Discussion
Among postoperative complications in parotid surgery, there is no doubt that facial nerve palsy is the most important. In a meta-analysis by Witt et al. 10), temporary and permanent postoperative facial nerve palsy reportedly occurred in 59.8% and 4.0% of cases, respectively, in total parotidectomy, and in 17.9% and 0.2% of cases, respectively, in partial parotidectomy. In our facility, a review of 902 patients who underwent surgery for benign parotid tumors revealed that 19.3% of patients had temporary palsy and 0.5% had permanent palsy; these results are in line with previous reports 3). Partial parotid lobectomy requires confirmation of the facial nerve, so even if the surgeon successfully preserves the facial nerve, transient facial paralysis may occur. Therefore, we examined whether it is possible to predict the incidence of postoperative facial nerve palsy, as well as the time required for recovery, by measuring the responses of the nerve to stimulation during surgery.
When measuring electrical potential in response to facial nerve trunk stimulation before and after tumor resection, significantly higher responses were observed after tumor resection than before for all branches in group A. We suspected the effect of muscle relaxant administered at the start of anesthesia remained before tumor resection. The median time from muscle relaxant administration at the time of anesthesia introduction to electrical potential measurement was 69 minutes before tumor resection and 117 minutes after resection. It was previously reported that in parotid surgery for which muscle relaxants were administered at the start of general anesthesia, changes in the numerical values of the muscle relaxant monitor differed from patient to patient, and as a result, differences in FNM response were also noted among patients 11). In addition, with muscle relaxant monitors, measurements are obtained at the ulnar nerve, and the ulnar nerve and facial nerve respond differently to muscle relaxants 12). Therefore, in this study, based on the values of the muscle relaxation monitor before tumor resection, it was assumed that there was no effect of muscle relaxants, but in fact, in some cases, the effect of muscle relaxation on the facial nerve may not have completely disappeared. In other words, we considered it likely that the electrical potentials in response to stimulation measured before tumor resection were lower than the true values. In group C, a decrease in electrical potential was observed for all branches after resection compared to before resection; in particular, there was a statistically significant difference for MB. This decrease in electrical potential is likely due to damage to the facial nerve during resection, but actual difference might have been even larger when considering the effect of the muscle relaxant before resection. On the other hand, in group B, no significant difference in electrical potential was observed before and after resection. In group B, we did not note any occurrence of muscle paralysis corresponding to the measured branches, but we presumed there was minor damage to the nerve because muscle paralysis corresponding to other branches occurred. Accordingly, we considered that the reason we observed no significant difference in electrical potential in response to stimulation before and after resection in group B was because the decrease in electrical potential after resection was offset by the effect of the muscle relaxant before resection.
In contrast, the measurements of electrical potential in response to nerve trunk and branch stimulation after tumor resection were performed when enough time had passed after administration of the muscle relaxant. It is therefore likely that the effect of the muscle relaxant had already disappeared. Given that there was no damage to the non-exposed part of the nerve, the electrical potentials in response to branch stimulation were considered normal. In contrast, the electrical potentials in response to trunk stimulation (based on the results of examinations before and after resection) were estimated to be normal in group A, slightly decreased in group B, and decreased in group C. In fact, the electrical potentials in response to trunk stimulation in group B were slightly lower than those of group A, although no significant differences were observed. In group C, responses to trunk stimulation decreased for all branches. Among the branches, a significant difference was observed only for BB and MB, but it is highly likely that other branches would also show a significant difference if the number of patients were increased.
To estimate the degree of postoperative nerve damage and the resulting facial nerve palsy, it was considered more appropriate to use the response ratio, rather than absolute values. We calculated the ratio of electrical potential in response to trunk stimulation before and after tumor resection (ARR1) and in response to trunk stimulation and branch stimulation after tumor resection (ARR2). ARR1 is the ratio that relates to the electrical potential in response to stimulation before tumor resection, whereas ARR2 relates to the electrical potential in response to branch stimulation. Considering that there might have been the residual effect of the muscle relaxant on stimulation before tumor resection, ARR2 is probably more reliable than ARR1. In fact, compared to ARR1, the number of outliers was clearly lower for AAR2. When we constructed ROC curves for the relationship between ARR and the presence or absence of postoperative facial nerve palsy, the cut-off values were 0.63 for ARR1 and 0.55 for ARR2. Using these cut-off values, the sensitivity, specificity, and accuracy were determined to be 92.7%, 77.8%, and 91.7%, respectively, for ARR1, and 96.5%, 93.1%, and 96.0%, respectively, for ARR2. We found that ARR2 can reduce false positives better than ARR1, indicating that ARR2 is the more reliable measure.
Together, these data show that by obtaining ARR during surgery, we can predict facial nerve palsy intraoperatively, and ARR can serve as an index for administering drugs such as steroids immediately postoperatively. A limitation of this method is that it is not applicable in cases where the tumor is resected without identifying the facial nerve, or where damage to the nerve involves an area more central than the identified facial nerve trunk, or more peripheral than the facial nerve branches that were traced.