What Can Cymbalta Teach Us About Pain?

April 29, 2011

You’ve probably noticed widespread TV advertisements lately for Cymbalta, Eli Lilly’s blockbuster antidepressant.  However, these ads say nothing about depression.  Sure, some of the actors may look a little depressed (the guy at right, from the Cymbalta web site, sure looks bummed), but the ads are instead promoting Cymbalta for the treatment of chronic musculoskeletal pain, an indication that Cymbalta received in August 2010, strengthening Cymbalta’s position as the “Swiss Army knife” of psychiatric meds.  (I guess that makes Seroquel the “blunt hammer” of psych meds?)

Cymbalta (duloxetine) had already been approved for diabetic neuropathy and fibromyalgia, two other pain syndromes.  It’s a “dual-action” agent, i.e., an inhibitor of the reuptake of serotonin and norepinephrine.  Other SNRIs include Effexor, Pristiq, and Savella.  Of these, only Savella has a pain [fibromyalgia] indication.

When you consider how common the complaint of “pain” is, this approval is a potential gold mine for Eli Lilly.  Moreover, the vagueness of this complaint is also something they will likely capitalize upon.  To be sure, there are distinct types of pain—e.g., neuropathic, visceral, musculoskeletal—and a proper pain workup can determine the exact nature of pain and guide the treatment accordingly.  But in reality, overworked primary clinicians (not to mention psychiatrists, for whom hearing the word “pain” is often the extent of the physical exam) often hear the “pain” complaint and prescribe something the patient says they haven’t tried yet.  Cymbalta is looking to capture part of that market.

The analgesic mechanism of Cymbalta is (as with much in psychiatry) unknown.   Some have argued it works by relieving the depression and anxiety experienced by patients in pain.  It has also been proposed that it activates “descending” pathways from the brain, helping to dampen “ascending” pain signals from the body.  It might also block NMDA receptors or sodium channels or enhance the body’s own endorphin system.  (Click on the figure above for other potential mechanisms, from a recent article by Dharmshaktu et al., 2011.)

But the more important question is:  does it work?  There does seem to be some decent evidence for Cymbalta’s effect in fibromyalgia and diabetic neuropathy in several outcome measures, and in a variety of 12-week trials summarized in a recent Cochrane review.

The evidence for musculoskeletal pain is less convincing.  In order to obtain approval, Lilly performed two studies of Cymbalta in osteoarthritis (OA) and three studies in chronic low back pain (CLBP).  All CLBP studies showed benefit in “24-hour pain severity” but only one of the OA studies showed improvement.   The effects were not tremendous, even though they were statistically significant (see example above, click to enlarge).  The FDA panel expressed concern “regarding the homogeneity of the study population and the heterogeneity of CLBP presenting to physicians in clinical practice.”  In fact, the advisory committee’s enthusiasm for the expanded indication was somewhat muted:

Even though the committee also complained of the “paucity of sound data regarding the pharmacological mechanisms of many analgesic drugs … and the paucity of sound data regarding the underlying pathophysiology,” they ultimately voted to approve Cymbalta for “as broad an indication as possible,” in order for “the well-informed prescriber [to] have the option of trying out an analgesic product approved for one painful condition in a patient with a similar painful condition.”

Incidentally, they essentially ignored the equivocal results in the OA trials, reasoning instead that it was OK to “extrapolate the finding [of efficacy in CLBP] to other similar musculoskeletal conditions.”

In other words, it sounds like the FDA really wanted to get Cymbalta in the hands of more patients and more doctors.

As much as I dislike the practice of prescribing drugs simply because they’re available and they might work, the truth of the matter is, this is surely how Cymbalta will be used.  (In reality, it explains a lot of what we do in psychiatry, unfortunately.)  But pain is a complex entity.  We have to be certain not to jump to conclusions—like we frequently do in psychiatry—when/if we see a “success story” with Cymbalta.

To the body, 60 mg of duloxetine is 60 mg of duloxetine, whether it’s being ingested for depression or for pain.  If a patient’s fibromyalgia or low back pain is miraculously “cured” by Cymbalta, there’s no a priori reason to think that it’s doing anything different in that person than what it does in a depressed patient (even though that is entirely conceivable).  The same mechanism might be involved in both.

The same can be said for some other medications with multiple indications.  For example, we can’t necessarily posit alternate mechanisms for Abilify in a bipolar patient versus Abilify in a patient with schizophrenia.  At roughly equivalent doses, its efficacy in the two conditions might be better explained by a biochemical similarity between the two conditions.  (Or maybe everything really is bipolar!  —sorry, my apologies to Hagop Akiskal.)

