Introduction

Minimally invasive oblique lateral interbody fusion (MIS-OLIF) has been reported to have potential advantages in minimally violating the paraspinal muscles and restoring segmental lordosis more efficiently than posterior and transforaminal lumbar interbody fusion (PLIF and TLIF, respectively) [1,2,3,4,5]. In OLIF or extreme lateral interbody fusion (XLIF), because a larger cage than in PLIF and TLIF can be used, optimal correction of decreased lumbar lordosis (LL) is easier. This plays a crucial role in improving clinical outcomes and preventing adjacent segment disease (ASD) [6, 7].

Addressing sagittal imbalance is critical even in short-segment fusion because optimal reduction of pelvic tilt (PT), enhancement of the C7 sagittal vertical axis (SVA), and harmonization of pelvic incidence (PI), and LL can reduce the incidence of postoperative lower back pain, lower extremity pain, and ASD [6,7,8]. Therefore, achieving adequate correction of lordosis at the index surgical level is important for improving clinical outcomes in patients undergoing MIS-OLIF.

Despite the superior corrective potential of OLIF and XLIF versus TLIF and PLIF, typical per-level corrections after OLIF or XLIF have been reported to be < 10° [9]. However, several studies have shown changes in regional LL >10° after XLIF [10,11,12,13]. This discrepancy implies that there is a specific condition in which sufficient correction of the segmental angle (SA) is expected; however, the clinical and radiological factors associated with sufficient correction remain underexplored. Therefore, this study aimed to identify the preoperative or postoperative radiological factors associated with adequate correction of the SA greater than 10° at each surgical level and to establish a predictive formula for SA correction following MIS-OLIF with percutaneous pedicle screw instrumentation.

Materials and methods

Patients

We retrospectively reviewed patients who underwent single- or two-level MIS-OLIF between August 2012 and March 2023. Given the retrospective nature of this study, the Institutional Review Board waived the requirement for informed consent. The inclusion criteria were as follows: (1) undergoing single- or two-level MIS-OLIF with percutaneous pedicle screw instrumentation and (2) preoperative C7 SVA >50 mm. The exclusion criteria were as follows: (1) unavailability of immediate postoperative spine radiographs, (2) concurrent compression fracture at the time of OLIF, and (3) congenital hip dysplasia, congenital stenosis, malignancy, inflammatory disease, or infection. We obtained demographic and clinical data, such as age, sex, body mass index (BMI), preoperative diagnosis, surgical level, fusion level, prior surgery at the index level, height and angle of the inserted intervertebral cage, and perioperative complications from electronic medical records and the picture archiving and communication system.

Surgery

Most patients underwent circumferential MIS-OLIF via the anterior retroperitoneal approach in the lateral decubitus position. A polyether-ether-ketone intervertebral cage filled with demineralized bone matrix was placed after discectomy to enhance the posterior disc space height. Cages with lordotic angles of 6°, 8°, or 12° were implanted in all patients, positioned between the middle and anterior third of the disc space. We aimed to maximize disc height without resorting to anterior column release (ACR) or posterior osteotomy, therefore neither ACR nor posterior osteotomy was not used in the OLIF procedure. Patients were then repositioned prone for posterior percutaneous pedicle screw insertion under intraoperative fluoroscopy. When the patient was in the prone position, we used a chest bar and two posts beneath both anterior superior iliac spines to achieve maximum lumbar sagging. The hip position was nearly extended to maximize LL in the lower segments. Screws were inserted in situ, without the use of any compression or distraction manoeuvers.

Radiological assessment

Two authors, blinded to the clinical data, reviewed preoperative and immediate postoperative standing lumbosacral and whole-spine standing radiographs, along with preoperative axial and sagittal T1- and T2-weighted magnetic resonance (MR) images. Preoperative radiographs were assessed for disc height, spondylolisthesis, angular and translational instability, spondylolysis, and disc vacuum at each surgical level. Instability was defined as a range of motion exceeding 10 degrees or vertebral body translation exceeding 3 mm in disc space when comparing flexion and extension lumbar lateral X-rays. The preoperative and immediate postoperative SAs were examined at each surgical level. Preoperative and immediate postoperative sagittal parameters, such as PI, LL between the L1 and S1 upper endplates, the sacral slope, PT, thoracic kyphosis (TK) between the T5 and T12 endplates, and C7 SVA, were evaluated.

The radiological characteristics of preoperative MR images at the index surgical level were examined. These included the preoperative Pfirrmann grade [14], osteoarthritis grading of the facet joints [15], presence of facet cyst and facet effusion [16], and the Goutallier grade of the paraspinal muscle at the L5–S1 disc level [17]. Central and foraminal stenoses were evaluated using the Schizas and Lee grades, respectively [18, 19].

