Prospective Study of Long‐term Effect between Multifidus Muscle Bundle and Conventional Open Approach in One‐level Posterior Lumbar Interbody Fusion

Abstract

Objective

To compare postoperative imaging results, clinical outcomes and complications between the multifidus muscle bundle (MMB) approach and the conventional open (CO) approach in one‐level posterior lumbar interbody fusion (PLIF).

Methods

Based on the inclusion and exclusion criteria, 201 of 351 patients in our hospital were enrolled in this prospective study and underwent MMB‐PLIF or CO‐PLIF randomly: 111 patients in the MMB‐PLIF group and 90 patients in the CO‐PLIF group. A total of 100 patients failed to be followed up in the following 7–9 years. Therefore, in this study, 52 patients of the MMB group and 49 patients of the CO group were included. We evaluated the differences in terms of multifidus atrophy rate, intervertebral disc height and segmental lordosis restoration of the operation segment, lumbar lordosis restoration, fusion rate, visual analogue scale (VAS) for back and leg pain, Oswestry disability index (ODI), complication rates, and patient satisfaction rates between the two groups. Correlation between multifidus muscle degeneration and the incidence of complications was investigated, and we compared the multifidus muscle degeneration rate between patients with or without intractable back pain or adjacent segment degeneration.

Results

There were no significant differences in age, sex, body mass index (BMI), diagnosis, segments distribution, and mean follow‐up time between the MMB‐PLIF group and the CO‐PLIF group. In addition, no differences regarding sex, age, or BMI were found between the lost follow‐up group and the successful follow‐up group. In regard to imaging and clinical evaluation, at the final follow‐up, there were significant differences in multifidus atrophy rates (27.0% ± 6.8% vs 38.7% ± 10.9%), lumbar lordosis restoration (4.6° ± 2.5° vs 3.0° ± 1.9°), postoperative VAS for back pain (1.1 ± 0.9 vs 1.8 ± 1.2), ODI (7.7 ± 5.0 vs 12.4 ± 6.7), and patient satisfaction rates (86.5% vs 61.2%) between MMB‐PLIF and CO‐PLIF groups. However, there were no significant differences in segmental lordosis, intervertebral height restoration, postoperative VAS for leg pain or fusion rate between the two groups. In regards to complications, there were significant differences in the incidence of adjacent segment degeneration (3.8% vs 14.3%), intractable back pain (3.8% vs 22.4%), and residual neurological symptoms (5.8% vs 20.4%) between the two groups (P < 0.05) at the final follow‐up. In addition, patients with adjacent segment degeneration and intractable back pain were observed with more significant multifidus muscle atrophy than those without these two complications (31.9% ± 1.1% vs 39.6% ± 2.1% and 30.9% ± 1.1% vs 42.8% ± 2.1%).

Conclusion

Compared with CO‐PLIF, MMB‐PLIF had advantages in relation to protection of the multifidus muscle, better maintenance of lumbar lordosis, reduced lower back pain and ODI score, fewer complications, and a higher patient satisfaction rate. Protection of the multifidus muscle in lumbar surgery is an important aspect of minimally invasive surgery.

Introduction

Conventional open posterior lumbar interbody fusion (CO‐PLIF) has become one of the main surgical procedures in the treatment of spine disorders that require decompression and reconstruction of spinal stability. However, it is usually associated with extensive muscle stripping and retraction, which results in ischemic necrosis and denervation of paraspinal musculature. The spinous processes and supraspinal and interspinal ligaments are resected for decompression in the surgical process, which also causes excessive iatrogenic injury of the lumbar posterior column1, 2. Both the retraction of paraspinal muscles and resection of the lumbar posterior column may lead to decreased trunk muscle strength, which can later lead to spinal instability and severe back pain termed “failed back syndrome”1, 2.

