Lumbar discectomy (LD) is performed in patients with radicular pain caused by a herniated lumbar disc (HLD) after failed conservative treatments including medication, physiotherapy, and various neural blocks [23].Considerable research has studied LD surgical techniques including open lumbar discectomy, microendoscopic discectomy, percutaneous endoscopic lumbar discectomy, and other modified techniques [4].

Results of surgical LD are unsatisfactory in 3% to 19% of patients due to outcomes such as sustained pain, caudae eqinae, and recurrence. By radiologic imaging, recurrent HLD is found in 5% to 15% of these patients, and unresolved symptoms might lead to a secondary operation [17,19, 23].

Risk factors of HLD recurrence including age, sex, body mass index (BMI), symptom duration, type of herniation, level of operation, degenerative changes, disc volume, and operative technique have been studied [3,22,37].

The purpose of this study was to investigate risk factors of secondary LD for recurrent HLD from a single surgeon’s data.

Owing to the retrospective nature of the study, this study was not approved by the Insitutional Review Board of Hanyang University Medical Center. Our study population included 160 patients who underwent open microscopic LD between 2009 and 2016 by a single surgeon in Hanyang University Medical Center.

Demographic information (age, sex, BMI, hypertension, diabetes mellitus, smoking, vertebral level of operation, and type of operation) and radiologic features (type of disc herniation, location of herniation, degree of disc degeneration, disc height, and height of vertebral body) were obtained retrospectively (Table 1). Postoperative magnetic resonance imaging (MRI) was performed in patients with symptoms to confirm recurrence.

By assessing axial and sagittal MRI, herniation type was classified as protrusion, extrusion, or sequestration. Protrusion was a focal protrusion with an intact annulus fibrosus. Extrusion was a penetration of the nucleus pulposus (NP) through a defect in the annulus fibrosus. Sequestration was defined as a particle-free spinal canal.

The degree of disc degeneration was graded according to the Pfirrmann disc degeneration grade from using sagittal T2- weighted MRI (Table 2). Disc height, vertebral body height and disc/vertebral body height ratio was estimated from the height of the disc and vertebral body in T2-weighted sagittal MRI. All patients were performed with discectomy and sequestration, and foraminotomy was performed in patients with foraminal stenosis on MRI finding. And the operation type was classified by degree of bone loss in the lamina and facet during primary surgery.

Statistical analysis was performed as follows. A logistic univariate regression was used to identify risk factors of HLD revision. Variables that were independent risk factors of secondary LD for recurrent HLD were analyzed by multivariate logistic regression, and the results were reported as an adjusted odds ratio (OR) and 95% confidence interval (CI). A p-value <0.05 was considered to indicate a risk factor of HLD recurrence. All data were analyzed with R version 3.3.2 (R Foundation for Statistical Computing, Vienna, Austria).

Probability of revision surgery was assessed using the Kalpan-Meier method and log-rank tests.

Variables with a p-value <0.10 were re-entered in the multivariable logistic regression model using a backward stepwise method. To evaluate risk factors for revision, a multivariate logistic regression model with a backward stepwise method was used. ORs and 95% CI were reported for statistically significant factors (p-value <0.05). All data were analyzed with R, version 3.3.2 (https://www.r-project.org/; R Foundation for Statistical Computing).

One hundred sixty patients underwent primary LD during the study period, and 24 patients underwent secondary LD. Patients with revision and non- revision HLD were of similar age (56.0 vs. 56.5 years). Also, sex did not differ between groups. In total, 51 patients had hypertension, 45 in the non-recurrent group (33.1%) and six in the revision group (25.0%).

In the revision HLD group, the incidence of diabetes mellitus and smoking was more than 1.5 times higher than in the non- revision group, but the difference was not statistically significant (p>0.05).

Herniated discs were classified as protrusion (54 patients), extrusion (69 patients), or sequestration (37 patients). In each group, eight patients underwent a second operation for revision HLD.

The highest revision rate (41.7%) occurred at level L4–5 in 10 of 85 patients, followed by level L5–S1 in six of 43 patients (25.0%). Four patients each had revision at level L2–3 or L3–4. Ten patients had a multi-level operation with no evidence of revision. Consequently, the level of operation was not statistically significant.

Demographic features and surgical information of our patients are listed in Table 1. Symptoms due to HLD recurrence were the only complication collected from our database.

A univariate analysis reported that Pfirrmann grade and BMI differed significantly between groups.

We estimated disc degeneration using Pfirrmann disc degeneration grade. Most patients were grade 3 (92 patients), followed by grade 4 (61 patients) and grade 2 (seven patients). No patients were grade 1 or 5. In the revision HLD group, 17 patients (70.8%) were grade 4 and the rest were grade 3. Further analysis demonstrated that higher Pfirrmann grade was associated with an increased revision rate.

The association between BMI and HLD revision was also statistically significant (p<0.05). The mean BMI of the revision HLD group was 29.8 (range, 27.1–32.8), and the mean BMI of the non-revision HLD group was 26.9 (range, 24.7–29.5). The total mean BMI was 27.2 (range, 24.7–32.8). We included overweight patients in the non-obese group. Obese patients were at 1.2-times higher risk for revision than were non-obese patients (OR, 1.20; 95% CI, 1.06–1.37) (Table 3).