Definition of D and C segments

Costanzo demonstrated that lateral lumbar interbody fusion typically resulted in average corrections of less than 10° per level, without the need for ACR or posterior osteotomy [9]. Therefore, the segments that exhibited adequate segmental correction >10° after MIS-OLIF without ACR were categorized as discontinuous segments (D segments), whereas those without such improvement were assigned as continuous segments (C segments) (Figs. 1 and 2). The term “discontinuous” refers to the acute interruption in a single, smooth lumbar lordotic curve, contrasting with the “continuous” type, which signifies a smooth curvature transition. This nomenclature is derived from mathematical distinctions between continuous and non-continuous function graphs.

Fig. 1
figure 1

A 66-year old female patient who underwent two-level oblique lumbar interbody fusion with both D and C segment. (a) Preoperative standing lumbar lateral radiograph showing C segment at L3–4 and D segment at L4–5 with acute disruption of smooth lumbar lordotic curve. (b) Preoperative magnetic resonance imaging showing bilateral facet effusion at L4–5 level. (c) Preoperative standing whole-spine lateral radiograph showing global sagittal malalignment. (d) Postoperative standing lumbar lateral radiograph and (e) standing whole-spine lateral radiograph showing restoration of smooth lumbar lordotic curve, resulting in adequate correction of preoperative global sagittal imbalance. D discontinuous, C continuous

Full size image
Fig. 2
figure 2

A 62-year old female patient who underwent single level oblique lumbar interbody fusion with D segment. (a) Preoperative standing lumbar lateral radiograph showing D segment at L4–5 with spondylolytic spondylolisthesis showing acute disruption of smooth lumbar lordotic curve. (b) Preoperative standing whole-spine lateral radiograph showing global sagittal malalignment. (c) Postoperative standing lumbar lateral radiograph and (d) standing whole-spine lateral radiograph showing restoration of smooth lumbar lordotic curve, resulting in adequate correction of preoperative global sagittal imbalance. D discontinuous

Full size image

Statistical analysis

Continuous variables showing differences between D and C segments were assessed using Student’s t test or the Kruskal–Wallis test, whereas categorical variables were evaluated using chi-squared, Fisher’s exact, or linear-by-linear association tests. Logistic regression identified preoperative factors predicting SA correction >10° after MIS-OLIF at each index surgical level. To optimize variable selection and reduce the risk of omitting potentially significant factors, variables demonstrating significant associations in the univariate logistic regression (p < 0.20) were included in the multivariate model. Backward elimination was used to calculate the odds ratios (OR) and 95% confidence intervals (CI) for stooping improvement. Receiver operating characteristic (ROC) curves were used to calculate the area under the curve (AUC), gauge sensitivity, and specificity of risk factors in predicting stooping improvement. AUC values of 0.5–0.7, 0.7–0.9, and >0.9 denoted low, moderate, and high discriminatory power, respectively. The relationship between the identified preoperative radiological risk factors, which were continuous variables, and the degree of SA correction was evaluated using Pearson’s correlation coefficient and simple linear regression analysis, resulting in a predictive formula for SA correction. All statistical analyses were performed using IBM SPSS Statistics (version 25.0; IBM Corp., Armonk, NY).

Results

Our study included 211 segments in 148 patients with a mean age (± standard deviation [SD]) of 71.4 ± 8.0 years (Fig. 3). Overall, 35 patients (23.6%) had D segments at the index surgical level. Thirty-eight segments (18.0%) were categorized as D segments, and 173 segments (82.0%) were categorized as C segments. Table 1 displays the demographic and radiological characteristics of patients with and without D segment. No significant intergroup differences were observed in age, sex, BMI, number of surgical levels, or the specific surgical levels. Among all patients, 96 (64.9%) showed an improvement in C7 SVA < 50 mm after surgery. Radiologically, patients with D segment exhibited significantly larger preoperative PI minus LL, PT, and C7 SVA and significantly smaller preoperative LL and TK compared with patients without D segment. There were no significant differences in the postoperative LL, PI minus LL, and C7 SVA between D and C segments after MIS-OLIF. In the OLIF procedure, the mean cage height was 11.5 ± 1.4 mm for the D segment and 11.1 ± 1.7 mm for the C segment, with no significant differences noted between the two groups (p = 0.136). Similarly, the mean cage angle was 8.6 ± 2.8° in the D segment and 7.7 ± 2.5° in the C segment, showing no significant difference between groups (p = 0.069).

