January 20, 2010

Petition to Ban Fibromyalgia Drug Milnacipran (Savella)

Margaret Hamburg, M.D., Commissioner
U.S. Food and Drug Administration
5600 Fishers Lane
Rockville, MD  20857

Dear Dr. Hamburg:

Public Citizen, representing more than 65,000 consumers nationwide, hereby petitions the Food and Drug Administration (FDA), pursuant to the Federal Food, Drug, and Cosmetic Act 21 U.S.C. Section 355(e)(3), and 21 C.F.R. 10.30, to immediately remove from the market the drug Savella (milnacipran; Cypress Bioscience, Inc. and Forest Laboratories, Inc.) because it has highly questionable clinical efficacy and has been found, in randomized controlled trials, to cause a large number of potentially serious adverse reactions including hypertension, increased heart rate, and increased suicidal ideation. On July 23, 2009, milnacipran’s approval for fibromyalgia was denied in the European Union for these very same efficacy and safety reasons.

Current FDA-approved Indication

Milnacipran was approved January 14, 2009, for the management of fibromyalgia in adults, its only indication in the U.S.; marketing began in May 2009.

Foreign approvals

Milnacipran was approved in Europe and Japan in 1997 for the treatment of major depressive disorder and generalized anxiety disorder.[1]

Mechanism of Action

Milnacipran is an inhibitor of the uptake of two important neurotransmitters, serotonin and norepinephrine, and as such, is referred to as a selective norepinephrine and serotonin re-uptake inhibitor or SNRI. Normally, serotonin and norepinephrine action would end with their reuptake into cells. However, when milnacipran is present, blocking reuptake, levels of serotonin and norepinephrine remain high for a much longer time, continuing their action on target tissues. Since serotonin and norepinephrine receptors are widespread in the body, the elevated levels of these neurotransmittor can cause reactions in many tissues.

Advisory Committee

There has never been an advisory committee for any of the three FDA-approved drugs for fibromyalgia (Lyrica, Cymbalta, and Savella); thus, none of the issues relating to the definition of the disease, endpoint measurements, risk/benefit ratio, length of studies, and adverse reactions have been discussed by experts in the field. Our first ability to analyze Savella occurred after the internal FDA reviews were posted online in March 2009.

Efficacy (Composite Endpoints)

Two composite endpoints were utilized for each of the two pivotal trials: Composite Pain and Composite Syndrome. Composite endpoints consist of at least two single endpoints that are evaluated separately and given equal weight when combined. Advantages include statistical precision along with smaller, less costly trials since the total number of events is increased. However, efficacy on the composite endpoint may hide a lack of efficacy in an individual endpoint, as was the case here.

Definitions of Terms Used in Defining Endpoints

VAS (Visual Analog Scale): a measurement of pain, ranging from 0 (no pain) to 100 (worst possible pain), recorded daily on each patient’s individual electronic device.

FIQ-PF (Fibromyalgia Impact Questionnaire- Physical Function): an 11-question section of the questionnaire rating the ability to perform physical activities.[2]

PGIC (Patients’ Global Impression of Change): a 7 point scale (rated from much worse to very much improved)[3],[4] that asks patients to describe the change, if any, in their sense of well-being as defined by activity limitations, symptoms, emotions, and overall quality of life.

SF-36 PCS (Short Form – 36 questions with Physical and Mental Component): asks questions concerning general health, vitality, physical and mental functioning.[5]

Definitions of Composite Endpoints

Definition of Composite Pain Endpoint

The composite pain endpoint combined patients’ pain data (VAS) with questions on how patients felt they were doing (PGIC).

Definition of Composite Syndrome Endpoint

The composite syndrome endpoint combined pain data (VAS), function data (SF-36 PCS), and patient global impression of change (PGIC).

Problems with approval:

Trial #031 (first clinical trial)

The original protocol had a single endpoint as measured between baseline and both three and six months.[6] Success required a ≥30% improvement above baseline in pain scores as measured on the VAS, a score of 1, 2, or 3 on PGIC, and an improvement of ≥20% in FIQ-PF and/or ≥5 units improvement in SF-36 PCS (see definitions above).

