Placebo and Opioid Analgesia—Imaging a Shared Neuronal Network Predrag Petrovic, Eija Kalso, Karl Magnus Petersson, Martin Ingvar

Remifentanil (4-(Methoxycarbonyl)-4-[(1-oxopropyl) phenylamino]-1 - piperidine propa-noic acid methyl ester) is an opioid acting on mu-type receptors. It has a rapid onset of action and a very short duration (see: J. Med. Chem. 1991, 34, 2202). Remifentanil is a potent opioid that is rapidly hydrolyzed by blood and tissue esterases (41). The time for a 50% reduction in the effect site concentration is about 3.65 min and the terminal elimination half-life 10.2 min at a dose of 2 g/kg (42). Because the interval between the scans was 10 minutes and the dose was smaller than in the study referred to above, we estimated that the possible residual effect on the next-coming condition would be negligible. More importantly, any possible bias would be the same for the no drug conditions (W and P) as for the placebo conditions (WPL and PPL) and could therefore not explain differences between these conditions. The optimal dose of remifentanil was determined in a pilot study where the analgesic effect and possible adverse effects were monitored.

Supplement B

The subjects that participated in this PET study were right-handed healthy males aged 20-27 years. Pain was induced by a continuos tonic heat stimulation of 48° C during 70 seconds on the dorsum of the left hand using a Medoc^® stimulator and a 2x3-cm probe. The control stimulation consisted of a tonic stimulation of 38° C during 70 seconds using the same probe. The stimulation onset time was 10 seconds before the scanning started and continued through the 60 second long scan. The drugs would be injected intravenously starting 40 seconds before each stimulation began (injection time was 30 seconds). The drugs (remifentanil (0.5 g/kg) or saline (placebo)) were given in a double-blind fashion.

In one third of the scans (both pain and warm stimulation) no drug was injected, and the subjects were told that these stimuli would be given without prior analgesics. The six conditions formed one block. The order of the conditions was randomized within the block and each subject was scanned in two blocks. Thus, there were a total of 18 experimental blocks in the study. Each subject tested the painful stimulation in a training session during the week preceding the PET-study. The subjects rated the pain intensity after each scan using a visual analogue scale (VAS). 0 mm on the VAS indicated that the subject perceived no pain intensity and 100 mm indicated that the subject perceived the highest imaginable pain intensity.

The present study represents a potential breach of the paragraph on truthfulness to subjects in scientific studies. However, the core of the study is to shed light on endogenous mechanisms of pain relief and the initiation of such mechanisms requires procedures like the presently used. We regard the misinformation as limited and part of the core question of the study. We do not see any alternative way around this ethical issue without loosing the analgesic effect of the placebo treatment. The design of the study has passed the local ethical committee.

Supplement C

Each subject participated in twelve measurements of the rCBF using a 3D Ecat Exact HR positron emission tomograph (PET) and bolus injections of 450 MBq [¹⁵O]-H₂O. The PET images used for the statistical analysis were realigned, spatially normalized, 3D isotropic Gaussian filtered (14 mm FWHM), proportionally scaled to account for global confounding factors and transformed into an approximate Talairach-Tournoux stereotatic space (43) as defined by the SPM99 template ((44) http://www.fil.ion.ucl.ac.uk/spm). We primarily investigated the activity in the rostral ACC and the brainstem (see introduction). However, also other regions that are known to be involved in the processing of pain (31) or opioid mechanisms (4-14) were analyzed. Any activation with a Z-score 3.09 was considered significant in the limeted search voume.

Supplement D

To explore whether the measured activity in the rACC covaried condition specifically with the brainstem activity, a regression analysis was performed. In the voxel showing the highest Z-value in the rACC during the main effect of opioids, defined as [(POP+WOP) - (P+W)], the adjusted activity was extracted from all conditions. The activity for each scan was subtracted with the average activity for each condition. These values were then used as covariates of interest for each of the six conditions in a general linear model and the six conditions were modeled as confounding covariates. We then studied the regressions between the rACC and the brainstem during the different conditions. In addition, differences in the observed regressions between the conditions, were studied by employing a simple contrast analysis in SPM99 (See also psychophysiological interactions (45-46)). Because the regression analysis was restricted to the brainstem as the region of interest, an uncorrected p-value of 0.005 was chosen as the level of significance.

