Effect of duloxetine on pain, function, and quality of life among patients with chemotherapy-induced painful peripheral neuropathy: a randomized clinical trial.

To my knowledge, this is the first Phase 3 study to show a statistically and clinically significant effect of any drug- versus chemotherapy-induced neuropathic pain (CINP) with moderate-term dosing. There have been several notable CINP failures, e.g., gabapentin and pregabalin. I hope that this study will give encouragement to the pharma industry and to other academic investigators to do additional studies for CINP using either “standard” or more creative study designs.

CINP remains a neuropathic pain condition without good treatment options. The best type of treatment would prevent neuropathic pain from occurring, for example, if the preventative product was administered before or during chemotherapy treatment. Lacking that, a product that treats the symptoms of CINP may not only be good for the palliation of pain, but it may also allow patients to remain in chemotherapy for longer periods of time and possibly achieve better outcomes. CINP is one of the leading reasons for patients to reduce doses, extend time between treatments, or discontinue chemotherapy prematurely.

This was an academically led effort. Although they reviewed the protocol and results, the footnotes to the article state that neither Eli Lilly nor NCI were involved in the conduct of the trial or the analysis of the results. This is unusual since, under Eli Lilly’s development efforts, duloxetine already has label indications for diabetic peripheral neuropathy (DPN), fibromyalgia (FMS), and chronic musculoskeletal pain, in addition to anxiety and depression. I would have expected that they [Lilly] would have been conducting studies of their own to gain a CINP label indication.

The crossover design may have contributed to the success of this treatment, but there could be other causes as well. I have never been a fan of crossover clinical trials in chronic pain due to my own experience with a PHN trial several years ago where it was clear that the baseline level of pain in the second half of the trial was substantially different (lower) than in the first half of the study. This is not uncommon with PHN where there may be a natural “healing” effect over time for some subjects, which may depend in part on how long they had had PHN before joining the trial. Secondly, for drugs that produce significant adverse effects (or any type of “cueing effect”), subjects will often have strong opinions about whether they received placebo or active medication in the first half of the study and gauge their responses accordingly when they figure out that they are getting the opposite treatment in the second half of the study. A third problem is that subjects who drop out in the first half of the study, for any reason including adverse effects or lack of efficacy, are simply not represented in the second half of the study, so there is self-selection bias (i.e., enrichment) for those who were either “responders” or who did not experience adverse effects in the first half, but who may drop out for non-response or adverse effects in the second half of the study.

Nevertheless, there are strong statistical arguments in favor of a crossover design when the study sponsors are reasonably certain that the level of pain will be constant and that they will have a highly compliant study population who are not likely to drop out over time. However, this does not seem to be the case in the study by Smith et al., since they reported an overall 19 percent dropout rate during the first half of the study. The dropout rate was not reported for the second half of the study.

Even though subjects in this study were selected for having completed their chemotherapy regimen three months or longer before this study began, those who were treated in the first half of the study with duloxetine were closer to the time when they completed their chemotherapy; subjects treated in the second half of the study were closer to the time when they might have begun to experience spontaneous remission of some of the CINP symptoms. It will be important to evaluate the natural time course of CINP pain versus treatment results in this study to determine whether treatment order, or time since last chemotherapy, or initial baseline pain was the dominant factor in providing a more robust response in the second phase of the trial. To make this point even more clearly, I was particularly impressed from a close inspection of Figure 2 that subjects in the second half of the study ended up with mean pain scores of 3.5 after six weeks of treatment, versus mean scores of 4.5 for those in the first part of the study. These are clinically significant differences! However, it should be appreciated that the baseline level of pain was nearly 0.5 points lower for duloxetine-treated subjects in the second half of the study compared to the first half.

The authors point out several important limitations of their trial in the Discussion section, not the least of which is a difference in the dropout rate between the active (duloxetine) and placebo subjects in the first phase of the clinical trial. As noted by the authors, differences in adverse event profiles may have led subjects to infer that they were in the active or in the placebo drug groups and make decisions that led to partial enrichment in the second phase of the study. Missing data are always problematic in analgesic clinical trials; the authors used some, but not all, potential methods to control for this.

Replication of these results and longer-term clinical trials (minimum three months' active dosing) will be needed to achieve FDA acceptance of these results in support of a label indication for CINP. This is the first step of many that will be required to fulfill the promise of reduced pain and improved quality of life for patients undergoing chemotherapy with taxane or platinum-containing chemotherapy drugs.

In summary, this contribution is clearly an advance in the field. As a hypothesis-generating exploratory study, it opens the doors for this group and for others to follow up with larger studies that are focused on groups who are more likely to respond to treatment (i.e., platinum-treated subjects who were the best responders in this study), with a flexible dosing design that is likely to encourage subjects to remain in the study for a longer period of time, and for a duration of time that is consistent with treatment patterns that would lead to a permanent change in the quality of life for CINP subjects. It will also be important to evaluate whether starting treatment with duloxetine or other neuropathic pain drugs at a much earlier time period, closer to the time of chemotherapy (or coincident with chemotherapy if there are not potential problems with drug-drug interactions) will yield an even greater impact on the pain status and quality of life of CINP subjects.