Antidepressant treatments have been used to manage pain syndromes (1). Duloxetine received US FDA approval in 2010 for the indication of chronic musculoskeletal pain, including osteoarthritis (OA) (2). The mechanism of action is thought to be related to the amelioration of central pain pathway dysfunction (3) through actions at serotonin and norepinephrine transporters (2); thus, the mechanism of action is notably different from those of opioids and non-steroidal anti-inflammatory drugs (NSAIDs).
We recently published elsewhere a systematic review of duloxetine monotherapy for osteoarthritic pain (4). We quantified the clinical relevance of the statistically significant results from the 13-week duloxetine studies using number needed to treat (NNT) and found that the NNT for duloxetine vs. placebo for treatment relief using a composite measure was six (95% CI 4–10), which overlaps with the 95% CI observed with the NNT for response for other treatments, such as etodolac after 4 weeks (NNT 5, 95% CI 3–25) and tenoxicam after 8 weeks (NNT 4, 94% CI 3–8), as calculated in systematic reviews for OA (5). Of interest is that amelioration in pain was not dependent on improvement in depressive symptoms (4).
The tolerability and safety profile for duloxetine may have some advantages over alternative therapies. For example, NSAIDs can lead to gastrointestinal bleeding, and opiates, such as once daily morphine, commonly cause constipation. The three most common adverse events associated with duloxetine, nausea, fatigue and constipation, have small effect size differences for duloxetine vs. placebo (number needed to harm (NNH) 16, 17 and 19, respectively). Even though the individual adverse events had small effect sizes, approximately 16.3% of the patients who received duloxetine in the 13-week placebo-controlled trials for chronic pain due to OA discontinued treatment due to an adverse reaction, compared with 5.6% for placebo, for a NNH of 10 (95% CI 7–20).
Although the use of duloxetine as a monotherapy for pain has been approved by regulatory agencies, it is quite common for patients to receive a combination of medication treatments. NSAIDs as a class are arguably the most frequently prescribed medications for pain associated with OA. Newly available are the results of a study examining the potential synergy of duloxetine with NSAIDs (6). In this 10-week double-blind trial, 524 adult outpatients who had persistent moderate pain (≥ 4 on a 0–10 numerical rating scale) because of OA of the knee despite having received optimised oral NSAID therapy, were randomised to receive adjunctive flexible-dose duloxetine (60 or 120 mg/day) (N = 264) or adjunctive placebo (N = 260). To be included in the study, patients were required to have had knee pain for at least 14 days/month in the 3 months preceding study entry and using oral NSAIDs on most days during that time period. Excluded were those with a history of gout; pseudogout; inflammatory arthritis; end-stage, bone-on-bone OA; or knee surgery or intra-articular injection in the last 6 months. Also excluded were patients who had another chronic painful condition such as fibromyalgia that could interfere with assessment of the knee, those with a body mass index greater than 40 kg/m2, and those who were non-ambulatory or required equipment other than a single cane to walk. Participants could not use opioid analgesics more than 3 days/week or be unwilling to taper and discontinue their use prior to randomisation. Subjects were required to take a proton pump inhibitor while using NSAID therapy during the study. Patients receiving duloxetine were titrated as follows: 30 mg/day for 1 week, followed by 60 mg/day for 2 weeks. After 3 weeks, subjects with a mean average pain severity rating ≥ 4 during the previous week had a blinded dose escalation to 120 mg/day. To minimise the possibility of efficacy ratings being influenced by the impending end to a study, a sham endpoint at week 10 was prespecified in the protocol’s IRB supplement; the prespecified primary efficacy endpoint, week 8, was unknown to subjects and investigators. In total, 74% of the patients completed the study. Mean age was 61 years, 57% were female, and 81% were white. Ibuprofen and naproxen were the most frequently used NSAIDs and mean doses of each NSAID given were typical of those used for treatment of patients with moderate-to-severe OA pain. There were no differences between duloxetine and placebo groups in the relative frequency or mean dose of each NSAID used.
In this study of adjunctive duloxetine vs. adjunctive placebo, duloxetine-treated patients had significantly greater pain reduction at week 8 (p < 0.001) than placebo-treated patients based on the weekly mean of the daily average pain rating as entered in a telephone-based diary.
