In our study, 10.05% of patients after LLIF underwent revision surgery, all within one year after the first operation. This is similar to the previous studies reported that the revision rate after LLIF is 3.4–26.0% and the most common complication is transient neurological symptoms [6, 7, 12, 13, 17,18,19]. The demographic characteristic data were similar between the revision group and the non-revision group, and there was a significant difference in the proportion of preoperative spinal canal stenosis and postoperative foraminal stenosis. In addition, the reasons for revision include vascular injury, unsatistactory implant placement, internal spinal instrumentation displacement, cage migration and infection. The most common cause of revision is indirect decompression failure, accounting for 47.6%. In the long-term follow-up, the revision did not affect the postoperative recovery and satisfaction of patients. We will now discuss our strategies for revision in different complications.
Indirect decompression failure
Overall reported incidence of indirect decompression failure (IDF) was 9%. The reasons for IDF included endplate collapse, osteoporosis, severe foraminal stenosis, inadequately restored disc and bony lateral recess stenosis [10, 20,21,22,23]. LLIF allows the surgeon to place a large cage in the intervertebral space to fully restore the height of the intervertebral space and intervertebral foramen, and at the same time tighten the ligament tissue and release the pressure in the spinal canal, so as to achieve a good effect of indirect decompression [10, 21, 24]. Since LLIF does not directly decompress the lamina, the symptoms may persist or even worsen after IDF. In the choice of revision mode, we need to choose the corresponding method according to the causes of IDF. In order to relieve neurological symptoms, most of the previous studies were supplemented by posterior direct decompression. In addition to the traditional posterior decompression, minimally invasive decompression can also be selected. When the diagnosis is clear, we choose intervertebral foramen endoscope to do intervertebral foramen plasty, and we can also use unlateral biportal endoscoic (UBE) to decompress the lamina. And posterior pedicle screw fixation can prevent further intervertebral space collapse due to osteoporosis and endplate injury after operation.
Postoperative infection
The probability of infection in LLIF is only 0.6–0.73% [25, 26]. The reasons for the lower incidence of infection may be related to less trauma, shorter operation time and passage operation. Infection usually occurs in patients with obesity or diabetes, and wound infection can be cured by conservative treatment. However, intervertebral space infection is rarely reported. Intervertebral space infection can be transmitted from the blood or gastrointestinal tract or directly. In order to prevent the infection from spreading to the posterior column of the spine, debridement through the original surgical incision is a better choice. In the early stage of infection, implants can be considered to be preserved. When the focus of infection is large and the infection is difficult to control, focal debridement, bone grafting and internal fixation are needed to maintain the stability of the spine and control the infection. Figure 2 shows a special case of spinal infection, and there is no obvious abnormality in hemogram and lumbar MRI before operation (Fig. 2a-b). Fever occurred after one-stage operation. Brucella agglutination test showed negative, lumbar nuclear magnetic resonance showed infection (Fig. 2c-d). After revision through the original surgical incision, pathology showed brucellosis infection.
Vascular injury
The incidence of major vascular injuries occurring for LLIF performed in a larger case series involving a total of 4,607 patients is 0.4% [26]. Although its incidence is not high, it is very dangerous. If the space between psoas major muscle and vascular sheath is very narrow, vascular injury is easy to occur during OLIF. Once suspected vascular injury occurs during operation, it is necessary to stop bleeding thoroughly, otherwise it is easy to cause delayed bleeding after operation. The key to revision lies in the early detection that once the wound is painful and the blood pressure drops rapidly, timely and effective measures should be taken to stop the bleeding.
Cage migration
In our study, the revision rate of patients with cage migration accounted for 1.44%. The causes of cage migration are mainly due to obesity, getting out of bed too early, small interbody fusion cage, abnormal vertebral body shape, improper bone graft material, osteoporosis, bony endplate injury and insufficient opening of contralateral annulus [27,28,29]. Cage migration usually occurs after SA-LLIF. Early stage of cage migration, because of small displacement and mild symptoms, the cage can be fixed by adding posterior internal fixation alone (Fig. 3a-d). If cage prolapse occurs, the prolapsed cage will oppress the lumbar plexus, and the protruded intervertebral space will collapse and cause lower limb symptoms. So, it is necessary to adjust the position of the cage from the original way and then apply internal fixation (Fig. 3e-h).