When this first completed pivotal trial (#031) failed to reach statistical significance at either three or six months, changes were made to the protocol of the ongoing second, pivotal trial (#02) which resulted in trial #02 reaching statistical significance. Changes to protocol #02 included replacing FIQ (which proved to be “unresponsive”, i.e. did not show an advantage for the drug) with PGIC, increasing the number of patients in the trial, excluding patients with moderate to severe depression, increasing the size of the trial, limiting the trial’s duration to three months, and changing the method for handling missing data. When study #031 was reanalyzed (post-hoc) with this modified protocol, results became statistically significant. Even with these manipulations, the absolute differences between placebo and treated groups, ranged only between 7% and 9% (i.e., more than 90% of patients received no benefit from the drug beyond what they would have gotten from a placebo).

Study #031 Per cent of patients achieving the composite endpoints (initial trial, reanalyzed)[7]

	Placebo N=223	100 mg/day N=224	200 mg/day N=441	Change (%)	Significance vs. placebo at 100 & 200 mg
Composite pain
3-month	43 (19%)	61 (27%)	118 (27%)	8%; 8%	P=0.048 & 0.033
6-month	39 (17%)	53 (24%)	104 (24%)	7%; 7%	P=0.11   & 0.070
Composite syndrome
3-month	27 (12%)	44 (20%)	85 (19%)	7%, 8%	P=0.035 & 0.020
6-month	27 (12%)	40 (18%)	73 (17%)	6%, 5%	P=0.10   & 0.13

Study #02 Per cent of patients achieving composite endpoints (Second clinical trial; analysis of modified protocol)[8]

	Placebo N=401	100 mg/day N=399	200 mg/day N=396	Change (%)	Significance vs. placebo at 100 & 200 mg
Composite pain
3-month	66 (16%)	91 (23%)	98 (25%)	7%; 9%	P=0.025 & 0.0037
Composite syndrome
3-month	35 (9%)	58 (15%)	55 (14%)	6%; 5%	P=0.011 & 0.015

Although the trials reached statistical significance at three months using the modified protocol, neither the sponsors nor FDA could detect any efficacy for the pain reduction measurement by itself (measured by VAS). Nor was there statistical significance in the function measurements by themselves. The FDA statistician concluded that, “When the domains are considered separately, there is no evidence in either phase 3 study that milnacipran is associated with improvements in pain (i.e., pain domain only) or improvements in function (i.e., function domain only) at three months of therapy.”[9] Only when pain scores were combined with a function measurement (such as SF-36 or PGIC)[10] could the sponsor obtain statistical significance.

The table below demonstrates the minimal amount of change observed in measured pain scores in the second trial, as measured by VAS.

Pain Measurement (Study #02)[11]

Visual Analog Scale (0-100 mm)

Treatment group	Baseline (mm)	Mean Change (mm)	Difference from Placebo (mm)
Placebo	65.8	10.0
100 mg/day	64.5	12.4	2.4
200 mg/day	64.3	12.9	2.9

Normally, composite endpoints are used to look at the effects of drug treatment when there is difficulty deciding which individual outcome should be primary as well as to increase the efficiency and statistical precision of a trial. However, it is known that a major problem with composite endpoints is that, “although the composite as a whole may appear to be affected by treatment, the evidence for benefit relating to its most important constituent may not exist.”[12] In this case, it is the pain component which is the main diagnostic feature of fibromyalgia, as defined by the American College of Rheumatology[13] and which, by itself, lacks statistical significance.

Short- vs. Long-term efficacy

Since fibromyalgia is a chronic condition, the original FDA reviewing division requested 6-month studies. However, when an FDA reorganization resulted in a transfer of milnacipran to a new division, that division dropped the 6-month requirement (no explanation given), very convenient for the sponsors since neither pivotal trial showed any statistical efficacy beyond three months.[14] Yet, this drug is for management of a chronic disease.