Table 4 shows the univariate and multivariate logistic regression analysis of revision surgery. Two variables remained statistically significant after adjusting for confounding factors : 1) Body mass index (OR, 1.30; 95% CI, 1.12–1.52; p=0.0007) and 2) Pfirrmann disc grade (OR, 2.13; 95% CI, 1.29–2.97; p=0.0002).

Fig. 1 shows the Kaplan-Meier curves of the cumulative hazards of revision surgery stratified by BMI categories. Probability of revision surgery in obese group was significantly higher than normal and overweight group (log‐rank test, p<0.001). With increased BMI over 30, cumulative revision risk gradually increased.

The rate of HLD recurrence has not been well defined. The rates of HLD recurrence have a wide range because they can include disc herniations at the same level on the ipsilateral or contralateral side as the previous LD [28,37,38]. However, our study only included patients with recurrent HLD at the same level and on the same side, which we considered to be more appropriate. And Patients with recurrent symptoms, with radiographic evidence of reccurent HLD at the same side were considerd to recurrence group. The revision group was defined as the group of patients who had secondary LD for persistent symptoms after conservative treatment (Fig. 2).

Various studies have reported HLD recurrence after LD, and 5% to 15% of patients undergo secondary LD [20,24]. Our study reported a 15% incidence of recurrence within a mean of 28.3 months. This study only includes patients with secondary LD. If we included patients who underwent spinal fusion due to back pain caused by instability, the recurrence rate would be much higher.

Proposed risk factors of HLD recurrence include age, sex, smoking, BMI, trauma, symptom duration, herniation type, level of the operation, degeneration, disc volume, and operative technique. Some researchers have found statistically significant differences based on age, sex, BMI, or degeneration. Our study found a statistically significant association with BMI and degeneration [3,22,37].

The intervertebral disc consists of an inner NP and an outer annulus fibrosus. The NP comprises mainly proteoglycans with type II collagen, and the annulus fibrosus comprises type I collagen. The proteoglycans, mostly aggrecans, bind with water to endure compressive loads and maintain tension on the annulus fibrosus [16,33]. When disc degeneration begins, cytokine levels increase and elevate aggrecans and matrix metalloproteinases. This process results in cleavage of NP substances and leads to loss of disc height [1]. Pfirrmann et al. [32] studied this signal change on T2-weighted MRI, where a loss of brightness reflects proteoglycan degradation and water content [10,12,33].

We found a statistically significant association between disc degeneration assessed with the Pfirrmann disc degeneration grade and HLD revision (p<0.01). Cinotti et al. [5] reported that, in severe degenerative disc patients, the annulus fibrosus showed a markedly low recovery rate after primary LD, and the attenuated portion of the annulus fibrosus would aggravate recurrent HLD.

However, Dora et al. [8] reviewed the relationship between disc degeneration assessed with five grading systems and HLD recurrence. They reported that patients with relatively low-grade disc degeneration had a greater risk of recurrence. However, they included 60 non-randomized patients into each of the study and control groups. The correlation between age and Pfirrmann grade was found to be significant by Okada et al. [25]. Dora et al. [8] did not consider older patients with a high Pfirrmann grade who underwent spinal fusion instead of LD due to spinal instability as a secondary operation. Therefore, a large randomized study with long-term follow-up should be considered [8].

High BMI is a well-known negative factor in spinal health and other general conditions [6,7,9,13,14,21,26-31,35,39]. Our study classified patients as normal (BMI under 25), overweight (BMI 25 to 30), or obese (BMI over 30). We compared obese and nonobese patients and found that obese patients were significantly more likely to have HLD revision (p=0.0045). The idea that compressive loading on a intervertebral disc with a single or multiple stimuli after primary LD could result in HLD recurrence was suggested by Kelsey et al. [18] and Meredith et al. [21]. In addition, biomechanical reviews have explained that repetitive increased tension on the annulus fibrosus greatly increases the risk of tearing in the posterolateral portion, leading to HLD. In obese patients, the axial load on the intervertebral disc is higher than in non-obese patients during weight bearing by the spinal column, consequently increasing HLD. Furthermore, a forward flexed position in obese patients could increase the HLD recurrence rate [2,11,15,34,36,40]. Using BMI, which only considers height and weight, we might have misclassified patients who had high BMI due to high muscular mass as obese.

One limitation of our study is the small number of patients from a single medical center. Moreover, the retrospective medical information could introduce selection bias. Further study is required to assess the previously discussed factors in multiple hospitals over the long term. In this study, LD was the only reoperation considered. However, other procedure to assess spinal instability due to previous operations or aging should also be considered.

In our study, Pfirrmann disc degeneration grade and BMI were statistically significantly associated with HLD revision. After LD, surgeons should counsel patients with high BMI and degeneration on risk of revision.

The total numbers of patients and patients who underwent secondary LD were small, so further evaluation should be planned to confirm these results.