Unsatistactory implant placement
Unsatistactory implant placement cases accounted for 14.3% of all revision cases. There are many reasons for unsatistactory implant placement, such as high iliac crest, psoas major hypertrophy, the establishment of lumbar bridge and other factors, it is difficult to straighten the channel, so the cage and inferior screws are not completely vertical. For the cases with persistent neurological symptoms, good results can be obtained by changing the position of implant placement and relieving nerve compression by two-stage operation. However, in some cases, due to the use of cage with fin device, it is difficult to remove it during revision, so we adopted posterior decompression or symptomatic side transforaminal lumbar interbody fusion (TLIF), which relieved the compression under direct vision and avoided the removal of cage. (Fig. 4)
Internal spinal instrumentation displacement
The incidence of internal fixation failure and revision after LLIF is about 0.96%. The failure of spinal internal fixation is often caused by improper selection of operation or erroneous installation. If the posterior internal fixation is loose or broken, it is naturally necessary to replace the internal fixation after being removed from the original posterior incision. However, the direction of stress and the curvature of the lumbar spine should also be taken into account during revision to prevent recurrence. When lateral internal fixation is selected for one-stage operation, in addition to removing the original internal fixation, it is also necessary to consider increasing posterior pedicle screw fixation to maintain spinal stability. Figure 5 shows a case of lateral nailboard loosening, due to more underlying diseases and no obvious lower limb symptoms, we did not add internal fixation and advised him to increase bed rest time. During the postoperative follow-up, although there was a small amount of endplate collapse (Fig. 5f), the operative segment tended to be stable one year after operation, and there were no obvious lower limb symptoms (Fig. 5 g). Therefore, it is necessary to make a clear diagnosis before operation and fully estimate the stress that the implant can bear in the body; master the installation mode and application purpose of each component of the implant. For those patients with severe osteoporosis, posterior internal fixation should be chosen to provide stronger support, and strengthen the treatment of anti-osteoporosis.
Although there are studies have suggested that LLIF can be applied to some patients with mild to moderate spinal canal stenosis [11]. However, for severe spinal canal stenosis, some scholars believe that it is a taboo of LLIF [30, 31], and severe spinal canal stenosis is a risk factor for secondary decompression after LLIF [32]. In our study, there was a difference in the proportion of preoperative spinal canal stenosis between the revision group and the non-revision group, but this was not an independent risk factor for revision. Similar to the study of Hiyama et al. [33], we believe that preoperative severe spinal stenosis is not an absolute contraindication of LLIF, but LLIF may not be suitable for severe bony stenosis.
By reviewing the data of all patients undergoing revision after LLIF, we found that postoperative foraminal stenosis was an independent risk factor for revision. It is also one of the factors affecting the revision decision of the operator. According to the study of Wang et al. [34], LLIF can not decompress the nerve roots of patients with severe intervertebral foramen stenosis before operation. In addition to bony stenosis, we believe that endplate collapse is one factor leading to intervertebral foramen restenosis. Since LLIF completely increases the intervertebral height by inserting large cage, once the graft subsidences or moves, it will lead to the loss of intervertebral height and foraminal stenosis again, resulting in lower limb symptoms [28, 35, 36]. Displacement, loosening and fracture of the internal fixation will also lead to the weakening of the stabilization effect, causing foraminal stenosis and low back pain. In addition, excessive test model and improper choice of internal fixation are also factors leading to postoperative foraminal stenosis, which need further study in the future.
For the surgical efficacy of LLIF, health-related quality-of-life measurements decreased in varying degrees after operation. Timely revision operation can achieve satisfactory results, but the rehabilitation time of the revision operation is longer. Due to the small sample size and follow-up period, our research has multiple limitations. First of all, the factors that we can explore in multivariate analysis are limited. In addition, due to the lack of follow-up data for some patients, they were not included in this study, even if they may affect the results. In the statistics of complications, because of the short follow-up time, we did not summarize or discuss the long-term complications such as adjacent segmental degeneration. In the future, we should expand the sample size, prolong the follow-up period, and explore the long-term efficacy of LLIF.