Post-fusion lumbar flatback deformity with global sagittal imbalance causes gait disturbances, chronic low back and refered leg pain, and reduced quality of life. The incidence of patients requiring lumbar corrective osteotomy for post-fusion lumbar flatback deformity with sagittal imbalance is increasing as the numbers of spine fusion surgeries and overall average life expectancy increases12). In patients with sagittal imbalance, adequate correction to restore an ideal lumbar lordosis relative to pelvic incidence (PI) in the sagittal plane is very important during corrective osteotomy3). According to Schwab et al.15), corrective osteotomy for adult spinal deformity (ASD) leads to good sagittal spinal balance if the range of correction is within lumbar lordosis (LL)=PI±9°, pelvic tilt (PT)<20°, and sagittal vertical axis (SVA)<5 cm. However, patients with post-fusion lumbar flatback are generally osteoporotic, elderly females who have poor paraspinal muscle condition due to fatty degenerative changes, atrophy of the back muscles, and thinning of the paraspinal and psoas muscles. Therefore, restoration of lumbar lordosis below the PI value might result in another (or secondary) postoperative or delayed sagittal imbalance due to the lack of the compensatory roles from these paraspinal muscles. Two previous studies reported theoretical values for PT and LL related to PI. These studies suggested that ideal lumbar lordosis consists of 67% LL at L4–S1, 85% LL at L3–S1, and 97% LL at L2–S1. The degree of PI was categorized into 6 groups (I to VI), and theoretical LL and PT values were recommended for each group14). In this study, radiographical outcomes of post-fusion lumbar flatback patients treated with pedicle subtraction osteotomy (PSO) or Smith-Petersen osteotomies (SPOs) with ALIFs were analyzed and compared at preoperative, postoperative 1 month, and at the final follow up and achieved mean PSO angles were evaluated. Clinical outcomes by Oswestry Disability Index (ODI), visual analog scale (VAS) score of back pain/leg pain, and Scoliosis Research Society-22 score (CRS-22r) were analyzed and compared. At the final follow-up, all patients were divided into 2 groups (SVA ≤5 cm : normal SVA, SVA >5 cm : positive SVA) and compared with both groups by according to radiological and clinical outcome parameters.

As for the overall radiological outcome for total 28 patients, the mean PI was 53±9°. The TK, LL, PT, PI-LL, SVA, and T1PA were significantly improved both immediate postoperatively and on the final follow-up (p<0.05). TK improved from 10±17° kyphosis preoperatively to 31±9° at postoperative one month and was well maintained through the final follow-up (37±8°). The LL improved from 2±23° kyphosis preoperatively to -60±8° lordosis at postoperative one month and maintained -53±23° at the final. As for the changes in the PI-LL mismatch, SVA, and PT over the due clinical courses (preoperative, postoperative one month, and final follow-up), the PI-LL improved from 55±24° preoperatively to -6±6° at postoperative one month and this improvement was well maintained through the final follow up (-4±7°). The SVA was 194±81 mm preoperatively, 7±27 mm at postoperative 1 month, and 19±35 mm at the final follow up. The T1PA improved from 43±14° preoperatively to 11±7° at postoperative one month and this improvement was maintained into 13±7° at final follow up. Similarly, the PT was 30±11° preoperatively, 14±7° at postoperative one month, and 17±7° at the final follow up (Table 2).

As for the clinical outcomes for this series between preoperative and final follow-up, all investigated parameters including ODI, VAS score, and SRS-22r score were significantly improved (p<0.05). ODI improved from 66±10 preoperatively to 28±14 at final. VAS of the back pain/leg pain were improved from 8.0±1.0/7.5±1.2 preoperatively to 2.5±1.4/2.3±1.5 at final. Mean SRS-22r score was improved from 2.1±0.4 preoperatively to 3.6±0.4 at final (Function; from 2.1±0.5 to 3.3±0.5 : Pain; from 2.1±0.5 to 3.9±0.5 : Self-image; from 2,0±0.6 to 3.5±0.5 : Mental health; from 2.3±0.6 to 3.7±0.5) (Table 3).

The SVA values at the final follow up were categorized as normal SVA [≤50 mm, n=23(82%)] and positive SVA [>50 mm, n=5 (18%)]. The mean SVA was 7±26 mm for the normal SVA group and 73±11 mm for the positive SVA group, a statistically significant difference between groups (p=0.001). The PI-LL was -5±6° for the normal SVA group and 3±7° for the positive SVA group (p<0.05). the T1PA was 11±6° for normal SVA group and 22±8° for the positive SVA group (p<0.05). However, the PT did not differ significantly between groups (Table 4). On the contrary, there were no significant differences in clinical outcome measures such as ODI, VAS score of the back pain/leg pain, and SRS-22r score including 5 main domains between the two groups at the final follow-up. All clinical outcome parameters were significantly improved for both groups on the final follow up from baseline (Table 5).

