The investigations were carried out following the rules of the Declaration of Helsinki of 1975, revised in 2013. Ethical approval for this study (approval number, 2019-S1205) was provided by the Institutional Ethics Committee of Tongji Medical College, Huazhong University of Science and Technology, Wuhan, China (Chairperson Prof Hui Chen) on 20 November 2019. Written informed consent was obtained from all patients.

We enrolled 112 patients undergoing hand surgery with brachial plexus nerve block at Union Hospital, Tongji Medical College, Huazhong University of Science and Technology from December 2019 to August 2020. They were not genetically related to each other. We adopted the following inclusion criteria: 1) written informed consent; 2) scheduled for elective hand surgery; 3) aged 18–80 years; 4) American Society of Anesthesiologists (ASA) Physical Status I or II; 5) a BMI <30; and (6) no history of surgery and drug addiction. The exclusion criteria included 1) allergy to any of the drugs used in the study; 2) liver or renal function dysfunction; 3) sinus bradycardia or moderate to severe atrioventricular block; 4) treated with vasoactive or other drugs during the perioperative period; and 5) pregnancy or lactation.

Given the Narcotrend monitoring (MT MonitorTechnik, Bad Bramstedt, Germany) and Observer’s Assessment of Alertness/Sedation (OAA/S) Scale, which have been used as validated tools to assess the depth of sedation (Kreuer et al., 2003; Schrier et al., 2017), the sedation condition was evaluated by a trained assessor with the Narcotrend index (NI) and modified OAA/S scores (Pastis et al., 2019).

All patients routinely received standard monitoring of electrocardiogram (ECG), noninvasive blood pressure (NIBP), peripheral pulse oximetry, and Narcotrend brain monitoring. The mean arterial pressure (MAP), heart rate (HR), pulse oxygen saturation (SpO₂), and NI were recorded continuously. Simultaneously, intravenous access was established. Subsequently, ultrasound-guided intermuscular groove brachial plexus block and axillary brachial plexus block were performed by the same experienced anesthesiologist. In our implementation, the method we used was partly based on Senel’s study (Senel et al., 2014) with the following modifications: 1) none of the patients received preoperative medication before the block; 2) the local anesthetic solution consisted of 20 mL 1% lidocaine and 20 mL 0.5% ropivacaine, were 20 mL local anesthetic was injected into the intermuscular groove and axillary groove, respectively; and 3) nerve stimulators were not used. When the sensory and motor functions were blocked completely, DXM was pumped within 10 min at a loading dose of 0.5 μg/kg and then continuously pumped at 0.4 μg/kg/h during the surgery. We documented the MAP, HR, and NI before the pump (T0) and 5 (T1), 10 (T2), 15 (T3), and 20 (T4) minutes after the pump started. Another researcher assessed the OAA/S scores at 5-min intervals up to 20 min. OAA/S score of 2 was believed to the ideal sedation state in the current study (Laporte et al., 2011). Once the patient did not arrive at ideal sedation within the first 20 min, we kept evaluating until it did. The total amount of DXM and total injection time were recorded as OAA/S scores of 4, 3, 2, 1, and 0 (Table 1). Supplemental oxygen by face mask was delivered continuously. Any patient treated with vasoactive or other drugs during the perioperative period was excluded from the analysis. Above that, all the patients were just given the local anesthetic solution and DXM.

The primary outcome was the relationship between the single nucleotide polymorphism (SNP) and DXM sedation. The indices of the sedative effect included OAA/S scores and NI values. And the secondary outcome was the relationship between the SNP and the hemodynamic consequence of DXM. The hemodynamic indices included MAP, HR, and percent changes in MAP and HR. The percent change in MAP was calculated by the formula [(actual measured MAP value - baseline MAP value)/baseline MAP value] × 100%; the percent change in HR was calculated by the formula [(actual measured HR value - baseline HR value)/baseline HR value] × 100%. Considering the small number of patients with the minor allele for some SNPs, all patients were separated into two groups depending on the genotyping results: 1) homozygous for the major allele and 2) either heterozygous or homozygous for the minor allele.

The total amount of DXM and total injection time at an OAA/S score of 2 were chosen to determine the susceptibility to DXM as the “effective dose” and “onset time”, respectively. On the other hand, the differences in MAP and HR at the four time points mentioned above (T1 to T4) were used to detect cardiovascular susceptibility to DXM.

The genetic sample was obtained from the dorsal hand vein of each patient and collected in an EDTA-containing blood tube at the end of the operation. Then, they were instantly stored frozen at −80°C, and we extracted the blood later using standard phenol–chloroform procedures. Thus, the DNA was isolated from the frozen blood. Sequenom Mass ARRAY SNP genotyping systems were then used to determine genotypes, which were based on detection through MALDI-TOF MS (Sequenom Inc., San Diego, CA, United States).

