If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Address reprint requests to Karin B. Jensen, PhD, Massachusetts General Hospital, Harvard Medical School, Psychiatric Neuroimaging Building, 120 Second Avenue, Charlestown, MA 02129.
Department of Psychiatry, Massachusetts General Hospital, Harvard Medical School, Boston, MassachusettsAthinoula A. Martinos Center for Biomedical Imaging, Boston, Massachusetts
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, SwedenOsher Center for Integrative Medicine, Karolinska Institutet, Stockholm, Sweden
Department of Clinical Neuroscience, Karolinska Institutet, Stockholm, SwedenOsher Center for Integrative Medicine, Karolinska Institutet, Stockholm, Sweden
Milnacipran responders reduced their pain sensitivity, compared to placebo responders.
•
Milnacipran responders had increased neural activity in the posterior cingulum.
•
A short history of pain predicted positive response to milnacipran, not to placebo.
Abstract
Antidepressant drugs are commonly used to treat fibromyalgia, but there is little knowledge about their mechanisms of action. The aim of this study was to compare the cerebral and behavioral response to positive treatment effects of antidepressants or placebo. Ninety-two fibromyalgia patients participated in a 12-week, double-blind, placebo-controlled clinical trial with milnacipran, a serotonin-norepinephrine reuptake inhibitor. Before and after treatment, measures of cerebral pain processing were obtained using functional magnetic resonance imaging. Also, there were stimulus response assessments of pressure pain, measures of weekly pain, and fibromyalgia impact. Following treatment, milnacipran responders exhibited significantly higher activity in the posterior cingulum compared with placebo responders. The mere exposure to milnacipran did not explain our findings because milnacipran responders exhibited increased activity also in comparison to milnacipran nonresponders. Stimulus response assessments revealed specific antihyperalgesic effects in milnacipran responders, which was also correlated with reduced clinical pain and with increased activation of the posterior cingulum. A short history of pain predicted positive treatment response to milnacipran. We report segregated neural mechanisms for positive responses to treatment with milnacipran and placebo, reflected in the posterior cingulum. The increase of pain-evoked activation in the posterior cingulum may reflect a normalization of altered default mode network processing, an alteration implicated in fibromyalgia pathophysiology.
Perspective
This study presents neural and psychophysical correlates to positive treatment responses in patients with fibromyalgia, treated with either milnacipran or placebo. The comparison between placebo responders and milnacipran responders may shed light on the specific mechanisms involved in antidepressant treatment of chronic pain.
The use of antidepressant drugs to treat fibromyalgia (FM) has rapidly increased with the increased reports of their efficacy. There is, however, little knowledge about the mechanisms underlying this treatment effect. The new insights from research focusing on central nervous system function in pain modulation have been of paramount importance for the understanding of chronic pain syndromes such as FM. Today, FM is commonly regarded as a syndrome of pain amplification,
Investigating the mechanisms behind successful treatment with antidepressants in FM may therefore contribute to the general understanding of treatment in related functional pain syndromes.
The diagnosis of FM used in the present study relies on the American College of Rheumatology 1990 criteria,
A randomized, double-blind, placebo-controlled trial of duloxetine in the treatment of women with fibromyalgia with or without major depressive disorder.
A randomized, double-blind, placebo-controlled trial of duloxetine in the treatment of women with fibromyalgia with or without major depressive disorder.
and at present the latter 2 are approved for treatment of FM in the United States. Despite the wide use of antidepressant drugs in chronic pain syndromes, little is known about their mechanisms of action. So far, clinical studies have shown that antidepressant and analgesic effects are independent of each other,
A randomized, double-blind, placebo-controlled trial of duloxetine in the treatment of women with fibromyalgia with or without major depressive disorder.
Further support comes from a neuropathic pain study where behavioral data indicated that the degree of baseline dysfunction of endogenous pain modulation predicted and was related to duloxetine treatment response, whereas ratings of depression were not.
Thus, if the treatment effect of antidepressants for chronic pain cannot be explained by reduced levels of depression, what are the neural mechanisms underlying reduced pain?