Perioperative complication was 29% (8 cases) and early reoperation were 18% (5 cases). Among them, 4 acute proximal junctional failures occurred ; 2 UIV fractures, one UIV+1 fracture, and one proximal screw loosening. The other perioperative complications included 4 cases (14%). 3 cases of deep vein thrombosis and one focal neurological deficit in spite of 3 times of neural decompression and exploration surgery of a left L3 nerve root compression after L3 PSO occurred. Late complications were 46% (13 cases) and late revision surgery were 29% (8 cases). 4 rod fracture were reoperated with rod change with/without additional 3 or 4 rods techniques. 3 fusion extension due to delayed proximal junctional kyphosis and one nonunion at L5S1 level were reoperated. Among 3 cases of proximal junctional kyphosis, one patient with delayed UIV+1 chronic compression fracture who had severe neurological deficit was addressed with neural decompression and fusion extension up to UIV+4. In addition, there were 5 cases of long-term complications; one distal screw loosening, 3 proximal junctional kyphosis, and one asymptomatic rod fracture.

It is well known that post-fusion lumbar flatback deformity is caused by hypolordotic lumbar fusion, adjacent segment disease after lumbar fusion, non-union with or without insufficient anterior column support, and post-laminectomy fusion. In addition to these conditions, atrophy or fatty degeneration in the paraspinal muscles might cause global sagittal imbalance in the lumbar spine that leads to increased fatigue and pain. To diagnose post-fusion lumbar flatback deformity, clinicians should review the patient's history, AP and lateral whole spine standing X-rays, and CT and MRI of the spine. The patient's history may include difficulty in standing upright and complaints of back or leg pain, difficulty in standing during large load bearing, difficulty bending, and frequently relying on the knee as buttress during gait or stair climbing. In whole spine standing X-rays, more than 5 cm of SVA, loss of lumbar lordosis, PI-LL mismatch of more than 10 degrees, a large PT of more than 20°, and an increase in the T1-pelvic angle of more than 15 degrees can be measured, and paraspinal muscle atrophy and fatty infiltration, including thinning and stretching of the psoas muscles and back extensor muscle atrophy, can be detected on lumbar spine MRI. Spine angiography 3-dimensional CT is very useful to confirm the condition of major abdominal vessels, including the abdominal aorta, inferior vena cava, and iliac vessels, as well as the previous bony fusion state, facet changes, intradiscal vacuum changes, and adjacent segment pathology.

Accordingly, the initial radiological parameters included in this series of post-fusion lumbar flatback deformity demonstrated similar phenomena; PI, 53±9°; LL, 2±22° kyphosis; PI-LL, 55±24°; SVA, 194±80 mm; T1PA, 43±13°, and PT : 30±10°.

The aim of surgery for correction of post-fusion lumbar flatback deformity is to achieve a stable, well-balanced lumbar spine centered over the pelvis. Surgical treatment involves correction of lordotic spinal alignment in the flat or kyphotic lumbar spine by corrective osteotomies, such as PSO or lower lumbar SPOs with ALIFs. In patients with post-fusion flatback whose main pathology was hypolordotic lumbar fusion (19 cases), PSO were performed at L2 in 4 cases, L3 in 8 cases, and L4 in 7 cases. The others 9 patients were corrected with lower lumbar SPOs with ALIFs. The posterior only approach, including PSO, was done by one-stage operation, but lower lumbar SPOs with multilevel ALIFs were usually segmented as staged operations with initial posterior lower lumbar SPOs followed by anterior ALIFs and rod placement posteriorly (9 cases). Regarding surgical methods and indications, posterior only approach (PSO) was performed in patients with evidence of satisfactory fusion at previous surgical levels on radiological follow ups whereas cases lacking the assured evidence of fusion, anterior approach such as ALIFs was combined for proper fusion achievement as well as restoration of lordosis. Similarly, in those cases in which satisfactory fusion up to S1 level was evident, LIV was not extended to iliac level. On the other hand in patients with insufficient fusion at L4–5 and L5S1 level, iliac screw fixation procedure was performed to augment L5S1 interbody fusion.

