The neurologic pain signature responds to nonsteroidal anti-inflammatory treatment vs placebo in knee osteoarthritis

Supplemental Digital Content is Available in the Text. fMRI-based measures, validated for nociceptive pain, respond to acute osteoarthritis pain, are not sensitive to placebo, and are mild-to-moderately sensitive to naproxen.

the start of the study; (v) subjects who have asthma or a history of asthma, (vii) subjects smoking more than 20 cigarettes per day; (viii) secondary causes of arthritis of the knee; (ix) had lower extremity surgery within 6 months prior to screening or scheduled for surgery of any kind during the study period in the affected knee; (x) significant prior injury to the index knee within 12 months prior to screening; (xi) disease of the spine or other lower extremity joints to affect the index knee or any other musculoskeletal or arthritic condition that may affect the interpretation of clinical efficacy and/or safety data; (xii) use of any analgesic, cyclooxygenase-2 (COX-2) inhibitor or nonsteroidal anti-inflammatory drugs (NSAIDs), other than study defined rescue therapy within 5× half-life prior to the first dosing day or during the study; (xiii) corticosteroid use prior to baseline; (xiv) received hyaluronan injections into index knee within the previous six months prior to baseline; (xv) initiation of or change to an established physiotherapy program within 2 weeks prior to baseline or during the study period. During the baseline clinical assessment, patients rated the severity of pain for left and right knees on a numerical rating scale (NRS) (0: "No pain"; 10: "Extreme pain"). The most severe knee was used in each session. fMRI parameters. A 1.5 T Signa Excite system (General Electric) equipped with an 8-channel, phasedarray head coil and single-shot echo planar imaging software was used. Functional sequences consisted of gradient recalled acquisition in the steady state (time of repetition [TR], 2000 ms; time of echo [TE], 50 ms; pulse angle, 90o; field of view [FOV], 24 cm; 64 x 64-pixel matrix; slice thickness, 4 mm (interslice gap, 1.5 mm)). Twenty-two interleaved slices, parallel to the anterior-posterior commissure line, were acquired to cover the whole-brain. The acquisitions were preceded by 4 additional dummy images allowing the fMRI signal to reach equilibrium.

Study 2
In order to further test NPS responses to pressure pain and the presence or absence of Placebo effects, we re-analyzed data corresponding to the Placebo arm (before [no treatment] and after 120 days of Placebo) of a previously published [10] randomized, double-blind, placebo-controlled trial (ClinicalTrials.gov. Identifier: NCT01226615) conducted at Hospital del Mar, Barcelona, Spain.
Patients. N = 32 patients were scanned, and 27 of these passed fMRI quality control and were included in the final analysis [10]. Patients' mean age was 65.6 ± 6.2 years, 19 were female, all were white Caucasian, had a knee pain severity at baseline (NRS) of 6.4 ± 1.3 (in a NRS from 0 "No pain" to 10 "Extreme pain"), 21 patients had radiological grade II whereas 3 were classified as radiological grade III. 11 had regular use of NSAIDs to control clinical pain. In brief, the study was approved by the local Ethics Committee (Clinical Research Ethical Committee-Institut Municipal d'Assistència Sanitària (CEIC-IMAS)) and run in compliance with the World Medical Association's Code of Ethics (Declaration of Helsinki), with written informed consent from all patients. Eligibility criteria. A subject was eligible for inclusion when all of the following criteria applied: (i) male or female with age between 40 and 75 years; (ii) diagnosis of primary osteoarthritis and suitability for the study as determined by the responsible rheumatologist, based on a comprehensive medical evaluation; (iii) radiological and clinical osteoarthritis based on the