Minimally invasive (MI) techniques have been developed to reduce iatrogenic injury of paraspinal muscles. The concept of “minimally invasive” means not only shorter incisions but also less soft tissue injury and more protection of paraspinal muscles3. Since 1986, when Leu and Hauser developed a percutaneous biportal endoscopic approach for interbody fusion, the tubular retractor system has been developed rapidly4. Short‐term and long‐term studies have confirmed that the tubular retractor‐assisted MI‐PLIF has advantages such as reduced intramuscular pressure and paraspinal muscle edema, less blood loss, lower serum creatinine kinase and inflammatory cytokine levels, faster recovery after surgery, less postoperative back pain, and even less economic burden compared with traditional CO‐PLIF2, 5, 6, 7, 8, 9, 10, 11, 12.

However, in clinical practice, the tubular retractor system has restrictions, such as a small view field and fixed surgical corridors. Alternative corridors are often needed when there are changes in surgical vision or internal fixation. Fixed surgical corridors delay the surgical process and cause increased surgical trauma13, 14. Moreover, the tubular retractor system is expensive and requires extensive training. These factors all restrict its application10, 13, 14, 15.

In recent years, improvements have been made on the tubular retractor system by our team13, 14. First, the multifidus muscle bundle method is chosen as the surgical approach and a micro‐laminectomy retractor is applied instead of an ordinary laminectomy retractor. By doing so, detachment of the paraspinal muscles can be avoided, and a more flexible surgical approach can be utilized. Second, adequate decompression is achieved by resecting part of the lamina, the whole inferior articular processes, and part of the superior articular processes, so the midline supporting musculo‐ligamentous complex is left as intact as possible. Preliminary short‐term results confirmed that compared with CO‐PLIF, the multifidus muscle bundle (MMB) PLIF has smaller incisions, less blood loss, less recovery time, lower incidence of both multifidus atrophy, and low back pain13, 14.

To evaluate the long‐term effect of the MMB‐PLIF, a prospective study for 7–9 years’ follow‐up of 101 patients with lumbar disease was carried out. Our aim was to investigate the advantages of MMB‐PLIF for postoperative imaging evaluation, clinical outcomes, and complications compared with CO‐PLIF.

Materials and Methods

Inclusion and Exclusion Criteria

From March 2006 to July 2009, 351 patients were enrolled in this study. The inclusion criteria were: (i) patients with segmental instability at the level of spinal stenosis, central or foraminal‐lateral herniated disc with herniated discs ossificated or posterior bony edge separation of lumber vertebral body, and degenerative spondylolisthesis (with all confirmed by plain radiographs, CT scans or MRI); (ii) a single lumbar intervertebral segment involved before surgery, without an obviously degenerated adjacent segment in MRI or CT before surgery; (iii) bilateral decompression according to the preoperative examination; and (iv) with severe lumbocrural pain, affecting daily working and life, and strict conservative treatment for more than 6 months is invalid.

Study exclusion criteria included: (i) scoliosis greater than 10 degrees; (ii) treatment for traumatic fracture; (iii) preoperative symptomatic lumbar segment disease more than two levels; (iv) prior lumbar infection; and (v) psychological factors preventing follow‐up. A total of 150 patients were excluded from this study.

Grouping and Follow‐up Method

A total of 201 patients with lumbar spine disorders underwent MMB‐PLIF or CO‐PLIF in our hospital and the operations were done by one medical team. Detailed information concerning the advantages and disadvantages associated with each surgical method was provided to all participants, and patients provided voluntary informed consent to participate in the study. Patients were assigned their surgical method according to odd or even number of their medical records. A total of 111 were assigned to the MMB‐PLIF group and 90 to the CO‐PLIF group. After surgery, all patients were strictly followed up; at the final follow‐up visit, neutral lateral lumbar plain radiographs, dynamic radiographs, lumbar MRI, and CT were assessed. In addition, the clinical effects, long‐term complications, and satisfaction surveys were also assessed. During the 7–9 years’ follow‐up, 100 patients were lost. Finally, 52 patients of the MMB‐PLIF group and 49 patients of the CO‐PLIF group were included (Fig. 1).