The selection of these SNPs was based on our knowledge of metabolic pathways, drug distribution, drug excretion, targets, and the mechanism of DXM action after an extensive literature study. And these SNPs were confirmed to be relatively common in the Asian population and were therefore considered to have potential clinical significance. Eventually, from essential proteins to important SNPs, 45 SNPs of 28 genes were chosen as candidates (Table 2). Among the total candidates, three genes are known to be responsible for the pharmacokinetics of DXM (CYP2A6, UGT1A4, UGT2B10) (Holliday et al., 2014; Ber et al., 2020), and ten analyzed genes were speculated to be associated with DXM pharmacokinetics (ABCG2, ABCB1, ABCC2, ABCC3, SLCO1B1, SLC22A1, CYP1A2, CYP2C19, CYP2D6, and WBP2NL). Additionally, the ADRA2A gene has been demonstrated to participate in the primary mechanism of DXM action (Zhu et al., 2019; Choi et al., 2021). The remaining fourteen genes were hypothesized to be involved in the pharmacodynamics of DXM (SLC31A1, KATP, KCNJ11, KCNMB1, KCNMA1, KCNN4, CACNA1D, CACNA1C, ABCC9, ADRB2, COMT, GNB3, PRKCB, and PRKCH).

Statistical analyses were performed using SPSS 23.0 or GraphPad Prism 8.0 software. All data are expressed as the means ± standard deviations (SD). We applied Akaike information criterion to select the best model of inheritance and inheritance modeling suggested a dominant model for all SNPs, that is all patients were divided into two groups based on genotyping results, the homozygous for the major allele and the group of heterozygous and homozygous for the minor allele. We adopted Shapiro–Wilk test to examine the normal distribution of data, and the homogeneity of variance was assessed by F test. The differences in clinical characteristics among different time points were compared by one-way ANOVA. Independent-sample two-tailed t-test or Mann–Whitney U test was utilized to analyze the differences in NI values, percent changes of MAP or HR, onset time, effective dose and clinical characteristics between the homozygous for the major allele and the group of heterozygous and homozygous for the minor allele. The chi-squared test was selected to evaluate the candidate SNPs’ Hardy-Weinberg equilibrium (HWE). Finally, a multiple linear regression analysis was added to measure the effect of significant SNPs and clinical factors (age, sex, and BMI) on the effective dose. Commonly, p < 0.05 was utilized to consider the deviation from equilibrium or statistical significance.

As 20 patients did not meet the inclusion criteria (2 patients were overweight, 15 did not have phenotypic data, and 3 were treated with vasoactive drugs due to excessively high preoperative BP), 92 of the 112 patients were recruited for the follow-up analysis in this study. The demographics of all study patients were referred to in Table 3.

As shown in Figure 1, the sedative effect induced by DXM was time- and dose-dependent. With the continuous injection of DXM, the OAA/S score decreased from 5 to 4.68 ± 0.73, 3.64 ± 1.14, 3 ± 1.16, and 2.52 ± 0.92 at T1, T2, T3, and T4, respectively (Figure 1A). Simultaneously, NI decreased from 98.51 ± 1.35 to 89.48 ± 22.73, 74.32 ± 29.87, 65.08 ± 29.28, and 58.89 ± 26.52 (Figure 1B). The total amount of DXM required for OAA/S scores of 4, 3, and 2 was 26.01 ± 10.34, 30.89 ± 8.38, and 36.43 ± 7.11 µg, respectively (Figure 1C). The onset time, NI and effective dose of each patient were displayed in Figures 1D–F, respectively. Figure 1D shows a marked individual variation in the dose-response under DXM sedation. First, 13.33-fold variability existed between the fastest and slowest onset times, which ranged from 6 to 80 min and averaged 22.72 ± 16.45 min. According to Figure 1E, NI value at OAA/S of 2 mainly fell into 30–50 in 71 patients despite the maximum and minimum, which were 22 and 99, differing by 4.5 times. The variable effective dose also showed a similar result, ranging from 18 to 90.67 µg when the OAA/S was 2 (Figure 1F).

DXM also has a statistically significant effect on systemic hemodynamics. HR and MAP both depicted a time- and concentration-dependent change tendency (Figure 2). With a greater degree of sedation, as shown in Figures 2A,B, the HR declined from 73.42 ± 10.32 to 70.54 ± 10.88 bpm, to 66.60 ± 10.32, to 65.09 ± 9.77 bpm, and to 63.70 ± 8.98 bpm from T1 to T4, and the percent alterations in HR were −3.72% ± 8.82%, −9.09% ± 8.31%, −11.03% ± 8.97%, and −12.78% ± 9.18% at the above time points. A similar phenomenon was observed in the MAP (Figures 2C,D), which was diminished from 98.88 ± 11.43 (T0) mmHg to 96.30 ± 11.01 (T2) mmHg, 91.54 ± 11.72 (T3) mmHg, and 88.90 ± 11.18 (T4) mmHg, except for a modest increase found at the start (elevated to 99.95 ± 10.59 (T1). Thus, the MAP was changed by 1.50% ± 7.95%, −2.08% ± 9.85%, −6.95% ± 10.42%, and −9.63% ± 9.99% from T1 to T4, respectively. Further analysis showed that at an OAA/S of 2, there were large individual differences in patients’ HR and MAP in response to DXM (Figures 2E–H). The data in Figures 2E,F suggested the HR ranged from 46 to 96 bpm, and the MAP ranged from 70 to 123 mmHg. The percent changes in HR and MAP ranged from −39.53% to 7.35% and from −39.84% to 11.36%, respectively (Figures 2G,H).

The genotype distributions of the 45 SNPs and the minor allele frequency (MAF) of each SNP are shown in Table 2. No mutations in CYP2A6 rs5031016, SLC22A1 rs4646277, or PRKCH rs2230500 were found in these patients. All the patients carried the major alleles. Except for UGT2B10 rs835309, CYP2A6 rs56113850, and CYP2D6 rs1065852, allele frequency distributions of the 22 polymorphic sites were in HWE (p > 0.05).

The NI, onset time, effective dose, and percent changes in HR and MAP at OAA/S scores of 2 were selected to assess the susceptibility to DXM. As indicated in Table 4, ABCG2 rs2231142, CYP2D6 rs16947, WBP2NL rs5758550, KATP rs141294036, KCNMB1 rs11739136, KCNMA1 rs16934182, ABCC9 rs11046209, ADRA2A rs1800544, and ADRB2 rs1042713 were found to be linked with sensitivity to sedative or hemodynamic effect of DXM.

Considering that the major pharmacological effect of DXM is sedation and the effective dose represents a comprehensive measure of the sedative effect observed in each participant after administration, we applied multiple linear regression analysis to further test the effect of independent variables on the susceptibility to DXM sedation (effective dose). The significant SNPs (CYP2D6 rs16947, KCNMA1 rs16934182, and ADRA2A rs1800544) that are associated with individual variation in response to DXM sedation, and clinical traits (age, sex, and BMI) were selected as independent variables. The effective dose of DXM was selected as the dependent variable. As illustrated in Table 5, sex, BMI, and ADRA2A rs1800544 were statistically significant associated with the individual sensitivity to DXM sedation (F = 6.685, R ² = 0.321, p < 0.001). Additionally, the significant clinical variables (sex and BMI) and one significant SNP ADRA2A rs1800544 in the model accounted for 27.3% of the variability in DXM dosage for sedation (adjusted R ² = 0.273).

ABCG2, the ATP-binding cassette, subfamily G, isoform 2 protein, encoded by the ABCG2 gene, is a key member of the ABC transporter superfamily, removing substrates from the cell in an energy-dependent manner (Kukal et al., 2021). ABCG2 is expressed in the gastrointestinal system, kidney, liver and blood–brain barrier and protects organisms against xenobiotic exposure (Zhang et al., 2018). Numerous studies have demonstrated that ABCG2 SNPs affect the pharmacokinetics of many therapeutic drugs, such as topotecan, sunitinib, and risperidone (Stacy et al., 2013; Werbrouck et al., 2019; Ganoci et al., 2021). Our study indicated that the rs2231142 polymorphism in the ABCG2 gene could influence the patients’ NI values at an OAA/S score of 2 and their onset times. As reported by previous studies (Bauer et al., 2016; Al-Majdoub et al., 2019), patients with GT or TT alleles in ABCG2 rs2231142 seem to have a shorter onset time and deeper sedation, because TT is associated with reduced function. The possible reason is mutant allele increased the rate of ABCG2 degradation in the endoplasmic reticulum via ubiquitination and proteosomal proteolysis (Furukawa et al., 2009). Contrary to expectations, our result found that patients with GT or TT alleles were more awake (higher NI values) and had a longer onset time. A possible explanation for this might be that ABCG2 rs2231142GT or TT allele tends to reduce more MAP or tends to have more pronounced vasodilation. After that, the blood concentration of DXM is reduced, and patients with GT or TT alleles are more awake (higher NI score) and have a longer onset time. Another possible explanation is that previous studies were conducted among white or black people. They did not include Asians. To our knowledge, this is the first study to investigate the correlation between the ABCG2 SNP and sensitivity to DXM. In other words, our current findings expand on prior work.