American College of Rheumatology (ACR) criteria; [1] (iv) osteoarthritis radiological grade II or III; [2] (v) stable knee symptoms for at least 1 month prior to screening; and (v) a minimum knee pain severity of 5 points on the 11-point numerical rating scales (NRS) at baseline. Exclusion criteria: (i) clinical evidence or history of drug/alcohol addiction; (ii) previous adverse effect to CS; (iii) relevant, non-controlled medical or psychiatric disease; (iv) formal MRI contraindication; (v) severe pain in other joints; (vi) inflammatory or systemic diseases with potential repercussion on joints; (vii) secondary causes of arthritis of the knee; (viii) the use of any analgesic, cyclooxygenase-2 (COX-2) inhibitor or nonsteroidal anti-inflammatory drugs (NSAIDs) within seven days prior to inclusion or during the study; (ix) systemic use or local corticosteroid injection three months prior study and during the study; (x) use of CS, diacereine, glucosamine or other symptomatic slow acting drugs for osteoarthritis three months prior the study; (xi) hyaluronan injections into the index knee within the previous six months prior to the study; (xii) subjects smoking over 20 cigarettes a day; (xv) initiation or change of a physiotherapy program in the 2 weeks prior to screening or during the study period; and (xvi) females of childbearing potential. Following screening, eligible subjects underwent a 2-week medication washout prior to the first fMRI assessment. Placebo treatment consisted of 2 hard gelatin yellow capsules taken together once a day during a meal. The capsule contained microcrystalline cellulose as a non-active excipient. During the baseline clinical assessment, patients rated pain severity using an 11-point NRS. Again, the most severe knee was used during the fMRI assessment. Here we provide a description of the common experimental details across both studies. fMRI task and painful stimuli. The same experimental paradigm was used in the scanner for both studies. The task consisted of a 6 min sequence alternating 11 baseline "rest" periods of 20 s (plus a final baseline "rest" period of 30 s) and 11 painful stimulation periods of 10 s. Pressure painful stimulation was applied using an MRI-compatible algometer developed in-house, with a pressure surface of 1 square cm [3]. Patients received pressure stimulation on the medial articular interline of the selected knee (most painful knee) at the most tender point in each subject with the knee in the position of 60° flexion. The tender point was established by palpation and marked using a permanent felt-tip pen. Pressure intensity to be applied during fMRI acquisition was calibrated to evoke pain between 5 and 7 on an 11-point NRS. Such a pressure adjustment was performed before dosing in each fMRI session day in study 1. There were no significant differences in applied pressure intensity between the No Treatment, Placebo, and Naproxen visits (ANOVA results, F = 0.38 and p = 0.687; mean pressure intensity for Naproxen was (SD): 2.6 Kg (1.0 Kg); Mean pressure intensity for Placebo was 2.5 Kg (0.8 Kg)), and mean pressure intensity for No Treatment was 2.6 Kg (0.9 Kg). In Study 2, the same pressure intensity was applied during the fMRI visits for the No Treatment and Placebo conditions (mean ± SD applied pressure was 2.5 kg ± 1.1 Kg). Each subject was asked to rate the subjective pain perceived during the entire fMRI sequence immediately after fMRI acquisition using a NRS.
Image preprocessing and motion analyses. fMRI time series for each study were preprocessed and analyzed using Statistical Parametric Mapping (SPM8) software, Welcome Department of Imaging Neuroscience, running on Matlab 7.1. Note that the processing and first-level model code is unchanged in SM12, and we confirmed that NPS responses did not differ as a function of whether contrast images were generated using SPM8 or SPM12.