Figure 1.

Flow chart for the enrollment of this study.

Surgical Procedure of Multifidus Muscle Bundle–Posterior Lumbar Interbody Fusion

The surgical procedure of MMB‐PLIF is briefly described in what follows13, 14.

Position and Incision

The patients were positioned prone and two paramedian skin incisions approximately 2.5 cm long were made 3.0 cm lateral to the midline (Fig. 2A).

Figure 2.

(A) Arrows indicating the ipsilateral upper and lower pedicle surface location where multifidus muscle bundle posterior lumbar interbody fusion (PLIF) skin incisions are located. (B) An ordinary laminectomy retractor (left) and micro‐laminectomy retractor (right) used for multifidus muscle bundle PLIF. (C) Unilateral multifidus muscle bundle PLIF incision via the multifidus muscle bundle approach. (D) Resecting part of the lamina, the whole inferior articular processes and part of the superior articular processes during decompression.

Multifidus Muscle Bundle Approach

Using a micro‐laminectomy retractor (Fig. 2B), the lamina and facet joint complex were exposed through the multifidus muscle bundle (Fig. 2C).

Bilateral Decompression

Pedicle screw instrumentation was performed first. Adequate decompression was achieved by resecting part of the lamina, the whole inferior articular processes, and part of the superior articular processes, while the lateral recess or nerve root canal could also be decompressed if necessary. The same procedure was conducted on the opposite side for bilateral decompression (Fig. 2D).

Impacting Bone Graft and Cage Placement

A complete discectomy was performed, after which the disc space was sequentially distracted. The endplates were then prepared for fusion. The anterior disc space was packed with autologous bone graft, then interbody cages packed with autograft were placed.

Pedicle Rod Instrumentation Placement and Compression

Once the interbody fusion had been completed, pedicle rod instrumentation was placed. Compression was applied to the instrumentation before final tightening, providing compression of the bone graft to guarantee a stable anchor and to recreate lordosis.

The pedicle screw locking system (BLACKSTONE, USA) and the polyetherether ketone (PEEK) interbody cage (Synthes, Switzerland) were applied.

Surgical Procedure of Conventional Open Posterior Lumbar Interbody Fusion

In the CO‐PLIF group, a skin incision of approximately 4–5 cm was made at the midline, and the paravertebral muscles were dissected away from the bony structures during the standard operating procedure. After complete bony exposure, pedicle screw instrumentation was performed first, then laminectomies and medial facetectomies were performed. The remaining steps were the same as those used for the MMB approach.

Radiographic Evaluations

Sagittal Parameters

For the sagittal parameters, lumbar segment intervertebral height, segmental lordosis and lumbar lordosis were measured. (i) For intervertebral height, as introduced by Ekman et al. (2009), the vertical distances from the anterior and posterior of the inferior endplate to the superior endplate are separately measured and then the average mean is weighted (Fig. 3A)16. (ii) For segmental lordosis, the Cobb angle between the superior endplate of the superior vertebra and the inferior endplate of the inferior vertebra was measured (L₅S₁ segmental lordosis: Cobb angle between L₅ superior endplate and S₁ superior endplate) (Fig. 3B). (iii) For lumbar lordosis, the Cobb angle between L₁ vertebral superior endplate and S₁ vertebral superior endplate was measured (Fig. 3C). The restore value of lumbar sagittal parameters of each patient = value after surgery − value before surgery.

Figure 3.

Measurement of (A) lumbar sagittal parameters of intervertebral disc height, (B) segmental lordosis and (C) lumbar lordosis. (D) Measurement of cross‐sectional area of multifidus in MRI.

Multifidus Atrophy

As described previously, measurements were obtained with a picture archiving and communication system (PACS) workstation (Jin Ye Xiang Software, Beijing, China)5. To determine the lean multifidus muscle cross sectional area (CSA), the region of interest (ROI) was drawn around the multifidus muscles bilaterally, while nearby fat, bony structures and other soft tissues were avoided (Fig. 3D). The sum of CSA of bilateral lean multifidus was calculated; then the average mean was weighted. The atrophy rate of multifidus of the patient = (average value after surgery − average value before surgery)/average value before surgery × 100%.