Or maybe the medication is not the important thing.  Maybe the patient’s perceived need for the medication matters more than the medication itself, and 60 mg of duloxetine for pain truly is different from 60 mg duloxetine for depression.  However, if our explanations rely on perceptions and not biology, we’re entering the territory of the placebo effect, in which case we’re better off skipping duloxetine (and its side effect profile and high cost), and just using an actual placebo.

Bottom line:  We tend to lock ourselves into what we think we know about the biology of the condition we’re treating, whether pain, depression, schizophrenia, ADHD, or whatever.  When we have medications with multiple indications, we often infer that the medication must work differently in each condition.  Unless the doses are radically different (e.g., doxepin for sleep vs depression), this isn’t necessarily true.  In fact, it may be more parsimonious to say that disorders are more fundamentally alike than they are different, or that our drugs are being used for their placebo effect.

We can now add chronic pain to the long list of conditions responsive to psychoactive drugs.  Perhaps it’s also time to start looking at pain disorders as variants of psychiatric disorders, or treating pain complaints as symptoms of mental disorders.  Cymbalta’s foray into this field may be the first attempt to bridge this gap.

Addendum:  I had started this article before reading the PNAS article on antidepressants and NSAIDs, which I blogged about earlier this week.  If the article’s conclusion (namely, that antidepressants lose their efficacy when given with pain relievers) is correct, this could have implications for Cymbalta’s use in chronic pain.  Since chronic pain patients will most likely be taking regular analgesic medications in addition to Cymbalta, the efficacy of Cymbalta might be diminished.  It will be interesting to see how this plays out.

21 Comments | medications, pain, psychopharmacology | Tagged: , , , , , | Permalink
Posted by stevebMD

The Painful Truth of Antidepressants

April 25, 2011

In a study published today, scientists at Rockefeller University proclaim that SSRI antidepressants (like Prozac and Celexa) may lose their efficacy when given with anti-inflammatory drugs like ibuprofen and aspirin.  Considering the high prevalence of depression and the widespread use of both SSRIs and anti-inflammatory medications, this result is bound to receive much attention.  As a matter of fact, it’s tantalizing to jump to the conclusion (as has been done in the Fox News and WSJ reports on this study) that the reason SSRIs may be so ineffective is because so many people with depression also use non-steroidal anti-inflammatory drugs (NSAIDs).

By my read of the data, it may be a bit too early to draw this conclusion.  Nevertheless, the study, by Paul Greengard, Jennifer Warner-Schmidt, and their colleagues, and published online in the Proceedings of the National Academy of Sciences, does propose some interesting mechanisms by which anti-inflammatory agents may affect antidepressant action.

The majority of the work was performed in mice, for which there are valid “models” of depression that are routinely used in preclinical studies.  In past work, Greengard’s group has shown that the expression of a small protein called p11 (which is associated with the localization and function of serotonin receptors) is correlated with “antidepressant-like” responses in mice, and probably in humans, too.  In the present study, they demonstrate that the antidepressants Prozac and Celexa cause an increase in expression of p11 in the frontal cortex of mice, and, moreover, that p11 expression is mediated by the ability of these antidepressants to cause elevations in interferon-gamma (IFN-γ) and tumor necrosis factor-alpha (TNF-α).  In other words, antidepressants enhance neural expression of these cytokines, which, in turn, increases p11 activity.

However, when mice are given NSAIDs or an analgesic (i.e., ibuprofen, naproxen, aspirin, or Tylenol), this prevents the increase in p11, as well as the increase in IFN-γ and TNF-α.  NSAIDs also prevent the “antidepressant-like” behavioral responses elicited by Celexa (as well as other antidepressants like Wellbutrin, Parnate, and TCAs) in mouse models of depression.

The group went one step further and even created a p11 “knockout” mouse.  These mice had no response to Celexa, nor did they have antidepressant-like responses to injections of IFN-γ or TNF-α.  However, the p11 knockout mice did respond to desipramine, an antidepressant that works mainly on norepinephrine, thus emphasizing the significance of serotonin in the p11-mediated response.

What does this mean for humans?  To answer this question, the group analyzed data from STAR*D, a huge multicenter antidepressant trial.  In the first stage of STAR*D, all patients (total of approximately 1500 individuals) took Celexa for a 12-week period.  The remission rate for patients who took an NSAID at any time during this 12-week period was only 45%, while those who took no NSAID remitted at a rate of 55%.