Fig. 3
figure 3

Flowchart of patient recruitment. OLIF oblique lumbar interbody fusion, D discontinuous, C continuous

Full size image
Table 1 Baseline and radiological characteristics of patients with D segment and patients without D segment
Full size table

Table 2 presents a comparison of the preoperative and postoperative radiographical parameters between D and C segments. D segments demonstrated a significantly larger proportion of decreased disc height (92.1% vs. 74.6%, p = 0.019), significantly smaller preoperative SA (mean ± SD, −1.1° ± 6.7° vs. 6.6° ± 6.3°, p < 0.001), and significantly larger postoperative SA (mean ± SD, 12.4° ± 6.0° vs. 9.7° ± 5.4°, p = 0.006) and change of SA (mean ± SD, 13.5° ± 3.4° vs. 3.1° ± 3.9°, p < 0.001) compared with C segments. Although a higher proportion of patients with D segment displayed spondylolisthesis, spondylolysis, angular instability, and translational instability compared with patients with C segment, these differences were not significant.

Table 2 Radiological characteristics of included segments in preoperative and postoperative radiography
Full size table

Table 3 compares the radiological parameters of preoperative MR images between D and C segments. D segments had a significantly higher proportion of sequestrated discs (35.1% vs. 16.3%, p = 0.009) and facet effusion (76.3% vs. 48.6%, p = 0.002) compared with C segments. No significant differences were noted in the rate of facet joint osteoarthritis or the Schizas and Lee grades.

Table 3 Radiological characteristics of included segments in preoperative MRI
Full size table

In the multivariate logistic regression analyses, preoperative SA (OR [95% CI]:0.733 [0.639–0.840], p < 0.001) and the presence of facet effusion (OR [95% CI]:14.054 [1.758–112.377], p = 0.027) were significant predictors for >10° SA correction after MIS-OLIF at each index surgical level. (Table 4, the full version of Table 4 in the Appendix).

Table 4 Univariate and multivariate logistic regression analyses about the identification of preoperative factors predicting >10° SA correction after MIS-OLIF
Full size table

ROC curve analysis identified threshold values predicting >10° SA correction after MIS-OLIF at each surgical level. Preoperative SA showed moderate discriminative power (AUC: 0.800, 95% CI: 0.717–0.882, p < 0.001) with a cutoff value of −0.05° (Fig. 4), implying that patients with preoperative kyphotic SA values are likely to achieve SA changes >10° after MIS-OLIF. The combination of facet effusion with preoperative kyphotic SA enhanced this predictability (OR = 10.5; specificity = 0.91; positive predictive value = 0.56). In segments displaying preoperative lordotic SA without facet effusion, the likelihood of failing to achieve an SA change >10° after MIS-OLIF was 95% (negative predictive value = 0.95).

Fig. 4
figure 4

ROC curves illustrating the diagnostic accuracy of preoperative SA with a threshold value of −0.05° in predicting >10° SA correction after minimally invasive oblique lumbar interbody fusion. SA segmental angle, AUC area under the ROC curve

Full size image

Strong correlations were observed between the preoperative SA and degree of SA correction (r = −0.618, p < 0.001). Predictive formulae for the degree of SA correction were established using preoperative SA and the presence of facet effusion. For segments with facet effusion, the formula was as follows: degree of SA correction = −0.560 × preoperative SA + 8.431 (R2 = 0.478, p < 0.001; Fig. 5a). For segments without facet effusion, the formula was: SA correction = −0.354 × preoperative SA + 5.853 (R2 = 0.226, p < 0.001; Fig. 5b).

Fig. 5
figure 5

Relationships between the degree of SA correction and preoperative SA in (A) segments with and (B) without facet effusion. SA, segmental angle

Full size image

Discussion

Our study revealed that the presence of preoperative kyphotic SA and facet effusion can predict >10° SA correction after MIS-OLIF at each surgical index level. These correctable segments were defined as D segments. However, in patients presenting with preoperative lordotic SA without facet effusion, the likelihood of achieving >10° SA correction after MIS-OLIF is limited. Hence, supplemental procedures such as ACR or posterior osteotomy might be required for additional correction of the decreased LL.

We defined segments that showed a >10° SA correction following MIS-OLIF as discontinuous segments (D segments). The term “discontinuous” was used because of the acute disruption of continuity within a single smooth lumbar lordotic curve. Examples of D segments are shown in Figs. 1 and 2, where the entire lumbar curve appears interrupted at the L4–5 level, which is the D segment. This disruption in D segment can be rectified by MIS-OLIF, resulting in a smooth lumbar curve. This concept parallels the disruption of the Shenton line at the hip joint [20].

Patients with D segment showed a more severe preoperative global sagittal imbalance with a larger preoperative PI minus LL and C7 SVA compared with patients without D segment; however, the difference disappeared after MIS-OLIF. Both patient groups, those with and without D segments, demonstrated acceptable postoperative LL, PI minus LL, and C7 SVA (Table 1). Patients with D segment showed a smaller postoperative C7 SVA after MIS-OLIF compared with patients without D segment, although the difference was not significant. These findings suggest that the D segment significantly affects global sagittal imbalance and that correcting the D segment can result in adequate correction of the preoperative global sagittal imbalance. Our results align with those of Huang’s study, indicating that a smaller preoperative PI, PT, or PI–LL predicts sagittal spinal realignment in older patients with LSS after short-segment decompression and fusion surgery [21].