Fusion Rate

For the fusion rate, 3D thin‐layer reconstruction CT and lateral lumbar spine X‐ray were used along with the modified Brantigan criteria (0–4 points), in which ≥3 points is defined as a successful fusion (Table S1)17, 18. Fusion rate = the number of successful fusion cases/the total number of cases at the final follow‐up × 100%.

One radiologist and one superior spine surgeon experienced in reading musculoskeletal images analyzed all selected images; both were blinded to the study design. Measurements were repeated after 3 weeks with the same protocol. Inter‐observer and intra‐observer repeatability were calculated using the intra‐class correlation coefficient (ICC) and formula (3,1)19. The reliability coefficients were determined as follows: <0.2 indicated slight agreement, 0.21–0.4 fair agreement, 0.41–0.6 moderate agreement, 0.61–0.8 substantial agreement, and 0.81–1.0 excellent agreement20.

Clinical Effect Assessment

Clinical Symptoms

To define clinical symptoms, a visual analogue scale (VAS) of back pain and leg pain as well as the Oswestry disability index (ODI) were applied for assessment.

Patient Satisfaction Rate

The patient satisfaction rate was measured with a questionnaire adapted as follows: “1–5 points separately represent the degree from dissatisfaction to great satisfaction, what is your degree of satisfaction towards this lumbar surgery?”2. Patient satisfaction rate = (the number of patients whose answers is ≥4 points)/(the total number of patients of the follow‐up) × 100%. In cases in which patients reported satisfaction levels of 1–3 points, the reason for dissatisfaction was asked.

Assessment of Complications

Adjacent Segment Degeneration

Adjacent segment degeneration (ASD) was defined as radiographic and symptomatic degeneration. Radiographic degeneration contains one or more of the following changes: disc height loss >20% in X‐ray; vertebral displacement of adjacent segment ≥3 mm in dynamic X‐ray or the change of angle between the upper and lower end plates ≥10% or vertebral slippage ≥10% of width; Pfirrmann scale ≥stage 3 in disc degeneration; hyperplasia and hypertrophy of articular process; osteophyte formation ≥3 mm; and symptomatic lumbar disc herniation or vertebral stenosis which were confirmed by CT or MRI21. The definition of symptomatic degeneration is ASD with clinical symptoms and is confirmed by images.

Intractable Back Pain

Intractable back pain was defined as persistent back pain around the incision after surgery from which patients could not recover through conservative treatment. This is, in other words, “fusion disease”7.

Residual Neurological Symptoms of the Lower Limb

Residual neurological symptoms of the lower limb were defined as cases in which there remained some neurological symptoms of the lower limb after surgery, or the symptoms repeatedly occured after temporary relief, or there were new symptoms occurring in locations where there was no pain or numbness preoperatively22.

Statistical Analysis

The continuous variables such as age, body mass index (BMI), mean follow‐up time, intervertebral height, segmental lordosis or lumbar lordosis restoration, VAS, and ODI score were measured as mean ± SD, and categorical variables such as sex, diagnosis, segments distribution, satisfaction rate, multifidus atrophy rate, and fusion rate were expressed by frequency or percentage. Student's t‐test was used to examine differences between groups of continuous variables. The χ²‐test was used to examine differences between groups of enumeration data. All analyses were performed using SPSS 19.0 (IBM, USA) and the significant level of P was defined as 0.05.

Results

General Information

The demographic data of the MMB and CO‐PLIF groups are compared as shown in Table 1. There were no significant differences in age, sex, BMI, diagnosis, segments distribution and mean follow‐up time between the two groups (P > 0.05). In the MMB‐PLIF group, the mean follow‐up time was 93.8 months, while for the CO‐PLIF group, it was 94.7 months. To increase the reliability of the follow‐up data, we analyzed the follow‐up bias in the patients that were lost to follow‐up. No difference in sex, age, and BMI were found in these patients between the two groups (P > 0.05) (Table S2).