Images were realigned to the first volume, co-registered and normalized to the Montreal Neurological Institute (MNI)-space provided in SPM8 (voxel size = 3x3x3 mm 3 ) and smoothed with a full width at half maximum (FWHM) Gaussian kernel of 8 mm. We verified that the included patients had head displacements of less than 2mm translation and 2º rotation. For both studies, we also computed mean framewise head displacement for each patient and condition following previously published methods [12]. In brief, motion was quantified using the realignment parameters obtained during image preprocessing, which included 3 translation and 3 rotation estimates. Average inter-frame motion measures (head position variations of each volume as compared to the previous volume) were used to capture head motion across the entire scan sequence [11,14,15]. A motion summary measure (mean framewise head displacement) that combined translations and rotations was computed in mm by adapting the formula of Van Dijk et al. [15] (for further explanations see [12], supplementary information). We ran a linear mixed effects repeated measures analysis to assess the effect of condition on mean framewise head displacement for Study 1 and showed no effect of treatment session on average framewise head displacement (F=0.47, p=0.63). Results from a paired samples t-test for Study 2 showed no effect of session (No Treatment, Placebo) on mean framewise head displacement (t=-0.93, p=0.36). Last, we performed a correlation analysis between mean framewise head displacement and NPS expression for each condition for each study and found no significant correlation for any comparison (all p's>0.1).

Signature pattern expression.
For both studies the same procedure was used to compute pattern response (or 'pattern expression') values for each patient and each condition (i.e., Placebo, No Treatment and Naproxen). For each patient we computed a single scalar value representing the expression of the NPS pattern (NPS in pro-nociceptive regions, and entire NPS pattern), SIIPS1 pattern, PINES pattern or Distress pattern for the contrast images representing "pain > rest" (as explained in detail in previous articles [5][6][7][8]). The NPS includes voxel weights in an a priori defined mask of brain regions that were significantly related to the term "pain" in the Neurosynth meta-analytic database (http://neurosynth.org/), see [16] for a detailed description. Data outside this mask did not contribute to the pattern expression value. For this analysis, we used a previously defined NPS component, the "Pronociceptive NPS" (NPSp), which comprised regions likely to be related to nociceptive pain (associated with pain-evoked activation in the NPS) [1,8]. In this subset of regions, which comprises most of the regions in the NPS, activity increases with increasing pain. These regions include the major targets of ascending nociceptive afferents, including the thalamus, secondary somatosensory regions (SI/SII), posterior, mid and anterior insula and adjacent opercula, midbrain, dorsal anterior cingulate cortex (dACC), inferior frontal gyrus and amygdala ( Figure 1). The SIIPS1, PINES, and Distress signatures are whole-brain patterns, i.e., they were developed including weights across the whole brain. For every contrast image of each patient participant, we computed the dot product of the vectorized activation contrast image (β map ) with the NPSp (or NPS, or SIIPS1, or PINES or Distress Signature) pattern of voxel weights (NPS-ω map ), i.e., β map Τ NPS-ω map , yielding a continuous scalar value for each patient, for the [pain vs. rest] comparison. To test for pressure pain responsiveness, drug effects, and placebo effects, we used planned contrasts specified a priori. The [Drug vs. Placebo] comparison and the [Placebo vs. No Treatment] comparison are the most widely accepted standard for testing drug effects and placebo effects, respectively. The former is the standard comparison in randomized clinical trials of pharmacotherapy [2,13], and the latter is the standard way of assessing active placebo effects (as compared with statistical artifacts or sampling biases) [4,9]. Therefore, we planned our analyses to test these two comparisons. For the sake of completeness, we ran linear mixed effects repeated measures analyses including treatment (categorical factor) as the independent variable and (a) NPSp and NPS, (b) SIIPS1, (c) PINES, (d) Distress, and (e) pain ratings as the dependent variable (in separate models) using SPSS (version 25).
Because we had strongly directional a priori hypotheses about these planned comparisons, statistical tests were performed one tailed.
We used one-sample t-tests to evaluate whether NPS responses to [pain vs. rest] were above zero, as pain beta maps are implicitly contrasted with the intercept (rest) in the 1 st -level model. To test within-subject contrasts for [Placebo < No Treatment], and [Naproxen < Placebo], we performed paired-samples t-tests across these conditions. Matlab software (MATLAB, MathWorks, v. R2016a, https://www.mathworks.com) and CanLab customized publicly available code (https://github.com/canlab) was used for all statistical comparisons and data illustration using bar plots with 95% confidence intervals.