

There was significant age subgroup heterogeneity in the treatment effect on nausea (P=0.008) and fatigue (P=0.023), as determined by the Breslow–Day test. In the less than 75-year subgroup, the incidence of nausea was significantly higher for duloxetine-treated patients compared with placebo-treated patients (12.1 vs. 2.0%; P<0.001), whereas in the at least 75-year subgroup, the incidence was similar between the treatment groups (8.7 vs. 9.1%; P=0.920). In the less than 75-year subgroup, the incidence of fatigue was similar between the treatment groups (4.2 vs. 4.1%; P=0.928), whereas in the at least 75-year subgroup, the incidence of fatigue was significantly higher for the duloxetine treatment group compared with the placebo treatment group (8.1 vs. 0.0%; P=0.010).

Vital signs

Overall

The mean changes from the baseline to the endpoint in standing and supine vital signs are presented in Fig. 1a for the treatment groups. Overall, the estimated mean changes in standing and supine blood pressure and pulse were modest, and there were no significant differences between the duloxetine-treatment and placebo-treatment groups in standing systolic or diastolic blood pressure, pulse, and supine systolic blood pressure and pulse. There were, however, significant differences between the treatment groups in supine diastolic blood pressure (1.36 vs. −0.49 mmHg; P=0.006), orthostatic differences (standing−supine) in systolic blood pressure (−1.41 vs. 0.60 mmHg; P=0.013), and orthostatic differences (standing−supine) in diastolic blood pressure (−1.52 vs. 0.19 mmHg; P=0.003).

<75 and ≥75 years of age

As shown in Fig. 1b, there were significant treatment×age subgroup interactions for standing diastolic blood pressure (P=0.032) and supine diastolic pressure (P=0.057). For the less than 75-year subgroup, duloxetine-treated patients compared with placebo-treated patients showed a mean increase in standing (0.55 vs. −0.70 mmHg; P=0.111) and supine (1.86 vs. −0.89 mmHg; P<0.001) diastolic blood pressure. Conversely, for the at least 75-year subgroup, duloxetine-treated patients compared with placebo-treated patients showed a mean decrease in standing (−1.65 vs. 0.17 mmHg; P=0.147) but not supine (0.32 vs. 0.30 mmHg; P=0.997) diastolic blood pressure.

Table 3 presents the incidence of treatment-emergent abnormal elevations in supine vital signs and treatment-emergent orthostatic hypotension. Treatment-emergent abnormal elevations in supine blood pressure and pulse at any time during acute treatment and at the endpoint were not significantly different between the treatment groups. Furthermore, the incidence of treatment-emergent orthostatic hypotension at any time was not significantly different between the treatment groups (19.5 vs. 21.6%; P=0.558).

Weight

Overall

The mean changes in body weight from the baseline to the endpoint are presented in Table 4 for the treatment groups and for the age subgroups. Overall, there was a significant difference between the treatment groups in the estimated mean change in weight, with duloxetine-treated patients losing more weight than placebo-treated patients (−0.76 vs. −0.02 kg; P<0.001).

There were no significant differences between the duloxetine-treatment and placebo-treatment groups in the frequency of abnormal weight gain (≥7% from baseline) at any time during acute treatment (1.8 vs. 0.4%; P=0.160) or at the endpoint (1.8 vs. 0.4%; P=0.160). Similarly, there were no significant differences between the treatment groups in the frequency of abnormal weight loss (≥7% from baseline) at any time during acute treatment (2.2 vs. 2.7%; P=0.707) or at the endpoint (2.2 vs. 1.8%; P=0.716).

There were no significant differences between duloxetine and placebo in treatment-emergent weight-related adverse events, such as decreased appetite (3.5 vs. 2.2%; P=0.357), increased appetite (0.4 vs. 1.8%; P=0.079), and decreased weight (0.9 vs. 0.0%; P=0.154). Furthermore, no patient in either treatment group discontinued because of a weight-related adverse event.

<75 and ≥75 years of age

There was a significant treatment×age subgroup interaction for weight (P=0.002). For the less than 75-year subgroup, duloxetine-treated patients lost weight in contrast to placebo-treated patients who gained weight (−0.93 vs. 0.19 kg; P<0.001). For the at least 75-year subgroup, both duloxetine-treated and placebo-treated patients generally lost the same amount of weight (−0.41 vs. −0.43 kg; P=0.950).