Problems with approval: Safety

It is clear from the list of Warnings and Precautions in the milnacipran label that the use of milnacipran requires both very careful patient selection as well as patient monitoring since some of the adverse effects are life-threatening. The sponsors list thirteen major problems with additional precautions in other sections of the label. We have combined them in the list below:

1) Risk of suicide

2) Serotonin syndrome

3) Increase in blood pressure

4) Increase in heart rate

5) Increase in seizures

6) Hepatotoxicity

7) Physical dependence and withdrawal symptoms

8) Hyponatremia (low blood sodium)

9) Abnormal bleeding

10) Activation of mania

11) Urethral resistance

12) Testicular and ejaculation problems

13) Controlled Narrow-Angle Glaucoma: Mydriasis (prolonged dilation of pupil)

14) Aggravation of liver disease with alcohol use

15) Increased fracture risk

16) Hazard to the fetus and newborn

17) Hazard to nursing infant

18) Risk to those with renal impairment

19) Interactions with other drugs

20) Gastrointestinal disorders

1) Risk of suicide: Milnacipran has been marketed in Europe and Japan as an antidepressant, and although not marketed as such in the U.S., it is required to have the class black box antidepressant warning concerning increased risk of suicides in children, adolescents, and young adults. Families and caregivers are advised of the need to carefully monitor patients for behavior changes which should be immediately reported to their health care providers.

The label states that, for previously depressed patients,“the incidence of suicidal ideation was 0.5% in patients treated with placebo, 0% in patients treated with Savella 100 mg/day, and 1.3% in patients treated with Savella 200 mg/day.” These data led the Medical Reviewer to state that, “The data suggest that among patients with depression, treatment with milnacipran – particularly at the higher dose – could increase the risk of suicidal ideation.”[15] One needs to remember that this data comes from two trials only three months in duration in which the second, larger trial specifically excluded patients with moderate or severe depression.[16]

2) Serotonin syndrome: This potentially life-threatening syndrome can occur with any drug that raises serotonin levels. Thus, concomitant use with monamine oxidase inhibitors is contraindicated and concomitant use with triptans requires careful observation. Symptoms include changes in mental state including hallucinations and coma; changes in blood pressure and heart rate; lack of coordination; and/or nausea, vomiting, and diarrhea. Other drugs that can also increase serotonin levels are listed in the label under “Interactions with Other Drugs.”

3) Increase in blood pressure: Inhibitors of norepinephrine and serotonin reuptake have been associated with increases in blood pressure. In study trials, patients taking milnacipran, who had normal blood pressure at baseline, became hypertensive at the end of study (7.2% placebo vs. 19.5% at 100 mg/day). There was also an increase in the percentage of patients who were pre-hypertensive at baseline and became hypertensive: 9% of placebo vs. 14% of treated patients.[17] The drug label warns that there were “Cases of elevated blood pressure requiring immediate treatment.”[18]

4) Increase in heart rate: Mean increases in heart rate were seen in patients taking milnacipran, a finding common to SNRIs. Palpitations were reported as an adverse event. The Clinical Pharmacology Reviewer noted that an increase in heart rate of >10 beats per minute occurred in 7%, 34%, and 40% (placebo, 100 mg and 200 mg patients respectively). An increase of >20 beats per minute occurred in 0.3%, 8%, and 8% of (placebo, 100 mg and 200 mg). According to the Clinical Pharmacology Reviewer, “Based on this data, it can be concluded that all patients on milnacipran had higher blood pressure and heart rate than placebo at all doses.”[19]

Although cardiovascular changes were considered “modest,” the Medical Reviewer stated that, “the clinical implication[s] . . . is that if the effects persist throughout the dosing interval and with chronic treatment, for every 6 mm Hg change in blood pressure there is doubling of cardiovascular risk (death, myocardial infarction, and stroke). In the case of MLN [milnacipran] this increased risk would be of an estimated 50%.”[20]

An interdisciplinary review team from the Division of Cardiorenal Products reviewed a study of the effect of milnacipran on heart rate.[21] The reviewers found many problems with the conduct of the studies (FDA redacted all figures on QT interval so we have no way to independently look at the data).[22] The Medical Reviewer stated that “the effect of MLN [milnacipran] on the QT interval has not been fully elucidated and the data does not provide the needed safety information for the NME [new molecular entity]”.[23]

5) Increase in seizures: Seizures have occurred in patients treated with milnacipran when prescribed as an antidepressant. The label warns that milnacipran should be “prescribed with care in patients with a history of a seizure disorder.” There is no information as to how that translates into clinical practice.

