Cement leakage into adjacent vertebra during percutaneous vertebroplasty: “Gild the Lily” or “Kill two birds with one stone”?

Percutaneous vertebroplasty (PVP) has been widely used to treat painful osteoporotic vertebral compression fracture (OVCF). Cement leakages (cement goes beyond the vertebra and appears in other locations such as disc space, paravertebral tissue, intervertebral foramen or epidural space) during PVP are common and reported ranging from 10 to 65%. Routes of cement leakage usually include four types (1) B-type: via the basivertebral vein; (2) S-type: via the segmental vein; (3) C-type: via a cortical defect; and (4) D-type: via a damaged endplate1. Cement leakage into adjacent vertebra is rare. Kim (2016) reported the first case of cement leakage into adjacent upper normal vertebra during PVP2. We (2021) reported two cases in which one was leakage into adjacent lower normal vertebra and the other was leakage into adjacent upper normal vertebra3. Here, we submitted a set of unique images of intraoperative cement leakage into “sandwich” vertebra, proposed the possible mechanisms and discussed its clinical significance. “Sandwich” vertebra is a special adjacent vertebra which refers to a well-preserved vertebra between two cement-augmented vertebrae and is more prone to fracture than dose ordinary adjacent vertebra in theory due to its bearing the double enlarged load transmission from the upper and lower augmented vertebra4. The rare interesting phenomenon further broadens the routes of cement leakage. A woman in her 70s with a 2-month history of painful OVCFs at T9 and T11 vertebra was treated with PVP by unilateral (right) transpedicular approach under local anesthesia and conscious sedation. Under G-Arm fluoroscopic monitoring, after two 11-gauge, hollow, straight, beveled needles were ideally located on anteroposterior and lateral views respectively, polymethyl methacrylate was injected at the paste-like stage using a set of 10 ml syringe and metal propeller. Firstly, T11 was augmented. During the procedure, cement was noted to migrate out of T11 vertebral body from its compressive superior endplate into T10/11 intervertebral disk space. After a 5-second pause, injection was continued and cement penetrated through T10 inferior endplate and dispersed into T10 vertebral body. The procedure was finished and cement injection volume was 7 ml. The residual cement was injected into T9 vertebral body uneventfully (Fig. 1). She had no neurological impairment and could walk with a thoracolumbar brace after 1 day. Outpatient anti-osteoporotic treatments were continued and she was asymptomatic at the 12-month follow-up. Figure 1.: G-Arm lateral view monitoring for the process of cement leakage from compressive T11 vertebra to upper adjacent normal-appearing T10 (sandwich) vertebra through T10/11 intervertebral disk space (A, B, and C) and the residual cement injected into T9 vertebral body uneventfully (D). Cement leakages during PVP are common and most are asymptomatic which are generally considered as normal phenomena during surgery rather than true complications like spinal cord compression, pulmonary embolism, and nerve root irritation5. Many established factors can contribute to the occurrence like cortical bone integrity, intravertebral vacuum cleft, basivertebral foramen sign, cement injected volume, etc6. Cement leakage into adjacent irresponsible vertebral body is rare and the mechanism is unclear. Perivertebral venous plexus is frequently interpreted as the anatomic basis for cement leakage; however, this mechanism is apparently inapplicable for the intravertebral cement leakage from the anatomical aspect. For the adjacent intravertebral cement leakage, we suggest the possible mechanisms are biologically latent vertebral endplate defects and biomechanically intervertebral stress discrepancy between the compressive condensing vertebra and the normal-appearing osteoporotic vertebra. This phenomenon maybe is ignored clinically and reported rarely because this intravertebral cement leakage does not weaken the overall therapeutic effect. It may be worthwhile in future to think about how to avoid this unexpected occurrence to the adjacent normal vertebra (if occurring, it may be “Gild the lily”) or how to utilize this occasional occurrence to the adjacent fractured vertebra or “sandwich” vertebra which is susceptible to fracture (if occurring, it may be “Kill two birds with one stone”). Though the latent vertebral endplate defect is our proposed anatomical mechanism for the adjacent intravertebral cement leakage during PVP, the endplate defect of adjacent vertebra was not found on the preoperative CT and MRI for this patient. So, this interesting phenomenon needs further study through more cases and preoperative deep multimodal imaging analysis or even artificial intelligence technology to help predict the possible occurrence of intraoperative cement leakage7.