Early reoperation means repeated MVD procedure within 1 month of the first MVD.

For experienced neurosurgeons, spasm resolution rates immediately after MVD are about 90–94 %, in other words, the fail rate is about 6–10 % (Engh et al. 2005). It’s very difficult to persuade the patient that the hemifacial spasm may automatically resolve after discharge, especially when the spasm was not relieved at all or even worsened after the first surgery.

It’s still controversial concerning about the treatment of such kind of patients. For example, Hyun et al. (Oh et al. 2008; Hyun et al. 2010) believed that any decision regarding retreatments should be drawn after postoperative 12 months or later, no matter how the outcomes of first MVD is. In contrast, Li (2005) reported that a second operation was indicated for patients with significant persisting or even worsening HFS after the first MVD; and if the spasm became less severe and less frequent after MVD, it’s better to wait for delayed resolution. Engh et al. (2005) reported that the patients with significant postoperative spasm undergo facial EMG; and if the postoperative abnormal muscle response (AMR) is still significant, they usually offered re-exploration, provided that the patient is well aware of the high risk of operative complications; otherwise, the patient is not offered a repeat MVD. Li et al. (2010) also considered AMR as an important reference to make the decision of early reoperation. It was observed that the long-term clinical outcomes of the intraoperative AMR disappearance before and after MVD were correlated. Thus, AMR may be considered an objective factor in the indication for reoperation (Ying et al. 2011).

Hughes et al. (2015) reported that re-imaging with high-resolution T2-weighted MRI will identify the missing culprit vessel after the failed MVD for hemifacial spasm. The imaging interpretation was concordant with the surgical results regarding artery versus vein in 86 % of cases and regarding the segment of the nerve contacted in 92 %. They found the unaddressed vascular compression is typically proximal to the previously placed Teflon felts.

Li Shi-ting team divided the facial nerve root into 4 zones (Zhong et al. 2010) and insisted that all zones should be explored carefully and decompressed adequately (Zhong et al. 2011). Otherwise, multiple neurovascular conflicts might be neglected and lead to failure. Especially, arteriole compression in zone 1 and “cross-type” AICA compression in zone 4 are the typical types of missing culprits (Li et al. 2010, 2013; Zheng et al. 2011).

In fact, the different percentages of delayed relief underlie this controversy. However, the delayed relief rate reported in literatures ranges from 90 % (Ishikawa et al. 2001) to <4 % (Zhong et al. 2015). The authors advocating early reoperation generally reported very low percentage of delay relief. And the authors reporting high percentage of delay resolution would accordingly recommend waiting. Therefore, unless an agreement as to the exact delayed relief percentage is reached in the future by multicenter prospective clinic trial, the debate will continue.

The different answers to if early reoperation should be performed reflect the surgeons’ different belief of the mechanism of hemifacial spasm. According to Moller et al., primary HFS may be due to hyperactivity of the facial nucleus, progressively induced by the continuous pulsation of the neurovascular conflict. If follows logically that the effect of surgical decompression takes time to decrease and normalize the clinical spasm, and the delayed spasm resolution may be due to the time required for remyelination of the compressed axons, as well as the recovery of normal excitability of the facial nucleus (Moller and Jannetta 1984, 1985a, b). Of course, a surgeon believing this hyperactivity hypothesis will firmly insist on the “waiting” principle. The hypothesis is not consistence with the fact that in most patients AMR vanished immediately when the compressing vessel was detached. It’s also challenged by Park’s report (Kim et al. 2009) that the severe indentation on the facial nerve were associated with good surgical outcomes, because the deduction of this hypothesis is that the more severe the compression is, the more hyperactivity there should be in the nucleus. In contrast, Zheng et al. hypothesized that the cross-transmission between the facial nerve fibers is bridged by sympathetic nerve fibers on the offending artery wall (Zheng et al. 2012a, b). Hence, this might explain most of patients who were relieved immediately after facial nerve and offending vessels wall were separated by Teflon. And they believe that the reasons for failure may include missing the multiple vessels, Teflon pled get movement after closure, or incomplete decompression. Accordingly, a neurosurgeon who believes the sympathetic bridge hypothesis tends to do early reoperation to find the missing offending vessels or the inadequate decompression. However, neither hypothesis has been fully proven by substantial direct evidence. Probably, some patients are of hyperactivity, but the others are mediated by sympathetic bridge. Thus, the debate as to whether early reoperation is necessary will continue until the mechanism of hemifacial spasm is clearly addressed.

