In the last decade, clinicians have been somewhat disillusioned after the promise that pharmacogenomics would substantially transform the way anesthesia and perioperative medicine are delivered. Indeed, recommendations based on pharmacogenetic testing to help anesthesiologists and pain doctors tailor opioid regimens for safe and effective analgesia are still awaited.
This turbulent journey has been depicted in the journals relevant to anesthesia and pain medicine, with a myriad of publications and more than a dozen editorials describing the quest for a better understanding of the impact of genetic variants on opioid action and ways to improve clinical research in the field of pain genetics and opioid response. In 2002, we had envisioned light at the end of the tunnel and imagined how the Human Genome Project would “infuse new life into human physiology.”1 Soon after, we were invited to embark on an effective “fishing expedition” to explore pain genes,2 and numerous reports examining the μ-opioid receptor gene as a target for postoperative opioids emphasized that “one size does not fit all”3 and that genes associated with analgesic effect are distinct from those associated with side effects and adverse reactions.4 After recognizing the need to educate the audience and inform anesthesiologists,5–7 the current premise is a bittersweet realization that the heritability of opioid responsiveness is probably less straightforward and predictable than previously foreseen.8,9 Therefore, tailoring opioid therapy based on genetic testing may not be that simple, because different effects of opioids seem to be controlled by different genes, and preexisting environmental influences may more significantly contribute to some opioid responses than genes per se, as elegantly identified in a recent twin study.10
In this issue of Anesthesia & Analgesia, readers will find an observational investigation in a pediatric cohort undergoing surgery,11 for which, once again, a variety of genes was examined to shed more light on the genetic basis of postoperative pain and the response to morphine; the novelty, however, resides in the fact that this is the first study performed in an ethnically diverse group of children. The authors selected 6 candidate genes, identified as being relevant to explain the variability in opioid disposition, either because of their involvement as a key target receptor for opioid binding (OPRM1), or their role in opioid metabolism (CYP2D6), endorphins synthesis (POMC), transport across the blood-brain barrier (ABCB1), nociceptors development (NTRK1), or inactivation of dopaminergic neurotransmitters (COMT). Some of these genes would therefore seem more important as modulating pain perception (POMC, NTRK1, and COMT), whereas the remainder are more directly implicated in the pharmacokinetics and pharmacodynamics of opioid action.
Compared with previous studies searching for an association between genetic variants of selected genes and postoperative pain as a phenotype, this Swiss study by Mamie et al.11 is undoubtedly sophisticated and elaborate. Clinical outcomes included pain scores and morphine consumption, and genotyping included 6 genes. In addition, to adjust for population admixture in this ethnically diverse population, parental mating type was considered. The study was designed to determine the effect of each genetic variant separately on pain outcomes in a sample of children sufficient to identify an increased risk (adjusted relative risk) for the more prevalent genes (NTRK1, COMT, ABCB1, and OPRM1). One hundred sixty-eight children completed the study, and genotyping for the triad (both parents and child) was available in just over 100 cases. Because of the low prevalence in this cohort for the considered genetic variants of CYP2D6 and POMC, further analysis accounting for these 2 genes was not performed. The total dose of IV morphine consumed over the first 24 hours postoperatively was recorded for each child, and as often required in a pediatric setting, children younger than the age of 6 years were placed on a patient-parent-nurse–controlled analgesia system, and those aged 6 years and older received an IV morphine patient-controlled analgesia system. Pain scores, assessed with the revised faces pain scale (FPS), were recorded up to 11 times over the 24-hour study period, at rest and during mobilization, and it was assumed that a polymorphism would be considered to “boost” the pain intensity if it significantly increased the number of peak pain scores ≥4 of 10. Based on the observation that 15% of children had >4 of 22 recordings ≥6 of 10, a dichotomous outcome of having >4 episodes of pain exceeding 6 of 10 was also considered. Finally, mixed modeling taking into account the longitudinal follow-up over 24 hours was applied to report the mean 24-hour FPS score.
The major finding in this study is that none of the genetic variants was found to influence the total morphine dose consumed over 24 hours. The authors emphasized that this lack of association or pharmacogenetic effect was contrary to their expectations, and because they concluded that 2 of the selected single nucleotide polymorphisms were associated with higher pain scores (OPRM1 and ABCB1), the assumption that children with higher pain scores would consume more morphine would seem valid.