Laboratory analytes

Table 5 shows the mean change from the baseline to the endpoint in laboratory analytes that were significantly different between the duloxetine and the placebo treatment groups (P<0.05). The mean changes in alkaline phosphatase, platelet count, potassium, random glucose, uric acid, and cholesterol were significantly different between the duloxetine and the placebo groups. There were no significant treatment differences for the mean changes from the baseline to the endpoint in the hepatic analytes: aspartate aminotransferase, alanine aminotransferase, creatine phosphokinase, γ-glutamyl transferase, and bilirubin. No significant differences existed between duloxetine and placebo groups in the incidence of treatment-emergent abnormal laboratory values at any time, with two exceptions. Significantly more duloxetine-treated than placebo-treated patients had low leukocyte counts (2.9 vs. 0.0%; P=0.015), whereas significantly fewer duloxetine-treated than placebo-treated patients had low platelet counts (0.2 vs. 2.0%; P=0.025). Moreover, the incidence of abnormal low sodium levels at any time was not significantly different between the duloxetine (1.19%) and the placebo (0.48%; P=0.395) groups.

Discussion

The safety and tolerability profile of duloxetine 60 mg/day in a large cohort of elderly patients 65 years of age and older with MDD was examined in the present study to better inform clinicians providing individualized care to this patient population. The overall discontinuation rates (patients who discontinued irrespective of the reasons) and discontinuation rates because of adverse events can be used as indicators of antidepressant tolerability. The discontinuation rate for any reason in the duloxetine group (25.2%) was less than the rate in the placebo group (29.8%), but not significantly different. These results are consistent with those reported previously for duloxetine 60 mg/day and placebo in patients aged 18 and older (^{Nelson et al., 2006}) and comparable to the point estimates for the discontinuation rates reported by ^{Mittmann et al. (1997)} for SSRIs (18.5%), tricyclic antidepressants (23.2%), and placebo (25.6%).

Overall, 10.7% of duloxetine-treated patients discontinued because of an adverse event in this study, a rate similar to that of placebo-treated patients (7.1%). The overall rate of discontinuation because of any adverse event was higher in the at least 75-year subgroup (18.1%) than in the less than 75-year subgroup (7.2%). These results for duloxetine in the overall treatment group and in the less than 75-year subgroup are consistent with the discontinuation rates because of adverse events for duloxetine at doses of 60 mg/day [13.1% in ^{Mallinckrodt et al. (2006)}] and at doses ranging from 40 to 120 mg/day [8.0% reported by ^{Nelson et al. (2006)}] in patients with MDD who were 18 years of age or older. Note that both ^{Mallinckrodt et al. (2006)} and ^{Nelson et al. (2006)} reported significantly lower rates of discontinuation because of adverse events for patients treated with placebo compared with those treated with duloxetine (3.4 and 4.0%, respectively). In a pooled analysis of six double-blind, placebo-controlled studies of duloxetine at doses ranging from 40 to 120 mg/day in patients 55 years of age and older, the rates of discontinuation because of adverse events for duloxetine-treated patients were higher (21.0%) than in the current study, but the rates for placebo-treated patients (6.7%) were comparable (^{Nelson et al., 2005}). The difference in the rates for duloxetine-treated patients in the current study versus ^{Nelson et al. (2005)} may be partly because of differences in the number of patients studied.

An important consideration in the choice of an antidepressant for an individual patient is its safety and tolerability. Moreover, the potential side-effect burden and the impact on the patient’s quality of life may be associated with an increased risk of treatment nonadherence and discontinuation (^{Pollock, 1999}). In the current study, the most frequently reported TEAEs were dry mouth, constipation, and nausea. The rates are comparable to those reported for some SSRIs in patients with late-life depression (^{Bondareff et al., 2000; Mottram et al., 2006}) and were consistent with or lower than duloxetine adverse event rates in patients with a mean age of 50.3 years (dry mouth, 12.2%; constipation, 10.3%; nausea, 24.8%) (^{Gahimer et al., 2007}) and in patients at least 55 years (dry mouth, 22.7%; constipation, 16.0%; nausea, 25.2%) (^{Nelson et al., 2005}). Nausea is frequently the most common TEAE in duloxetine-treated patients with MDD who are at least 18 years of age [37.8% in ^{Mallinckrodt et al. (2006)}; 29.3% in ^{Pigott et al. (2007)}; 34.7% in ^{Perahia et al. (2006)}]. In the current study, however, dry mouth was the most common TEAE following duloxetine treatment and nausea was the third most common. The inconsistency in the nausea rates between the current duloxetine study (lower nausea rates) and others that included younger patients 18 years of age and older (higher nausea rates) may be attributed to differences in the ages of the patient populations studied. A previous pooled analysis (^{Greist et al., 2004}) focused specifically on antidepressant-induced nausea in duloxetine-treated patients and reported a trend toward reduced rates of treatment-emergent nausea in patients 65 years of age and older (16.7%) than in patients younger than 65 years of age (20.1%), but the difference was not significant. Chemotherapy-induced nausea is probably the most recognized form of medication-induced nausea, and age is an identified risk factor with older patients experiencing less chemotherapy-induced nausea and vomiting than younger patients (^{Pollera and Giannarelli, 1989; Hesketh et al., 2010}).

