Association Between Vertebral Segmental Artery Occlusion and Kummell Disease: A Prospective Cohort Study Based On MRA


 Purpose

We designed a prospective cohort study based on magnetic resonance angiography (MRA) to identify whether the degree of Kummell disease (KD) vertebral artery occlusion is more severe than that of simple vertebral compression fractures.
Methods

We enrolled elderly patients with VCFs who met the established criteria from January 2019. MRA was used to determine the degree of vertebral artery occlusion. We defined the lesion segmental occlusion rate (LSOR) as the sum of the ischemic values of the bilateral vertebral segmental arteries (0 means expedite, 1 means stenosis, 2 means occlusion; range from 0–4). The average LSOR is the sum of LSORs divided by the number of vertebrae. Follow-up outcomes included VAS and ODI scores after surgery. X-rays were re-examined at 1 year after surgery to determine whether the vertebral body had recollapsed.
Results

25 cases of KD segments and 37 cases of non-KD segments were included. The average LSOR in KD segments was significantly higher than that in non-KD segments (1.44 vs 0.32, P<0.01). The recollapse rate of the KD segments after one year was significantly higher than that in non-KD segments (56% vs 27%, P=0.03). In non-KD segments, 57.1% of segments with a high LSOR (1–2) recollapsed, and 20% of segments with a low LSOR (0) recollapsed (P=0.045)
Conclusions

The degree of artery occlusion in the KD segment is more serious than that in the non-KD segment. KD segments or non-KD segments with a high degree of artery occlusion will have a higher recollapse rate.


Introduction
Kummell disease (KD) is a late complication of osteoporotic vertebral compression fractures (VCFs) [1] , rst described by Hermann Kummell in 1895 [2] . KD manifests as low back pain, neurological de cits, or kyphosis that occur several weeks or months after an asymptomatic period after minor trauma [3] . It is characterized by collapse of the vertebral body and intravertebral cleft (IVC) after trauma [4] . The complications of recollapse [5,6] and bone cement leakage [7][8][9] after vertebroplasty in KD patients are signi cantly increased, than that of simple VCFs.
Currently, there are many hypotheses about the pathogenesis of KD [10] , including ischemic osteonecrosis [11][12][13] , air formation [14,15] , changes of bone biomechanics [16,17] and some other hypotheses with supportive clinical or biomechanical evidence [10] . Among them, the hypothesis of ischemic osteonecrosis is the most studied hypothesis and recognized by most researchers [10] .
Researchers believe that the formation of IVC is caused by direct factors, including fat embolism or fracture compression, or secondary factors, such as corticosteroid use, alcoholism, and vasculitis, resulting in the destruction of the blood supply of the vertebral body after the initial minimal trauma. The vertebral body gradually undergoes osteonecrosis, and the vertebral body collapses, which further destroys the blood supply of the vertebral body, forming a vicious cycle that eventually progresses to KD [18] .
Lin et al. [13] reported that in patients with osteoporotic VCFs before vertebroplasty, decreased bone marrow perfusion, as measured by dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI), was associated with IVC formation. Kim et al. [12] found that most patients with IVC have unilateral or bilateral segmental artery occlusion through magnetic resonance angiography (MRA). These studies provided evidence for the hypothesis of vertebral avascular necrosis. However, these studies were singlearm observational studies. We designed a prospective cohort study based on MRA to identify whether the degree of KD vertebral artery occlusion is more severe than that of simple VCFs.

Methods
We enrolled elderly patients with VCFs who met the established criteria from January 2019. The study was approved by the ethics committee of Peking University People's Hospital, number 2019PHB240. All patients had the contents of the study fully explained and signed the informed consent form before enrollment. Inability to cooperate to complete the required follow-up period.

Data Extraction and de nition
The characteristics of the patients were recorded, including age, sex, body mass index (BMI), and preoperative medical history that may be related to the formation of IVC, such as smoking, alcohol abuse, glucocorticoid use and diabetes. The time interval between the development of back pain to the time of MRA, fracture segment, preoperative visual analog scale (VAS) score and Oswestry disability index (ODI) score were recorded.
Follow-up outcomes included VAS and ODI scores at 1 week, 1 month, and 12 months after surgery. Xrays were re-examined at 1 year after surgery to calculate the vertebral compression rate (VCR) and kyphotic angle (KA) to determine whether the vertebral body had recollapsed.
We de ned the lesion segmental occlusion rate (LSOR) as the sum of the ischemic values of the bilateral vertebral segmental arteries (0 means expedite, 1 means stenosis, 2 means occlusion; range from 0-4). The average LSOR is the sum of LSORs divided by the number of vertebrae. We de ned the normal vertebra as no fracture observed by MRI at the time of enrollment, and no collapse found on X-ray reexamination after 1 year.
The diagnosis of KD requires a comprehensive analysis of the patient's medical history, clinical symptoms, and characteristic imaging ndings, as shown in Figure 1.