In this double-blind, placebo-controlled, randomized trial, 92 FM patients were treated with the serotonin-norepinephrine reuptake inhibitor milnacipran or with placebo for 12 weeks. Behavioral and brain imaging assessments were performed before and after treatment. The aim was to characterize pain-related cerebral activations of milnacipran responders as compared to placebo responders. Another aim was to compare the sensitivity to evoked pain in milnacipran and placebo responders.
Treatment success varies in FM patients. Here, we investigated all patients who responded positively to either milnacipran or placebo and compared their cerebral function before and after treatment. We hypothesized that the route to symptom reduction would be different and that segregated mechanisms of action would be elucidated by functional magnetic resonance imaging (fMRI).
Methods
Patients
A total of 157 patients were screened, of whom 92 were randomized and included in the trial (46 to receive milnacipran and 46 to receive placebo); see Fig 1. The average age was 44 years (standard deviation = 8.2), and the average duration of pain symptoms was 135 months (standard deviation = 95). Patients eligible for inclusion were female, aged 18 to 55 years, fulfilling the American College of Rheumatology 1990 criteria for FM,
and with a self-reported average weekly pain intensity of at least 40 mm on a 100-mm visual analog scale (VAS), ranging from “no pain” to “worst imaginable pain.” Exclusion criteria included presence of severe psychiatric illness; significant risk of suicide; a history of substance, drug, or alcohol abuse; significant cardiovascular/pulmonary disease (including electrocardiograph abnormalities and hypertension); liver disease; renal impairment; pregnancy; or breastfeeding. All therapies that might interfere with the current treatment were prohibited, for example, other antidepressants and mood stabilizers, analgesics (tramadol, codeine, dextropropoxyphene), strong opioids including patches, anesthetic transdermal patches, anticonvulsants, centrally acting relaxants, joint injections, trigger/tender point injections, biofeedback, and transcutaneous electrical nerve stimulation. Analgesics such as paracetamol and dipyrone were allowed as rescue medication, and short-term use of zolpidem was allowed as the only treatment of insomnia. Nonsteroidal anti-inflammatory drugs were allowed under investigators' control. All rescue medications and nonsteroidal anti-inflammatory drugs had to be discontinued 48 hours before the assessments of pain sensitivity and fMRI. The reason for not including men in the study was the high female-to-male ratio of FM prevalence in the population
Treatment outcome data were available from 84 patients (44 milnacipran and 40 placebo), among whom 70 were study completers (32 milnacipran and 38 placebo). In total, 10 patients were excluded from fMRI analyses because of poor image quality, movement during scans, or findings of intracranial anomalies, leaving 60 patients for the fMRI analysis before and after treatment (30 milnacipran and 30 placebo). The full data set of 60 study completers with intact fMRI data was used for analyses in the present study (see Fig 1). The demographics are listed in Table 1. This study was approved by the ethical committee at each site, and informed consent was obtained from each participant before inclusion.
Figure 1Patient disposition and PGIC ratings. (A) Flow chart of the total number of patients included in the study and the number of patients who were categorized as treatment responders. A more detailed description of the patient disposition and reasons for treatment discontinuation can be found in Petzke et al.
(B) Subjective ratings of PGIC after 12 weeks' treatment with milnacipran or placebo represented in percentage. The numbers 1 to 7 correspond to the following response categories: 1 = very much improved, 2 = much improved, 3 = minimally improved, 4 = no change, 5 = minimally worse, 6 = much worse, and 7 = very much worse. Responders were defined as patients rating 1 to 3, and nonresponders as patients rating 4 to 7.