6) Hepatotoxicity:
Human data
“There have been cases of increased liver enzymes and reports of severe liver injury, including fulminant hepatitis with milnacipran from foreign postmarketing experience.”[24] Elevations of a liver enzyme occurred in 3%, 6%, and 7% (placebo, 100 mg/day, and 200 mg/day). Patients with “substantial” alcohol use (not defined) or who have chronic liver disease are advised not take milnacipran. Even though a single 50 mg dose (half of the low dose) was used in studies testing patients with of hepatic impairment, “Overall, the data from both studies indicates that in moderate and severe hepatic impairment, AUC increased by 46% and 60%, respectively.” The FDA reviewer added, “Caution should be exercised when administering milnacipran in patients with moderate and severe hepatic impairment.”[25]

Animal Data
There was slight to mild focal necrosis (areas of dead tissue) in the livers of male rats after one year at all doses tested (beginning at 0.05 times) and slight to mild vacuolation (at 1.5 times the human exposure at the maximum recommended human dose). At the higher doses and longer exposures of the carcinogenicity study, eosinophilic focus and centrilobular vacuolation occurred at 0.73 times the human exposure at the maximum recommended human dose in male rats.[26]

7) Physical dependence and withdrawal symptoms: An FDA consult from the Controlled Substance Staff, “concluded that milnacipran can induce physical dependence based on the presence of a withdrawal syndrome in non-fibromyalgia patients following MLN [milnacipran] discontinuation.”[27]

The Medical Reviewer added that:

a) the sponsors had not submitted any prospective studies in humans on physical dependence;

b) the sponsor acknowledged that milnacipran produced a withdrawal syndrome in non-fibromyalgia patients;

c) the reviewing division had “reports of withdrawal-like adverse events following discontinuation of milnacipran in clinical trials with fibromyalgia patients;”

d) milnacipran is an SNRI and the “ability of other SNRIs and serotonin selective reuptake inhibitors (SSRIs) to produce physical dependence is well-known . . . Thus, the ability of milnacipran to produce physical dependence is consistent with what is known about drugs in this pharmacological class.”

The Controlled Substance Staff concluded that the label should state: “milnacipran has the ability to produce physical dependence.”[28] This information was put into the label which states that “withdrawal symptoms can be severe.”

8) Low blood sodium: Milnacipran treatment can cause low blood sodium – a potentially life-threatening condition – in the elderly, with those taking diuretics being at greater risk. Patients are advised to discontinue drug if headache, memory impairment, confusion, or weakness is seen. The seriousness of this condition is underscored with the warning that severe cases can lead to coma, respiratory arrest, and death.[29]

9) Abnormal bleeding: Milnacipran can increase bleeding, especially in patients taking aspirin, nonsteroidal anti-inflammatory drugs, warfarin, or other anti-coagulants. Bleeding can range from discolorations under the skin to life-threatening hemorrhages.[30]

10) Activation of mania: “Activation of mania has and hypomania have been reported in patients with mood disorders who were treated with other similar drugs treated with other similar drugs for major depressive disorder.” and “should be used cautiously in patients with a history of mania.”[31]

11) Urethral resistance: Because SNRIs can cause urinary retention, “caution is advised”, especially in men with prostatic hypertrophy or other urinary tract obstructive disorders. The label includes the following: “dysuria occurred more frequently in patients treated with Savella (1%) than in placebo-treated patients (0.5%).” However, this hides the large percentage of men (24%) who had dysuria compared to only 0.1% of women treated with milnacipran.[32]

12) Testicular and ejaculation problems: The label states that men who are taking milnacipran are “more prone” to testicular pain or ejaculation disorders. Omitted are the numbers: ejaculation problems occurred in 7.3% of treated men compared to 0% of placebo; erectile dysfunction occurred in 5.5% vs. 0% of placebo.[33]

13) Controlled narrow-angle glaucoma: SNRIs can cause prolonged dilation of the pupil and must not be used in patients with uncontrolled narrow-angle glaucoma. Blockage of the fluid outflow in the eye increases intraocular pressure thereby damaging the optic nerve and causing permanent damage to vision.[34].