So does this mean that people taking antidepressants should avoid NSAIDs, and just deal with their pain?  Probably not. (In fact, one might ask the opposite question:  should people with chronic pain avoid SSRIs?  Unfortunately, the study did not look at whether SSRIs inhibited the pain-relieving effects of NSAIDs.)

In my opinion, some of the mouse data need to be interpreted carefully.  For instance, the mice received extremely high doses of NSAIDs (e.g., ibuprofen at 70 mg/kg/d, which corresponds to 4200 mg/d for a 60-kg man, or 21 Advil pills per day; similarly, the mice drinking aspirin received 210 mg/kg/d, or 12,600 mg = ~39 pills of regular-strength aspirin per day for a typical human).  Also, in the behavioral studies the mice received NSAIDs for an entire week but received only a single injection of Celexa (20 mg/kg, or about 1200 mg, 60 pills) immediately before the behavioral experiments.

The human data, of course, are equally suspect.  Patients in the STAR*D study were counted as “NSAID users” if they described using NSAIDs even once in the first 12 weeks of the study.  It’s hard to see how the use of ibuprofen once or twice in a three-month period might interfere with someone’s daily Celexa.  (Not to mention the fact that the “remission” data from STAR*D have come under some scrutiny themselves – see here and here).  Moreover, as the authors point out, it’s quite likely that patients with more severe forms of depression also had concurrent pain syndromes and used NSAIDs more frequently.  In other words, NSAID use might not attenuate SSRI activity, but may be a sign of depression that is more resistant to SSRIs.

In the end, however, I find the study to be quite provocative.  Certainly the correlation of antidepressant effect with expression of the p11 protein and with TNF-α and IFN-γ activity suggests a novel mechanism of antidepressant action—as well as new markers for antidepressant activity.  Moreover, the potential roles of NSAIDs in reducing antidepressant effects (or, in some cases, enhancing these effects), need to be explored.

But it raises even more unanswered questions.  For one, how do we reconcile the fact that antidepressant effects are associated with increased TNF-α and IFN-γ activity in the brain, while increases in these cytokines in the periphery are thought to cause depression?  Also, how can we explain the fact that other analgesic compounds, such as tramadol and buprenorphine, might actually have an antidepressant effect?  Finally, what does this mean for our treatment of pain symptoms in depression?  Should we avoid SSRIs and use other types of antidepressants instead?  Do NSAIDs inhibit the effects of SNRIs like Cymbalta, which has recently been FDA-approved for the treatment of chronic musculoskeletal pain (and whose users are most certainly also taking medications like NSAIDs)?

It’s great that the interface between mental illness and physical syndromes is receiving some well-deserved attention.  It’s also exciting to see that the neuroscience and pharmacology of depression and pain may overlap in critical ways that influence how we will treat these disorders in the future.  Perhaps it may also explain our failures up to now.  With future work in this area, studies like these will help us develop more appropriate antidepressant strategies for the “real world.”

[Finally, a “hat tip,” of sorts, to Fox News, which first alerted me to this article.  Unfortunately, the story, written by Dr. Manny Alvarez, was fairly low on substance but high on the “wow” factor.  It drew some broad conclusions and—my biggest pet peeve—did not refer the reader to any site or source to get more detailed information.  Alas, such is the case with much public science and medicine reporting: Alarm first, ask questions later.]

28 Comments | depression, medications, pain, psychopharmacology | Tagged: , , , , , , , | Permalink
Posted by stevebMD

Curbing Prescription Addiction

April 21, 2011

It should come as no surprise to anyone that prescription drug abuse is a serious problem.  As if we needed any reminder, a flurry of articles recently hit the press, showing just how serious the problem is.  Opioids (narcotic pain medications like Vicodin, Oxycontin, methadone, etc) are the most widely prescribed drugs in America, according to IMS and to a recent survey by the National Institute on Drug Abuse (NIDA), and prescriptions can lead to misuse, abuse, and dependence.

Predictably, the government plans to get involved.  As the New York Times reported earlier this week, the Obama administration wants to create legislation “requiring doctors to undergo training” before being permitted to prescribe opioid pain meds.

Hearing “government” and “training” in the same sentence doesn’t exactly inspire confidence.  What will the “training” consist of?   An online seminar?  A paper-and-pencil exam from the DEA?  A separate section on “managing pain patients” in our Board Certification exams?