Radiological Assessment
Two authors individually and independently assessed the imaging data twice to eliminate intraand interobserver bias. If there was a noticeable difference in any result, the two authors and a senior doctor conferred to determine the nal result. We de ned VCR =1-the height of the maximum compression of the vertebral body*2/ (anterior edge height of vertebral body + posterior edge height of vertebral body), as shown in Figure 2. We de ned the KA as the angle formed between the cranial endplates and the caudal endplates of the vertebral body. Two conditions were de ned as a recollapse of the vertebral body [6,19] : (1) ≥ 15% progression of VCR between the immediately postoperative and last follow-up period; (2) ≥ 10° progression in local KA between the immediately postoperative and last follow-up period.

Statistical Analysis
All analyses were performed using IBM SPSS Statistics for Windows, version 26.0 (IBM Corp., Armonk, N.Y., USA). Between-group comparisons of variables were performed by chi-square tests and Fisher's exact tests, whereas between-group comparisons of continuous variables were performed by t-tests or Mann-Whitney U tests.

Results
Characteristics of the population A total of 55 patients and 62 thoracolumbar or lumbar fracture segments (T10-L4) were included in the study, including 25 cases of KD segments and 37 cases of non-KD segments. There was no signi cant difference between the two groups in terms of age, sex, BMI or risk factors for IVC formation, as shown in Table 1.

Results of MRA
In normal vertebra, the average LSOR was 0.27. For KD segments, the LSOR of each segment were categorized as 0 (9/25), 1-2 (11/25), and 3-4 (5/35). The average LSOR was 1.44. In the non-KD segments, the LSOR of each segment were categorized as 0 (30/37) and 1-2 (7/37). The average LSOR was 0.32. We found that there was a signi cant difference in the average LSOR between KD and non-KD segments (P<0.01), and there was no simultaneous occlusion of the bilateral vertebral segmental arteries in non-KD segments.

Postoperative results
The recollapse rate of the KD segments after one year was 56%, and the recollapse rate of the non-KD segments was 27% (P=0.03). In non-KD segments, 57.1% of segments with a high LSOR (1)(2) recollapsed, and 20% of segments with a low LSOR (0) recollapsed (P=0.045). At the nal follow-up, there was no signi cant difference in the VAS and ODI scores between the KD and non-KD groups, as shown in Table 2.

