In vivo animal models of spinal metastasis
- First Online:
- Cite this article as:
- Cossigny, D. & Quan, G.M.Y. Cancer Metastasis Rev (2012) 31: 99. doi:10.1007/s10555-011-9332-x
- 495 Views
The vertebral column is the commonest site for skeletal metastases, with breast, prostate and lung cancers being the most common primary sources. The spine has structural and neural-protective properties thus involvement by metastatic cancer often causes bony instability and fracture, intractable pain and neurological deficit. In vivo animal models which resemble the human condition are essential in order to improve understanding of the pathophysiology behind the spread of metastatic cancer to the spine and its subsequent local growth and invasion, to enable in-depth analysis of the interaction between host and tumour cells and the molecular processes behind local cancer invasion and barriers to invasion as well as to allow assessment of novel treatment modalities for spinal metastases. This review summarizes the current status of the animal models specifically used for the study of spinal metastasis, their relevance, advantages and limitations, and important considerations for the development of future in vivo animal models.
The spine is the commonest site of metastatic disease to the bony skeleton. The majority of secondary bone cancers originate from breast, prostate and lung cancers which are known for their bone-metastasizing properties and patients with advanced cancer will more often than not develop spinal metastases [1, 2]. Since the vertebral column is essential for structural support of the bony skeleton, destruction by cancer often causes intractable local pain, pathological fracture and spinal deformity. Furthermore, each vertebra encases and protects the spinal cord and segmental nerves, such that metastatic epidural spinal cord compression inevitably leads to radicular pain, weakness, inability to walk and if untreated ultimately causes incontinence, paraparesis or quadraparesis. Palliative surgery in the form of decompression and stabilization has been shown to improve pain, function and the quality of life in selected patients with symptomatic spinal metastases . However, surgery is preferentially reserved for patients anticipated to have greater than 6 months survival, has a 25% overall morbidity which may require prolonged hospitalization and local tumour regrowth causing symptom recurrence is a difficult problem if patients continue to live. Radiotherapy is very commonly used in patients with spinal metastases with the aim of inducing rapid cellular necrosis and stopping tumour progression but the adjacent neural structures have a radiation threshold, moreover radiotherapy is associated with significantly increased surgical wound breakdown and infection rates if administered preoperatively. Advances in chemo- and immunotherapeutic agents have enabled improved local and systemic control of several different cancers types and, depending on the primary cancer, have given cancer patients the potential to survive longer. However, each of these therapies has its own toxic side effects and options for patients with primary or metastatic tumours poorly responsive to current medical treatment modalities are very limited. Furthermore, both chemotherapy and radiotherapy are ineffective in the presence of spinal instability, pathological fracture and neural and spinal cord compression. An optimal and universally accepted treatment algorithm for patients with vertebral metastases remains elusive.
Novel therapies are urgently needed as an adjunct to conventional treatments for patients with spinal metastases in order to delay pathological fracture and paralysis and to prevent local recurrence after surgery. Any improvement in the local control of spinal cancer should ultimately improve pain and neurological symptoms, functional status and quality of life in these patients. The sequence of events that occurs during the process of metastatic cancer establishment in the vertebral column to pathological fracture and spinal cord compression is poorly understood. Reproducible, clinically relevant in vivo animal models are essential in order to improve understanding of the temporospatial pattern and mechanisms of cancer growth in the spine and of its local host environment influencing the spread and barriers to spread of cancer. Moreover, in vivo models are required to develop and assess novel therapies prior to translation into human clinical trials.
The relevance of any animal spinal metastasis model to the human condition assumes a cancer which can establish, grow and invade bone, similar host spinal anatomy and physiology to humans, and the onset of neurological decline with tumour progression and spinal cord compression. Therefore, important considerations include the origin, molecular characteristics and metastatic capability of the experimental cancer cell lines used, the genetic profile of the chosen host animal, the method of tumour establishment in the spine, the assessment of clinical behaviour of the animal with advancing cancer, and the biological behaviour and radiological and histological appearance of the cancer upon establishment and growth in the vertebral column. Specific to the spinal location of the metastasis and the morbidity it causes, neurologic decline observed in the animal model should correlate with the degree of spinal cord compression. This review summarizes the current literature on animal models specifically used in the study of metastatic cancer of the spine. We address their relevance, advantages and limitations and outline future directions in this area of cancer research, highlighting the importance of a reproducible, clinically relevant in vivo animal model of spinal metastasis.
