Objectives. To describe and test a model to explain the biomechanical basis for persistent pain after compression fractures of the vertebral body.
Methods. The biomechanics model was derived axiomatically from a consideration of the anatomy of vertebral column when affected by compression fractures. Proof of principle was provided by performing controlled diagnostic blocks in six patients.
Results. The biomechanics model shows that the posterior elements of the vertebral column must subluxate cephalad or caudad in response to deformity of a vertebral body. The model implies that pain may arise from the posterior elements, and predicts that anesthetizing the posterior elements should relieve the pain of compression fractures. Six cases are described in which controlled medial branch blocks relieved the pain of compression fractures of thoracic or lumbar vertebral bodies.
Conclusions. In some patients with vertebral compression fractures, the pain may arise from posterior elements and not the fracture itself. This phenomenon has implications for the interpretation of the outcomes of vertebroplasty in both the active and control arms of sham-controlled studies.
Compression fractures of a lumbar or thoracic vertebral body are one of the complications of spinal osteoporosis. In younger individuals, they may occur as a result of trauma. These fractures may be asymptomatic, or the patient may report spinal pain in the region of the fracture. That pain may be temporary, or it may persist.
For self-limiting pain, the mechanism is probably inflammation around the fracture site. More elusive is the mechanism of persistent pain following vertebral body fracture. Two schools of implied thought apply. Both are implied because neither has been outrightly articulated.
One school implies that it is the fracture site itself that is the source of pain. Although not expressly stated, this mechanism is implied by those who use vertebroplasty to treat the pain. The injection of cement into the vertebral body ostensibly stabilizes the fracture, and relieves pain by preventing micromovements at the fracture site. An alternative, or additional, explanation is that the heat generated by the cement upon injection coagulates nerve endings in the vertebral body that mediate the pain. However, no direct evidence of either if these mechanisms has been forthcoming. They remain post hoc explanations of how pain is relieved by vertebroplasty.
The other school is less specific. It implies that vertebral body fractures compromise the biomechanics of the affected region; whereupon the resultant pain is biomechanical in nature, but no further explanation has been forthcoming. The nature of the biomechanical disturbance and how it produces pain has not been stated.
The present study seeks to redress the deficiency of this latter school of thought. In the first instance, an explanatory biomechanical model was developed. Second, six case studies were recruited to test the predictions of this model.
The model was developed in the manner of a theorem. Step by step, the biomechanical consequences of a vertebral body were derived logically by considering and applying anatomical facts.
Compression of a vertebral body results in loss of height in the anterior column of the spine. This loss of height may be unilateral in the case of lateral wedge fractures, or bilateral in the case of anterior wedge fractures or vertical compression fractures.
When the fracture is isolated to the vertebral body, there is no loss of height in the posterior elements. This results in a dissonance between the length of the posterior column and the length of the anterior column. The vertebral column must adjust in order to accommodate this dissonance. The nature of this adjustment, and how it affects the posterior elements, will differ depending on whether the fracture is wedge or vertical in nature.
In the case of an anterior wedge fracture, the vertebral body above the fractured vertebra must tilt forward (into kyphosis) if it is to remain in apposition with the top of the affected vertebra below (Figure 1). Meanwhile, the posterior elements cannot elongate in order to accommodate the tilt. All that they can do is separate. Consequently, its inferior articular processes must subluxate cephalad and forwards, in the direction of the tilt. The axial subluxation distracts the zygapophysial joint, and the forward subluxation results in weight being transmitted through point contact between the inferior articular process and the apex of the superior articular process below (Figure 1). If the angulation is of sufficient magnitude, either or both of two mechanisms might cause pain. The cephalad distraction could strain the capsule of the zygapophysial joint, or even disrupt it. The point contact with the apex of the superior articular process could cause periosteal irritation and pseudarthrosis formation. Alternatively or additionally, the disruption of the posterior elements might affect the behavior of the posterior back muscles, and result in tension myalgia.