Table 1.

General information of patients

	MMB‐PLIF group	CO‐PLIF group	P‐value
Number of patients	52	49	—
Sex (male/female)	25/27	26/23	P = 0.479
Age	53.0 ± 6.5	52.1 ± 7.1	P = 0.559
BMI (kg/m2)	21.9 ± 2.1	22.7 ± 2.0	P = 0.116
Mean follow‐up time (months)	93.8 ± 6.9	94.7 ± 8.7	P = 0.593
Diagnosis			χ 2 = 0.322, P = 0.956
Lumbar disc herniation with herniated discs ossificated	21	19	—
Spinal stenosis with segmental instability	9	8	—
Posterior bony edge separation of lumber vertebral body	5	6
Lumbar spondylolisthesis	17	16	—
I degree	7	8	—
II degree	10	8	—
Segment			χ 2 = 1.262, P = 0.532
L3‐4	6	6	—
L4‐5	26	25	—
L5‐S1	20	18	—

BMI, body mass index; CO‐PLIF, conventional open approach for posterior lumbar interbody fusion; MMB‐PLIF, multifidus muscle bundle approach for posterior lumbar interbody fusion.

Imaging Evaluation Outcomes

The inter‐observer and intra‐observer ICC for intervertebral height, segmental lordosis, lumbar lordosis, lean multifidus muscle CSA, and interbody fusion were between 0.85 and 0.95 and demonstrating excellent agreement.

Sagittal Parameters

Detailed data was shown in Fig. 4 and Table S3. Three months after the surgery, there were no significant differences in intervertebral height, segmental lordosis or lumbar lordosis restoration between the MMB‐PLIF and CO‐PLIF groups. However, at the final follow‐up, the difference became significant in lumbar lordosis restoration (4.6° ± 2.5° vs 3.0° ± 1.9°, improving by 53.3%) rather than in intervertebral height or segmental lordosis restoration (Fig. 4A–C).

Figure 4.

Comparison of intervertebral disc height restoration (A), segmental lordosis restoration (B) and lumbar lordosis restoration (C) between multifidus muscle bundle approach for posterior lumbar interbody fusion (MMB‐PLIF) and conventional open approach for posterior lumbar interbody fusion (CO‐PLIF) at 3 months and at final follow‐up after surgery. Data represents the mean ± SD (n = 52 in MMB‐PLIF group and 49 in CO‐PLIF group).

Multifidus Atrophy

Compared with bony structures, soft tissue protection is the most important aspect of MMB‐PLIF. Compared with the extensive fatty change of multifidus shown in the CO‐PLIF patient, the MMB‐PLIF patient showed obvious multifidus protection after surgery, and the multifidus atrophy rates were much lower in the MMB‐PLIF group at the final follow‐up (27.0% ± 6.8% vs 38.7% ± 10.9%, reducing by 30.2%) (Fig. 5A and Table S3). Representative cases are shown in Fig. 5B.

Figure 5.

(A) Comparison of multifidus atrophy rate between multifidus muscle bundle approach for posterior lumbar interbody fusion (MMB‐PLIF) and conventional open approach for posterior lumbar interbody fusion (CO‐PLIF). (B) Typical MRI images of multifidus in MMB‐PLIF group (before and 88 months after surgery) and CO‐PLIF group (before and 90 months after surgery). Data represents the mean ± SD (n = 52 in MMB‐PLIF group and 49 in CO‐PLIF group).

Fusion Rate

At the final follow‐up, there was no significant difference in the fusion rate between the MMB‐PLIF and CO‐PLIF groups (Fig. 6A and Table S3). One representative successful fusion case in the MMB‐PLIF group is shown in Fig. 6B.

