Background
It is well reported that the patient’s age plays an important role associated with proximal junctional failure (PJF) development. Various characteristics of adult spinal deformity (ASD) patients were different between younger and older age groups. We hypothesized that the radiographic risk factors for PJF would different according to younger and older age groups. This study aimed to evaluate different radiographic risk factor of PJF according to the two age groups undergoing thoracolumbar fusion for ASD.
Methods
ASD patients aged ≥ 60 years who underwent thoracolumbar fusion from low thoracic level (T9~T12) to sacrum were included. The minimum follow-up duration was two years. PJF was defined as proximal junctional angle (PJA) ≥ 20°, fixation failure, fracture, myelopathy, or necessity of revision surgery. Using various radiographic risk factors including age-adjusted ideal pelvic incidence (PI)-lumbar lordosis (LL), univariate and multivariate analyses were performed separately according to the two age groups : <70 years and ≥70 years.
Results
A total of 186 patients were enrolled (mean age=68.5 years old, 90.3% female). Mean follow-up duration was 67.4 months. PJF developed in 98 patients (32.0%). There were fracture in 53 patients, PJA ≥ 20° in 26, fixation failure in 12, and myelopathy in 6. PJF developed more frequently in patients older than 70 years than in younger than 70 years. In patients aged less than 70 years, preoperative LL, PI-LL and change in LL were significant risk factors in univariate analysis. Multivariate analysis showed only change in LL was significant for PJF development (Odds ratio [OR]=1.025, p=0.021). On the other hand, in patients older than 70 years, postoperative LL, postoperative PILL, overcorrection relative to conventional PI-LL target (within ±10°) as well as age-adjusted ideal PI-LL target were significant. On multivariate analysis, only overcorrection of PI-LL relative to age-adjusted ideal target was a single significant factor to cause PJF (OR=5.250, p=0.024).
Conclusions
In patients younger than 70 years, greater change in LL was associated with PJF development regardless of PI-related value. However, in older patients, overcorrection of PI-LL relative to the age-adjusted PI-LL target was important to cause PJF.

Spinal sagittal balance is influenced by thoracic kyphosis, lumbar lordosis, as well as the position and angle of the pelvis. The abnormal position of sagittal imbalance causes easy fatigue and pain on the back muscles, so that the body voluntarily takes action to compensate for the imbalance. Compensatory mechanism occurs over several steps. It begins primarily in the mobile lumbar segments. When sagittal imbalance occurs due to decreased lumbar lordosis, the disc space at the mobile segments is hyperextended, allowing the C7 plumb to be restored within near normal range. As patients are older, subsequent degenerative disc changes develops at the hyperlordotic segments leading to loss of lumbar lordosis, which results in sagittal imbalance again. If the compensation in the spine is no longer able to restore the sagittal imbalance, posterior rotation of pelvis can reestablishes sagittal balance. This finding shows that the rotation of pelvis is important in the compensatory mechanisms. Another thing to consider is that compensation gets involved in the actual lumbar lordosis. During the compensation process rotation of pelvis determines the sagittal imbalance. This is also very difficult to estimate how much the rotation of pelvis is, because the rotation is variable depending on the position. Pelvic incidence (PI) is a suitable index to determine how much correction of lumbar lordosis requires, as this does not change depending on the position of the pelvis. Patients with high PI are able to compensate the sagittal imbalance more than them with low PI. As PI is the sum of pelvic tilt and sacral slope, PI is closely related to lumbar lordosis. Patients with high PI needs more correction of lumbar lordosis to restore sagittal imbalance than the patients with low PI.

In patient with loss of lumbar lordosis, lumbar lordosis restoration is very important to get a normal sagittal balance. It is difficult to know the ideal lumbar lordosis, because it is a individual-unique pattern and is influenced by numerous variables. To restore a more ideal lumbar lordosis, careful analysis and planning will be needed because incomplete restoration of lumbar lordosis may result in poor clinical outcomes and serious complications. Lumbar lordosis has strong relationship with pelvic orientation, so there are many reports about predictive equation of lumbar lordosis using pelvic parameters. Pelvic incidence is major determinant of pelvic orientation, therefore, influence on lumbar lordosis. So pelvic incidence is a useful value to determine the ideal lumbar lordosis, and we recommend that the degree of restoration of lumbar lordosis is just enough to pelvic incidence at least. And also, a close study and understanding of the numerous factors affecting lumbar lordosis was needed to effective lumbar lordosis restoration.