2 Cancer cell lines
Bone forms a unique environment in which cancer cells may establish, compared with other common sites of metastatic spread such as the lung and liver. A key difference is that bone is an inherently stiff tissue, due to the strong extracellular matrix composed of collagens, proteoglycans and glycoproteins forming a complex structural scaffold. It uniquely contains osteoblasts which produce the extracellular bone matrix, and osteoclasts which resorb bone during normal bone remodelling as well as in pathological conditions associated with bone loss or lysis. Important regulators of growth and invasion of bone metastases include vascular endothelial growth factor (VEGF) and basic fibroblast growth factor (bFGF), which are potent tumour growth and pro-angiogenic factors, transforming growth factor-β (TGF-β), parathyroid hormone related protein (PTHrP), receptor activator of nuclear factor kappa-B ligand (RANKL) and interleukin-11 which induce the formation of the osteoclasts responsible for bone destruction, and matrix metalloproteinases which are proteolytic enzymes that degrade the strong extracellular matrix that forms the scaffold of bone . Experimental cancer cell lines used in in vivo models of spinal metastasis should exhibit the molecular characteristics necessary for their establishment and survival in the bone microenvironment in order to mimic the human condition. That is, they should inherently be able to stimulate growth and angiogenesis, degrade collagen fibres, and induce osteoclastogenesis to cause osteolysis and bony destruction.
The cancer cells used in early in vivo models of spinal metastasis were of the same species origin as the animal model. Ushio et al.  used the Walker-256 rat mammary carcinoma cell line in their rat model, Ikeda et al.  used rabbit VX2 virus-induced epidermoid carcinoma cells in their rabbit model, and Arguello et al.  used the B16–G3.26 murine melanoma cell line in their mouse model. Although these cancer cells were able to establish tumours in the spine, the use of cancer cell lines of human origin eliminates the disparity that exists between human cancers versus those of other species. More recently, three well-characterized human breast and lung cancer cell lines have been used in animal spinal metastasis models [5–7]. The MDA-MB-231 breast cancer cell line was derived from malignant pleural effusion in a 51-year-old woman with breast adenocarcinoma, the MT-1 breast cancer cell line was derived from a breast carcinoma surgical specimen xenotransplanted subcutaneously into nude mice, and the PC-14 lung cancer cell line was derived from a patient with pulmonary adenocarcinoma [8–10]. Histological analysis has demonstrated the propensity of each of these cancer cell lines to establish within and destroy bone, infiltrate the vertebral body and extend out into the epidural space to cause spinal cord compression [5–7]. It is important to note, however, that the use of human-derived cancer cells requires the use of immunocompromised animals, which precludes the ability to analyse the important immune responses that may influence the progression of cancer.
3 Host animal species
In vivo animal models used in the study of spinal metastasis
Cancer cell line
In vivo imaging
Walker 256 rat mammary gland carcinoma
Female Wistar rats
Orthotopic inoculation into paraspinal area of T12 or T13 via posterior approach
Grading of weakness of hindlimbs;
1: Slight weakness, hip instability observed only when running or jumping
2: Mild weakness, able to run
3: Moderate weakness, able to walk but not run
4: Marked weakness, able to stand but unable to walk
5: Severe weakness, unable to stand and only slight movement of the legs
6: Paraplegia, no movement
VX2 virus-induced rabbit epidermoid carcinoma
5-month-old female rabbits
Orthotopic inoculation into paraspinal area of T12 or T13 via posterior approach
Grading of weakness of hindlimbs;
1: Unstable running
2: Weak, but able to run
3: Attempts to walk
4: Hindlimb movement only on pinprick
B16–G3.26 murine melanoma,
8- to 12-week-old female C57B1/6, Balb/c and athymic Balb/c nu/nu mice
Development of hindlimb paraparesis and progression to complete paraplegia
4526 murine mammary carcinoma,
SK-N-MC human neuroepithelioma,
CHO-K1 hamster ovary,
HeLa human cervical adenocarcinoma,
SKsarc human undifferentiated sarcoma
VX2 virus-induced rabbit epidermoid carcinoma
3- to 4-month-old Japanese white rabbits
Orthotopic inoculation into L3 pedicle via posterior approach
Grading of weakness of hindlimbs;
Normal: Animals can run
Incomplete paraplegia: Animals can stand but not run
Complete paraplegia: Animals cannot stand
CRL-1666 rat mammary adenocarcinoma
10-week-old female Fischer 344 rats
Orthotopic inoculation into L6 vertebral body via transabdominal approach
BBB Locomotor Rating Scale ;
Local chemotherapy (paclitaxel dissolved in biodegradable polymer), radiotherapy, surgical resection (anterior corpectomy)
Score between 0 (complete paralysis) to 21 (normal) based on gait and hindlimb characteristics
VX2 virus-induced rabbit epidermoid carcinoma
New Zealand White rabbits
Orthotopic inoculation into T12 vertebral body via posterior approach
Signs of paraparesis including abnormal gait and inability to stand
MT-1 human breast carcinoma
4- to 6-week-old female athymic rats
BBB Locomotor Rating Scale 
PC-14 human lung adenocarcinoma
Athymic nude mice
Orthotopic inoculation into L3 vertebral body via transabdominal approach
BBB Locomotor Rating Scale 
MDA-MB-231 human breast adenocarcinoma
3-week-old female athymic rats
Orthotopic inoculation into L5 vertebral body via transabdominal approach
One of the most popular rodent species used in tumour metastatic research is the ‘nude’ mouse. Nude mice are born without a thymus gland and have no T-lymphocytes, thus are unable to elicit an immune response. They are commonly used to study the in vivo behaviour of allotransplanted or xenotransplanted tumour cells as they are able to receive tissue and tumour grafts from different species without mounting a rejection response. Severe combined immunodeficiency (SCID) mice are another strain of mice commonly used for cancer research. These mice are severely deficient in both B and T lymphocytes and also do not immunologically reject human cancer cells and tissue [6, 18–20].