Several different methods exist for preoperative measurement for aimed correction9). Most of these methods are geometrical and merely take into account the numerical amount of correction angles required to shift the location of the head properly over the pelvis. However, recent studies have shown that sagittal spinopelvic parameters among patients with spinal deformity play a role in clinical outcomes. These studies rated the severity of adult spinal deformity using the Scoliosis Research Society (SRS)-Schwab ASD classification and compared the results to HRQOL, PI-LL mismatch, SVA, and PT values19). The radiological parameters are important, as they can predict the patients' prognosis and satisfaction levels. Among these parameters, LL can be modified through surgical intervention, but PI is a constant value that is fixed for each individual10). Similarly, the PI-LL mismatch can adversely affect HRQOL, regardless of actual surgical performance15161719). Schwab et al.14) reported threshold values for sagittal spinopelvic alignment to achieve during reconstructive surgery for improved HRQOL. Surgical outcomes are generally measured using PI-LL values. According to the SRS-Schwab classification, good surgical outcomes are achieved if the PI-LL value ranges within±10°. According to Schwab, the equation LL=PI+9° (±9) is a basic calculation in asymptomatic adults. He suggested that realignment objectives in the sagittal plane should aim for a postoperative SVA <50 mm, PT <20°, and LL=PI±9° to achieve spinopelvic realignment14). Lafage et al.5) have proposed that SVA is associated with HRQOL and pelvic retroversion. Berjano et al.2) have suggested that the rule LL=-(PI+10°) can be used to plan for good alignment. However, the authors note that is not easy to predict optimal alignment when correcting deformity13).

Due to the variability of SVA value according to the different degree of pelvic retroversion or knee flexion even in a single subject, the T1PA value has been frequently cited and well acclaimed as a reliable, consistent radiological parameter that could be correlated with surgical outcome recently. This T1PA value accounts for both SVA and pelvic tilt, and according to the several references, this might be reduced to less than 10° in order to achieve optimal surgical outcome both radiologically and clinically811). The representability of T1PA instead of SVA was also reflected in our series; 11° for the normal SVA group and 22° for positive SVA group even after correction.

The authors' concept or suggestion according to past several years of surgical experiences in post-fusion lumbar flatback deformity would be that postoperative LL in these cases should be greater than PI to restore and maintain normal sagittal spinal balance. Because Asian patients with post-fusion lumbar flatback are generally osteoporotic, elderly females with poor paraspinal muscle condition due to fatty degenerative changes, atrophy of the back muscles, and thinning of the paraspinal and psoas muscles, the compensatory role from the paraspinal muscles is insufficient. There is an obvious difference in the volume of the paraspinal muscles between Asian and Western populations and since its volume is usually smaller for Asians, even if the initial or background PI-LL value might be similar, loading on the spine would be relatively higher. Paraspinal muscle atrophy due to surgery should be considered, as well67). Therefore, restoration of lumbar lordosis below the PI value might result in another or secondary postoperative or delayed sagittal imbalance consequently. Roussouly110) and Barrey et al.1) have also suggested theoretical values for PT and LL related to pelvic incidence. They categorized PI into 6 groups (I to VI) and recommended theoretical LL and PT in each PI group.

In this study, at the final follow-up, the normal SVA group (82%) had SVA of 7 mm, while the positive SVA group had SVA of 73 mm, a significant difference between groups. The PI-LL was -5° for the normal SVA group and 3° for the positive SVA group and T1PA was 11° for the normal SVA group and 22° for positive SVA group. The PI-LL, SVA, and T1PA were significantly different between groups. This pattern was exceptional for the PT. At the final follow-up, the ODI score, VAS scores for back and leg pain, and SRS-22r score showed improvement in all 28 patients. But between two groups for comparison, there were no significant differences in ODI, VAS score of the back pain/leg pain, and SRS-22r score including 5 main domains.

Due to the intense, multilevel- involving, corrective procedure in revisional fashion, the expected occurences of untoward consequence could surge during the management of this post-fusion lumbar flatback deformity [in our series, the reoperation rate was 46% (13 patients)] compared to other spinal procedures. Most of these necessity of revisional procedure were related to the proximal junctional problems (UIV or UIV+1 fractures) or instrument failure frequently brought up at the UIV level during early postoperative period except one case of severe neurological deficit. Fortunately, a positive prognosis through revision surgery was anticipated in most of compromised cases.

In addition, patients with all early revision and late revisions due to rod fractures or nonunion did not change significantly in the SVA, only delayed PJK patients progressed positive SVA.

Despite the slightly higher complication rate, most of patients were categorized as improved clinical outcomes with higher satisfaction, in accordance with radiological parameter improvements.

Fixed sagittal imbalance after lumbar spinal fusion could limit or disable quality of life. PSO or SPOs combined with multilevel ALIFs at the lower lumbar spine are significantly improves sagittal balance in post-fusion lumbar flatback patients with sagittal imbalance. For restoration and maintenance of normal sagittal spinal balance in these patients, PI-LL might be made negative value and T1PA might be less than 11° even though positive SVA group was also significantly improved clinical outcomes.