Clinical and genetic factors associated with post‐operative nausea and vomiting after propofol anaesthesia

The incidence of post‐operative nausea and vomiting (PONV) remains at about 30% despite all therapeutic efforts to reduce it. The clinical risk factors guiding the prophylactic treatment are well established, but genetic factors associated with PONV remain poorly known. The aim of this study was to explore clinical and genetic factors impacting PONV by performing a genome‐wide association study (GWAS) together with relevant clinical factors as covariates, and systematically attempt to replicate previously reported PONV associations. Relevant clinical factors are explored with logistic regression model.

Conclusions: Our GWAS approach did not identify any high-impact PONV susceptibility variants. The results provide some support for a role of dopamine D 2 receptors in PONV.

K E Y W O R D S
DRD2, genome-wide association study, GWAS, PONV, post-operative nausea and vomiting, propofol, SNP, TaqIA

Editorial Comment
Genotype contribution to risk for perioperative nausea and vomiting is not completely understood. This study assessed genetic factors and post-operative nausea and vomiting events in a large breast surgery cohort, along with other patient specific and treatment factors. This broad genetic analysis did not identify any new genetic variant major susceptibilities, and did confirm some already recognised associated variants and treatment approaches.

| INTRODUCTION
Despite current methods of prevention, the incidence of post-operative nausea and vomiting (PONV) remains at about 30%, 1,2 but varies from 10% to 80% in different risk groups. [3][4][5] Well-established risk factors are type of surgery, female gender, young age, previous history of motion sickness or PONV, family history of PONV, non-smoking, volatile anaesthetics, and use of post-operative opioids. 3,4,6,7 PONV occurs when prophylactic antiemetic treatment fails or has not been given. PONV has a multifactorial origin, with individual and surgery-and anaesthesia-related factors. Genetic susceptibility also plays a role.
Based on known PONV pathophysiology and effective pharmacotherapy, hypothesis-driven candidate gene association studies (CGASs) have been performed. CGASs explore associations between phenotype and selected genetic variant, like single-nucleotide polymorphism (SNP), within the gene that is under interest. SNPs are nucleotide substitutions distributed throughout the genome, accounting for up to 80% of individual features, including susceptibility to disease. 8 Whilst most SNPs do not have a direct biological effect, some of the SNPs modify the function of the gene in which it is located, or near it, by altering its expression or the structure of the protein for which it codes. CGASs on susceptibility for PONV have explored SNPs in genes HTR3A and HTR3B (encoding 5-HT 3 receptors), [9][10][11] SLC6A4 (serotonin transporter 5-HTTLPR), 12 TARC1 (NK-1 receptors), 13 OPRM1 (MOR), [14][15][16][17] and DRD2 (dopamine D 2 receptors). 18,19 Two PONV genome-wide association studies (GWASs) with a limited sample size have been published. 20,21 GWASs explore associations between the phenotype studied and hundreds of thousands of SNPs throughout the whole genome. 22 Janicki et al. reported that in a sample of 251 individuals, rs2165870, located near CHRM3 (encoding muscarinic acetylcholine receptor M3), was associated with PONV, 20 a result replicated in two independent CGASs 23,24 but not in two others. 12,21 Another PONV risk SNP discovered by Janicki et al. was rs349358 in KCNB2, 20 replicated in an independent study. 25 The second GWAS investigating PONV, based on 24 samples, found that variant rs11232965 near a long non-coding RNA MIR4300HG had a suggestive, unverified, association ( p < 1 Â 10 À5 ) with PONV. 21 Our aim was to identify clinical and genetic risk factors for PONV through logistic regression analysis and GWAS in a clinically wellcharacterised cohort of Finnish women receiving standardised propofol and remifentanil anaesthesia, and antiemetic prophylaxis with ondansetron and droperidol.

| METHODS
BrePainGen is a prospective study to examine the role of genetics in acute and persistent postsurgical pain, anaesthesia, and PONV. Its protocol was approved by the coordinating ethics committee Patients who had metastases other than axillary lymph nodes, bilateral disease, clinically relevant kidney or liver malfunction, or immediate breast reconstruction, were excluded.

| RESULTS
The final number of participants was 815 (Table 1A and B). Exclusions are shown in Figure 1 and Figure S1.  surgery. Distribution of Apfel points 3 are shown in Table S2. We focused on the interval 2-24 h after surgery.  Tables S3-S5. After discharge, we had complete PONV information from 755 patients (71 lacked information, Figure S1). Both models were additionally adjusted for age and ASA status because this improved model performance. Receiver operating characteristic (ROC) curves for both models are presented in Figure 2. DNA sample quality control procedures have been described before. 35 Mean genotyping success rate was 0.997. After quality control, genotyping data were available for 926 participants. After exclusions for clinical factors and missing data, the final count was 815 ( Figure 1 and Figure S1).

T A B L E 3
The results of the logistic regression Models 1 and 2.  Figure 3 and Table 4, and accompanying LocusZoom plots in Figure S2A-F. No studied SNP showed genome-wide significant association with PONV. Suggestive evidence for association (p < 1 Â 10 À5 ) was seen between PONV and six variants, five of which were independent. Summary statistics of our GWAS are available upon request.
Next, we systematically tested whether previously reported associations would replicate in our study ( Based on earlier findings, the most convincing PONV risk variant is rs2165870 (r 2 = 1, rs2355230) in CHRM3. We did not see any evidence of association with that SNP in our study (OR 0.90; p = .45).
Neither could we replicate the association of rs349358 near KCNB2 (OR 0.85; p = .39). Power calculations revealed that our study had sufficient power to detect association, assuming similar effect sizes than have been reported earlier.

| DISCUSSION
This study explored PONV in 1000 women undergoing surgery for breast cancer. As previous studies have established genetic susceptibility to PONV associated with volatile anaesthetics, our study provides new information about clinical and genetic risk factors for PONV after propofol anaesthesia in patients with high risk for PONV and who received double antiemetic prophylaxis. To our knowledge, this is the first GWAS exploring PONV in such patients, and the largest PONV GWAS performed thus far. Highest incidence of PONV, 23%, was at 2-24 h after surgery. Prophylaxis decreased the incidence as expected. 36 Our study revealed that consumption of oxycodone in PACU, current smoking, previous PONV, and history of motion sickness were associated with PONV incidence, in line with previous findings. 3 revealed the T allele to be associated with lower D 2 receptor density and availability in striatal and caudal areas of the brain. 41 D 2 receptors are also located in the area postrema, so perhaps the polymorphism also affects receptor density in brain areas critical for PONV.
There are some limitations in our study. Ideally, the GWAS setting would have required a larger cohort to reach sufficient power to detect association with variants that are less frequent or have a moderate impact on PONV risk. Furthermore, although our PONV case definition is based on strong clinical expertise, also somewhat different definition could have been used since there is no clear consensus and variability between the studies exist. The antiemetic prophylaxis the patients received decreased the intensity of PONV, making it harder to make a clear distinction between cases and controls. Furthermore, since we did not record the intensity of PONV symptoms, it is likely that our patient cohort contains both severe PONV patients and patients who had only mild discomfort. Also, we exposed our patients to nitrous oxide, a drug well-known for its emetic properties.
Nitrous oxide was standard clinical practice at our hospital at the time and by including it to our protocol, we made the study protocol easy and safe to use as it was familiar to all personnel.
Overall, our extensive study with clinical and genetic risk factors of PONV does not support the idea of a strong genetic risk factor explaining majority of the PONV susceptibility. Our logistic regression model confirms many of the previously established clinical risk factors.
Our results also provide some limited support for a role of dopamine