14) Aggravation of liver disease with alcohol use: Because milnacipran approximately doubled the percent of patients who had increases in ALT, a liver enzyme and a sign of liver toxicity, and because milnacipran may aggravate pre-existing liver disease, it should not be prescribed in patients with “substantial alcohol use”  [not defined],[35]

15) Increased fractuse risk: A Medline analysis of clinical trials looking at antidepressant medications, bone mineral density, and fractures, concluded that, “SSRIs are unique among the classes of antidepressants because they potently and selectively inhibit the 5-HTT [5-hydroxytryptamine transporter]. Therefore, the 5-HTT provides a potential explanation for differences in bone outcomes seen between SSRI users and users of TCAs [tricyclic antidepressants] or other psychotropic medications.”[36]

16) Hazards to the fetus and newborn: 
a) Fetus 
(all drug exposure comparisons are the animal exposure compared to that in humans at the maximum recommended human dose)

Milnacipran was toxic to the fetuses of three species (rat, mouse, and rabbit) at very low drug exposures. Toxic effects were seen not only at low doses but even when tests were of low sensitivity, i.e, done in small numbers of animals.

Rabbits: Fetuses had a single extra rib at 1.5 times the human exposure.[37]

Rats: Fetuses died in utero at only 0.25 of the human exposure. There were also reduced rat pup weights and viability at only 0.2 the human exposure. The per cent of rats with dead fetuses at Caesarean section was 0%, 15%, 20%, and 50%[38] [at 0.0, 0.24, 0.73, and 3.0 times the human, respectively].

Mouse: Fetal weights were reduced at 0.6 times the human exposure.[39] Although no other fetal effects were seen, the pharmacology reviewer stated that the doses chosen for this study were not high enough.

b) Newborn
In rats dosed day 15 of pregnancy through day 21 postpartum, “the treatment with milnacipran had an adverse effect on survival and weight of F1 pups that was not due to alterations in maternal behavior.”[40] There was slower weight gain at 0.25 times and retarded development in pups of treated rats at 4 times the human exposure.[41]

According to the Pharmacology Reviewer, milnacipran exposure (as low as 0.7 that of humans), “might have an impact on the development of neuro-endocrine systems in rats when exposed to milnacipran in utero.”[42]

The Division Director acknowledged the dangers: “I concur with Drs. Mellon and Mukherjee [the pharmacology reviewers] that the findings of embryofetal lethality and reduced pup weights and viability appear to be treatment related, and the labeling for Savella should clearly discourage the use of this drug during pregnancy. . .”[43] Although the European label (for its approved use as an anti-depressant) states that “it is preferable” not to administer Milnacipran during pregnancy[44], the U.S. label says “use only if the potential benefit justifies the potential risk to the fetus.”

17) Hazard to nursing infant: The Japanese label states that if mothers are taking milnacipran, they “must be instructed not to breast-feed their baby”[45]; the European label states that, “breast-feeding is contraindicated.”[46] Based on these recommendations, the Cross Discipline Team Leader recommended similar labeling for the U.S. product.[47]

The Division Director added that, “the labeling for Savella should clearly discourage the use of this drug . . . in breast feeding women.”[48] Unfortunately, the current U.S. label merely states that a “decision should be made whether to discontinue the drug” but does add that “nursing while on Savella is not recommended.”

18) Risk to those with renal impairment: Excretion of milnacipran was mainly (93%) urinary with only about 4% in the feces, based on radioactive label.[49] Two pharmacokinetic studies compared 50 mg single doses (only half of the lower daily dose) in healthy subjects to patients with mild to severe renal impairment. There were increases in pharmacokinetic measurements of drug exposure (Cmax, AUC, T½) as a function of increasing renal impairment.[50] The Biopharmaceutical Reviewer concluded her review stating that, “As in the previous study [of renal failure and plasma drug levels], the main conclusion from this study is that milnacipran exposure is dramatically increased in patients with renal impairment. Dose adjustment is necessary, possibly in all stages of renal impairment.”[51]

In spite of this warning from the Biopharmaceutical Reviewer, the label says, “no dose adjustment is necessary in patients with mild renal impairment,” and advises only “caution in patients with moderate renal impairment”, and suggests a 50% dose reduction in patients with severe renal impairment.