Figure 6.

(A) Comparison of lumbar fusion rate between MMB‐PLIF and CO‐PLIF groups at final follow‐up after surgery. (B) Female, 57 years old, multifidus muscle bundle L_4/5 posterior lumbar interbody fusion was performed. 3D reconstruction CT 1 year after surgery shows a sclerotic line between the graft and vertebral bone with no lucency existed. Data represents the mean ± SD (n = 52 in MMB‐PLIF group and 49 in CO‐PLIF group).

Clinical Effect Evaluation Outcomes

Clinical Symptoms

Before the surgery, there was no significant difference in the VAS of back pain, VAS of leg pain or ODI score between the MMB‐PLIF and CO‐PLIF groups (Fig. 7A–C and Table S4). At the final follow‐up, no significant difference in leg pain between the two groups was found, while there were significant differences in low back pain (1.1 ± 0.9 vs 1.8 ± 1.2) and ODI score (7.7 ± 5.0 vs 12.4 ± 6.7) between the two groups.

Figure 7.

Comparison of clinical symptoms of back pain (A), leg pain (B), and ODI score (C) between multifidus muscle bundle approach for posterior lumbar interbody fusion (MMB‐PLIF) and conventional open approach for posterior lumbar interbody fusion (CO‐PLIF) groups before surgery and at final follow‐up after surgery. Data represents the mean ± SD (n = 52 in MMB‐PLIF group and 49 in CO‐PLIF group).

Patient Satisfaction Rate

At the final follow‐up, there was a significant difference in patient satisfaction rates between the MMB‐PLIF and CO‐PLIF groups (86.5% vs 61.2%, improving by 41.3%, χ ² = 8.173, P = 0.004), as shown in Table 2.

Table 2.

Patient satisfaction survey

Score	MMB‐PLIF group (52)	CO‐PLIF group (49)
1	0	5
2	3	11
3	4	3
4	13	24
5	32	6
Satisfaction rate	86.5% (45/52)	61.2% (30/49)
Statistics	χ 2 = 8.173, P = 0.004

CO‐PLIF, conventional open approach for posterior lumbar interbody fusion; MMB‐PLIF, multifidus muscle bundle approach for posterior lumbar interbody fusion.

Assessment of Complications

Figure 8A shows that at the final follow‐up, there were significant differences in the incidence of ASD (3.8% vs 14.3%, reducing by 73.4%, P < 0.05), intractable back pain (3.8% vs 22.4%, reducing by 83.0%, P < 0.05), and residual neurological symptoms (5.8% vs 20.4%, reducing by 71.6%, P < 0.05) between the two groups. Interestingly, by comparing the multifidus atrophy rate between patients with or without intractable back pain or ASD, we found that patients who suffered from ASD (31.9% ± 1.1% vs 39.6% ± 2.1%) and intractable back pain (30.9% ± 1.1% vs 42.8% ± 2.1%) showed more significant multifidus muscle degeneration (Fig. 8B,C).

Figure 8.

(A) Evaluation of long‐term complications between multifidus muscle bundle approach for posterior lumbar interbody fusion (MMB‐PLIF) and conventional open approach for posterior lumbar interbody fusion (CO‐PLIF) groups. (B) Comparison of multifidus atrophy rate between patients with (n = 9) or without (n = 92) adjacent segment degeneration. (C) Comparison of multifidus atrophy rate between patients with (n = 13) or without (n = 88) intractable back pain. Data represents the mean ± SD.

Discussion

Minimal invasion is a trend in modern spine surgery. In our MMB‐PLIF, the multifidus bundles method was chosen as a surgical approach. Compared with CO‐PLIF, the short‐term advantages of MMB‐PLIF have been widely reported5, 13, 14. To learn more about the long‐term effects of these two procedures, we launched a prospective study and found that MMB‐PLIF had obvious advantages over CO‐PLIF in relation to imaging, clinical effects, and postoperative complications.