In general, the impact of these TEAEs on patient adherence and overall tolerability seems to be modest in the current study, given the relatively low rate of discontinuation because of adverse events in duloxetine-treated patients and the incidence of any individual adverse event leading to discontinuation being not significantly different between duloxetine and placebo. The overall rate of adverse events was similar between the less than 75 and at least 75-year subgroups (73.9 vs. 73.2%, respectively), with no significant heterogeneity in duloxetine effect. There was, however, significant heterogeneity between the age subgroups in treatment effects for nausea and fatigue, which are two common side effects of duloxetine in adults 18 years of age and older (^{Mallinckrodt et al., 2006; Perahia et al., 2006; Pigott et al., 2007; Duloxetine Hydrochloride Prescribing Label, 2011}). In the present study, among patients less than 75 years of age, duloxetine-treated patients had a significantly higher incidence of nausea than placebo-treated patients (12.1 vs. 2.0%); this finding is consistent with the results from duloxetine safety studies, including adults 18 years of age and older (^{Nelson et al., 2006; Nierenberg et al., 2007}). Yet, there were no treatment group differences in the rate of nausea between duloxetine-treated and placebo-treated patients in the at least 75-year subgroup (8.7 vs. 9.1%). Conversely, in the current study, duloxetine-treated patients who were at least 75 years had a higher incidence of fatigue compared with placebo-treated patients (8.1 vs. 0.0%), whereas there were no treatment group differences in patients less than 75 years (4.2 vs. 4.1%), which is consistent with ^{Mancini et al. (2009)}.

There are few reports assessing antidepressant safety in patients at least 75 years of age (^{Roose et al., 2004; Nelson et al., 2006; Raskin et al., 2008; Nelson and Devanand, 2011}), and only one with a comparison of safety in patients less than 75 years with those at least 75 years of age with MDD who were treated with an antidepressant (^{Raskin et al., 2008}). In the current study, the significant age subgroup differences in individual TEAEs (e.g. nausea and fatigue) suggest that there may be differences in the experience of specific adverse events between the two age subgroups. However, the lack of significant age subgroup differences in the rates of discontinuation for any reason and the rates of discontinuation because of adverse events suggest that the overall tolerability or the perceived severity of adverse events is similar.

Treatment-emergent falls were not pooled between the two trials in the current study because of the differing methodologies used in study I and study II. Incidents of falls were solicited in study II, but not in study I, and may have influenced the rate of treatment-emergent falls reported for both duloxetine and placebo in the former study. In study I, 2.4% of duloxetine-treated patients and 2.9% of placebo-treated patients reported treatment-emergent falls during the combined acute and taper phases (^{Raskin et al., 2008}). In study II, a higher number of duloxetine-treated than placebo-treated patients reported treatment-emergent falls during acute treatment [16.1 vs. 9.9%; ^{Robinson et al. (2012)}]. Therefore, to fully understand these treatment-emergent fall results, a separate detailed report of data from the Falls Assessment Questionnaire in study II is planned.

Many antidepressant medications have potential adverse cardiovascular effects that restrict use, particularly among older patients, including increases in blood pressure and pulse, orthostatic hypotension, and other cardiovascular effects (^{Thase et al., 2005; Unutzer, 2007}). In the current study, the estimated mean changes in standing and supine blood pressure and pulse were modest, and did not differ significantly between duloxetine and placebo, with the exception of significant mean increases in supine diastolic blood pressure. These results appear somewhat inconsistent with previous meta-analyses of placebo-controlled studies in adults reported by ^{Thase et al. (2005)}, which showed a significant increase in supine systolic blood pressure following duloxetine treatment at 60 mg/day. However, the actual magnitude of change for supine systolic blood pressure was similar for the duloxetine-treated groups in the current study and that of ^{Thase et al. (2005)} (0.17 vs. 0.2 mmHg, respectively). Furthermore, in both the current study and that of ^{Thase et al. (2005)} there was a mean increase in diastolic blood pressure (1.36 vs. 1.9 mmHg, respectively) and pulse (0.58 vs. 1.8 bpm, respectively), although the magnitude of change was greater in the study of ^{Thase et al. (2005)}. Given that the rates of treatment-emergent abnormal elevations in systolic and diastolic blood pressure, pulse, and orthostatic hypotension at any time were similar between the duloxetine-treatment and placebo-treatment groups in the current study, these results may reflect natural variability in blood pressure. Furthermore, in both study I and study II, there were no significant differences between duloxetine and placebo in the mean changes in Bazett and Fridericia-corrected QT intervals (^{Raskin et al., 2008; Eli Lilly and Company and Boehringer Ingelheim Pharmaceuticals, 2012; Robinson et al., 2012}). Taken together, these results did not identify any significant cardiovascular risks in the elderly population; however, clinicians should consider the current prescribing information on cardiovascular risks for duloxetine, and evaluate and monitor individual patients accordingly.