This study raises several interesting questions. The most fundamental questions that come to mind when postoperative pain is the phenotype of interest are (1) What parameters should be evaluated? and (2) Which outcome should be considered as clinically meaningful? In other words, how do we avoid pitfalls when designing a clinical study investigating postoperative pain and pain relief? In most studies that evaluate postoperative pain as a model of acute pain, pain scores recorded at rest and upon mobilization over a predetermined period of time, usually 48 hours, are used as the primary outcome.12,13 In addition, most research protocols use a cutoff point to identify moderate and severe pain, which may also serve as a threshold for pain treatment.14 Secondary outcomes often include the total opioid analgesic consumption, opioid-related side effects, and any morphine-sparing effect if a multimodal regimen is tested. Studies designed to assess such outcomes have in the past reported differences in morphine consumption based on genotype that may have been considered statistically significant; however, these differences surely seemed inconsequential to the clinician, because the dose differences were trivial and had no measurable impact on side-effect profile.15,16 Therefore, to ensure an adequate study design when the primary outcome is a clinical outcome (e.g., morphine consumption or pain scores), identifying an outcome that is truly meaningful from a clinical standpoint is crucial. In addition, the surgical model should ideally be consistent. Whereas abdominal procedures may result in pain that is more visceral in nature, orthopedic surgical pain may be quite different, often requiring different pain management for similar pain scores. Hence, mixing these 2 subpopulations may not be optimal when assessing pain and pain relief when the analgesic regimen is essentially the same.
Another question that has not often been discussed in the context of postoperative pain evaluation is When should pain recordings occur? Obviously, there is no need to record pain scores if the subject is sleeping. But there is probably also not much value to record pain scores at specific predefined time points (i.e., at precisely 6, 12, 18, and 24 hours postoperatively) when a subject may simply not be in pain, or may have just recently taken pain medication; indeed, breakthrough pain may occur between these time points and may be missed. In our era of smart technology, allowing patients, and even children, to record pain scores when it is relevant to them and when pain is present and requires medication would seem more useful and more accurately reflect their pain levels and subsequently the response to opioid medication they consume. With such a study design, patients could be instructed to record their pain scores (1) each and every time their pain at rest or upon mobilization was above a certain cutoff, and (2) each and every time they chose to take analgesic medication, recording their pain right before and then after taking the medication to evaluate its effect (and side effects). In their study, Mamie et al. attempted to calculate the mean 24-hour FPS score, using mixed linear regression models with random intercept for repeated measures. Although this analysis would in theory provide a good estimation of the average pain scores over 24 hours, it still only takes into consideration pain scores recorded at the arbitrary predefined time points, rather than real peak pain scores using the 24-hour study period as a continuum. This may explain why overall, mean FPS score, irrespective of genotype, was in the 2 to 3 of 10 range at rest, and 3 to 4 of 10 range during mobilization. The fact that children A118 homozygous for OPRM1, when compared with those A118G heterozygous, had a slightly lower mean 24-hour FPS score seems overall inconsequential because this was not associated with higher morphine consumption, and side effects are surprisingly not reported in this study.
Finally, the premise that higher pain scores will necessarily result in higher analgesic requirements/consumption needs to be revisited, particularly in the light of what we now know on the genetics of pain and opioid analgesia. Zubieta et al.17 described a decade ago the impact of the COMT gene on the relationship between pain perception and the μ-opioid neurotransmitter response, and found that volunteers with the Met/Met genotype exhibited higher sensory and affective ratings of experimental pain, and also had a higher regional density of μ-opioid receptors. This finding served as a basis for Rakvåg et al.18 to explain why they found lower oral morphine consumption in cancer pain patients with the Met/Met genotype, because an increase of μ-opioid receptor density will result in morphine being more effective in individuals carrying this genotype. In other words, lower pain tolerance (or higher pain scores) does not necessarily imply that higher morphine doses will be needed to manage pain. The reverse assumption also calls for caution; patients consuming higher morphine doses do not necessarily experience more pain, they may simply be “enjoying” more the rewarding effects of opioids, with a more favorable “liking/disliking” ratio and a better aversive opioid-effect profile. Angst et al.19 in their recent twin study found that genetic effects accounted for a striking 56% to 59% response variance for opioid-induced nausea, and 36% response variance for drug disliking, whereas familial effects accounted for 23% to 26% response variance for opioid liking. In addition, because liking and disliking was correlated with subjective aversive opioid effects (sedation, dizziness, nausea, and pruritus), it is unfortunate that Mamie et al. did not report those in their study. It is unlikely that liking or disliking opioids was a confounder for morphine consumption among the younger children managed with patient-parent-nurse–controlled analgesia, but it is possible that parents’ own response to opioids may have influenced the amount of morphine that was delivered. Evaluating opioid response in younger children certainly raises different questions, and may offer the rather unique setting and opportunity to study the impact of parenting on opioid analgesia, side effects, and the overall painful experience.
Last but not least, input from both biostatisticians and statistical geneticists is crucial for a clinical genetics trial to result in a meaningful finding. In determining group size, the power analysis needs to include not only the expected effect size, but also the allele frequency of the gene(s) to be studied. Because a single nucleotide polymorphism may be present in different frequencies in individuals with different ethnic backgrounds, variation in allele frequency across the expected population must also be considered, and when 2 or more mutations are studied, different analytical approaches are possible, including a haplotype-based method. The potential benefit of simultaneously analyzing a group of genetic markers is that a disease locus may be in linkage disequilibrium with a haplotype of 2 or more of these markers, and hence a method that can look at the combined effect of markers may be more powerful for detecting association than one that only considers markers individually. Given the allele frequency differences and population heterogeneity in this Swiss cohort, the trial conducted by Mamie et al. may ultimately be underpowered to reach a definite conclusion in the area of pediatric pain and opioid response.
