Dose of rhBMP-2
rhBMP-2 has been proven to be effective in relieving donor site pain and reduce morbidity. However, rhBMP-2 also shows a series of inflammatory dose-related complications, including radiculitis, osteolysis, and ectopic bone formation. Owens K et al. demonstrated a modest complication rate in MIS-TLIF procedure, with rhBMP-2-associated complications occurring infrequently. Even though the optimal rhBMP-2 dose in TLIF procedures remains a matter of debate, Villavicencio AT et al. reported that there was no correlation between the dose of rhBMP-2 and the incidence of rhBMP-2-induced radiculitis. Alan T et al. confirmed the effectiveness and safety of a combination of MIS-TLIF and rhBMP-2 in lumbar fusion whereas no significant difference was found in clinical outcome between MIS-TLIF and open procedures. A few years later, they adopted a wide dose range (2–12 mg) of rhBMP-2. Their retrospective study did not determine the relationship between the rhBMP-2 doses and the radicular symptoms.  To date, the 4-6 mg/level dose of rhBMP-2 is widely adopted in clinical practice.  Here, our adoption of low rhBMP-2 dosage (2 mg/ level) was effective and safe in the MIS-TLIF procedures and yielded satisfactory radiological and clinical outcomes.
Many studies assessing long-term follow-up data have shown that poor bone fusion results in degeneration of the vertebral body, functional disability, lumbar spine instability, and pain. [16, 17] We observed a higher fusion rate in the rhBMP-2 group than in the non-rhBMP-2 group, which was consistent with prior studies reporting a higher fusion rate compared with autologous bone grafts.  Vaidya et al. demonstrated that radiologic signs of fusion in patients receiving allografts and rhBMP-2 appeared earlier than those in patients not receiving rhBMP-2, with the signs appearing at 6 and 19 months, respectively. However, the exact time of fusion is difficult to determine because it is unrealistic for return visit patients to undergo surgical reexploration in the short time interval after surgery. We confirmed that application of low doses of rhBMP-2 leads to a significantly different fusion rate between both groups at 3 months postoperatively. Radiographic evidence of reexploration for recurrence of pain suggests that the cage was securely fused by 8-12 weeks, which indicates that the actual process began sooner than that detected on radiography. 
Previous studies demonstrated that disc degeneration causes an increase in ADH and PDH. [20, 21] Hsieh et al. found that SL decreased at the operated segment.  The compensatory increase in PDH relative to the increase in ADH led to the maintenance of the SL. Biomechanical research has suggested that procedures improve the disc height and FH deficits caused by disc degeneration.  We found a significant increase in ADH, MDH, and PDH at the operated level in both groups. The disc height and FH in the rhBMP-2 group decreased more significantly than that in the non-rhBMP-2 group.
Further, recent studies involving MIS-TLIF evaluated SL and presented variable results. Some reported an insufficient ability of MIS-TLIF to restore SL at the surgical level, whereas others showed substantial increases in SL. We found a significant increase in LL in patients with DLD, which was consistent with a previous study that confirmed that the interbody fusion is highly effective in improving LL. Some investigators demonstrated an association between postoperative LL and better clinical outcomes.
Previous studies evaluated short- and long-term complications associated with off-label rhBMP-2 use with TLIF. Some studies suggest that these complications may be dose-positive, and the higher the dose, the greater the probability of complications.  Complications and results were analyzed by BMP dose and primary versus revision surgery. Based on these results, surgical technique and rhBMP-2 dose recommendations were proposed.
Disc height restoration
No consensus regarding the results of the restoration of disc height was reached. Liu et al. suggested that the improvement in disc space height plays a vital role in facet joint subluxation.  Kaito et al. proposed that the distraction of disc space was a potent risk factor for the progression of adjacent segment disease (ASD), which may affect clinical outcomes of patients.  Michael C et al. found that disc height restoration was positively associated with segmental lordosis.  Our results are the first to investigate the effect of rhBMP-2 in the restoration of disc height. However, even though rhBMP-2 slightly slowed down restoration of disc height and FH to some extent, we did not find any significant differences in the two groups.
Biomechanical studies have proven that diminished LL may increase the risk of ASD.  Kepler et al. reported that the decrease in LL was related to worse clinical outcomes, which was consistent with our results.  Hence, our results showed that rhBMP-2 increased postoperative SL and LL by 0.9° and 3.1°, thus improving clinical outcomes by restoring sagittal alignment. Some researchers found a relationship between restoration of LL and improvement in clinical outcomes.  However, the relationship between radiographic improvement and clinical outcomes was not found in the current study. Postoperative maintenance of the SL may lead to a compensatory gain of adjacent segment lordosis, which is confirmed by the correlation between the postoperative LL and the SL of the adjacent levels.
Subsidence and MC
Several factors, including age, sex, and bone mineral density were reported to affect cage subsidence in MIS-TLIF. Tokuhashi et al. demonstrated that the average cage subsidence was 2.7 ± 3.4 mm on the caudal surface and 4.0 ± 2.3 mm on the cranial surface.  Their results were higher than those reported in the current study. The differences in cage subsidence may be attributable to the different measurement methods. CT could precisely measure the amount of subsidence during the bone remodeling process, so the average subsidence and prevalence of subsidence measured by plain radiographs were lower than those reported in other studies that used CT. David et al. reported that intact endplates increased strength and resistance to subsidence.  MC was associated with vertebral body structural changes such as disc herniation and vertebral endplate defect.  We found that cage subsidence in a different type of MC showed no significant difference except for normal endplate. Our result indicates that MC in the vertebral body may affect endplate preparation, which plays a vital role in preventing cage subsidence. However, our data did not show an improvement in cage subsidence in the rhBMP-2 group.
Even though all clinical outcomes improved significantly in the two groups, we found no significant difference between the groups. We demonstrated that the degrees of sagittal correction, fusion rates, and incidence of cage migration were not related with clinical outcomes in the first 12 months postoperatively. Compared with prior studies, our procedure integrated the advantages of MIS-TLIF and rhBMP-2 and confirmed that MIS-TLIF combined with rhBMP-2 is a better choice for lumbar degenerative disease.
Some limitations could not be avoided in this retrospective study. First, as the properties of L4-5 and L5-S1 are different in terms of their contribution to disc height, FH, SA, and the dimensions of the disc space, the bias resulting from the different numbers of operated levels could not be avoided. Second, our sample size was too small to obtain an accurate evaluation of the results. Third, long-term postoperative clinical assessment data were unavailable because some patients were lost to follow-up, making it impossible to reliably assess the clinical outcome of surgery. Fourth, our results did not identify the optimal rhBMP-2 dose in DLD surgeries. Prospective, randomized clinical trials are needed to determine the optimal rhBMP-2 dosage and complications.