Introduction

Lumbar disc herniation is the most common spinal degenerative pathology, and discectomy is the most frequently performed spinal surgery [7, 13, 14]. Reherniation after lumbar discectomy (RLD) is a known frequent failure and has been described extensively in the literature [5, 9, 10, 12, 20, 23, 30, 31, 53]. The reported reherniation rate after lumbar discectomy ranges from 3 to 18% [3, 41, 55]. RLD is defined as recurrence of symptoms after a postoperative pain-free period caused by a new disc herniation at the same ipsilateral level [43]. If RLD is diagnosed, reoperation may be required. The SPORT study described reoperation rates of 2% at 90 days, 4% at 1 year, and 10% after 4 years post-surgery [54]. Moreover, only 78% patient satisfaction was reported at 1 year post-surgery in the Swedish Spine Registry [46]. The reoperation rate after 3 years in our current trial was 19.3% in the control group and 11% in the group with annular closure device implantation [21].

Many risk factors have already been reported, such as age, sex, body mass index (BMI), smoking, type and size of annular defect, amount of removed disc volume, grade of disc degeneration, disc height, and range of motion [1, 8, 25, 26, 32,33,34,35, 42, 51]. In the study by Meredith et al., obese patients were 12 times more likely to suffer RLD compared with non-obese patients [33]. Several authors have confirmed the findings that smoking and increasing age are associated with RLD [1, 22, 25, 32, 34, 42]. A Finnish study found no difference between genders concerning risk for reoperation [20]. Carragee et al. demonstrated a trend towards higher reherniation rates with limited discectomy compared with the more aggressive discectomy group [9]. This was confirmed in a review article by Watters et al. which described a significant increase in RLD rate following conservative versus aggressive discectomy [52]. Furthermore, it was shown that patients with annular defects > 6 mm account for most of the clinically relevant reherniations [8]. Kim et al. recently acknowledged the correlation between a wider defect and a higher reherniation rate [25]. Moreover, a larger mean annular defect area (46 mm ± 18 mm2) and a lower percentage of removed disc volume were obtained in patients with reherniations in the study by McGirt et al. [33].

A recently introduced annular closure device (Barricaid®, Intrinsic Therapeutics, Inc., Woburn, MA) was developed to occlude the annular defect and prevent reherniation [56]. It has been reported that Barricaid® might reduce the risk of reherniation in high-risk patients. These findings were observed in two prospective, single-arm trials with a total of 75 patients. A low rate (1.4%) of symptomatic RLD was reported, and Barricaid® use was associated with a good overall outcome [6]. The 2- and 3-year follow-up results of a European prospective, multicenter randomized controlled trial including 554 patients in 21 centers have already been published [21, 50]. The frequency of symptomatic reherniation after 2 years was lower in the annular closure device group compared with the control group (12% vs. 25%, p < 0.001) [50]. Similar results have been reported for the 3-year follow-up (14.8% vs. 29.5%) [21].

The aim of the present data analysis was to identify factors associated with very early reherniation within the first 3 months after surgery in patients undergoing one-level lumbar discectomy, with or without annular closure.

Methods

Trial design

This study includes data of a post-marketing, prospective, multicenter, randomized controlled trial (RCT) of limited discectomy—with and without use of an annular closure device (Clinicaltrials.gov NCT01283438). The study protocol was published previously [27].

Participants

This European study included 554 patients in 21 centers located in Germany, Switzerland, Austria, Belgium, The Netherlands, and France from 2010 to 2014 and was conducted in compliance with the Ethics Committee, GCP, and ISO requirements. A total of 276 patients were randomized to ACG and 278 patients to CG. Only 267 patients in the ACG with a device implanted (no implantation n = 4, failures to implant n = 5, implantation on second attempt n = 3) were included in our analysis. Informed consent was obtained for all patients. Study oversight regarding safety was provided by an independent data safety monitoring board.

Randomization and masking

Patients were randomized web-based intraoperatively 1:1 to either standard discectomy (control group, CG) or standard discectomy with implantation of an annular closure device (annular closure group, ACG). The main criteria for study enrollment were an annular defect of 4 to 6 mm in height and 6 to 10 mm in width. Neither the surgeons nor the participants were blinded for the treatment allocation.

Early reherniation after lumbar discectomy

RLD was reported as adverse event in the RCT. RLD was reported as ipsi- or contralateral new disc herniation at index level and had to be confirmed by either magnetic resonance imaging (MRI) or during reoperation.

Annular closure device

The ACD (Barricaid®, Intrinsic Therapeutics, Inc., Woburn, MA, USA) was developed to prevent reherniation after lumbar discectomy, through closure of the annular defect. European Conformity (CE) marking was received in 2009 and FDA approval in 2019. The device consists of a flexible mesh, a mechanical barrier to the annular defect, and a bone anchor.

Surgical technique

The majority of surgeons preferred intraoperative microscopy. A limited nucleotomy described by Spengler [45] was performed. The volume of removed disc material varied according to the surgeon’s implementation of limited discectomy and was measured in cubic centimeters. The height and width of the annular defect were measured using sizing paddles. Randomization occurred after discectomy. No further nucleus removal was allowed thereafter. Device implantation and correct alignment were verified by intra- and X-ray.