19) Interactions with other drugs: There are a number of drugs that must be eliminated from patients’ use because they can cause serotonin syndrome when combined with milnacipran. These include monoamine oxidase inhibitors such as phenelzine (Nardil); migraine drugs such as sumatriptan (Imitrex); selective serotonin reuptake inhibitors such as citalopram (Celexa); other antidepressants such as clomipramine (Anafranil); and some herbal and dietary supplements, such as ginseng, St. John’s wort and tryptophan^.[52] 

20) Gastrointestinal disorders: All healthy subjects receiving a single 400 mg dose vomited, precluding study of that dose (since the drug couldn’t be absorbed). In this same group, after a single 100 mg dose, vomiting occurred in 58% and nausea in 71%. After a lower (single 50 mg dose), approximately 50% of subjects had nausea and 10-13% vomiting.[53] These rates declined over time to 7%-14%, perhaps, due in part, to acclimation and/or a drop out rate which ranged between 3.5 and 7%.

Public Citizen’s Analysis and Recommendations

Milnacipran’s approval was based on an analysis of two randomized, placebo-controlled trials. However, it was a twisted path that led to this conclusion: when the results of the first trial failed to achieve statistically significant results, the sponsors performed a post hoc re-analysis of the data and used those more favorable results to plan a second trial. Changes necessary to obtain significant results in the second trial included replacing a function endpoint, increasing the number of patients in the trial, excluding patients with moderate to severe depression, increasing the size of the trial, limiting the trial’s duration to three months, and changing the method for handling missing data. Statistically significant results also required employing composite, rather than single, endpoints.

When significant results were achieved in the second trial, the reviewing division agreed to allow the sponsors to reanalyze the first non-significant study using the “analyses and patient population criteria that were specified” for the second study (as stated above).[54] This reanalysis did result in statistical significance, but only at three, not the six months originally studied.

The fact that significant results could be seen only at three months is not very useful with a chronic disease like fibromyalgia. In addition, in spite of employing a post hoc analysis, limiting treatment to three months, and using composite endpoints (not single endpoints like pain), the sponsors were still only able to obtain a benefit of between 7% and 9% vs. those receiving placebo (more than 90% received no benefit at all).

The Division Director, the Cross Discipline Team Leader, and the Statistical Reviewer were all puzzled over the lack of effect on pain, the main issue for patients with fibromyalgia. The Division Director stated that, although, “the dominant feature of FM [fibromyalgia] is pain, . . . analyses of the individual pain endpoints did not demonstrate a statistically significant treatment for Savella on the pain endpoint in either of the clinical trials.”[55]

Nor was there efficacy on the other components of the composite endpoints when analyzed singly. The Statistical Reviewer concluded, “There is no evidence in both Phase 3 studies that milnacipran is associated with improvements in pain (i.e. pain domain only) or improvements in function (i.e. function domain only) at three months of therapy.”[56]

In addition to milnacipran’s questionable efficacy, there are a multitude of precautions and warnings relating to milnacipran use, many capable of causing great harm, including increases in blood pressure and heart rate and a risk of suicide. The possible occurrence of these adverse events is a high a price to pay for a such a small amount of possible efficacy.

A recent study demonstrated a statistically significant uptake of infused serotonin into a multitude of tissues including kidney, bladder, prostate, seminal vesicle, femur, liver, and heart.[57] The fact that these tissues take up significant amounts of serotonin provides a possible basis for some of the toxicity observed in these and other organs.

The European Medicine Agency, as of July 23, 2009, turned down the applications for both Lyrica (pregabalin)[58] and Savella (Impulsor; milnacipran) [59] for the fibromyalgia indication citing both the lack of efficacy and data on long-term effects. After the drugs’ sponsors challenged that decision, the European Medicine Agency reconfirmed its original decision.[60] FDA should never have approved Savella for fibromyalgia, and should now immediately undo its error by removing it from the market before large numbers of people in this country suffer serious harm from this marginally effective drug.

ENVIRONMENTAL IMPACT STATEMENT

Nothing requested in this petition will have an impact on the environment.

CERTIFICATION

We certify that, to the best of our knowledge and belief, this petition includes all information and views on which this petition relies, and that it includes representative data and information known to the petitioners which are unfavorable to the petition. 

Sincerely,

Elizabeth Barbehenn, Ph.D.
Research Analyst

Sidney Wolfe, M.D.
Director, Health Research Group
Public Citizen