[And didn’t we do this already?  As a matter of fact, yes, we did:  Back in 2000, JCAHO (the “Joint Commission” which accredits healthcare organizations) required doctors to undergo training to recognize and treat pain disorders.  Back then, we were told that we weren’t treating pain often enough.  Maybe the pendulum has swung too far in the other direction?  Maybe we’ve done our job too well?]

With respect to the prevention of opioid abuse, I agree it’s a good idea for doctors to recognize the warning signs of addiction, to implement monitoring procedures (like random urine tests and treatment contracts), to deny early refills, and to inquire about other risk factors for abuse.  Sadly, many doctors don’t take these measures and need encouragement to do so.  But something tells me that simply providing government-mandated “prescriber education” won’t fix the problem.

In my opinion, there are two other important issues to be addressed before this “training” will prove to be useful.

The first is to get rid of existing inefficiencies.  The truth is, most doctors already know the proper steps for prescribing potentially abusable opioids to pain patients.  Some clinics (particularly pain clinics) follow these steps with all patients, simply as a matter of course.  But in most treatment settings these steps are difficult to take.  Regular urine monitoring is cumbersome and intrusive (although relatively inexpensive); generating a treatment contract takes time (although it’s arguably the most important “paperwork” of the appointment); and reviewing a patient’s full medication history is a challenge.  Moreover, most of our non-patient-care resources and personnel are devoted to billing and data entry, rather than in these ancillary services that, in the long run, are far more important to cost-effective patient care.

[A side note: many states provide a “prescription drug monitoring” service to permit doctors to view prescriptions for controlled substances that any patient has filled in that state.  However—at least in California—the application process takes 3-6 months, the data are typically delayed about 2-3 months, it does not include non-controlled drugs, and not all pharmacies participate.  It still blows my mind that for the last 10-15 years it has been easier to purchase airplane tickets online or to send computer files halfway around the globe than to determine whether the patient sitting in my office has filled a prescription for OxyContin in the last 90 days.  Simply improving the existing technology would be the most immediately beneficial step.]

The second—and, in my opinion, more important—item is for doctors to understand what is the goal of treatment.  Not just “relief of pain,” but when (if at all) can the treatment be said to be complete?  I’ve written about this before (see “When Does Treatment End?”), and I’m convinced it’s an important question not just in the treatment of pain but in the management of all conditions, even those we consider “chronic.”  I believe that all prescribers need to ask themselves, “How long will the patient need this medication?” and engage the patient in this discussion, too.

I frequently see patients who have been prescribed opioid pain medications, or benzodiazepines or stimulants (not to mention SSRIs or other psychiatric meds), who have no idea how long they’ll need to take them.  They just “got a script.”  And because these medications are highly reinforcing (they relieve pain or anxiety, and sometimes have a pleasant psychoactive effect as well), they’ll continue to ask for more.  Why shouldn’t they, since they were never told they should stop?

In any treatment setting, the patient and doctor should have a mutual understanding of the goals and likely duration of treatment.  This plan can (and should) be flexible, but it should always have some realistic end point.  Moreover, we should always measure our progress relative to that goal, rather than “kicking the can down the road” and letting someone else deal with the discontinuation of care later.  I don’t think doctors should be in the business of denying care to patients, but if we’ve already had the discussion of when treatment might end, the issue of “no” has already been raised, and the patient understands this.

How would this minimize the abuse of addictive medication?  For one thing, it would limit access to the drug because we, a priori, are refusing to provide an endless supply.  In turn, this helps the patient recognize that everything is being prescribed for a particular purpose, whether for the transient relief of post-op pain or the longer-term management of cancer pain.  If and when other symptoms emerge, they need to be discussed and treated separately—or a more comprehensive treatment plan should be developed, if the evolving symptoms fit a characteristic pattern.

I know this is a tall order, and these suggestions may be hard to implement in many of the places where narcotic prescribing is common (ERs, urgent care clinics, etc).  But they are important measures to take.  We need to take the steps we know we should take (rather than wait for the government to tell us to do so—because we know how that will turn out).  And we need to think about patients as people with the power to heal, and plan for the healing process to take place, rather than give knee-jerk reactions (i.e., prescriptions) for symptoms.  If we do this, patients will be less likely to take matters into their own hands and “self-medicate,” and the outcome of treatment will be better for all.

19 Comments | addiction, medications, pain | Tagged: , , , , | Permalink
Posted by stevebMD

%d bloggers like this: