Patients diagnosed with VS or MCS for ≥4 weeks scheduled to undergo surgery in Beijing Tiantan Hospital or the Seventh Medical Center of PLA General Hospital from October 2021 to June 2022 were enrolled in this study. We excluded participants with pre-existing neurological conditions, who took long-acting sedative drugs before the study, and those with liver and kidney failure or serious complications. The overall research protocol was approved by the Ethics Committee of Beijing Tiantan Hospital and the Seventh Medical Center of the Chinese PLA General Hospital. Informed consent was obtained from the legal representatives of the subjects.

The demographic and clinical data of all included patients were recorded at baseline. The collected data included age, gender, weight, disease course, and etiology of DoC. Consciousness status was assessed by experienced raters using the revised JFK Coma Recovery Scale (CRS-R) (Kalmar and Giacino, 2005). The scores of six subscales of the CRS-R scale evaluating auditory, visual, motor, oromotor, communication, and arousal functions were recorded. Preoperative CRS-R scores were assessed at least five times within 2 weeks during awake and stable (without complications, including fever, and epilepsy) periods to avoid potential errors caused by fluctuations in responsiveness. The highest score was used to assess each patient’s baseline consciousness level.

The operation methods, the classification of the American Society of Anesthesiologists (Kotake, 2016), and the anesthesia methods were documented. Intraoperative medications, including propofol, sevoflurane, remifentanil, sufentanil, and rocuronium, and their doses were recorded for further analysis. The doses of sufentanil and remifentanil were converted to the morphine dose (Arnold and Weissman, 2003) and added to analyze the total analgesic effect.

After awakening from anesthesia, two experienced physicians screened and recorded each patient’s consciousness state for 48 h. Postoperative consciousness states were also evaluated by the CRS-R scale, and the highest scores were used to describe the patient’s postoperative consciousness level. No treatments or drugs for cortical excitability were used during the first 2 days after surgery. Patients with improved postoperative diagnosis (VS to MCS or MCS− to MCS+) or those who reached the criterion of “localization to sound” or “visual fixation” [reflecting higher-order processing (Weaver et al., 2022)] were classified as the improved group. The remaining patients were classified as the non-improved group. The same assessment was also performed 3 months after surgery.

The difference between preoperative and postoperative CRS-R scores and subscores were analyzed with the Wilcoxon signed rank test. To determine the cause of postoperative improvement, explorative data analyses were performed between the improved and non-improved groups. Results are presented as proportions (%), medians and interquartile range, or arithmetic means and standard deviations depending on their scale. Tests for statistical significance were performed with Fisher’s exact test, chi-square test, Wilcoxon–Mann–Whitney U test, or two-tailed Student’s t-test. Multivariate logistic regression analysis was used to identify independent risk factors for postoperative improvement. A p-value < 0.05 was considered statistically significant.

To analyze the causes and influencing factors of postoperative consciousness improvement, we divided DoC patients into improved and non-improved groups (Table 2). Patients with or without postoperative improvement were significantly different in etiology, preoperative diagnosis, and preoperative CRS-R score (mainly in the auditory, visual, and motor subscales) compared with the non-improved group. Patients with short-term postoperative consciousness improvement also performed better in long-term outcomes compared with the non-improved group. More importantly, patients with short-term postoperative improvement were more likely to regain the ability to communicate (28.6%, two accurate and four intentional) than those without postoperative improvement (7.1%, two intentional). However, the occurrence of postoperative improvement did not predict whether patients would benefit from other treatments (Supplementary Table 2).

We also found a significant difference in intraoperative opioid consumption between the two groups. Considering the similar pharmacologic effect of remifentanil and sufentanil, only the converted morphine dose was used in subsequent analyses. By contrast, there were no differences in the other intraoperative medications, surgical methods, anesthetic methods, and anesthetic times (Supplementary Table 2). We further investigated the relationship between the etiology, preoperative diagnosis, and analgesic dose. However, there were no differences in analgesics between patients with different etiologies or diagnoses.

Multivariate logistic regression analysis for the endpoint of postoperative consciousness improvement (improved vs. non-improved) was adjusted for etiology (TBI and ICH vs. ABI), preoperative diagnosis (MCS+ and MCS− vs. VS), and analgesic dose (high vs. low). Because of the close relationship between the diagnosis and the CRS-R score (as well as the subscores), we only included the diagnosis in the multivariate logistic regression. To simplify the evaluation, the analgesic doses were divided into high and low groups using the median as the boundary. We found an 8.2-fold higher odds ratio (OR) for postoperative consciousness improvement with a higher analgesic dose [OR, 8.22; 95% confidence interval (95% CI), 1.56–43.23; p = 0.013]. Furthermore, etiology (TBI vs. ABI: OR, 10.77; 95% CI, 1.31–88.43; p = 0.027) and preoperative diagnosis (MCS− vs. VS: OR, 9.6; 95% CI, 1.63–56.73; p = 0.013; MCS+ vs. VS: OR, 12.26; 95% CI, 1.06–142.07; p = 0.045) were independently associated with postoperative consciousness improvement (Figure 1).

We found that differences in preoperative diagnosis and etiology of DoC were associated with the incidence of postoperative consciousness improvement. The preoperative CRS-R scores and the auditory, visual, and motor subscores differed between patients in the improved and non-improved groups. Moreover, the patients with short-term consciousness improvement had higher CRS-R scores and the scores of visual, motor, communication, and arousal subscales at 3 months after surgery. Finally, the probability of regaining communication ability was significantly higher in patients in the improved group.

A diagnosis of minimal conscious state was previously defined as reproducible evidence of environmental- or self-awareness (Bender et al., 2015), while a diagnosis of minimal conscious state plus was characterized by the presence of linguistically mediated behaviors (Thibaut et al., 2020). Both states indicated a better-preserved residual consciousness (Laureys et al., 2004). Compared with VS patients, MCS patients demonstrated more widespread brain activation following simple sudatory stimulation and had more robust functional connectivity between the secondary auditory cortex and the temporal and prefrontal auditory-related cortices (Boly et al., 2004). An imageology study also revealed that the dorsomedial body volume of the thalamus was significantly lower in DoC patients, and that this atrophy was more extensive in VS than MCS patients (Fernández-Espejo et al., 2010). Furthermore, significant differences in the N200 and P300 waves of event-related potentials were found between MCS and VS patients (De Salvo et al., 2015). It is generally accepted that traumatic etiology is associated with more favorable outcomes (Ní Lochlainn et al., 2013; Giacino et al., 2018), indicating more complete neural networks in traumatic DoC. The higher rate of postoperative consciousness improvement in patients with traumatic etiology, better preoperative diagnosis, and higher CRS-R scores suggest that short-term improvement may be correlated with better residual consciousness. Given that significant differences were only found in the auditory, visual, and motor subscales in the present study, these subscales may have better predictive power for residual consciousness, which is consistent with our previous findings (He et al., 2014).

We also found that the short-term consciousness improvement was related to the level of consciousness at 3 months after surgery. More importantly, patients in the improved group had significantly higher scores on the communication subscale at 3 months after surgery, and this group had a higher reestablishment rate of communication ability. Although it is unclear whether this long-term improvement was caused by the intraoperative opioid application, the appearance of short-term postoperative improvement may help predict outcomes in DoC patients and guide more aggressive treatments. Communication ability is the most concerning issue for doctors and families of DoC patients, with a great deal of work performed to detect residual consciousness and potential communication abilities (Owen and Coleman, 2008; Gui et al., 2020). Complex methods, including structural (Sattin et al., 2021) and functional (He et al., 2015; Aubinet et al., 2020) magnetic resonance imaging, electroencephalography (Bai et al., 2021; Porcaro et al., 2022), and brain–computer interfaces (Müller-Putz et al., 2013), are used to evaluate residual consciousness and prognosis. However, changes in patient performance can affect the assessment and prognosis accuracy in DoC (Murovec et al., 2020). Our findings raise the possibility that a short-term administration of opioids (or remifentanil) may improve patients’ performance and assist in assessing residual consciousness.

We found no significant differences in postoperative consciousness improvement for any intraoperative medications except remifentanil. When combining the doses of remifentanil and sufentanil by converting to the morphine dose, the difference was even more marked. The lack of a difference with sufentanil treatment between the two groups may be related to use of a similar dose to that for induction drugs by the anesthesiologists. However, in subsequent multivariate regression analysis, the dose of opioid analgesics was an independent factor affecting short-term postoperative improvement.

Both remifentanil and sufentanil are opioid agonists, which produce an analgesic effect by activating the μ-opioid receptor (MOR) (Ziesenitz et al., 2018). Opioid receptors are widely distributed in the central nervous system. MORs located in the dorsal horn of the spinal cord, a major center of pain information processing (Braz et al., 2014), are essential for both the analgesic effects and the sensory input potentiation of opioids (Sun et al., 2019). The opioid system also potently modulates the mesolimbic circuitry, limbic circuitry, cortical and hippocampal circuitry, and various brain regions underlying motivation, fear responses, and cognitive functions (Puryear et al., 2020). MORs are predominantly expressed on GABAergic inhibitory interneurons and exert a potent disinhibitory effect on excitatory neurons (Nam et al., 2021).

There is a paradoxical hyperalgesia response in patients receiving opioids for pain control, whereby some patients become more sensitive to painful stimuli (hyperalgesia) and have a painful reaction to innocuous stimuli (allodynia)—this is termed opioid-induced hyperalgesia (OIH) (Velayudhan et al., 2013). The mechanism of OIH is not fully understood but is generally considered related to sensitization of the pronociceptive pathway caused by peripheral and central neuroplastic changes (Lee et al., 2011). OIH has been widely reported after the perioperative use of opioid analgesia (Colvin et al., 2019). A meta-analysis of OIH after surgery comparing 1,494 patients from 27 studies found that a higher intraoperative opioid dose (mainly remifentanil) was correlated with higher postoperative pain intensity and morphine use (Fletcher and Martinez, 2014). A biphasic time-dependent effect of fentanyl was also reported (Célèrier et al., 2000), with a nociceptive threshold increase lasting 2–5 h after fentanyl injection, followed by a sustained descending nociceptive threshold for up to 5 days.

The degree of hyperalgesia is dose-dependent. The occurrence of postoperative consciousness improvement was only found in patients receiving higher doses of opioids. Moreover, the time points of the consciousness improvement (within 48 h post-operation) and OIH were similar. The pattern between the onset of hyperalgesia and opioid analgesia use was similar to that between the short-term postoperative consciousness improvement and the use of opioids found in the present study, suggesting a relationship between OIH and consciousness improvement. However, we found no difference between minimally invasive and open surgery, indicating that the incision pain was not responsible for the consciousness improvement. Sensitization of the pronociceptive pathway is achieved via sensitization of primary afferent neurons and second-order neurons, an increased concentration of excitatory neurotransmitters (through enhanced production and release and diminished reuptake), and activation of descending facilitation of the rostral ventromedial medulla (Lee et al., 2011). These mechanisms ultimately lead to enhanced sensory afferent signals, which cause algesia in conscious patients. However, in patients with DoC, the same physiological responses to opioids strengthen the external environmental stimulation, enhancing the signal input to the ascending reticular activating system. This activation maintains cortical neurons in a state of facilitation and excitation, leading to a better clinical manifestation in DoC patients. Although this side effect of opioids is unwanted by anesthesiologists, it has the potential to become a therapy to promote recovery of consciousness in DoC patients. Further studies assessing the mechanisms of hyperalgesia in DoC are required.

Another potential mechanism underlying the postoperative improvement in consciousness with opioids involves enhanced dopamine release caused by the disinhibitory effect of activated GABAergic neurons. Opioids can activate MORs in the reward circuitry of the brain, which suppresses inhibitory neurotransmission in the ventral tegmental area and reduces the inhibitory postsynaptic event frequency of GABAergic interneurons, further increasing the release of dopamine into the striatum and prefrontal cortex (Colvin et al., 2019). This opioid-mediated disinhibition of dopaminergic neurons in the ventral tegmental area and substantia nigra pars compacta is hypothesized to cause the arousing and rewarding effects of opioids (Steidl et al., 2017). Recently, the occurrence of forebrain dysfunction in DoC was found to be caused by death and disconnection of neurons as well as “circuit-level” functional disturbances, which could be modulated to promote the recovery of consciousness (Schiff, 2010). The “mesocircuit” hypothesis suggests that normal anterior forebrain function depends on activation of thalamocortical projections in the central thalamus, which is inhibited by globus pallidus internal tonic signals during DoC (Schiff, 2008). High levels of dopaminergic activity maintain striatal firing rates, which inhibits the tonic signals of the globus pallidus and further activates the central thalamus (Grillner et al., 2005). The only validated treatment for DoC, amantadine (Giacino et al., 2012), is an indirect dopamine agonist. Increased striatal D2 dopamine-receptor availability and prefrontal cortical metabolism were found after amantadine treatment (Kraus et al., 2005; Schnakers et al., 2008). The same pattern of dopaminergic neuron activation is also found after opioid administration.

There are a number of other potential mechanisms involved in the beneficial effects of opioids. The latest clinical guidelines recommend evaluating and treating pain in patients with DoC (Giacino et al., 2018). However, as a subjective experience, it is difficult to recognize pain in DoC patients when no self-report is available. Nevertheless, recent advances in neuroimaging techniques have provided the capacity to perceive pain in DoC patients (Schnakers and Zasler, 2015). The high prevalence of spasticity in DoC patients and the positive correlation between the level of spastic muscle overactivity and pain (Thibaut et al., 2015) suggest that a large proportion of these patients suffer pain. Persistent pain may affect patients’ responses to the external environment, and pain relief via administration of analgesics may improve the clinical manifestations in DoC patients. Recent studies have shown that the pain-related brain circuit is incomplete in VS patients, with less evidence of painful conscious experiences. By contrast, MCS patients have sufficient cortical integration to process nociceptive stimuli, and the patterns of brain activation to painful stimulation in MCS patients were similar to those in healthy controls (Schnakers and Zasler, 2007). These differences in pain experience between VS and MCS patients may partly explain the different rates of postoperative improvement between these two groups.

This preliminary study included a small sample of patients. Further studies with more patients and standardized within-group differences are required. Furthermore, we only assessed the level of consciousness during the first two postoperative days because additional treatments introduced more distractors on the following days. It should also be noted that because of the lack of responsiveness in DoC patients, a wide range of the anesthetic agents were applied to our patients. Standardized anesthesia protocols should be stipulated in future studies.