Multifidus belongs to the major muscle groups, which maintains the dynamic balance of the lumbosacral region and spine–pelvis complex23. In CO‐PLIF, obvious atrophy and fatty degeneration are common symptoms for multifidus after the surgery5, 24. Multifidus atrophy is related to factors such as muscle stripping and retraction with a self‐retaining retractor, denervation changes, and musculo‐ligamentous complex damage25; however, the main factors are muscle retraction and muscle stripping26, 27, 28, 29. Fan et al. found that prolonged wide muscle retraction leads to intramuscular nerve injury and muscle ischemic necrosis, finally resulting in atrophy25. The effect of muscle stripping from multifidus has also been a hot topic during recent years. Kim et al. (2008) found that by retaining the tendon origin of the multifidus at the spinous process and reducing muscle stripping, postoperative multifidus atrophy could be prevented27. By comparing different posterior approaches towards multifidus, Liu et al. (2010) found that with the same muscle retraction time and strength, the degree of multifidus atrophy was mainly related to muscle stripping range28. Hu et al. (2013) found that the denervation and disuse of multifidus caused by muscle stripping and retraction may be important factors in multifidus muscle atrophy in rabbit models29. Our previous studies have found that using a micro‐laminectomy retractor and choosing multifidus bundles as a surgical approach, stripping and retraction of multifidus could be avoided and the multifidus atrophy rate was significantly reduced in short‐term and medium‐term follow‐up5, 13, 14. Through the long‐term prospective study we discuss here, the same conclusion is verified.

Lumbar lordosis is a key postural component in maintaining spinal sagittal balance and reducing spinal concussion30. In CO‐PLIF, lumbar lordosis is reduced or even disappears, resulting in an iatrogenic “flatback” deformity and intractable back pain, seriously affecting the outcome of surgery31. Gilad et al. (2008) showed that failure to correct a sagittal plane deformity intraoperatively predisposes the patient to instrumentation failures32. The posterior tension band is closely related to the function of the spine. In CO‐PLIF, the tension band is severely damaged, which results in not only imbalance between the posterior tension band of spine and the abdominal muscles, leading to the reduction in lumbar lordosis, but also decreased stability of the posterior column of the spine, resulting in the compensatory decrease of lumbar lordosis33, 34. In MMB‐PLIF, the posterior tension band was completely preserved; therefore, it has advantages over CO‐PLIF in lumbar lordosis maintenance. In this study, we compared the relative change in lumbar lordosis before and after surgery to minimize individual differences and found that 3 months postoperatively, no difference was found in lumbar lordosis between the two groups, while in the long term, MMB‐PLIF allows greater lumbar lordosis maintenance.

As for the complications of surgery, cases of intractable back pain reached 11 (22.4%) in the CO‐PLIF group, while in the MMB‐PLIF group, the number was only 2 (3.8%). These patients mainly complained about pain and stiffness around the incision, especially after activities, and that the pain could not be completely resolved by resting or use of pain medications, which severely restricted their daily life. Some studies have found that intractable back pain after surgery is associated with paraspinal muscle atrophy, lumbar lordosis decrease, and ASD31, 35, 36. We found that multifidus atrophy was much more serious in patients with intractable back pain. As such, we propose that multifidus atrophy is a risk factor for intractable back pain after PLIF. After the multifidus muscle atrophies accompanies epidural fibrosis and nerve root adhesion, intractable back pain occurs17. After multifidus atrophy, sagittal plane instability occurs and shear force around the facet joint greatly increases, stimulating pain‐sensing nerve endings and causing intractable back pain37. MMB‐PLIF takes normal bundles of multifidus as a surgical approach. There is limited facet joint and lamina resection, multifidus and posterior spinal structure are preserved, and the stability of the spine is maintained; therefore, the incidence of intractable back pain is greatly reduced.

We also note that more severe multifidus atrophy was found in patients with ASD after surgery. ASD is an inevitable complication after lumbar fusion surgery. Research showed that the ASD rate reached 8%–100% in postoperative imaging evaluation and 5.2%–18.5% in postoperative clinical symptoms38.