In the case of vertical compression, angulation of the affected vertebral body does not occur, and the anterior column does not rotate; but the anterior column loses height. Consequently, the posterior elements adjust in a different manner from that after wedge fractures. The loss of axial height in the anterior column obliges the posterior elements to subluxate inferiorly (Figure 2). The inferior articular process of the fractured vertebra distracts its zygapophysial joint inferiorly, and the lamina settles on the apex of the superior articular process below (Figure 2). Pain could result from strain or disruption of the zygapophysial joint or from periosteal irritation and pseudarthrosis between the lamina and superior articular process.
According to this model, in the case of anterior wedge fractures, the zygapophysial joint that is affected is the one above the fractured vertebra, whereas in vertical fractures, it is the joint below. However, in lateral wedge fractures, aspects of both mechanisms could apply. Because of the coronal tilt that is produced by the lateral wedge fracture, joints on the contralateral side would be distracted, below or above the affected segment or both; and joints on the ipsilateral side could be compressed, below or above the affected segment, or both.
Irrespective of the actual segmental location of the biomechanical change, this model predicts that pain arises not from the vertebral body, but from the affected posterior elements. This possibility can be tested by applying controlled local anesthetic blocks of the medial branches of the dorsal rami at or above the affected segment. If the predictions of the model are correct, such blocks would relieve patients of their pain. If the predictions are not correct, or do not apply in a given patient, then medial branch blocks will not relieve their pain.
Patient 1 is a 35-year-old woman who developed posterior thoracic spinal pain when leaning forwards at work. Initial radiographs were reported as normal, and she was treated with tramadol. When the pain persisted, a bone scan revealed a compression fracture of the T10 vertebral body, which was confirmed by MRI. She consulted a spine surgeon, who found no indication for surgical intervention. Her general practitioner continued treatment with oxycodone. Physical therapy was of no benefit. A pain clinic recommended continuing the analgesic regimen. A meeting of spine surgeons, rheumatologists, and radiologists offered no other recommendations. Vertebroplasty was considered unlikely to help after two years since the onset of symptoms.
The conjecture was explored that this patient's pain might arise in the posterior elements of her thoracic spine. Accordingly, controlled, diagnostic blocks were performed of her T8 and T9 medial branches bilaterally (Figure 3), according to the guidelines prescribed by the International Spine Intervention Society (ISIS) . These blocks completely relieved her pain. Repeat blocks again relieved her pain (Figure 4). Subsequently, thermal radiofrequency neurotomy of her T8 and T9 medial branches relieved her pain.
Patient 2 is a 75-year-old lady with spinal osteoporosis who suffered a severe compression fracture of her L3 vertebral body. She was admitted to hospital for bed rest and possible management. There were no indications for surgery, and her compression was too severe for vertebroplasty. Her pain persisted, and was managed with tramadol and oxycodone. After 7 weeks in the hospital, her pain had not improved, and she could barely ambulate to the toilet. Her pain was principally in her anteromedial left thigh. The orthopedic and neurosurgery team considered that this might be L3 radicular pain and requested a transforaminal injection of steroids. This was declined by the interventional pain physician on the grounds that her pain had no features of radicular pain, and appeared more likely to be somatic referred pain. Examination of the hip was normal.
Controlled diagnostic blocks of the L2 and L3 medial branches were performed according to ISIS guidelines  (Figure 5). These blocks completely relieved her pain, and she was able to walk around the ward with no pain (Figure 4). Repeat blocks, performed four days later, again relieved her pain, sufficiently so, that she could be discharged home. This patient was scheduled for lumbar medial branch radiofrequency neurotomy, but on presentation, she reported that her pain had remained only mild since the blocks, and that is was manageable with paracetamol alone. Neurotomy was not undertaken.
Patient 3 is an 81-year-old man who developed sudden pain across his lower abdomen while he was moving a heavy, old television set. He had persisting lower abdominal pain, which was aggravated by activities such as dancing and gardening. These activities did not precipitate pain in his back, but he did have intermittent back pain in response to sitting for prolonged periods. An ultrasound scan, arranged by his general practitioner, excluded an inguinal hernia, and he was referred for a specialist opinion 5 months after the onset of symptoms. The possibilities considered were that he may have suffered from a compression fracture at the thoracolumbar junction, and that his pain constituted somatic referred pain into the lower abdomen. Plain radiographs demonstrated significant anterior wedging of the L1 vertebral body. An MRI revealed a marked biconcave compression fracture of L1, with residual bone edema, but an interventional radiologist was of the opinion that the degree of compression was such that vertebroplasty was unlikely to help.