Many cytochrome P450 enzymes are encoded in the human genome. They are involved in a wide range of metabolic pathways, including the biotransformation of endogenous and exogenous molecules, dietary components, and drugs (Saravanakumar et al., 2019). Among the CYP superfamily members, CYP2D6 is one of the most important and extensively studied drug-metabolizing enzymes (Owen et al., 2009). Previous studies have described that highly genetic polymorphisms in CYP2D6, influencing the enzymatic function of CYP2D6, were associated with the efficacy of some drugs, including some antidepressants, antineoplastic agents, and analgesics (Nofziger et al., 2020; Taylor et al., 2020). Moreover, clinical guidelines have adopted CYP2D6 metabolism status into therapeutic recommendations for CYP2D6-substrate drugs (Taylor et al., 2020). Rs16947, one of the common SNPs in CYP2D6, is known to reduce CYP2D6 mRNA expression two-fold by affecting exon 6 splicing (Wang et al., 2014). Apart from itself, rs5758550, located 115 kbp downstream of the CYP2D6 promoter, resides within a pivotal enhancer region directing CYP2D6 expression. The “enhancer” SNP was suggested to affect overall CYP2D6 mRNA expression (Wang et al., 2015; Dinh et al., 2021). These findings provide support for our current results.

Based on our results, the rs16947 polymorphism in the CYP2D6 gene was related to the less effective dose of DXM, perhaps because of decreased CYP2D6 expression, thus, prolonging the duration of drug action. And rs5758550 significantly elevated the patients’ NI values and percent change in MAP. The data indicated that the two genes had a different impact on DXM’s efficacy, which contradicts the phenomenon above (Dinh et al., 2021). However, the two SNPs related to the CYP2D6 yield alteration in the same direction in every clinical index (Table 4). Considering those similarities and differences, more research is needed to explore the role of rs5758550. Similar to ABCG2, this is the first report of an association between CYP2D6 genetic polymorphisms, rs5758550, and DXM clinical efficacy.

Among three highly homologous subtypes (α2A-, α2B-, and α2C-adrenergic receptors), DXM produces sedation, analgesia, and sympathetic nervous system inhibition mainly by activating the central pre- or post-synaptic α2A-adrenergic receptor (ADRA2A), which is encoded by the ADRA2A gene (Zhu et al., 2019). Much effort has been made to explore the linkage between several genetic polymorphisms in the ADRA2A gene and susceptibility to DXM (Fu et al., 2020). The present study also demonstrated that patients with the mutation (G>C) in ADRA2A rs1800544 need more DXM to reach the ideal sedative extent. Our result agrees with Yagar’s conclusion that patients with the mutant allele in ADRA2A rs1800544 showed higher bispectral index and Ramsay sedation scores, signifying a longer period before falling asleep (Yağar et al., 2011). Small’s work consistently pointed out that the mutant of rs1800544 decreased ADRA2A expression (Small et al., 2006). Additionally, Zhu and colleagues reported that the rs1800544 polymorphism could manipulate heart rate changes in patients after rocuronium infusion (Zhu et al., 2019), which expands our work. Collectively, ADRA2A rs1800544 mutation plays an indispensable role in differences in individual sensitivity to DXM, even in other efficacy of other drugs.

The ADRB2 gene encodes the β2-adrenoreceptor, which plays a vital role in regulating the cardiovascular system (Dai et al., 2021). Several SNPs in the ADRB2 gene could impact cardiovascular activity, thereby altering the risk of hypertension (El-Menyar et al., 2015; Huang et al., 2018). Nielsen and colleagues found that carriers of the minor allele (G) showed a tendency toward vasodilation and high cardiac output during anesthesia. The effect could result in receiving more ephedrine (Nielsen et al., 2016). In addition, our research proved that homozygosity or heterozygosity of G allele in ADRB2 rs1042713 would demand a longer onset time. One possible explanation for this discrepancy is that the rs1042713 mutation would dilate blood vessels and ultimately require infusion of more DXM to gain the same effect-chamber concentration. Moreover, recent studies have identified the SNP to alter the expression and conformation of β2-adrenoreceptor (Xie et al., 2019). Thus, when determining the amount of DXM to produce the appropriate sedation status, the hemodynamic effect should be taken into consideration.

There are a few limitations in our study. First, the study sample size may limit our ability to identify more significant associations. And we are prepared to perform further research to expand the cohort. In the current study, multiple testing correction would likely be ineffective in capturing potentially relevant SNPs associated with the phenotype. Considering that our study was aimed to identify more potential genetic markers associated with the variability in individual response to DXM, we did not do adjustments like FDR or Bonferoni. Second, we did not measure the plasma concentrations of DXM, and we used the total infusion dose instead. As a result, the pharmacokinetic analysis of this study was inadequate. We would take the question into account in further research and then provide a better distinction between pharmacokinetic and pharmacodynamic variations.