Antidepressants have been associated with weight gain, which in turn can lead to patient nonadherence (^{Fava et al., 2000; Gahimer et al., 2007}). In the current study, duloxetine-treated patients lost slightly more weight than placebo-treated patients, as determined by the mean changes from baseline. These results are generally consistent with analyses of 10 clinical studies in patients 18 years of age and older, which concluded that, on average, duloxetine-treated patients experienced weight loss after short-term treatment (^{Wise et al., 2006}). Interestingly, in the current study, a significant age subgroup difference in the duloxetine treatment effect was observed for the estimated mean change in weight. In the less than 75-year subgroup, duloxetine-treated patients showed a decrease in weight compared with a small weight gain in placebo-treated patients, whereas in the at least 75-year subgroup, both duloxetine-treated and placebo-treated patients showed a small, similar decrease in weight. Importantly, there were no significant differences in the frequency of abnormal weight gain (≥7%) or loss (≥7%) for duloxetine compared with placebo at any time during acute treatment or at the endpoint.

Both SSRIs and SNRIs have been associated with hyponatremia, particularly in elderly patients (^{Sand et al., 2002; Coupland et al., 2011}). Postmarketing reports of hyponatremia have also occurred following treatment with duloxetine and have appeared to be reversible following discontinuation (^{Duloxetine Hydrochloride Prescribing Label, 2011}). In many cases, this hyponatremia appeared to be the result of the syndrome of inappropriate antidiuretic hormone secretion. In the current study, the incidence of abnormal low sodium levels for duloxetine was not significantly different from placebo, which is consistent with duloxetine safety studies in younger adults (^{Hudson et al., 2005; Nelson et al., 2006; Pigott et al., 2007}). Nevertheless, in light of postmarketing reports of hyponatremia following treatment with duloxetine, discontinuation of duloxetine should be considered in patients with symptomatic hyponatremia and the appropriate medical intervention should be instituted.

The only abnormality in liver-related enzymes in this pooled study of elderly patients with MDD was an elevation in alkaline phosphatase in duloxetine-treated patients compared with placebo-treated patients (1.1 vs. −1.7 U/l). In clinical trials of patients 18 years of age and older, increases in liver enzymes (e.g. aspartate aminotransferase, alanine aminotransferase, alkaline phosphatase, creatine phosphokinase, γ-glutamyl transferase) have been associated with duloxetine treatment (^{Gahimer et al., 2007}), and have generally manifested as transient, self-limiting transaminase elevations. Rare events characterized as hepatocellular injury, cholestatic injury, or mixed type of hepatic injury have been reported (^{Wernicke et al., 2008}). These hepatic risks are described in labeling, such as the United States Package Insert and Summary of Product Characteristics. In the current study, there were also significant mean changes in other laboratory analytes (i.e. platelet count, potassium, random glucose, uric acid, and cholesterol), but none of these small changes were considered clinically relevant.

The present study should be interpreted with respect to a number of potential limitations. The analysis of safety outcomes presented in this paper was based on data from two clinical trials that exclusively enrolled elderly patients; therefore, there were no formal statistical comparisons between older and younger cohorts. Other limitations of the study included the varying duration of the acute treatment periods (8 vs. 12 weeks). Furthermore, the ability to generalize these results to typical outpatients is somewhat limited, as the patients in both studies had relatively fewer comorbid medical conditions and concomitant medications compared with their general outpatient counterparts.

The results presented here are generally consistent with safety findings in adults 18 years of age and older. Although the rates of some TEAEs were different in the elderly, these rates were generally lower than those observed in younger patients. These safety results may better inform clinicians providing individualized care to elderly patients with MDD.

Acknowledgements

The original trial was funded and sponsored by Eli Lilly and Company and/or any of its subsidiaries, Indianapolis, Indiana, USA. The authors would like to thank PharmaNet/i3 for their help with writing, editing, and formatting.

Conflicts of interest

Dr Oakes, Dr Liu, Dr Raskin, and Dr Robinson are full-time employees of Eli Lilly and Company, and minor stockholders of Eli Lilly and Company. Dr Katona has provided consultancy for and received speaker honoraria from Lilly and Lundbeck and has also recently been in receipt of a grant from Lundbeck. Dr Greist receives research support from AstraZeneca, Forest, Eli Lilly and Company, Otsuka, Takeda, Transcept, and UCB. This study was funded by Eli Lilly and Company.

References