Follow-up

Clinical examination and X-ray were performed at the consultation after 6 weeks and 3 months. Assessment included Visual analog Scale (VAS) leg (LP-VAS) and low back pain (BP-VAS), Oswestry Disability Index (ODI), and Medical Outcomes Short Form-36 (SF-36) physical and mental component summary scales [15, 37].

Radiographic assessment

Preoperative imaging of the lumbar spine included anterior-posterior /lateral X-ray with extension/flexion, computed tomography (CT), and MRI. Analysis of the images was performed by a team of independent radiologists.

An MRI was performed in cases with clinical evidence of early RLD. Disc degeneration was graded according to Kellgren and Lawrence (K-L) [18] and Pfirrmann [38].

Statistical analysis

Descriptive statistics was used for demographic characteristics, intraoperative findings, and pain scores. Univariate logistic regression analysis was applied to investigate correlation between potential risk factors and early RLD in three different populations: (1) all subjects, (2) subjects implanted with the ACD, and (3) control subjects.

The multivariate logistic regression analysis was applied on the “all subjects” population only. Any variable with a p value < 0.1 in the univariate regression analysis was utilized in a multivariate logistic regression, stepwise, backward elimination. At each step, the variable with the highest insignificant p value was dropped, until all p values were less than 0.05.

Results

Preoperative demographic characteristics

There were no significant differences between the groups (Table 1). Follow-up rates at 3 months were 97.4% for ACG and 95.3% for CG. Ten patients (CG n = 6, ACG n = 4) withdrew from the study before their 3-month visit (Fig. 1).

Table 1 Preoperative demographic characteristics
Full size table
Fig. 1
figure 1

Flow diagram of randomization and patient follow-up

Full size image

Preoperative radiological findings

The majority of patients had a Pfirrmann grade III (423 patients, 76.3%) or IV (100 patients, 18%) disc degeneration (DD). In comparison, DD classified according to Kellgren and Lawrence presented mostly as doubtful (235 patients, 42.4%) or minimal (229 patients, 41.3%) changes (Table 2). There was no significant difference in Pfirrmann grades (p = 0.398) or Kellgren and Lawrence classification (p = 0.441) between ACG and CG groups. Average disc height was 8.9 ± 2.1 mm (ACG: 8.9 ± 2.1 mm, CG: 8.8 ± 2.2 mm), translational range of motion (ROM) 0.4 ± 1.5 mm (ACG: 0.5 ± 1.4 mm, CG: 0.3 ± 1.6 mm) and angular ROM − 4.2 ± 4.5° (ACG: − 4.0 ± 4.5°, CG: − 4.4 ± 4.5°).

Table 2 Disc degeneration according to Pfirrmann’s and Kellgren and Lawrence classification—with the corresponding reherniation rates
Full size table

Intraoperative findings

A full thickness defect was present in 51.4% of patients. Approach through the existing annular defect was used in 62.4% of patients. The overall mean defect size was 38.8 mm2, and the mean amount of nucleus removed was 1.3 mL (Table 3).

Table 3 Herniation type, annular defect type, surgical approach, defect geometry, and annular defect size at initial surgery
Full size table

Postoperative clinical outcome

We observed postoperative at the 3-month follow-up a significant improvement in VAS leg pain, VAS back pain, and ODI in ACG and CG compared with the preoperative baseline (< 0.0001).

Risk factors for RLD

Twenty-two patients (4%) suffered a symptomatic index level reherniation during the first 3 months, 4 (1.5%) occurred in ACG and 18 (6.5%) in CG. Type of RLD is shown in Table 4. RLD occurred ipsilateral in 91% (ACG 75%, CG 94%). Ipsilateral reherniations in ACG were found lateral to the device mesh. RLD due to device malfunction did not occur. No patient described an event that precipitated the recurrence. Reoperation due to RLD was necessary in 12 patients (n = 3 ACG, n = 9 CG). Median time from pain recurrence to reoperation was 17 days (1 day ACG, 20 days CG).

Table 4 Recurrent herniation type
Full size table

Regarding risk factors for RLD, a significant correlation was found for age ≥ 50 years and DD. Control subjects ≥ 50 years of age were more likely to have a symptomatic reherniation (CG p = 0.051, OR = 2.61) (Table 5). In CG, reherniation rate of patients aged < 50 years was 4.6% compared with 9.9% for patients aged ≥ 50 years. In ACG, all RLD occurred in patients < 50 years. Disc degeneration according to the K-L classification was a predictor for RLD in CG (CG p = 0.013, OR = 2.15). The RLD rate in CG was 13.2% for moderate degeneration. Pfirrmann grade IV degeneration showed a reherniation rate of 9% overall and 13.2% in CG. In summary, subjects with a 1 point higher Pfirrmann grade were 3.2 times more likely to have a symptomatic RLD.

