Osteoporotic VCF lead to significantly reduced life expectancy in both men and women 5. In postmenopausal women the risk of subsequent VCF is much greater than for other fractures (relative risk = 4.4) 2021. This risk increases with the severity of the deformity 22. Therefore surgical correction of deformity has the ability to reduce morbidity and mortality in these patients. Conventional open interventions require anterior open-wedge or posterior closing-wedge techniques with long distance posterior fusions because of poor bone quality. Due to the surgical access and co-morbidities these operations go along with severe complications, but they were found to have excellent 2-year results in ODI and VAS improvement 23. Much less invasive is the kyphoplasty procedure, which also has the capability of reconstructing VCF height. Voggenreiter et al 9 found a reduction of 3.1 degrees Cobb angle in standing radiographs after kyphoplasty (n = 30). Pradhan et al 24 found a local correction of the fractured vertebra of 7.2 degrees, but only 2.4 degrees of the Cobb angle, when measured 1 level above and below (n = 65). Interestingly they found a greater improvement of kyphosis with multilevel kyphoplasty of 7.7 degrees Cobb angle.

The comprehensive meta-analysis of Taylor et al 19 summarized all published kyphoplasty complications. Cement leakages occurred in 8.1% of all cases, but only 0.09% were symptomatic. New vertebral fractures occurred in 11.1%, and 9.4% were adjacent vertebrae. Pulmonary embolism occurred in 0.17% of all cases. Spinal stenosis with spinal cord compression occurred in 0.16% of all cases. Radiculopathy was found in 0.17% of all cases. The overall mortality was 4.4%, perioperative mortality was 0.13%.

Adjacent fractures are the most common adverse event found after kyphoplasty. The occurrence of adjacent fractures is known from vertebroplasty, where 12.4% had subsequent VCF after two years (n = 177) 25. Kyphoplasty was thought to have lesser adjacent fractures due to correction of kyphotic deformity 22, but kyphosis is not the only reason for adjacent fractures. Lin et al 26 correlated the incidence of adjacent fractures to cement leakage into the disc in vertebroplasty (n = 38, p < 0.005). Komemushi et al 27 found cement leakage into the disc to be a significant predictor of adjacent VCF (n = 83, p < 0.001). These findings will apply to cement leakage in kyphoplasty, too.

In our study population after 6 months adjacent fractures were found in 7.8% of all cases (n = 102). Fifty percent of these patients had secondary osteoporosis due to corticoid medication. Fribourg et al 17 found subsequent fractures after kyphoplasty in 26% of all cases (n = 38), 21% occurred during the first two months. Harrop et al 28 found subsequent fractures in 22.6% of all patients after a mean follow-up of 11 months (n = 115). Sixty-five percent of these had secondary steroid-induced osteoporosis. Another investigation by Lavelle & Cheney 29 found 17% recurrent fractures within one year after kyphoplasty, 11.7% occurred during the first 90 days (n = 94). They did not find any impact of secondary osteoporosis on subsequent fractures. A prospective investigation by Moon et al 30 found an incidence of subsequent VCF in 15.5% patients after one year (n = 111). Interestingly they could correlate the appearance of adjacent VCF to the amount of PMMA cement applied during the procedure (p < 0.05).

The available data reveals that subsequent VCF of adjacent vertebrae occur in 7.8% to 26% of all patients treated with kyphoplasty. Kyphosis, secondary osteoporosis, and cement leakage into the intervertebral disc facilitate the occurrence of subsequent VCF. Identifying risk factors for subsequent VCF, several authors discussed the indication for prophylactic cement augmentation of adjacent vertebrae [313233]. Especially in cases of "sandwich-kyphoplasty" with an osteoporotic non-fractured vertebra between two kyphoplasties some recommend a prophylactic kyphoplasty 32. Until now there is no clinical evidence for the effectiveness of prophylactic vertebro- or kyphoplasty 32, but the biomechanical investigation of Sun and Liebschner 31 using finite-elements found a significant reinforcement of high-risk vertebral bodies with prophylactic vertebroplasty. Despite some promising experimental data, in the light of the present evidence it rather seems that the prophylactic use of PMMA reduces the safety of the procedure 31.

Fearing secondary spinal stenosis due to further sintering of burst fractures after kyphoplasty several surgeons perform posterior instrumentation of the adjacent vertebrae to protect the posterior wall 34. This can be done using percutaneous posterior instrumentation or with a conventional open technique. Verlaan et al 35 investigated the use of kyphoplasty after posterior instrumentation in burst fractures in 20 patients (mean age 41.8 years). No bone fragment displacement was found. Unsymptomatic cement leakage occurred in 5 cases. Vertebral anterior height could be restored to 91% of the estimated intact height. Nöldge et al 34 performed kyphoplasty with posterior instrumentation in 9 patients with burst fractures. They found a reduction of mean VAS of 6.2 preoperatively to 2.0 after one year. Unfortunately the evidence supporting the additional instrumentation after kyphoplasty is very low. Therefore this procedure is promising, but it has to be evaluated prospectively in the future.

A feared complication of all vertebral augmentation techniques is PMMA cement leakage. The systematic review of the literature by Hulme et al 12 found rates of cement leakage in vertebroplasty of 41% (n = 2,283 levels) and in kyphoplasty of 9% (n = 1,486 levels) of treated vertebrae.