The study was a 12-week, double-blind, placebo-controlled, randomized trial assessing the effects of milnacipran 100 mg twice daily on the sensitivity to stimulus-evoked pain; see Fig 2. Patients were mainly recruited from primary care at the 3 study sites: London (England), Cologne (Germany), and Stockholm (Sweden). The screening visit was scheduled 7 to 28 days prior to inclusion and consisted of a clinical examination, confirmation of the FM diagnosis, laboratory tests, electrocardiographs, and questionnaires. Patients eligible for study inclusion were scheduled for a second visit after at least 7 days or the time needed for washout of medications. Required periods off medication were dependent on the actual pharmacologic characteristics. During the second visit, a final check of exclusion and inclusion criteria was made and stimulus response assessments of pressure pain sensitivity were performed. The following day, patients returned for an fMRI scan and then started the treatment (milnacipran/placebo). All randomized patients participated in the baseline assessments and received at least 1 dose of study treatment. Following a 3-week dose escalation phase, patients had a 9-week fixed-dose phase of milnacipran 100 mg twice daily (or placebo). Two follow-up visits were scheduled between baseline and study end, including checks of compliance, adverse events, pain ratings, and vital signs. Patients returned in week 12 (day 83 ± 1 day) for the evaluation of treatment effects followed by a 9-day down-titration phase. At week 12, patients also rated their subjective impression of treatment effect, using the Patient Global Impression of Change (PGIC) questionnaire: very much improved (1), much improved (2), minimally improved (3), no change (4), minimally worse (5), much worse (6), and very much worse (7). Treatment responders in the present study were a priori defined as patients reporting any type of improvement, that is, PGIC 1, 2, or 3. Nonresponders were defined as patients having no change or worsening of symptoms, that is, PGIC 4, 5, 6, or 7. PGIC is one of the most commonly used scales for evaluating patients' subjective impression of improvement in clinical trials, and the discrete ratings give a clear cut-off for treatment response.
Figure 2Study scheme. Screening was performed at least 1 week (and maximum 4 weeks) prior to treatment onset, depending on the length of the phase when all prohibited medications were washed out. Before treatment onset, there were several behavioral measures: average weekly pain rated on a VAS, FIQ, and stimulus response measures of pressure pain sensitivity (S-R). Neuroimaging assessments were performed by means of fMRI during calibrated pressure pain. Directly after the pretreatment fMRI session, patients were given their first dose of medication. After 12 weeks of treatment, the exact same behavioral and neuroimaging assessments were performed with the addition of the PGIC survey, assessing patient's subjective impression of treatment effect.
The data presented in the present study are from a pharmacologic clinical trial (EudraCT 2004-004249-16), and the results directly relating to the outcomes of the clinical trial (without subgrouping of responders and nonresponders) are presented elsewhere.
Stimulus Response Assessments and Calculation of P50
Pressure stimulations were applied to the left thumbnail using an automated, pneumatic, computer-controlled stimulator with a plastic piston that applies pressure via a 1-cm2 hard rubber probe.
Each subject was assessed for subjective pain ratings by administering 1 ascending series of pressure stimuli and 1 randomized series. Subjects rated the intensity of the pain evoked by each stimulus by putting a mark on a 100-mm horizontal VAS ranging from “no pain” to “worst imaginable pain.” A polynomial regression was used to determine each individual's representation of VAS 50 mm, built on a randomized series of 15 stimuli. The amount of pressure required to evoke pain at VAS 50 mm in each individual is referred to as P50 throughout this article.
Brain Imaging
Images were collected using 3 different 1.5-Tesla scanners: in London, a General Electric HDx scanner (GE Healthcare, Cleveland, OH) was used, in Stockholm a General Electric Twinspeed Signa Horizon (GE Healthcare) was used, and in Cologne a Philips scanner (Philips, Amsterdam, The Netherlands) was used. Multiple T2*-weighted single-shot gradient echo planar imaging sequences were used to acquire blood oxygen level–dependent contrast images. The following parameters were used: repetition time, 3,000 milliseconds (35 slices acquired); echo time, 40 milliseconds; flip angle, 90°; field of view, 24 × 24 cm; 64 × 64 matrix; 4-mm slice thickness with a .4-mm gap; sequential image acquisition order; and voxel size 2 × 2 × 4 mm. In the scanner, cushions and headphones were used to reduce head movement and dampen scanner noise. Visual distraction during scans was minimized by placing a blank screen in front of the patient's field of view. In addition to the functional scans, high-resolution T1-weighted structural images were acquired in coronal orientation. Parameters were as follows: spoiled gradient recalled 3D sequence; repetition time, 24 milliseconds; echo time, 6 milliseconds; flip angle, 35°, with a voxel size of .9 × 1.5 × .9 mm.
Cerebral pain representations were obtained by using pressure to the thumb during fMRI scanning. Two types of stimulations were used: the previously determined P50 and a nonpainful pressure perceived only as touch. All stimulations were randomly presented over the scanning time, preventing subjects from anticipating the onset time and event type. The time interval between consecutive events was randomized with a mean stimulus onset asynchronicity of 15 seconds (range, 10–20 seconds). The total duration of the scans was approximately 35 minutes. Before scanning, subjects were instructed to focus on the pressures on the thumb and to not use any distraction or coping techniques.