The question of which animal species, the rabbit, rat or mouse, is best adopted to study the human condition remains controversial. Rabbits are the largest of the three animals, potentially making surgical procedures less technically demanding and vertebral tumour analysis easier due to their larger size. On the other hand, rabbits are more costly and difficult to maintain compared to rodent species and there is a paucity of cancer cell lines available for rabbits. Mice share a high degree of gene sequence homology with humans and share similar organ systems. Furthermore, mouse genes can be manipulated to generate knockout, transgenic or over-expressing strains that can be used to investigate specific cancer pathways. For example, Sparc (osteonectin) knockout mice display increased tumour-induced osteolysis , and mice overexpressing mutant alleles of the p53 oncogene display a high incidence of lung, bone and lymphoid tumours . In comparison, although rats also share a similar sequence homology with humans, it is more difficult to achieve transgenic or genetically modified strains. Together with their widespread and relatively cheap availability, high sequence homology and similar organ systems to humans, it appears that the mouse offers a suitable in vivo model for studying human metastasis. This is reflected by the popular usage of mouse models of spinal metastasis in recent years [6, 7, 19, 23–25].
4 Establishment of the spinal metastasis
Since metastatic spread of cancer occurs spontaneously in humans, an ideal animal model would involve the spontaneous onset of spinal metastasis. However, in vivo models of spontaneous bone metastasis that arise in rodents or small mammals are uncommon . Spontaneous development of mammary carcinomas can occur in mice and rats but most do not metastasize and metastasis to bone is very rare . Therefore, the reliable and reproducible establishment of spinal metastases in experimental animal models requires either systemic or local inoculation of cancer cells. The method of administration of cancer cells into the host animal is an important consideration in any in vivo model of tumour metastasis.
Methods for systemic administration of cancer cells directly into the circulation include intravenously via the rodent lateral tail vein and intra-cardiac via the left ventricle of the heart. Tail vein inoculation requires immobilization of the animals but is otherwise relatively simple to perform [25, 28, 29]. Intra-cardiac inoculation requires animal general anaesthesia, with or without an anterior chest skin incision and exposure of the ribs, and passage of a 25- to 30-Ga. needle into the heart, the pulsatile entrance of bright red blood into the needle hub indicating correct positioning of the needle [24, 28, 30–32]. Systemic administration of cancer cells resembles the human metastatic process but invariably results in multiple-site metastasis, including to bone, liver and lung, and the spine is one of many areas that can be affected by tumour metastatic spread [33–41]. Limitations of systemic inoculation include the prolonged time required before the onset of metastasis to the spine, meaning that animals usually need to be sacrificed prior to developing any neurologic deficit because of extensive tumour burden and associated morbidity [24, 30, 31, 42, 43]. Furthermore, the timing, location and number of metastases, particularly to the spine, are extremely variable and unpredictable following systemic inoculation. Some investigators have observed a greater than 75% spinal metastasis rate [42, 43] but others have observed significantly less depending on the cancer cell line administered [24, 25]. These limitations may render a systemic tumour administration spinal metastasis model unsuitable for the purpose of investigating potential therapeutic interventions.
