European Radiology

, Volume 17, Issue 9, pp 2248–2252

The intravertebral vacuum phenomen as specific sign of osteonecrosis in vertebral compression fractures: results from a radiological and histological study


    • Department of RadiologyUniversity of Cologne
  • Andreas Appelt
    • Department of Trauma SurgeryUniversity of Heidelberg
  • Irina Berger
    • Department of PathologyUniversity of Heidelberg
  • Martin Baier
    • Department of Trauma SurgeryUniversity of Heidelberg
  • Peter-Jürgen Meeder
    • Department of Trauma SurgeryUniversity of Heidelberg
  • Ingo Grafe
    • Department of Internal Medicine, Division of OsteologyUniversity of Heidelberg
  • Katharina DaFonseca
    • Department of Trauma SurgeryUniversity of Heidelberg
  • Gerd Nöldge
    • Department of RadiologyUniversity of Heidelberg
  • Christian Kasperk
    • Department of Internal Medicine, Division of OsteologyUniversity of Heidelberg

DOI: 10.1007/s00330-007-0684-0

Cite this article as:
Libicher, M., Appelt, A., Berger, I. et al. Eur Radiol (2007) 17: 2248. doi:10.1007/s00330-007-0684-0


This study investigated the prevalence of the intravertebral vacuum phenomenon (IVP) and osteonecroses in vertebral compression fractures (VCFs). We therefore performed an histological analysis of biopsies obtained from VCFs prior to balloon kyphoplasty. Computed tomography (CT) scans were reviewed regarding the presence of an IVP (i.e. cleft sign, Kümmell disease). We reviewed the data of 266 consecutive patients treated by balloon kyphoplasty in 501 procedures from 2002 to 2004. From 180 patients (68%) we obtained adequate bone tissue for histological evaluation. Biopsy specimens were analysed regarding the presence of osteoporosis, infection, malignancy and osteonecrosis. CT scans of all 180 patients were reviewed for presence of an IVP. Histological examination revealed 135 (75%) osteoporoses, 20 (11%) neoplasms, 12 (7%) trauma cases and 13 (7%) osteonecroses. An IVP was present in 12 (7%) patients. There was a significant association of osteonecrosis and IVP (P < 0.0001). Eleven of 12 patients with a vacuum phenomenon showed an osteonecrosis on histology, wheras 11 of 13 patients with osteonecrosis showed an IVP on CT. The IVP is a specific sign of osteonecrosis in vertebral compression fractures (sensitivity 85%, specificity 99%, positive predictive value 91%). Our findings strongly support the thesis that an IVP indicates local bone ischemia associated with a non-healing vertebral collapse and pseudarthrosis.


Intravertebral vacuum phenomenonKümmell diseaseOsteonecrosis of the vertebral bodyBalloon kyphoplasty


The presence of an intravertebral gas cleft (i.e. cleft sign, intravertebral vacuum phenomenon) in vertebral compression fractures (VCFs) is considered indicative of osteonecrosis, also referred to as “Kümmell disease” [13]. It is regarded as a specific sign for ischemic vertebral collapse because it is not known to occur in malignant or inflammatory disease [3]. The two most important risk factors for development of an intravertebral vacuum phenomenon (IVP) are osteoporotic vertebral fractures and steroid medication [4].

However, to the best of our knowledge, there are no studies that adressed the question of a common underlying histological process in a larger population of VCFs. Therefore, the specificity and sensitivity of the intravertebral vacuum phenomenon as indicator of osteonecrosis is not known yet.

This is due to the fact that conservative or surgical treatment of presumed osteoporotic VCF usually does not require bone biopsies. This has changed with the development ballon kyphoplasty as a new treatment option for osteoporotic VCF [57]. Balloon kyphoplasty offers the opportunity of a bone biopsy to confirm the diagnosis or to detect a prior unknown malignant or metabolic vertebral disease [8].