Pain Ratings and Questionnaires
At baseline, patients rated the duration of their widespread pain symptoms, depressive symptoms,
were rated at baseline and at the end of the study. The FIQ is an instrument designed to quantify the overall impact of FM over many dimensions (eg, function, pain level, fatigue, sleep disturbance, psychological distress). It is scored from 0 to 100 and represents increased severity with higher numbers.
Statistics for the fMRI Analysis
Data were analyzed in a blinded manner, using only patient codes that did not contain any information about treatment type or response. Only in the final stage of statistical processing was the PGIC score for each patient unblinded and the statistical files divided into 2 groups: responder and nonresponder. At that stage in the process, there was no possible subjective bias that could have affected the data analysis. Preprocessing and analyses of imaging data were performed using the Statistical Parametric Mapping 5 (SPM5) software (http://www.fil.ion.ucl.ac.uk/spm/; SPM, Wellcome Trust Centre for Neuroimaging, University College London, London, United Kingdom) and Matlab 7.1 (Mathworks, Natick, MA). During preprocessing, all functional brain volumes were realigned to the first volume, spatially normalized to a standard echo planar imaging template, and finally smoothed using an 8-mm full width at half maximum isotropic Gaussian kernel.
Data analysis was performed using the general linear model and modeling of the 2 different conditions (P50/nonpainful stimuli) convolved with a canonical hemodynamic response function as implemented in SPM5. A file containing the movement parameters for each individual (6 directions) was included in the model. Brain activation during nonpainful pressures was individually subtracted from activity during P50 in order to create a pain-specific contrast image and to control for individual differences in cerebral responsiveness.
Second-level analyses were used when testing for statistical effects on the group level. To determine if there was any variance in pain-evoked brain activity that could be explained by the site factor, an analysis of variance (ANOVA) was performed within SPM5, including the factors site and time point. Also, pain-evoked brain activity in a commonly activated anatomic location (secondary somatosensory cortex coordinate) was extracted for each individual and analyzed by a univariate ANOVA, using site as the between-subjects factor. These analyses were only performed in order to assess the possible contribution of the site effect on pain-evoked brain activations and did not relate to any experimental hypotheses.
Comparisons of pain-evoked brain activity between treatment groups were performed using 2-sample t-tests in SPM5. A significant result led to further exploration of individually extracted brain signals using an ANOVA within SPSS, version 16.0 (SPSS Inc, Chicago, IL) where the degree of response was a between-subjects factor with 3 levels: response (ie, very much improved, much improved, minimally improved); treatment as the second between-subjects factor (milnacipran/placebo); and brain signal intensity as the dependent variable.
All fMRI analyses were performed using an initial image threshold of P < .005 (uncorrected) with a spatial extent threshold of 20 contiguous voxels, and all reported results were family-wise error corrected at the cluster level (P < .05). Conjunction analyses were exploratory and performed with a more liberal initial threshold (P < .05). We did not use any region-of-interest analyses. Anatomic locations were expressed in Montreal Neurological Institute (MNI) stereotactic atlas coordinates (x, y, z).
Statistical Analysis of Behavioral Data
Baseline differences in the duration of pain symptoms were analyzed by a univariate ANOVA with the between-subjects factor group (placebo responders, milnacipran responders, and nonresponders). The effect of milnacipran treatment on P50 was analyzed by a 2-way repeated measures ANOVA. The within-subject variable was treatment with 2 levels (before and after), and the between-subjects variable was group. In case of significant overall effects, individual within-group comparisons were performed using a paired t-test, and between-group comparisons were performed using a 2-sample t-test. Because of the nonparametric properties of VAS ratings, the differences in VAS clinical pain ratings between groups were analyzed using the Kruskal-Wallis test. When overall significant differences were present for VAS ratings, individual comparisons were performed using Mann-Whitney's U test with correction for ties. Within-group correlations between the changes in pain ratings (VAS) and changes in P50 were analyzed by Spearman's rank-order correlation coefficient. A probability level of P <.05 was considered to represent a significant difference (2-tailed tests).