Supplemental Figure 1. The average VAS rating of pain intensity for the P, PPL and POP condition is shown for each subject. The behavioral results showed that remifentanil provided effective analgesia in the heat pain condition as the pain intensity ratings were lower in the POP vs. the P condition in all of the 18 blocks. The subjects rated the painful stimulation as less intense in 16 of the total 18 blocks in the PPL condition compared with the P condition. In two blocks the rating was lower for the P condition compared with the PPL condition. Five of the subjects were classified as high placebo responders since their average placebo analgesia was more than 10 %. Four subjects were classified as low placebo responders since their placebo analgesia was less then 10 % (one of the subjects showed an increased average pain intensity rating during the PPL vs P).

Medium version | Full size version

Supplemental Table 1. Regions which showed increased activity in the main effect contrast of pain [(POP+PPL+P)-(WOP+WPL+W)]. The Talairach coordinates ([x y z]) for the maximally activated voxel in each of the regions and their Z-value are shown to the right in the table. For the abbreviations of the conditions see paper.
	[x y z]	Z-score
Main effect of pain (POP + PPL + P) - (WOP - WPL - W)
Insula dx	[36-26 14]	4.43
	[22 10 8]	3.86
Insula sin	[-30 -20 12]	3.62
	[-22 -12 14]	3.19
Anterior cingulate cortex (ACC)	[4 -6 40]	3.22
Thalamus dx	[16 -4 12]	4.06

Supplemental Table 2. Regions which showed increased activity in the main effect contrast of opioids [(POP+WOP) - (P+W)]. The Talairach coordinates ([x y z]) for the maximally activated voxel in each of the regions and their Z-value are shown to the right in the table. For the abbreviations of the conditions see paper.
	[X Y Z]	Z-value
Main effect of opioids (POP + WOP) - (P + W)
Anterior cingulate cortex (ACC)	[0 8 34]	6.76
	[8 44 12]	6.75
	[6 -12 40]	5.82
Insula dx	[48 8 -8]	9.40
Insula sin	[-40 14 0]	6.58
Orbitofrontal cortex (Obfc) / Temporopolar cortex	[18 12 -18]	4.05
Orbitofrontal cortex (Obfc) / Temporopolar cortex	[-10 10 -18]	3.91
Pons	[2 -20 -40]	4.70
Relative decreases in regions processing pain during opioid treatment (P - W) - (POP - WOP)
Insula dx	[-32 8 10]	4.08
Insula sin	[32 10 10]	3.53

Supplemental Table 3. Regions that showed increased activity in the placebo effect during pain (PPL - P) and the interaction effect of ([(PPL - P) - (WPL - W)]). The Talairach coordinates for the maximally activated voxel in each of the regions and their Z-value are shown to the right in the table. To control for unspecific placebo effects we also investigated the interaction between the factors of placebo and noxious stimulation ([(PPL - P) - (WPL - W)]). This analysis indicated an effect in the ACC, somewhat caudal to the rACC effect in (PPL - P) but rostral to the ACC activation during pain.
	[X Y Z]	Z-value
Effects of placebo (PPL - P)
Orbitofrontal cortex (Obfc) dx	[30 30 -6]	3.54
	[30 46 -14]	3.49
Orbitofrontal cortex (Obfc) sin	[-24 42 -16]	3.09
Anterior cingulate cortex (ACC)	[18 32 14]	3.28
Effects of placebo (interaction) ([(PPL - P) - (WPL - W)])
Anterior cingulate cortex (ACC)	[18 12 38]	3.82

Supplemental Table 4. Regions in the brainstem, which showed increased covariation with rACC during the POP, the PPL and the P condition. Also the difference between the regressions, i.e. the functional connectivity of (POP-P) and (PPL-P) condition are shown. The Talairach coordinates for the maximally activated voxel in each of the regions and their Z-value are shown to the right in the table. For the abbreviations of the conditions see paper. n.s. denotes sites of subsignificant changes.
	[X Y Z]	Z-value
Areas in the brainstem covarying with rACC (during POP)
Pons	[-6 -22 -42]	3.48
	[4 -24 -36]	3.26
PAG	[8 -30 -10]	3.20
Areas in the brainstem covarying with rACC (during PPL)
Pons	[6 -22 -36]	4.14
PAG	[0 -34 -12]	2.52 (n.s.)
Areas in the brainstem covarying with rACC (during P)
-	-	-
Differences in regressions between the conditions (POP vs. P)
Pons	[-4 -22 -40]	3.02
PAG	[4 -22 -10]	3.04
Differences in regressions between the conditions (PPL vs. P)
Pons	[10 -18 -40]	3.09
	[2 -20 -24]	2.99
PAG	[0 -34 -12]	2.42 (n.s.)