We therefore evaluated bone biopsies of patients with VCF prior to balloon kyphoplasty and compared histological diagnosis with CT findings. Thus, we could measure the frequency of osteonecrosis and intravertebral gas in our study population. Additionally, we could evaluate sensitivity and specificity of the vacuum phenomenon as indicator for osteonecrosis.

Materials and methods


From 2002 to 2004 we performed 501 balloon kyphoplasty procedures in 266 consecutive patients. The 266 cases included 179 osteoporotic (67%), 67 traumatic (25%) and 20 malignant (8%) VCFs. We were able to obtain adequate bone biopsies for histological evaluation in 180 of 266 patients (68%).

This study group of 180 patients (125 women, 55 men; 65 ± 10 years) included 135 osteoporotic (75%), 20 malignant (11%), 13 osteonecrotic (7%) and 12 traumatic (7%) VCFs proven by bone biopsy. There was no patient with inflammatory spinal disease.

The largest group of osteoporotic VCFs included chronic and acute fractures, wheras the trauma cases did not contain patients with an underlying osteoporosis. Diagnosis of osteoporosis prior to balloon kyphoplasty was based on bone density measurements of all 180 patients. We used DXA scan technology (Hologic Discovery) by analysing lumbar spine and left hip. In addition, any secondary causes for osteoporosis were excluded by an endocrinological work up. All patients were evaluated by an interdisciplinary team of endocrinologists, orthopedic surgeons and radiologists and were recruited by a standardised approach [5, 6]. For this retrospective study we did not require an approval from our institutional review board since all patients considered for kyphoplasty are managed routinely in the described way of imaging and clinical work-up.

Balloon kyphoplasty and bone biopsy

After anesthesia and intubation, patients were positioned in the prone position with lordosis of the lumbar spine. Balloon kyphoplasty was performed with fluoroscopy using a bilateral transpedicular approach. Two cannulae were inserted transpedicularly and cylindric bone biopsies with 3-mm diameter were obtained to allow insertion of the balloon tamps. Each cylinder was immediately fixated in formalin and labelled according the location of the vertebra. Two cavities were then created by two balloon tamps that were inserted through the cannulae. After removal of the balloon tamps, polymethylmethacrylate (PMMA; KyphX; Kyphon) was injected into the created cavities. The technical procedure itself followed the same standardised approach published previously in detail [911].

Histomorphological analysis

The cylindrical biopsy specimen measured 0.8–0.5 cm in length and 0.3 cm in diameter. The tissue was fixed in formalin and embedded in Techvit 9100. Three-micrometre slices were cut and mounted on glass slides (Mikrotom HM 355). Slides were stained with Masson-Goldner, Kossa, modified Giemsa and haemotoxilin-eosin stains to visualise new bone formation, osteonecrosis and define bone turnover. All biopsies were reviewed by a pathologist experienced in musculoskeletal diseases (I.B.). The pathologist was blinded to the radiological findings and the presence of an intravertebral vacuum phenomenon.

Computed tomography (CT)

All patients were examined preoperatively by CT using an identical imaging protocol [12, 13]. CT scans of all 180 patients were reviewed for the presence of an intravertebral vacuum phenomenon within the treated vertebrae by an experienced musculoskeletal radiologist (M.L.). Because of the unequivocal diagnosis of an intravertebral gas accumulation, there was no consensus reading necessary. The reading radiologist was blinded to the histological diagnosis.

Statistical analysis

Variables (osteonecrosis, vacuum phenomenon, sex and age distribution) were compared using the Fisher’s exact test or the Wilcoxon rank sum test with a significance level of P < 0.05. Because some patients had several VCFs, we included only one fracture for each patient to avoid statistical independence problems [14]. Statistical analysis was performed using JMP (SAS Inst., Calif., USA).