Local orthotopic inoculation involving direct implantation of cancer cells into the desired site for the metastasis overcomes some of the limitations of intra-cardiac administration with regard to predictability and the timing of tumour establishment in the spine. Ushio et al.  first described the injection of mammary carcinoma cells percutaneously into the paraspinal region of the anterior T12 or T13 vertebral body of Wistar rats, establishing tumours that grew through the intervertebral foramina to cause spinal cord compression and paraplegia within 4 weeks of inoculation. Takahashi et al.  described direct orthotopic inoculation of carcinoma cells via a posterior approach into the L3 vertebra of rabbits, by drilling a 2-mm-diameter hole in the left L3 pedicle, implanting the tumour cells within the hole in the pedicle and subsequently sealing it with bone wax. The majority of animals developed paraplegia within 5 weeks and the remainder died of lung metastases prior to the development of neurologic deficit. Amundson et al.  described a similar technique whereby a 2-mm diameter hole was made in the posterior elements of the lowermost thoracic vertebra in New Zealand White rabbits using a fine burr and a 20-Ga. needle was then passed 5 mm through the pedicle into the vertebral body for cancer cell administration, such that tumours formed in the vertebral bodies which are the commonest anatomical site for spinal metastases, rather than the pedicle. Mantha et al.  described orthotopic implantation of mammary adenocarcinoma tissue directly into the L6 vertebral body of Fischer rats. This novel in vivo model required animal general anaesthesia, an anterior surgical approach via a midline skin incision over the abdominal wall, retraction of abdominal contents and bowels, drilling of a 1 mm diameter cavity in the L6 vertebral body, tumour implantation and sealing of the bone with bone cement. Tumours developed consistently and all rats developed paraparesis within 16 days of inoculation. The surgical procedures involved in orthotopic inoculation of cancer cells into the spine may be technically demanding, especially for implantation in the small vertebral bodies of the mouse and can result in secondary complications including infection, haemorrhage and death [7, 44]. Additional criticisms of the technique of orthotopic inoculation in in vivo cancer models are that it fails to encompass the molecular mechanisms by which cancer cells are attracted to bone by bypassing this stage and the injection process itself stimulates bone remodelling. However, advantages are that a known number of cancer cells or defined quantity of tumour tissue can be orthotopically implanted, tumours develop consistently and reproducibly and grow within the bone microenvironment, and neurologic deficit due to spinal cord compression occurs within a relatively short timeframe.
5 Clinical assessment of animals
The inevitable outcome of untreated metastatic epidural spinal cord compression is paraparesis, signs of which in animals include abnormal gait or inability to stand on hindlimbs. In simplest form, animals may be assessed on whether or not they are able to run or stand [14, 15]. Bladder and bowel incontinence has also been observed in some animals in vivo [7, 11, 12]. In a rat model of spinal cord compression by tumour, Ushio et al.  graded the strength of rat hindlimbs from 0 to 6: Grade 0, normal; Grade 1, slight weakness with hip instability only noticed when running; Grade 2, mild weakness but able to run; Grade 3, able to walk but not run; Grade 4, able to stand but not walk; Grade 5, unable to stand with only slight movement of legs; Grade 6, paraplegia. Using this grading scale, the authors assessed the effect of dexamethasone, radiation therapy, laminectomy, dimethyl sulfoxide and cyclophosphamide after rats were unable to walk due to spinal cord tumoral compression [11, 12]. They found that animals treated with cyclophosphamide, dexamethasone and radiotherapy improved neurological function. Ikeda et al.  similarly graded the function of rabbit hindlimbs from 0 to 5: Grade 0, normal; Grade 1, unstable running; Grade 2, weak but able to run; Grade 3, attempts to walk; Grade 4, moves hind limbs only on pinprick; Grade 5, paraplegia. The Basso, Beattie and Bresnahan (BBB) Locomotor Rating Scale, initially described in a rat spinal cord contusion model , is a more descriptive and comprehensive rat hindlimb functional scale for neurologic injury and has been used by several investigators in rat models of spinal metastasis [30, 42, 44, 46–48]. This scoring system involves placement of the rat in an open field and gait observation for 4 min, to derive a score between 0 to 21, where 21 is the highest level of function (normal) and 0 (complete paralysis) is the lowest score, and examines degree of movement of the three hindlimb joints, plantar stepping and weight support, and forelimb and hindlimb coordination. Despite its complexity, the original authors of this scoring system found satisfactory inter-observer reliability and improved scoring consistency with experience. Using this functional scoring system in an orthotopic mouse model of lung cancer spinal metastasis, Tatsui et al.  observed four key neurological events during deterioration from normal coordinated walking; tail weakness and dragging, followed by inability to walk on the plantar aspect of the foot (dorsal stepping), leading to hindlimb sweeping and finally to paralysis. They were able to correlate the neurological decline with degree of spinal cord compression by tumour on axial histological analysis. Notably, the BBB scale for rats has been modified to be more relevant to mice, since mouse recovery from spinal cord injury differs from rats with regards to coordination, paw position and trunk instability . This resulted in the Basso Mouse Scale for Locomotion, giving a score between 0 and 9 based on ankle movement, plantar stepping, coordination, position of paws at initial contact and lift off, trunk stability and tail position. The authors suggested that this scale was a more sensitive, valid and reliable method for measuring locomotor recovery following spinal cord injury specific to mice, however to date it has not been adopted in any mouse model of spinal metastasis.