Controlled diagnostic blocks of the right and left T12 and L1 medial branches were performed according to ISIS guidelines [1,2] (Figure 6). These blocks completely relieved his pain, and repeat blocks, performed 6 weeks later (Figure 4), also relieved his pain. He underwent treatment with lumbar medial branch thermal radiofrequency neurotomy . At the time of treatment, his pain was perceived predominantly on the right, and only the right T12 and L1 medial branches were treated. The neurotomy resulted in almost complete and ongoing relief of pain. He experiences intermittent pain of low intensity only in response to vigorous activity, and he has been able to resume dancing and gardening. The residual pain could potentially be relieved by neurotomy of the left T12 and L1 medial branches, but he has insufficient pain to justify this intervention.
Patient 4 is a 26-year-old man who fell backwards into a digger bucket when he was 19 while working as a builder's apprentice. X-rays revealed a moderate compression fracture of T12 and minor compression of T8. He had constant pain in the midline of his back at the level of the thoracolumbar junction since the original injury, and he had been unable to continue working as a builder. His pain was investigated by an orthopedic surgeon, who arranged an MRI, a radionuclide bone scan, and blood tests. The only abnormality found was the compression deformity of T12 ,and it was felt that no cause for the patient's persisting thoracolumbar pain had been identified.
Diagnostic blocks of the right and left T10 and T11 medial branches were performed according to ISIS guidelines , but his pain was relieved at rest only, and for a period shorter than the expected duration of action of the local anesthetic that was used. Subsequent diagnostic blocks of the right and left T11 and T12 medial branches (Figure 7) relieved his pain completely, and repeat blocks again relieved his pain (Figure 4). His pain was subsequently relieved by T11, T12 medial branch thermal radiofrequency neurotomy.
Patient 5 is a 51-year-old farmer who presented with bilateral low abdominal and groin pain, worse on the left. The pain arose several weeks previously after an episode of shearing sheep in a flexed posture for many hours. Provoked by physical activity, the pain would initially be sharp and shooting in quality, but then replaced by residual dull aching that would last for hours or days. Physical examination revealed no neurologic abnormalities, and no remarkable physical signs other than tenderness in the left iliac fossa. Pressing this region reproduced his pain. Spinal range of movement was normal. Previous assessment by a gastroenterologist, including abdominal CT scan, ultrasound, and colonoscopy, had revealed no cause for his pain.
This patient had a history, 8 years previously of a motor vehicle accident, in which he sustained multiple fractures of his pelvis, and compression fractures of the vertebral bodies T12 and L1, with T12 sustaining more compression than L1. Initially, he suffered pain in the low back, abdomen, and pelvis, but as time passed, he was left only with the complaint of episodic abdominal pain. An MRI scan, taken after the recent presentation, revealed longstanding compression fracture of the T12 vertebral body, and mild compression of the L1 vertebral body.
Diagnostic blocks were performed (Figure 8). For the first block, the left T11, T12, and L1 medial branches were targeted. This resulted in an immediate decrease in his index groin pain from 80/100 to zero, for a period concordant with the expected duration of action of the local anesthetic used. On the second occasion, blocks were performed at the T12 and L1 nerves only, and again his index pain fell from 80/100 to zero, with restoration of his activities of daily living for a period concordant with the local anesthetic used.
In the light of these responses, thermal radiofrequency neurotomy of the left T12 and L1 medial branches was undertaken using a 16-gauge electrode, according to the protocol of the International Spine Intervention Society . Following the operation, the patient's pain fell to zero, with full restoration of his activities of daily living. At follow-up 18 months postprocedure, he remains pain-free, and fully able to carry out his activities as a farmer.