Categorical outcomes for tolerability
Discontinuation because of an adverse event was observed in 40/264 (15.2%) of subjects taking adjunctive duloxetine vs. 23/260 (8.8%) of subjects receiving adjunctive placebo, for an NNH of 16 (95% CI 9–130). Adjunctive duloxetine differed from adjunctive placebo on the frequency of several different adverse events. The published report provides the percentage of patients with each adverse event: nausea (duloxetine, 15.5% vs. placebo, 4.6%; NNH 10), dry mouth (9.5% vs. 2.7%; NNH 15), constipation (8.7% vs. 3.1%; NNH 18), dizziness (6.4% vs. 2.7%; NNH 27), fatigue (6.8% vs. 1.5%; NNH 19) and decreased appetite (5.7% vs. 0.4%; NNH 19). These are consistent with the monotherapy trials; however, nausea, dry mouth and decreased appetite were more frequently encountered in the combination trial (4). Bleeding-related adverse events occurred in four patients in the duloxetine group (epistaxis, haematochezia, decreased haematocrit, easy bruising) vs. one patient in the placebo group (haematoma), with no bleeding adverse event classified as serious. Duloxetine, similar to other serotonergic antidepressants, can be associated with platelet dysfunction. Although the NNH for adjunctive duloxetine vs. placebo for bleeding as an adverse event was 89 and not statistically significant, the study was short (10 weeks). It can be expected that with longer exposure to duloxetine in combination with NSAIDs, or exposure to NSAIDs alone, additional bleeding events could occur.
Categorical outcomes for efficacy
Categorical outcomes for response are provided in the original report (6) with NNTs and 95% CIs. Moderate improvement in pain (defined by ≥ 30% improvement on the diary-based measure of pain severity) was observed in 139 of 259 (53.7%) subjects in the duloxetine group and 86/255 (33.7%) subjects in the placebo group, resulting in an NNT of 5.01, which was rounded up by the authors to six (95% CI 4–9). Substantial improvement (≥ 50% improvement on the diary-based measure of pain severity) was observed in 92 of 259 (35.5%) subjects in the duloxetine group and 41/255 (16.1%) subjects in the placebo group, resulting in an NNT of 6 (95% CI 4–9). These and other efficacy outcomes are displayed in Figure 1, contrasting NNT and NNH.
The authors of the study acknowledged that the short duration of this study (10 weeks) limits conclusions about the long-term use of adjunctive duloxetine and also the potential for bleeding-related complications. Whether or not the results of this study can be generalisable to other antidepressants, particularly to those with similar mechanisms of action on serotonin and norepinephrine transporters, will need to be confirmed by the conduct of appropriately designed clinical trials. Appraisal of sexual dysfunction will also need to be carefully performed as the emergence of this adverse event can be unacceptable for many patients; clinical trials have underestimated the frequency of this, as they usually measure this outcome by spontaneous adverse event reporting rather than by specific inquiry.
Although the duloxetine monotherapy trials included path analyses to demonstrate duloxetine’s direct analgesic effect rather than it being dependent on improvement in depression or anxiety symptoms (4), there is no mention of this being done for the data collected in the adjunctive duloxetine study (6).
NNT and NNH are clinically intuitive measures of effect size differences, but there are limitations to their use (7). When comparing results from different studies when no head-to-head data are available, rates of events (and their respective NNTs and NNHs) are dependent on the particular study population from which they are derived. When comparing outcomes from different studies, it is important assess whether the study populations are “similar enough” or whether significant differences (aside from the drug) may be accounting for the observed outcomes. This also applies when determining whether or not the study (or studies) in question is (are) generalisable to patients treated in the clinic.
It is not uncommon in clinical practice to treat OA with combinations of medications that have differing mechanisms of action. This new data supports this approach and further illustrates that “antidepressants” are not just for depression. Clinicians managing patients suffering from OA should also consider adjunctive antidepressants that can effectively impact central pain pathways.
No writing assistance or external financial support was utilised in the production of this article. In the past 12 months, Leslie Citrome, was a consultant for, has received honoraria from, or has conducted clinical research supported by the following: Alexza, Alkermes, AstraZeneca, Avanir, Bristol-Myers Squibb, Eli Lilly, Janssen, Lundbeck, Merck, Novartis, Noven, Otsuka, Pfizer, Shire, Sunovion and Valeant. Amy Weiss-Citrome has nothing to disclose.