Results
Patient Data and Compliance
Among patients randomized to milnacipran treatment, there were 14 cases of premature withdrawal (30.4%): 11 due to adverse events (23.9%) and 3 due to therapeutic failure (6.5%). In the placebo group there were 8 cases of premature withdrawal (17.4%): 4 due to adverse events (8.7%) and 4 due to therapeutic failure (8.7%). Adverse events were reported by 45 patients treated with milnacipran (97.8%) and by 40 patients treated with placebo (87%). Most adverse events were of a mild or moderate type, and none of the serious adverse events (4 cases in the milnacipran group and 8 in the placebo group) were deemed related to the study drug.
Based on the patient's subjective impression of treatment effect (PGIC), 21 patients were classified as milnacipran responders and 16 as placebo responders. The nonresponders were distributed equally across treatment groups: 23 milnacipran nonresponders and 24 placebo nonresponders (see Fig 1).
Functional Imaging Data
Validation of the Pain Task and Use of Multiple Scanners
The main effect for pain-evoked brain activity at baseline was calculated using a 1-sample t-test (n = 60). Results showed significant increases of activity in 9 regions, including primary somatosensory cortex, secondary somatosensory cortex, bilateral insulae, anterior cingulum, and cerebellum, thus reproducing the commonly reported pain-processing regions previously described in the literature.
Results from the analysis of site-related differences in pain-evoked brain activity revealed no brain regions with significant variance that could be attributed to any of the sites (Stockholm, London, or Cologne) at any of the 2 time points (baseline/after treatment). Results from the SPSS analysis, where measures of pain-evoked signal intensities were extracted from a predefined coordinate in secondary somatosensory cortex (x = −42, y = −20, z = 20), revealed no significant difference between the 3 sites (Stockholm, London, and Cologne): F(2, 82) = .69, P = .51, ns.
Baseline Results
There was no baseline difference in pain-evoked brain activity between milnacipran responders and placebo responders, indicating that the cerebral response to pain at baseline would not confound the subsequent analyses of treatment effects between the 2 groups. A conjunction analysis of baseline data showed significant overlaps in regions implicated in pain processing, that is, the anterior cingulum, bilateral insulae, primary somatosensory cortex, and cerebellum (see Fig 3). For exploratory reasons, a 2-sample t-test between milnacipran responders and milnacipran nonresponders was performed, but it did not reveal any baseline differences either, indicating that subsequent treatment response to milnacipran was not influenced by pain-evoked cerebral activations at baseline.
Figure 3Pain activations before treatment in milnacipran and placebo responders. Illustration of the similarity between pain activations in milnacipran (blue color) and placebo responders (red color) before treatment. Purple color represents voxels where red and blue overlap completely. There were no regions with significant differences in pain activation before treatment between the 2 groups (P < .005, uncorrected). The image is thresholded at P = .01, uncorrected, for illustration purposes.
A 2-sample t-test was used to assess the difference in pain-evoked brain activity after treatment between milnacipran responders and placebo responders. Results demonstrate that milnacipran responders had significantly higher activity in the posterior cingulum after treatment compared to placebo responders (t = 3.99, MNI coordinates x = −4, y = −30, z = 46). The cluster included coordinates in both cerebral hemispheres and also extended into the precuneus (Fig 4A). There were no regions where placebo responders had significantly higher pain-evoked brain activity after treatment. A conjunction analysis of milnacipran and placebo responders revealed that the right and left amygdala were similarly activated after treatment, with higher signal intensity on the left side (MNI coordinates x = −28, y = −2, z = −12; z-value= 3.73, P < .001; family-wise error corrected at cluster level).
Figure 4Posttreatment fMRI and brain activations for different degrees of treatment response. (A) The anatomic region where pain-evoked brain activity was higher in milnacipran responders, compared to placebo responders, after treatment. The cluster is located in the posterior cingulum (x = −4, y = −30, z = 46). The exact anatomic location (x, y, z) is given in MNI coordinates. (B) Pain-evoked signal intensity in the posterior cingulum (x = −4, y = −30, z = 46) after treatment in milnacipran responders and placebo responders, clustered according to the magnitude of clinical improvement, rated by PGIC (±2 standard errors). ∗∗Significant main effect of treatment on posterior cingulum activity. (C) Correlation between pain-evoked signal intensity in the posterior cingulum after treatment and the difference in experimental pain tolerance, calculated as a subtraction of post minus pre P50 values (Δ).