Discussion
To our knowledge, this is the rst prospective cohort study of segmental artery occlusion in KD. Our results suggest that the degree of occlusion of KD segmental arteries was signi cantly higher than that of non-KD segmental arteries. Fractured segments with a high degree of artery occlusion had a signi cantly higher probability of recollapsing after vertebroplasty. There was no signi cant difference in postoperative VAS and ODI scores between KD and non-KD patients.
There are many hypotheses about the pathogenesis of KD formation [4,11,12] , among which ischemic osteonecrosis is the most widely supported and studied hypothesis [10] . Avascular osteonecrosis after VCFs has its own anatomical theory. Ratcliffe [20] used microangiography to study the arterial anatomy of the adult lumbar vertebral body and proposed that the upper and lower anterior segments of the vertebral body are the watershed areas of the vertebral body. These areas are susceptible to ischemia. Prakash [21] also reported that the central anterior branch of the pair of segmental arteries supplies the ventral side of one vertebral body, while the central posterior branch supplies the dorsal side of two adjacent vertebral bodies. Therefore, the dorsal side of the vertebral body receives collateral blood ow, but the ventral side does not; this will theoretically make the ventral part of the vertebral body have a higher risk of blood supply destruction.
Lin et al. [13] reported that in patients with osteoporotic VCFs before vertebroplasty, decreased bone marrow perfusion, as measured by DCE-MRI, was associated with IVC formation. Kim et al. [12] found that most patients with IVC have unilateral or bilateral segmental artery occlusion through MRA. In our study, compared with normal vertebral bodies and simple VCFs, we found that the degree of KD segmental artery occlusion was signi cantly increased. All these results suggest that the destruction of blood supply may be an important reason for the formation of IVC, indicative of KD.
In many research reports, IVC has been suggest to be an important high-risk factor for recollapse after vertebroplasty [9,19,22] . Our results suggest that the recollapse rate of KD segments was signi cantly higher than that of non-KD segments. At the same time, the recollapse rate of VCFs with a high degree of artery occlusion in non-KD segments was also signi cantly increased. Therefore, we believe that in addition to IVC being a high-risk factor for recollapse, segmental artery occlusion is also an important high-risk factor for recollapse after vertebroplasty.
At present, the main treatment for KD is vertebroplasty [23][24][25] . Complications such as recollapse after vertebroplasty are relatively high in patients with KD, and therefore, some clinicians have tried to use short-segment vertebral xation to treat KD [26][27][28] . Our research suggests that there was no signi cant difference in VAS and ODI scores between KD and non-KD patients. This is also consistent with the results of many clinical reports that vertebroplasty is an effective treatment for KD [23][24][25] . There are also studies comparing the prognosis of vertebroplasty and short-segment vertebral xation, and the results showed no signi cant differences in VAS and ODI scores [26][27][28] . We found that no non-KD segments progressed to KD after surgery, regardless of the degree of segmental artery occlusion. This may be because vertebroplasty restores the stability of the vertebral body, disrupts the vicious cycle mechanism leading to KD formation, and avoids the further blood supply damage.
It takes an amount of time for VCF to progress to KD [13] , and the time duration is currently inconclusive.
Our ndings suggest that the average course of disease in KD patients was signi cantly higher than that in non-KD patients. We also had a pretty interesting nding. The average LSOR of KD patients within 1 week, 1 month, 3 months, and over 3 months were 2.71, 1.00, 0.60, and 1.67, respectively. The average LSORs within 1 week and more than 3 months were signi cantly higher than those in the other two stages. For the short course within 1 week, we believe that it is precisely because of the high occlusion rate that VCF rapidly progresses to KD. Regarding the long course of more than 3 months, due to the long course of the disease, the degree of compression of the vertebral body increased, destroyed the blood supply and resulted in aggravation of segmental artery occlusion. Therefore, the destruction of blood supply and vertebral compression fracture are mutually promoting and vicious cycle processes that ultimately produce KD.

Limitations
First, although the number of target segmental arteries was obtained, the number is still small. Second, we could not determine the exact time or sequence of artery occlusion and formation of KD. This would require multiple MRA inspections from the initial time of injury, but this is di cult to achieve.

Conclusion
The initial damage or fracture progression insults the segmental arteries, which eventually leads to osteonecrosis and progresses to KD. Therefore, the degree of artery occlusion in the KD segment is more serious than that in the non-KD segment. At the same time, KD segments or non-KD segments with a high degree of artery occlusion will have a higher recollapse rate, indicating segmental artery occlusion is also an important high-risk factor for recollapse after vertebroplasty. For KD segments without nerve compression, vertebroplasty is still an effective treatment.

Declarations
Funding This study was funded by Ministry of Education Key Laboratory of trauma treatment and nerve regeneration.
Con icts of interest SZ.Z, TY.Z, F.X, DY.Z and BG.J declare that they have no con icting interests Availability of data and material The datasets generated and analyzed during the current study are not publicly available due to the data also forms part of an ongoing study but are available from the corresponding author on reasonable request  Figure 1 Characteristic radiological imaging of KD. An 82-year-old female who had low back pain when she woke up 20 days ago. She visited the emergency department on the same day, X-ray examination revealed old VCF of T12 and L1 (past medical history). Later, the symptoms gradually worsened, and reexamination of X-ray revealed a fresh VCF of L2. Further CT and MR examinations revealed a new VCFs of L2 and IVC formation.

Figure 2
Legends for calculating vertebral compression rate (VCR) and kyphotic angle (KA). Figure A shows the illustration of VCR. VCR =1-the height of the maximum compression of the vertebral body*2/ (anterior edge height of vertebral body + posterior edge height of vertebral body). Figure B shows the illustration of the KA. The KA is de ned as the angle formed between the cranial endplates and the caudal endplates of the vertebral body. Figure C shows the reexamination results of the same vertebral body after 1 year, the progress of KA is greater than 10°, which meets the diagnostic criteria of recollapse. Typical imaging of MRA. Figure A shows the MRA results of compression fractured vertebral body (L1) from the sagittal and horizontal planes: bilateral segmental arteries are expedite. Figure B shows the MRA results of compression fractured vertebral body (L1) from the sagittal and horizontal planes: the artery of the right segment is blocked, and is a signi cant difference between the imaging results of the MRA of the adjacent vertebral body.