Based on the literatures mentioned above and our own experience, here we propose indications for early reoperation.

If the above listed five criteria are matched, an early reoperation is recommended.

On the contrary, if there is initial relief after surgery and then subsequent recurrence, the decision to reoperate should be considered very carefully (Park et al. 2006).

If the first MVD surgery was aborted because of neuroelectrophysiological monitoring alert, a reoperation should be scheduled within 1 week.

As described in our previously reported papers (Zhong et al. 2010), the early reoperation was performed via the original approach. Craniectomy is very easy by removing sutures and titanium wire mesh. There is no or very little adhesion.

If arachnoid membrane is not dissected adequately, additional dissection is needed in reoperation. The dissection was started from the caudal cranial nerves. While the arachnoid membrane between facial nerve and the caudal cranial nerves being opened fully, the cerebellum as well as flocculus was gently raised until the pontomedullary sulcus was exposed.

For reoperation, the previously decompressed site was double checked at first to confirm that a satisfied decompression was completed. Then the full course of facial nerve was inspected through different angles. Exploration began from zone 2 and then moved to zones 3, 1, and 4 (zone 1, where the nerve root emerges to the brainstem surface from the parenchymal and goes through the pontomedullary sulcus; zone 2, where the nerve root attaches to the surface of the pons; zone 3, where it gradually detached from the brain stem; zone 4, where the nerve runs free in the subarachnoid cistern and extends to the internal meatus).

Any arteries or big veins in contact with the facial nerve were detached. After the offending vessel was moved away from the nerve, pieces of Teflon sponge were gently placed between the vessel and the brainstem, so as to transpose the course of the vessel. However, when transposition is impossible, Teflon felt can be inserted into the gap between the nerve and the vessel. The suspicious venulein contact with the nerve was coagulated with low power.

According to Hughes MRI (Hughes et al. 2015) study and Zhong clinical observation (Zhong et al. 2015), the missing culprit vessels usually lie proximal to the previous Teflon felt, especially at zone 1. Therefore, early reoperation should be mainly focused on this portion; sometimes, endoscope may help find small vessels in the pontomedullary sulcus.

With the aid of intraoperative AMR and ZLR monitoring, we found a new type of vascular compression pattern, the cross type, which involves the AICA passing through the gap between cranial nerves VII and VIII and compressing the cisternal portion of facial nerve, that is, zone 4. For this type, the treatment is to interpose Teflon felt between the facial nerve and AICA (Li et al. 2013; Zheng et al. 2011).

If new vascular compression was not found in reoperation after careful scrutinizing, it’s very likely that the shape and tension of Teflon felt are not suitable. The old Teflon felt should be replaced by a new one (Li et al. 2010). If the Teflon felts previously inserted should be pulled out, great caution should be paid as pulling large sponges may break some perforating arterioles.

In the early reoperation, some small perforator arteries turned out to be the culprit vessels, which were hidden beneath the big artery. To this subset, determination of the culprit vessels in the reoperation was completely relied on the intraoperative AMR monitoring. These small vessels looked just ordinary, and did not seem “compressing” in appearance; however, when they were lifted with a nerve dissector, the AMR disappeared in no time. After these small perforator arteries were decompressed by small piece of Teflon sponges, these patients were cured (Li et al. 2010).

There are higher risks of operative complications related to the early second MVD (Li et al. 2010). Of course, there is nothing with adhesion. When performing the early reoperation, the surgeon is under great psychological stress and always tries his best to find additional culprit vessels; thus, the entire course of facial nerve will be explored over and over. We think this is the main reason for high risk of early reoperation (Li et al. 2010). Therefore, each step should be performed carefully. For example, the dissection of the arachnoid membrane, exploration of the facial nerve, detachment of the offending vessel, and insertion of the Teflon sponge all should be done gently under direct vision. While dissecting the arachnoids around the nerve, the feeding arterioles for the nerve should be preserved. While mobilizing the offending artery, retraction of the nerve should be avoided. When vasospasm was observed, the operation should be paused and sometimes narceine was administered on the vessel surface. When the coagulation of the small vein is strongly required, it should be kept far away from the nerve. Meanwhile, real-time neuroelectrophysiological monitoring of the nerve function is mandatory (Zhong et al. 2010).