Ebeling et al. [9] examined 92 patients who underwent revision surgery after primary lumbar discectomy and reported that disc herniation recurred at the same level and site in 43% of patients, whereas it recurred at a different level in 15% of patients. A combination of small disc herniation and epidural granulation formation was observed in 23% of patients, whereas epidural granulation tissue had an important role in the development of symptomatic radicular pain in only 5% of patients [9]. Furthermore, Nakajima et al. [10] reported that 92 patients underwent revision surgery because of recurrent herniation (57%), stenosis owing to granulation (13%), osseous stenosis (16%), postoperative spinal instability (11%), and adhesive arachnoiditis (3%). In addition, Hanakita and Suwa [11] reported regarding 15 patients who underwent revision surgery at the same level and site and noted recurrent herniation in 33%, compression by epidural granulation tissue in only 13%, and a combination of small recurrent herniation and epidural granulation in 47%. Several studies have reported regarding the significance and role of postoperative epidural granulation. According to Sen et al. [12], epidural scar tissue is formed at 6 weeks to 6 months postoperatively. After 12 months, further scar formation is not observed. Ross et al. [13] reported that patients with a marked amount of epidural scar formation tend to experience a 3.2-fold greater radicular pain compared with those with less extensive scar formation. In contrast, Annertz et al. [14] reported no difference in the presence and extent of epidural fibrosis between symptomatic and asymptomatic patients. In summary, the actual significance and role of epidural postoperative scar formation remain controversial.

We analyzed the intraoperative status of the affected root, and investigated the association between revision surgery time and associated pathological findings (Fig. 2). As mentioned, epidural granulation was observed after several months. Therefore, we suggest that pathological findings and reoperation timing have some association, but there was no statistical difference.

The occurrence of dural lacerations with or without root injury is one of the most frequent complications of lumbar discectomy. The frequency of dural lacerations during lumbar discectomy has been reported to be 1.1%–3.7% among patients who undergo initial discectomy and 8.1%–21% among those who undergo revision surgery [45678]. The sites of dural laceration were examined using intraoperative video footage and operative records. As mentioned in the Results section, dural lacerations in the revision surgery group mainly occurred at the nerve root shoulder, where marked epidural scar formation and adhesion were observed. We assessed and drew the location of dural lacerations using the operative records and surgical video footage (Fig. 3). A meticulous detachment procedure must be performed at this site to avoid dural lacerations. Furthermore, at our department, foraminotomy is performed first by following the nerve root sleeve proximally to identify the intact nerve root sleeve. Takahashi et al. [15] assessed the locations of dural laceration and noted that in patients with lumbar disc herniation, tears mainly occurred near the herniated disc, similar to that observed in our study. In the revision surgery group, dural lacerations were observed adjacent to the root-shoulder region, where severe granulation formation and marked adhesive changes occurred [1617].

Several studies have reported the results and reoperation rates of lumbar discectomy [3491618]. Caspar et al. [18] compared MSD and conventional lumbar discectomy (CD) in 418 patients. According to their study, the complication rate was lower and the length of hospitalization was shorter in the MSD group than in the CD group; thus, patients returned to work faster in the MSD group than in the CD group. The reoperation rates in the MSD and CD groups were 3.4% and 7.6%, respectively [18]. In a randomized controlled study, Tureyen [19] compared 114 patients and reported that surgical outcomes were equal in the MSD and CD groups, and there was no significant difference regarding hospitalization length in both the groups. However, the MSD procedure allowed patients to return to their work and normal lives earlier, without excessive reliance on postoperative narcotic analgesics. Righesso et al. [1] compared CD and microendoscopic discectomy (MED) in 40 patients. Although the outcomes were the same in both the groups, the size of the skin incision, hospitalization length, and operative time were significantly less in the MED group. Several findings have been reported regarding the recurrence rate of MED. Hashimoto et al. [20] and Wu et al. [3] reported recurrence rates of 16% and 0.7%, respectively. Table 2 shows previously reported recurrence rates and follow-up periods for several surgical procedures [34916171819202122].

The operative time and length of hospital stay were longer, but blood loss was lesser in our study than in other studies [2324]. In our study, primary and revision surgeries were thoroughly performed; therefore, no significant differences were noted between the two groups. It appears that patients have a longer hospital stay in Japan compared with other countries. Although the length of hospital stay for such patients is usually shorter at present, the length of hospital stay for such patients approximately 10 years ago would extend to a month, as mentioned in a previous study [22].

The inherent limitation of this study was its retrospective design. Therefore, we could not assess all patients who underwent a primary surgery. Moreover, some patients may have undergone revision surgery at another hospital, and hence, the recurrence rate may be slightly higher. Gadolinium-enhanced MRI should be used to evaluate epidural scar formation. By evaluating MRI and the intraoperative status of patients, the accuracy of imaging findings can be determined.