In order to unpack the posterior cingulum activation that separated milnacipran responders from placebo responders after treatment, the signal intensity in the posterior cingulum was individually extracted (MNI coordinates x = −4, y = −30, z = 46) and grouped according to degree of treatment response in terms of clinical pain (Fig 4B). An ANOVA was performed in SPSS that showed a significant effect for treatment, F(1, 24) = 6.5, P < .05; that is, milnacipran responders exhibited higher activity in the posterior cingulum after treatment compared to placebo responders. We also found a significant correlation between the degree of improvement of experimental pain (P50) and posterior cingulum signal intensity after treatment in milnacipran responders (P = .04, 2-tailed) but not in placebo responders (P = .09, 2-tailed); see Fig 4C.
In an exploratory 2-sample t-test, we compared pain-evoked brain activity in milnacipran responders and milnacipran nonresponders after treatment in order to find out if there were activations uniquely expressed by responders to the drug. Results revealed that milnacipran responders demonstrated higher activity in the posterior cingulum after treatment compared to milnacipran nonresponders (t = 3.97; MNI coordinates x = 10, y = −28, z = 46). The opposite contrast did not yield any significant results when correcting for multiple comparisons; however, there was a trend toward increased activations in the left lateral prefrontal cortex in nonresponders (t = 3.5; MNI coordinates x = −34, y = 44, z = 16).
Behavioral Measures
The magnitude of the calibrated experimental painful stimulation (P50) did not differ between groups at baseline, F(2, 48) = .24, P = .79, ns. Following treatment, there was a significant effect for group, F(2, 48) = 3.8, P < .05, where P50 increased significantly in milnacipran responders (P < .001) but not in placebo responders (P = .35) or milnacipran nonresponders (P = .51), indicating a specific analgesic effect in milnacipran responders that was not caused by mere exposure to milnacipran (see Fig 5A). Moreover, there was a positive correlation between the reduced sensitivity to pressure and decrease in clinical pain ratings in milnacipran responders, r(21) = .59; P < .01, but not in placebo responders, r(16) = −.15, ns, or milnacipran nonresponders, r(23) = −.21, ns, indicating a specific correlation between reduced pressure sensitivity and reduced clinical pain in milnacipran responders (Fig 5B).
Figure 5Stimulus response assessments and clinical pain ratings. (A) The mean pressure (± standard error of the mean) required to achieve pain ratings of 50-mm VAS (P50) before and after treatment in milnacipran responders (n = 21) and placebo responders (n = 16). No statistically significant differences in P50 were seen between the 2 groups at baseline. Following treatment, P50 increased in milnacipran responders (∗∗P < .001) but not in placebo responders. (B) P50 improvement, calculated as a subtraction of posttreatment minus pretreatment values (Δ), and improvement of clinical pain (VAS), calculated as a pretreatment minus posttreatment difference (Δ), were positively correlated among milnacipran responders; r(21) = .59; P < .01.
The duration of widespread pain had a significant effect on treatment outcome, demonstrated by a significant overall effect for group, F(2, 59) = 5.78, P < .01. Milnacipran responders had a significantly shorter duration of widespread pain before entering the study (mean 85 ± 18 months) compared to milnacipran nonresponders (mean 169 ± 17 months), P < .005. There was no statistically significant difference between placebo responders (mean 115 ± 21 months) and milnacipran responders concerning this measure (P = .764, ns).
The intensity of the average weekly clinical pain (VAS) at baseline did not predict treatment outcomes, r(62) = −.018, P = .89, ns, suggesting that patients with all severities of pain symptoms were likely to respond to treatment. Not surprisingly, there was an overall difference in weekly VAS ratings between milnacipran responders and nonresponders after treatment (UA = 429, P < .001). There was no statistically significant difference in pain ratings between placebo responders and milnacipran responders after treatment (UA = 942, P = .828, ns), but both groups had significantly lowered pain ratings compared to milnacipran nonresponders (P < .001 and P < .001, respectively).