Patient 6 is a 58-year-old caretaker who sustained a heavy fall into a seated position some two years previously, and developed excruciating pain in her back, radiating into her abdomen. Investigation 10 days later revealed a compression fracture of her L1 vertebral body (Figure 9). The pain persisted. In quality, the pain was deep and aching, located in the lower thoracic and thoracolumbar region, on the left, with episodic sharp, lancinating pain around her lower ribs. Because of the pain, she was unable to work, and could not stand straight, dry her hair, hang out washing, go hiking, or sit for prolonged periods. Six months after her accident, a local pain clinic had not been able to help.
On examination, she stood with a stoop, leaning to the left, and with a thoracic kyphosis. She walked hesitantly. Spine extension was significantly restricted. There were no neurological findings in the lower limbs. She was tender on springing the spinal column from the level of T7 down to L2. An MRI scan revealed a longstanding compression deformity of the superior endplate of the L1 vertebral body.
Diagnostic blocks of the left T11 and T12 medial branches were performed. After the first block, her pain decreased from 60/100 to zero, with full restoration of activities of daily living. The same result was achieved when the blocks were repeated. Both responses were concordant with the expected duration of the local anesthetic used. Thermal radiofrequency neurotomy of the left T11 and T12 medial branches abolished her pain and restored her activities of daily living. She remained with some residual low back discomfort, but far from disabling.
The biomechanics model advanced in this study is based on a self-evident truth. Vertebrae are not elastic; they are rigid bodies. Consequently, in the face of loss of height in the anterior column, the posterior elements must tilt or subluxate in proportion to that loss of height. This model provides an explanation of the hitherto ambiguous “biomechanical” consequences of vertebral body fractures.
The model does not necessarily explain the mechanism of pain in all cases of vertebral body fractures. There may be additional biomechanical aberrations that the model has not addressed. However, it does provide a possible explanation for some cases of pain following vertebral body fractures. Moreover, that explanation is testable clinically.
Controlled medial branch blocks are an innocuous, readily applied diagnostic test for pain stemming from the zygapophysial joints and the medial back muscles. A positive response does not, of itself, implicate any particular of these structures, but it does place the source of pain among the posterior elements and therefore not in the vertebral body.
The six cases reported illustrate this principle in practice. The cases were fortuitous. None of the three centers from which they were drawn specializes in osteoporosis or spinal fractures. All were small pain clinics dealing largely with nonspecific back pain. In the patients described, medial branch blocks were performed because other measures had failed to relieve the patient's pain, and, because they were something that the investigators could offer. Consequently, the cases are not intended to serve as a measure of the prevalence of posterior element pain following vertebral body fractures. That would have to be tested by clinics that had access to more representative populations. Nevertheless, the cases do serve as proof of principle for the biomechanics model.
The cases reported serve to draw attention away from the anterior elements and to the posterior elements as possible sources of pain associated with vertebral body fractures. On the one hand, this opens up the possibility of providing a definitive diagnosis, and the prospect of treatment by radiofrequency medial branch neurotomy, for patients who face no proven options for treatment.
On the other hand, the present model notionally offends the current fashion for vertebroplasty as a means of treating the pain of vertebral body fractures. The rationale for vertebroplasty implies a source of pain in the anterior elements, not the posterior elements. But vertebroplasty is not universally successful. Perhaps there are patients with anterior column pain and patients with posterior column pain. The latter would not benefit from vertebroplasty.
Yet there is a more intriguing consideration. Vertebroplasty involves obtaining access to the vertebral body by driving a cannula through the pedicles of the fractured vertebra. It may be more than coincidental that this route of access passes through the zygapophysial joint, and, in some trajectories, through the course of the medial branch. It may be that vertebroplasty achieves relief of pain, not by affecting the vertebral body, but by inadvertently disrupting the zygapophysial joint compromised by the fracture, or its nerve supply.
This consideration is not without relevance to the contemporary dispute about the efficacy of vertebroplasty [4–10]. Two placebo-controlled trials have shown no attributable effect of vertebroplasty [11,12]. Yet in both, the control treatment was an injection of local anesthetic over the insertion site of the vertebroplasty cannula. Issues of precision aside, such injections would have been tantamount to medial branch blocks, as used in the present study. Therefore, rather than a strict placebo, these injections might unwittingly have actively relieved the patients of pain stemming from the posterior elements behind their vertebral body fractures. In that event, the controlled trials provide circumstantial evidence of the model proposed in this study.