Table 5 Multivariate analysis of predictors for early symptomatic reherniation
Full size table

All other variables including sex, BMI, smoking status, amount of nucleus removed, herniation or defect type, defect size, surgical approach, and radiological data—including range of motion and disc height—were not significant in the uni- and multivariate logistic regression analysis.

Smokers had a trend for a higher symptomatic RLD rate overall (p = 0.07, OR = 2.26). Although sex was not a significant predictor for reherniation, the RLD rate for females in CG was higher (F: 7.5% vs. M: 5.9%), and all reherniations in ACG occurred in females.

Implantation of ACD correlated with an overall reduction of symptomatic RLD (p = 0.007, OR = 0.23). In the multivariate logistic regression model, the following three variables remained after use of backward stepwise elimination: current smoker (p = 0.049, OR = 2.57), Pfirrmann grade (p = 0.010, OR = 3.26), and implantation of ACD (p = 0.007, OR = 0.23). Reherniation rate was 0.9% for non-smokers with grade III degeneration and implantation of ACD, compared with 19.2% for smokers with grade IV degeneration and no ACD device. In summary, the multivariate model predicts that a smoker with one point worse Pfirrmann grade without ACD is 36.4 (2.57 × 3.26 × 1/0.23) times more likely to suffer a symptomatic reherniation within 3 months.

Discussion

In our multicenter randomized controlled trial, the early RLD rate until the 3-month follow-up was 4% overall. These rates are in line with the literature [9, 20, 23, 30, 31, 49]. Implantation of an ACD was a significant predictor for RLD risk reduction in the overall population.

The median interval until RLD in the study by Moliterno et al. [35] and Wera et al. [55] was 3 and 2.8 month, respectively; McGirt et al. identified two postoperative time periods when RLD most frequently occurred, < 4 months or > 11 months after surgery [32].

In our current study, RLD reoperation rate was 2.2% within 3 months. The reported reoperation rate was 2% at 90 days in the SPORT trial [54] and 4.9% at 3 months after discectomy in the study by Kim et al. [23], where 46% of reoperations were performed within half a year. In Aizawa et al.’s trial, reoperation for RLD was performed significantly more often within the first 6 months [2].

Patients with grade IV Pfirrmann DD in CG and overall had the highest reherniation rate. Several authors found that patients with low to moderate DD were at increased risk of RLD [1, 11, 26]. Dora et al. presented a reduced RLD rate for each Pfirrmann grade increase [11].

A trend for RLD and increasing age has been reported [1, 23, 24, 32]. We confirmed these findings with a significant correlation between age ≥ 50 years and RLD in CG. Martin et al. presented results showing older age associated with higher short-term and lower long-term reoperation risk [31].

Our study identified smoking as a risk factor for RLD. Earlier studies have also shown that smokers have a higher recurrence rate [4, 19, 26, 34, 42, 47]. The lumbar disc space is known to be an avascular structure and absorbs all nutrients by diffusion [16, 17]. It was proposed that smoking-associated hypoxia inhibits closure of the discogenic and ligamentous defect after discectomy [28, 29, 34, 36, 44].

Similar to our findings, Martin et al. noted that female sex was associated with a higher risk for RLD and reoperation [31]. In contrast, Suk et al. reported male sex to be a risk factor for RLD [47]. However, other studies showed no difference between the genders [20, 33, 35, 48].

As with other studies [39, 40, 57], BMI did not prove to be a risk factor for RLD in our trial. However, patients with a BMI greater than 40 kg/m2 were excluded, and the mean BMI was 26.3 kg/m2. Several other authors have found higher BMIs to be associated with RLD rates [1, 24, 33, 42]. In contrast, Moliterno et al. found that patients with RLD had on average significantly lower BMIs [35].

It was shown by Carragee et al. that patients with large annular defects (> 6 mm wide) account for most of the clinically significant RLD [8]. Kim et al. [25] confirmed these results in a patient cohort of 467 patients, whereby large annular defect (> 6 mm) was a risk factor for RLD.

McGirt et al. [32] observed a greater mean annular defect area for patients with symptomatic RLD. In patients with early RLD (within 4 months), mean annular defect was larger compared with patients with RLD at a later time point [32]. An annular defect size of > 54 mm2 correlated with an 18% RLD rate, which is 4 times higher compared with 4.7% rate for patients with defect size of 36 mm2 [32].

A minimum of 6 mm and maximum of 10-mm defect width were necessary for our study enrollment. This might explain why we did not observe correlation between RLD and defect size. Awareness of a maximum allowed width and height made the surgeons more careful in their incision of the annulus. Only a small number (7%) of patients in our study had a defect size of ≥ 54 mm2, and the RLD rate in this group was 7.9%.

We were not able to observe any correlations between removed nucleus amount, disc height, type of herniation, sagittal ROM, and RLD rate.

Conclusion

Significant risk factors for early recurrent disc herniation after lumbar microdiscectomy in this study included age ≥ 50 years (OR = 2.6) and moderate disc degeneration (OR = 3.2). The annular closure device was a predictive factor that significantly reduced the risk of early reherniation. These results suggest that implantation of an ACD can prevent early reherniation after lumbar discectomy.