The incidence of ASD depends on multiple factors. Generally speaking, the risk factors can be divided into two classes. One is patient‐specific risk factors, like age, female gender, postmenopausal status, obesity, postoperative pelvic incidence, and postoperative pelvic tilt39, 40, 41, 42. The other is surgery‐related factors, including sagittal alignment, length of spinal fusion, lateral bending, and axial rotation43, 44, 45, 46, 47, 48, 49.

As reported in our previous work, the MMB procedure can effectively protect paraspinal muscles and posterior column structure, preventing the occurrence of ASD after lumbar fusion38. Unlike CO‐PLIF, MMB‐PLIF uses the multifidus muscle bundle method as a surgical approach and the detachment of paraspinal muscles is avoided. Adequate decompression is achieved by resection of part of the lamina, the whole inferior articular processes and part of the superior articular processes, while the midline musculo‐ligamentous complex is left as intact as possible to maintain spinal stability. In this study, as shown in Table 1, there were no differences in sex, age or BMI between patients in the MMB‐PLIF and CO‐PLIF groups. A single lumbar intervertebral segment was involved in both groups of patients; as a result, there was no difference in the length of spinal fusion between the two groups. Based on this, we speculate that patients have better spinal stability after MMB‐PLIF, resulting from less paraspinal muscle atrophy and maintenance of the midline supporting complex, which may be one important cause of low ASD incidence rates.

Residual neurological symptoms of the lower limb were defined as occurring when: (i) neurological symptoms of the lower limb remain after surgery; (ii) the symptoms repeatedly occur after temporary relief; and (iii) there are new symptoms occurring in locations where there was no pain or numbness preoperatively22. In CO‐PLIF, the retraction range of the nerve root is bigger, which can more easily cause a tension ischemic injury of the nerve root. In MMB‐PLIF, due to bilateral decompression and a smaller nerve root retraction range, the incidence rates of temporary nerve root injury and temporary recurrent neurological symptoms are much lower than in CO‐PLIF. Moreover, in CO‐PLIF, laminectomies and medial facetectomies were performed, which expose the nerve root and spine cord completely and cause scar adhesion, nerve root peripheral fibrosis, and arachnoid fibrosis. This can cause neurological symptoms to repeatedly occur after temporary relief. However, in MMB‐PLIF, the midline musculo‐ligamentous complex is preserved, which lowers the incidence of temporary recurrent neurological symptoms. New onset postoperative neurological symptoms are another type of residual neurological symptoms. In our study, new onset postoperative neurological symptoms often co‐existed in patients with ASD. As discussed above, reduced paraspinal muscle atrophy and preservation of the midline supporting complex are important causes of the lower ASD incidence rate in MMB‐PLIF surgery. Therefore, the new onset of neurological symptoms may also be lower in MMB‐PLIF patients.

Above all, we found that compared with CO‐PLIF, MMB‐PLIF has advantages such as reduced multifidus atrophy, better maintenance of lumbar lordosis, greater remission of postoperative back pain and ODI, fewer complications such as intractable back pain or ASD. and higher patient satisfaction rates in long‐term follow‐up. Protection of the multifidus muscle in lumbar surgery should be an important aspect of minimally invasive surgery.

There were a few limitations to this study. First, considering the sample size of this study, multi‐center prospective randomized trials are still needed. Second, more subgroups could be included for comparisons, such as between single‐level and multi‐level PLIF, young patients and old patients, and diseases types. Thus, more precise indications could be revealed for proper use of CO‐PLIF and MMB‐PLIF.

Supporting information

Table S1 The modified Brantigan criteria (0‐4 points) for lumbar interbody fusion.

Table S2 General information of patients lost to follow‐up.

Table S3 Lumbar sagittal parameters, multifidus atrophy rate and fusion rate of two groups.

Table S4 Clinical symptoms of two groups.