Ratings of FM impact (FIQ) did not differ between groups at baseline, F(2, 59) = .3, P = .74, ns. After treatment, however, there was a significant effect for group regarding the FIQ measure, F(2, 56) = 8.391, P < .001. FIQ ratings improved in milnacipran responders (P < .001) and placebo responders (P < .001), but not in milnacipran nonresponders (P = .068, ns), indicating no separation between milnacipran responders and placebo responders regarding FIQ.
Discussion
Our results suggest that the treatment response to milnacipran or placebo depends on partially segregated mechanisms. The assessment of blood oxygen level–dependent fMRI signals during evoked pain revealed that milnacipran responders had higher activations in the posterior cingulum after treatment compared with placebo responders. The posterior cingulum has repeatedly been linked to analgesic response to nonopioidergic treatment in chronic pain patients
Results revealed a graded effect for milnacipran response and posterior cingulum activity, demonstrating a pattern where strong treatment effects corresponded to large changes in posterior cingulum activity, moderate effects corresponded to moderate posterior cingulum activity, etc. Furthermore, we could show a correlation between the antihyperalgesic effect of milnacipran by means of increased tolerance of experimental pain, and increased posterior cingulum activity after treatment. The posterior cingulum is a core hub in the default mode network (DMN), a set of brain regions that forms an intrinsic network through synchronized spontaneous signal fluctuations.
Decreased activity in the posterior cingulum and precuneus during painful stimulation indicates that the DMN plays an important role in central processing of pain. For example, patients with chronic low back pain exhibit decreases in activity within the DMN,
and studies in acute pain suggest that an alteration of the DMN connectivity pattern, if repeated across time, might induce permanent disruptions of DMN function.
The results of the present study, including findings of increased posterior cingulum and precuneus activity in response to pharmacologic treatment, could reflect reduced ongoing pain signaling and concomitant normalization of DMN activity. The presence of clinical pain in patients could be compared with a goal-directed behavior that requires attention from the brain, resulting in lowered DMN activity. With increased pain relief, the activity of the DMN may be enhanced because of an increase of internal cognitive capacity that is available for self-referential mental processes,
detached from the directed focus on pain. In a recent study of a 4-week acupuncture treatment intervention in FM, Napadow and colleagues found normalized DMN connectivity after successful treatment, suggesting that intrinsic brain connectivity can be used as an objective marker for changes in FM pain.
The result from the fMRI contrast between milnacipran responders and milnacipran nonresponders strongly suggests that the posterior cingulum region is specifically related to drug response. Hypothetically, the mere exposure to 12 weeks of milnacipran could have generated differences in pain-evoked brain activity because of nonspecific changes in noradrenergic and serotonergic activity. However, the significant brain activity observed in milnacipran responders was not seen in nonresponders, suggesting that the posterior cingulum/precuneus region might specifically reflect drug response (because activity was also higher compared to placebo responders). There was a trend toward increased activation in the PFC in milnacipran nonresponders, compared to responders, reflecting the effect of being exposed to milnacipran for 12 weeks, without any improvement of FM symptoms. It is clear that the action of milnacipran affects patients differently, and we speculate that the increased activation in the PFC reflects a difference in the serotonergic and noradrenergic function of milnacipran, where the noradrenergic effect is mainly thought to affect pain processing (as seen in milnacipran responders) whereas the serotonergic effect may be more related to affective/cognitive evaluation, as antidepressant drugs target serotonergic receptors in the prefrontal cortex.
Our stimulus response assessments indicate that milnacipran responders had a specific antihyperalgesic effect that was not found in placebo responders or in milnacipran nonresponders. The reduced pressure pain sensitivity seen in milnacipran responders was therefore not secondary to the improvement of the condition itself (since no increase in P50 was found in placebo responders) and was not a nonspecific effect of mere exposure to milnacipran, unrelated to the clinical efficacy of the drug (because no increase of P50 was seen in milnacipran nonresponders). Moreover, the increase in P50 in milnacipran responders was correlated to the change in clinical pain ratings measured by weekly VAS and also to increased posterior cingulum activation after treatment.