6 In vivo animal imaging
In vivo animal imaging is important in monitoring tumour establishment in the spine, its subsequent growth and invasion of adjacent tissues, and correlating radiological evidence of bony destruction and epidural spinal cord compression by tumour with clinical neurological deficit, since this is what occurs in the human condition. The first and simplest in vivo animal imaging modality was plain radiography , however although radiography is capable of detecting gross lesions, it is not sensitive at detecting early or small tumours in the absence of significant osteolysis or mineralized-matrix formation. Using plain X-rays centred on the site of orthotopic tumour inoculation in a rabbit spinal metastasis model, osteolytic tumours were detected only after complete paraplegia had occurred but remained undetected in rabbits with incomplete paraplegia . Similar to the human condition in which radiologically evident spinal lesions usually occur late in the evolution of metastatic cancer spread to the spine, radiologic evidence of vertebral metastasis in animal models is a late event and commonly associated with significant spinal cord compression, although plain X-rays remain useful for showing gross morphologic changes.
Bioluminescent reporter imaging and fluorescent imaging are emerging imaging modalities in animal tumour models and are significantly more sensitive than plain radiography for the detection of vertebral tumours [31, 50]. Cancer cell lines are permanently transfected with either the firefly luciferase or the green fluorescent protein and their expression in vivo may be detected and quantified to monitor tumour growth . Bioluminescence imaging can detect minimal metastatic deposits at a volume of 0.5 mm3 in the bone marrow prior to osteolysis . Furthermore, it can be repeated at regular intervals and by detecting viable tumour cells may potentially be used to confirm tumour establishment, monitor tumour growth and assess treatment effects. However, it does require an injection of a chemical substrate (luciferin) every time imaging is performed, and multiple metastatic lesions in different parts of the body and heterogenous tumours may be difficult to quantify due to photon characteristics and light emission . Green fluorescent protein (GFP) imaging does not require an injection of substrate at time of image acquisition but is not as sensitive as bioluminescence imaging. Fluorescence can be assayed using tissue sections via microscopy and dissociated cells can be analysed through flow cytometry.
Other more expensive and sophisticated animal imaging modalities include computed tomography (CT), positron emission tomography (PET) and magnetic resonance imaging (MRI) which are all routinely used in the diagnosis, staging and treatment planning of spinal metastases in human patients. Similar to bioluminescence imaging, micro-CT has the ability to detect spinal tumours earlier than conventional radiographs in animal models . It is also particularly useful for imaging the local effect of tumour on bone, with the possibility of performing quantitative bone density assessment, bone volume calculations and 3D reconstruction as well as enabling monitoring and evaluation of tumour progression over time . In a rabbit orthotopic tumour model of spinal cord compression, micro-CT and MRI did not reveal gross tumour prior to the onset of paraparesis. However, at the onset of paraparesis, CT demonstrated osteolytic tumour and MRI showed tumour arising from the vertebral body and compressing the spinal cord, corresponding with macroscopic and histopathologic examination . PET functions via detection of the increased uptake of glucose in malignant cells, making it useful to investigate tumour viability and metabolism but it lacks the anatomical reference frame is therefore often used in conjunction with CT . Signal from tumours in the spine can be detected by PET from day 1 post-inoculation , thus it can be useful in the early monitoring of tumour growth and progression. MRI is the most sensitive imaging modality used in patients for assessment of soft tissue tumoral extension and metastatic epidural spinal cord compression and is integral in the planning of surgery in these patients. Although MRI has shown some promise in the detection and assessment of epidural tumour in rabbit models [14, 53], its use has not been common in animal spinal metastasis models to date.
This work was supported by the National Health and Medical Research Council of Australia (Fellowship No. 558418)
Conflict of interest
The authors declare that they have no conflict of interest.