Histological examination of all 180 patients revealed 135 (75%) osteoporoses, 20 (11%) neoplasms, 13 (7%) osteonecroses and 12 (7%) trauma cases. There was no case of inflammatory disease. Data are summarized in Table 1.
Table 1

Biopsy proven diagnosis of all 180 patients



















The group of 135 osteoporotic fractures included acute and chronic fractures. Diagnosis of osteoporosis was established by bone density measurements. The 20 patients with malignant disease included 18 patients with multiple myeloma and two patients with metastatic disease. The malignant nature of the VCF was suspected before biopsy because of the patients’ history and the presence of osteolytic lesions on CT. There was no unexpected diagnosis of malignancy in our study population.

The 13 patients with diagnosis of an aseptic osteonecrosis included nine women and four men [67 ± 9 years (mean±SD)] with a mean age of 66 ± 8 years and 68 ± 11 years, respectively (P > 0.05).

The group of 167 patients without osteonecrosis showed a comparable sex and age distribution (116 women, 51 men, 63 ± 12 years) without significant difference to the patients with osteonecrosis (P > 0.05).

CT results

CT examinations prior to balloon kyphoplasty showed an intravertebral vacuum phenomenon in 12 of 180 patients (7%). The remaining 168 patients (93%) did not show intravertebral gas within the collapsed vertebrae on CT.

Association of the intravertebral vacuum phenomenon with osteonecrosis

We found in 11 of 12 patients with a positive vacuum phenomenon an ischemic osteonecrosis; one case was negative, possibly due to a sampling error. There were 168 patients without a intravertebral vacuum phenomenon (IVP) and two of them showed an ischemic necrosis on histology (Figs. 1, 2). One patient suffered from an osteoporotic VCF due to chronic steroid medication. The other patient showed a collapsed vertebra with sclerosis due to radiation of the involved spine.
Fig. 1

Intravertebral vacuum phenomenon. This 80-year-old woman presented with severe osteoporotic VCFs at levels L1, L2 and L4. a There is a horizontal intravertebral vacuum phenomenon (white arrows) present on lumbar vertebra L1 on coronal CT images. b On MRI, the gas-containing cleft was filled with fluid surroundoud by a hypointense rim (white arrowheads) as demonstrated on coronal STIR images. Biopsy of the L1 vertebral segment revealed an osteonecrosis shown in Fig. 2
Fig. 2

Biopsy specimen of the same patient with an intravertebral vacuum phenomenon. The histological slide shows necrotic fragments (black arrows) as well as areas of fibrosis. There are adjacent vital bone trabeculae (white arrowhead) with restricted osteoid synthesis at the bone surface (asterisk) (Masson-Goldner stain; 7.6×)

In summary, we detected a prevalence of 7% osteonecroses in our study population (13 of 180 patients). In 11 of these 13 patients with osteonecroses a gas containing cleft was apparent on CT, indicating a sensitivity of 85%. Only one of the 167 patients without osteonecroses showed a IVP, indicating a specificity of 99%. The resulting positive predictive value (PPV) of the IVP as indicator for osteonecrosis is 91%. The association of osteonecroses and the IVP was statistically significant (P<0.0001, Fisher’s exact test). Data are summarised in Table 2.
Table 2

Contingency table (prevalence of osteonecroses and an intravertebral gas cleft). There is a significant association of osteonecrosis and the intravertebral vacuum phenomenon (P<0.0001, Fisher’s exact test;  n=180 patients)

Radiology (CT)




No osteonecrosis

Intravertebral gas




No intravertebral gas









Since the description of Maldague et al. [3], a gas-containing cleft within a collapsed vertebra is considered a sign of ischemic necrosis. However, the authors could only prove the diagnosis in one of ten patients. There are different synonyms for this sign: “intravertebral vacuum cleft” [3], “intravertebral vacuum sign” [15], “intravertebral vacuum phenomenon” [4, 16, 17], “linear intravertebral vacuum” [18], “intravertebral vacuum cleft sign” [19], “intravertebral clefts” [20, 21], “intraosseous vacuum phenomenon” [22], or “Kümmell disease” [1, 23].