In the presented study we found radiographically confirmed cement leakage in 6.9% of all cases. These results are lesser than average, but cement leakage was only identified on plain radiographs. CT-scans identify more leaks than radiographs by a factor of 1.5 36. In kyphoplasty of the 65 leakages reported in the literature most were paraspinal (48%), intradiscal (38%), epidural (11%), pulmorary (1.5%) and foraminal (1.5%) 12. Paraspinal and intradiscal leakages generally are asymptomatic, even though intradiscal leakage is blamed to promote adjacent fractures 27. Intradural cement leakage has only been described for vertebroplasty so far 37, but epidural leakage had devastating neurological effects both in vertebroplasty 38 and in kyphoplasty 39. These complications can require immediate surgical intervention with decompression and, if possible, removal of the cement causing stenosis 40.

Pulmonary embolism of PMMA cement was found in 4.6% of the 65 patients treated with either vertebroplasty (n = 88) or kyphoplasty (n = 25) by Choe et al 14. No correlation between the occurrence of pulmonary cement embolism and the type of procedure was found. This is remarkable since kyphoplasty has a much lesser rate of cement leakage than vertebroplasty 12. Pulmonary cement embolism rarely requires intervention and mostly remains asymptomatic. Often they are accidental findings in chest radiographs, but there are several case reports with clinically relevant cement embolisms. Jang et al 41 presented three cases of cement embolisms after vertebroplasty, of which two had mild dyspnoea. No pulmonary perfusion defects were seen and intervention was not necessary. More severe is the case of François et al 42 with a large PMMA cement embolus floating in the right pulmonary artery menacing pulmonary function. Therefore the embolus was removed by open heart surgery. In a case report by Yoo et al 43 a 5 cm long PMMA cement embolus in the right pulmonary artery after vertebroplasty lead to acute respiratory distress syndrome, requiring intensive care treatment and open embolectomy under cardiopulmonary bypass. The patient did not recover and died ten days after vertebroplasty. A further fatal pulmonary embolism after vertebroplasty has been described by Monticelli et al 44. There is only one report on cement embolism after kyphoplasty by Garfin et al. 45. Reports on lethal pulmonary cement embolism after kyphoplasty do not exist.

With proper surgical techniques the risk of cement leakage can be minimized. Correct placement of the balloon, high viscosity of the PMMA cement, controlled application of the cement in to the vertebra, and limitation of the applied volume reduce the risk of leakage. A popular technique to reduce the risk of cement leakage in kyphoplasty is the eggshell-technique, where after primary reduction with the balloon a small amount of doughy cement is applied into the cavity followed by re-inflation of the balloon 46. Using this technique a cement "eggshell" prevents further leakage when the rest of the cement is applied with radiographic control.

The severity of pulmonary PMMA cement embolism and the urgent need of immediate decompression in relevant spinal stenosis after cement leakage, questions the common practice of vertebroplasty and kyphoplasty in an outpatient practice, without any spinal surgeon on call, without any available operating theatres, and without an intensive care unit. Even though fatal embolisms are few kyphoplasty should not be regarded as minor intervention which can be performed without the availability of the above mentioned conditions and requirements.

Only two cases of infections after kyphoplasty have been described so far in the available literature. Nussbaum et al 47 found two cases of infection (discitis/osteomyelitis) in the large "Food and Drug Administration" (FDA) database for adverse events related to kyphoplasty devices. The two presented cases with infection after kyphoplasty in this study reflect the risk of infection being adherent to every surgical intervention 48. To our knowledge with vertebroplasty only seven cases have been described with postoperative infections. The vertebroplasty pioneers Deramond et al 7 presented one case of postoperative spondylitis in an immunosuppressed patient, which could be treated successfully by bedrest and antibiotics. Kallmes et al 49 described a case of a postoperative infection in an immunocompromised patient. Another case was presented by Yu et al 50 with severe pyogenic spondylitis one month after vertebroplasty which was performed while the patient had urinary tract infection. The treatment was surgical with anterior corporectomy and bisegmental fusion after multisegmental posterior instrumentation. A further case of spondylitis after vertebroplasty was presented by Schmid et al 51, which treated the patient conservatively with a 3-month antibiotic regimen. Walker et al 52 and Mummameni et al 53 present two additional cases of spondylitis after vertebroplasty treated by anterior corporectomy and multisegemental fusion. Alfonso Olmos et al 54 report a case of spondylitis after vertebroplasty requiring corporectomy and 360 degrees fusion. An unusual case report of spondylitis after vertebroplasty with epidural abscesses containing mycobacterium tuberculosis was published by Bouvresse et al 55. This patient was under immunosuppression because of a liver transplantation and an inactive tuberculous lesion was obviously activated. Successful treatment implied posterior decompression, abscess evacuation, and long-term antituberculotic therapy.

Even though the risk of infection after kyphoplasty is extremely low, infection does occur, as we have demonstrated in this investigation. The available clinical experience from infection after vertebroplasty stresses the safety limits of vertebral cement augmentation in immunocompromised patients. In these cases a standardized antibiotic prophylaxis is recommended. Nevertheless the rate of postoperative infections is lower both in kyphoplasty and in vertebroplasty than in any other spinal surgical procedure in general 48.