Predictors of residual low back pain in patients with osteoporotic vertebral fractures following percutaneous kyphoplasty

Abstract

Objective: Patients with osteoporotic vertebral fractures (OVFs) often suffer from residual low back pain (LBP) after percutaneous kyphoplasty (PKP). The purpose of this study was to identify risk factors for postoperative residual LBP and to develop a nomogram to predict the occurrence of residual LBP.

Methods: We retrospectively reviewed 236 patients who underwent PKP for OVFs and had a minimum follow-up of 12 months. The mean age was 72.1 ± 6.3, 74.3% were female and 25.7% were male. Patients with LBP VAS scores ≥ 3.5 at the 12th month postoperatively were considered to have residual LBP. Risk factors for residual LBP were identified by univariate and multifactorial logistic regression analysis. Then, a predictive nomogram was constructed and validated using the bootstrap method. The discrimination, calibration, and clinical utility of the nomogram were assessed using a receiver operating characteristic curve (ROC), a calibration curve, and a decision curve analysis (DCA).

Results: univariate and multifactorial logistic regression analysis identified depression (P = 0.02), intravertebral vacuum cleft (P = 0.01), no anti-osteoporosis treatment (P < 0.001), cement volume <3 ml (P = 0.02), and cement distrubution (P = 0.01) as independent risk factors for residual LBP. The area under the ROC was 0.83 (0.74-0.93) and further validated by bootstrap method was 0.83 (0.73-0.92). The calibration curve illustrated the consistency between the predicted probability and the observed results. DCA showed that nomogram exhibits clinical utility and net benefit when the threshold probability is between 6% and 73%.

Conclusions: Our study found that depression, intravertebral vacuum cleft, no anti-osteoporosis treatment, cement volume <3 ml and cement distribution represent independent risk factors for residual LBP. The nomogram containing the above five predictors can accurately predict the risk of residual LBP after surgery.

Keywords: low back pain; nomogram; osteoporotic vertebral fractures; percutaneous kyphoplasty; risk factors.

Figure 1

Patients selection flowchart.

Figure 2

Radiographic evaluation of compressed vertebrae. Preoperative VHR (left,%):2b/a + c, VKA (left,°):∠d; Postoperative VHR (right,%):2B/A + C, VKA (right,°):∠D.

Figure 3

MRI images of intravertebral vacuum cleft. Hypointense on T1-weighted images (A, arrows); hyperintensity or hypointensity on T2-weighted images (B) or STIR images (C), depending on fluid (dotted arrows) or gas (solid arrows) filling.

Figure 4

Preoperative MRI images showed that the fracture area was located in the upper third of the vertebral body (A: T1, B: T2, C: STIR, red arrow); postoperative CT images showed insufficient cement filling in the vertebral fractured area (D: left in median sagittal section; E: median sagittal section; F: right in median sagittal section, red arrow). Posterior fascia edema:hypointense on T1-weighted images (A, yellow arrow); hyperintensity on T2-weighted images (B, yellow arrow) or STIR images (C, yellow arrow).

Figure 5

Predictive nomogram for residual LBP.

Figure 6

ROC curves of the nomogram for the assessment of capable of accurate prediction and discrimination.

Figure 7

Calibration curve of the nomogram for the assessment of the consistency between the actual diagnosed residual LBP and the projected likelihood of residual LBP.

Figure 8

DCA curve of the nomogram for the assessment of clinical practice based on the net benefit and threshold probabilities.