On his initial report in 1895, Kümmell described a condition of post-traumatic kyphosis developing months to years after a symptom-free period, but did not mention the presence of intravertebral gas because his findings were made before discovery of radiography [1]. However, this characteristic radiological finding was integrated in this still poorly understood entity and named after the first description of Kümmel. Another eponym is Calvé-Kümmell-Verneuil disease (

The cleft can be detected on radiographs as horizontal accumulation of gas (90% nitrogen) along the fracture line [24]. In some cases, the cleft only appears in extension stress views and disappearing in flexion. On magnetic resonance imaging (MRI) the signal intensity might differ from the expected low signal intensity band. After prolonged supine positioning, the clefts can be filled with fluid, resulting in a high intensity band on T2-weighted sequences [25].

However, the presence of a fluid collection on MRI is not indicative of osteonecrosis [14, 26, 27].

There are known risk factors for developing intravertebral vacuum phenomena in VCF. Stäbler et al. [4] found an increasing age and osteoporosis as risk factors in their study of 84 patients. The age distribution of their patients (67 ± 9 years) is concordant to our study population (68.3 ± 10.5 years).

According to the study of Maldague et al. [3], we also found osteoporosis, trauma, steroid medication and radiation therapy in the history of patients with osteonecrosis.

The incidence of an intravertebral cleft sign is dependent on the patients that are recruited for a specific study. Kumpan et al. [17] calculated a frequency of 0.85% by reviewing radiographs of 2,000 patients with VCFs. When CT is used, a tenfold higher frequency can be expected as shown by Stäbler et al. [4]. In selected patients with only osteoporotic VCFs that were recruited for cement augmentation (vertebroplasty), an even higher frequency of 10–48% is possible [2022].

However, to the best of our knowledge, there is no study that directly compared patients with gas-containing vertebral clefts and histology. In our study population, we observed an incidence of 7% for the intravertebral vacuum phenomenon (12 of 180 patients). Biopsies revealed the same incidence of 7% (13 of 180 patients) of osteonecrosis.

It is not clear, if ischemic osteonecrosis or the vertebral collapse is the first step in the sequence of developing an intravertebral cleft. Both pathways may coexist. Especially in elderly patients, a vertebral compression fracture seems to represent the initial event, followed by a delayed fracture healing due to impaired vascular proliferation, which is followed by osteonecrosis. Flexion and extension forces along the horizontal fracture plane might prevent union of the fragments. Thus, the gas-filled cleft represents a type of vertebral pseudarthrosis, i.e. fracture non-union. The dynamic mobility can be demonstrated by a more pronounced gas accumulation in extension that leads to a partial restoration of vertebral body height [3, 20, 28, 29].

One limitation of our study is that we did not examine patients with inflammatory vertebral disease because this is a contraindication for balloon kyphoplasty. However, the distribution of gas differs from the typical cleft pattern seen with osteonecrosis in VCFs [2, 17, 19, 30].

Another limitation is the retrospective design of our study. However, because of the standardised work-up of all patients that are scheduled for balloon kyphoplasty, there was no additional bias in patient selection. Our study population showed comparable risk factors and age distribution as published in the cited publications. Additionally, the radiological and pathological reading was performed in a blinded fashion regarding the presence of a gas containing cleft or an osteonecrosis.

In conclusion, the prevalence of aseptic osteonecrosis and gas-containing clefts was 7% in our study population of 180 patients. There was a significant association of osteonecrosis and the vacuum phenomenon. The IVP is a specific sign of osteonecrosis in vertebral compression fractures (sensitivity 85%, specificity 99%, PPV 91%). Therefore, our findings strongly support the thesis that an intravertebral vacuum cleft indicates local bone ischemia associated with a non-healing vertebral collapse and pseudarthrosis.

Copyright information

© Springer-Verlag 2007