D segments showed more severe preoperative regional loss of lordosis with a significantly smaller preoperative SA compared with C segments. However, after MIS-OLIF, D segments showed significantly larger postoperative SA compared with C segments. Furthermore, the change in SA in D segments after MIS-OLIF was significantly higher, with a mean value of 13.5°, than that in C segments (Table 2). Previous research demonstrating a >10° change in regional LL after XLIF often involved the use of 20° or 30° hyperlordotic cages [10, 13]. Youn et al. also recommended using hyperlordotic angle cages in short-segment lumbar fusion for better correction of LL [22]. However, in our clinical experience, without performing ACR, inserting a cage with an angle significantly larger than the endplate angle did not lead to widening of the disc space in accordance with the cage angle. Instead, a discrepancy between the endplate and cage angles reduced the contact area between them, often resulting in endplate fracture. This phenomenon was particularly pronounced in patients with osteoporosis. Therefore, we primarily used cages with a 6° or 12° lordotic angle without additional ACR or posterior osteotomy to maximize disc space height with intact anterior longitudinal ligament. Thus, our findings represent the inherent correction potential for lordosis in D segments.

In logistic regression, the identification of preoperative factors predicting >10° SA correction after MIS-OLIF revealed that a smaller SA and the presence of facet effusion were significant predictors. This is corroborated by Alahmari and Zhu, who demonstrated that preoperative kyphotic discs acquire more lordosis with interbody cage use [23, 24]. They suggested that kyphotic discs possess greater corrective potential for lordosis, making positive lordotic alterations easier to attain than lordotic discs. Their explanation is also consistent with our results of smaller preoperative SA with a mean value of −1.1° and a larger change in SA in D segments, compared to C segments.

The presence of facet effusion, which is often considered an indicator of segmental instability [16], was a significant factor in predicting >10° SA correction after MIS-OLIF. No previous studies have investigated the association between the presence of facet effusion and SA correctability. Our findings suggest that facet effusion can serve as a potential indicator of satisfactory segmental correction after MIS-OLIF. The joint space occupied by fluid represents the space available within the posterior element for LL restoration, a concept that is diametrically opposed to an ankylosed facet joint [25].

Through ROC curve analysis, we established quantifiable criteria for predicting >10° SA correction after MIS-OLIF. Patients with preoperative kyphotic SA and facet effusion can expect an adequate SA correction of >10° with a 56% probability. Using our predictive formulae for the degree of SA correction, surgeons can calculate the expected degree of SA correction after MIS-OLIF. If sufficient correction is anticipated in a D segment using MIS-OLIF, surgeons can initially consider MIS-OLIF, preventing unnecessary extensive surgeries, such as posterior osteotomy or long-level fusion, based solely on preoperative sagittal imbalance in the SVA or LL. Conversely, patients with preoperative lordotic SA and no facet effusion have a 95% chance of failing to achieve an adequate SA correction >10°. Thus, for patients with lordotic SA and no preoperative facet effusion, if additional LL correction is required for remnant global sagittal imbalance, supplementary procedures, such as ACR or posterior osteotomy, may be employed during OLIF.

This study has some limitations. First, this was a retrospective study, and there is possibility of confounder bias. Second, this early-stage study did not evaluate potential mechanical complications or other long-term outcomes such as adjacent segment degeneration and cage subsidence. Moreover, we did not incorporate clinical scores such as the visual analogue scale, Oswestry disability index, or implement the Japanese Orthopaedic Association Back Pain Evaluation Questionnaire. Nevertheless, our primary aim was to identify the preoperative factors predicting >10° SA correction after MIS-OLIF. Consequently, the absence of clinical scores is unlikely to have significantly influenced our analysis. Despite these limitations, the strength of this study lies in its novelty and comprehensive examination of preoperative factors associated with satisfactory SA correction after MIS-OLIF.

Conclusion

Our study revealed that preoperative kyphotic SA and the presence of facet effusion can predict >10° SA correction following MIS-OLIF at each surgical index level. In patients with D segments showing preoperative sagittal imbalance, kyphotic SA, and facet effusion, satisfactory SA correction exceeding 10° can be expected after MIS-OLIF without ACR. Consequently, initially considering MIS-OLIF as a solitary procedure may be prudent rather than resorting to extensive interventions, such as posterior osteotomy or long-level fusion. However, for patients presenting with preoperative sagittal imbalance and lordotic SA without facet effusion, the likelihood of achieving >10° SA correction after MIS-OLIF is limited. For patients with lordotic SA and no preoperative facet effusion, supplemental procedures, such as ACR or posterior osteotomy, should be prepared for additional LL correction required for remnant global sagittal imbalance after MIS-OLIF.