Finally, this study highlights the complexity of studying postoperative pain as a phenotype whether genetics are accounted for or not. Given the marginal findings reported here, this lends more credence to the hypothesis that the contribution of common genetic variants to explain interindividual variability in postoperative pain perception and opioid consumption, whether in adults or in children, is not particularly important and that tailoring opioid analgesia based on selective genotyping is unlikely to occur any time soon. Perhaps future studies on the genetics of pain and pain relief in young children should include the impact of parental genotype on the amount of postoperative opioid consumption, and the interplay among empathy, epigenetics, and pharmacogenomics.
Name: Debra Schwinn, MD.
Contribution: This author helped write the manuscript.
Attestation: Debra Schwinn approved the final manuscript.
Name: Ruth Landau, MD.
Contribution: This author helped write the manuscript.
Attestation: Ruth Landau approved the final manuscript.
This manuscript was handled by: Peter J. Davis, MD.
1. Schwinn DA, Booth JV. Genetics infuses new life into human physiology: implications of the human genome project for anesthesiology and perioperative medicine. Anesthesiology. 2002;96:261–3
2. Eisenach JC. Fishing for genes: practical ways to study genetic polymorphisms for pain. Anesthesiology. 2004;100:1343–4
3. Landau R. One size does not fit all: genetic variability of mu-opioid receptor and postoperative morphine consumption. Anesthesiology. 2006;105:235–7
4. Nackley A, Diatchenko L, Maixner W. Perspectives on the genetic basis of opioid-induced hyperalgesia. Anesthesiology. 2006;104:909–10
5. Eisenach JH, Schroeder DR. Genomic discoveries in perioperative medicine: educating the audience. Anesthesiology. 2009;110:693–5
6. Lee FH, Raja SN. Should anesthesiologists be equipped as genetic counselors? Anesthesiology. 2010;113:507–9
7. Galinkin JL, Demmer L, Yaster M. Genetics for the pediatric anesthesiologist: a primer on congenital malformations, pharmacogenetics, and proteomics. Anesth Analg. 2010;111:1264–74
8. Mogil JS. Are we getting anywhere in human pain genetics? Pain. 2009;146:231–2
9. Walter C, Lötsch J. Meta-analysis of the relevance of the OPRM1 118A>G genetic variant for pain treatment. Pain. 2009;146:270–5
10. Fillingim RB. Genetic contributions to opioid side effects: helix me, helix me not. Anesthesiology. 2012;117:6–7
11. Mamie C, Rebsamen MC, Morris MA, Morabia A. First evidence of a polygenic susceptibility to pain in a pediatric cohort. Anesth Analg. 2013;116:170–7
12. Kolesnikov Y, Gabovits B, Levin A, Voiko E, Veske A. Combined catechol-O-methyltransferase and mu-opioid receptor gene polymorphisms affect morphine postoperative analgesia and central side effects. Anesth Analg. 2011;112:448–53
13. Coulbault L, Beaussier M, Verstuyft C, Weickmans H, Dubert L, Trégouet D, Descot C, Parc Y, Lienhart A, Jaillon P, Becquemont L. Environmental and genetic factors associated with morphine response in the postoperative period. Clin Pharmacol Ther. 2006;79:316–24
14. Gerbershagen HJ, Rothaug J, Kalkman CJ, Meissner W. Determination of moderate-to-severe postoperative pain on the numeric rating scale: a cut-off point analysis applying four different methods. Br J Anaesth. 2011;107:619–26
15. Sia AT, Lim Y, Lim EC, Goh RW, Law HY, Landau R, Teo YY, Tan EC. A118G single nucleotide polymorphism of human mu-opioid receptor gene influences pain perception and patient-controlled intravenous morphine consumption after intrathecal morphine for postcesarean analgesia. Anesthesiology. 2008;109:520–6
16. Chou WY, Wang CH, Liu PH, Liu CC, Tseng CC, Jawan B. Human opioid receptor A118G polymorphism affects intravenous patient-controlled analgesia morphine consumption after total abdominal hysterectomy. Anesthesiology. 2006;105:334–7
17. Zubieta JK, Heitzeg MM, Smith YR, Bueller JA, Xu K, Xu Y, Koeppe RA, Stohler CS, Goldman D. COMT val158met genotype affects mu-opioid neurotransmitter responses to a pain stressor. Science. 2003;299:1240–3
18. Rakvåg TT, Klepstad P, Baar C, Kvam TM, Dale O, Kaasa S, Krokan HE, Skorpen F. The Val158Met polymorphism of the human catechol-O-methyltransferase (COMT) gene may influence morphine requirements in cancer pain patients. Pain. 2005;116:73–8
19. Angst MS, Lazzeroni LC, Phillips NG, Drover DR, Tingle M, Ray A, Swan GE, Clark JD. Aversive and reinforcing opioid effects: a pharmacogenomic twin study. Anesthesiology. 2012;117:22–37