Our results suggest that milnacipran could exert an analgesic effect by direct antinociceptive action and/or strengthening of the endogenous pain modulatory system, resulting in decreased sensitivity to stimulus-induced pain (ie, antihyperalgesia). However, our fMRI data did not reveal any increase in activity in brain regions pertaining to the most central regions of the pain inhibitory system, arguing against strengthened endogenous pain modulation in response to milnacipran. One possible explanation could be that placebo analgesia and milnacipran-related pain relief operate through activation of endogenous pain inhibition, possibly canceling out the effects of endogenous pain modulation when contrasting milnacipran and placebo responders. A conjunction analysis of milnacipran and placebo responders revealed overlapping activity in the amygdala, a pain defensive region implicated in the motivational aspect of pain inhibition. We conclude that the endogenous pain inhibitory system is likely strengthened in milnacipran responders but with a specific interaction of the posterior cingulum/precuneus region that differentiates milnacipran from placebo responses.
There were no baseline characteristics of FM severity (VAS, FIQ) that predicted treatment response either to placebo or to milnacipran, suggesting that it was unlikely that responders in this trial were less seriously affected by their FM compared to nonresponders. Interestingly, milnacipran responders were characterized by a shorter duration of widespread pain compared to milnacipran nonresponders, and it is possible that early treatment with milnacipran might increase the response rate in patients with FM. It also raises the question of secondary prevention. Prospective studies have revealed that localized/regional pain is the major risk factor for the development of chronic widespread pain,
Lack of pressure pain modulation by heterotopic noxious conditioning stimulation in patients with painful osteoarthritis before, but not following, surgical pain relief.
Treating these conditions with drugs that target central pain modulation, such as serotonergic-norepinephrine inhibitors, is probably going to be an effective strategy in the future, as indicated by previous trials of chronic low back pain
The serotonergic-norepinephrine inhibitor drug duloxetine is approved for treatment of chronic musculoskeletal pain in the United States, indicating that similar drugs will soon be used in various musculoskeletal pain syndromes. Hypothetically, these drugs might also be useful as a means of secondary prevention of widespread pain. Studies addressing this clinically relevant issue are urgently needed.
Our study suffers from several limitations. The mechanistic nature of this study required very stringent screening criteria, which means that patients with comorbidities and patients unable to withdraw from analgesic medication could not be enrolled. Another limitation to our study was that we studied cerebral mechanisms in response to a pharmacologic intervention without having the tools to directly investigate the neurochemical dynamics of the brain. The blood oxygen level–dependent fMRI results only let us speculate about the action of different neurotransmitters involved in the brain changes we see after treatment. In a future study, the use of positron emission tomography could possibly shed more light on the different transmitter substances involved in milnacipran treatment of patients with FM.
Conclusion
Our results point toward differences in the cerebral mechanisms underlying treatment response to placebo and milnacipran reflected in the posterior cingulum, a key region for the brain's DMN activity. There was also significantly reduced sensitivity to experimentally evoked pressure pain in milnacipran responders, an antihyperalgesic effect that was not seen in placebo responders.
References
Arnold L.M.
Goldenberg D.L.
Stanford S.B.
Lalonde J.K.
Sandhu H.S.
Keck P.E.
Welge J.A.
Bishop F.
Stanford K.E.
Hess E.V.
Hudson J.I.
Gabapentin in the treatment of fibromyalgia: A randomized, double-blind, placebo-controlled, multicenter trial.
A randomized, double-blind, placebo-controlled trial of duloxetine in the treatment of women with fibromyalgia with or without major depressive disorder.
Lack of pressure pain modulation by heterotopic noxious conditioning stimulation in patients with painful osteoarthritis before, but not following, surgical pain relief.
The study was sponsored by and performed in collaboration with Pierre Fabre. E.C. acknowledges financial support from the Department of Health via the National Institute for Health Research (NIHR) comprehensive Biomedical Research Centre award to Guy's & St Thomas' NHS Foundation Trust in partnership with King's College London and King's College Hospital NHS Foundation Trust. K.B.J. receives support from the COFAS Marie Curie Fellowship and Osher Center for Integrative Medicine at Karolinska Institutet. E.K. received support from the Swedish Rheumatism Association.
The rest of the authors report no conflicts of interest.
00938-9&id=gr1.jpg){kind=link}
00938-9&id=gr2.jpg){kind=link}
00938-9&id=gr3.jpg){kind=link}
00938-9&id=gr4.jpg){kind=link}
00938-9&id=gr5.jpg){kind=link}