J Manipulative Physiol Ther 2004 (Nov); 27 (9): 539–546 ~ FULL TEXT
Gregory N. Kawchuk, DC, PhD, Shari Wynd, DC, Todd Anderson, MD
Faculty of Rehabilitation Medicine,
University of Alberta,
BACKGROUND: Cervical spine manipulation is most often performed to affect relief of musculoskeletal complaints of the head and neck. Performed typically without complication, this modality is thought to be a potential cause of cerebrovascular injury, although a cause-effect relation has yet to be established. To explore this relation, an experimental platform is needed that is accessible and biologically responsive.
OBJECTIVE: To establish an animal model capable of accommodating (1) direct study of its vertebral arteries and (2) creation of controlled interventions simulating arterial injury.
STUDY DESIGN: Descriptive.
METHODS: Under fluoroscopic guidance, an ultrasonic catheter was inserted into the left vertebral artery of 3 anesthetized dogs. The ultrasonic probe was then drawn proximally through the artery at a specific rate, and cross-sectional images of the vessel were collected. These images were then reconstructed to provide a variety of 2- and 3-dimensional representations of the vessel. This procedure was repeated after the overinflation and/or displacement of an angiographic balloon within the vertebral artery itself.
RESULTS: The resulting ultrasonic images were able to delineate the structural layers that constitute the vertebral artery. Analysis of 2- and 3-dimensional reconstructions before and after angiographic intervention revealed the creation of discrete vascular injuries (aneurysm or dissection).
CONCLUSIONS: For the first time, an animal model has been established that permits direct interrogation of the internal structures of the vertebral artery. This model can also be manipulated to create "preexisting" vascular injuries that are thought to be possible prerequisites for cerebrovascular injury associated with manipulation. As a result, an experimental platform has been established that is capable of providing investigators of all backgrounds with the ability to quantify biologic and mechanical outcomes of cervical manipulation.
From the Full-Text Article:
The ability of a canine model to support direct interrogation of its vertebral arteries was investigated in this study. Intravascular ultrasound was used to generate 2-dimensional images of vessel structure from within the vessel itself. These images displayed the layers which constitute the vessel wall, whereas various reconstructions of this image data revealed the longitudinal nature of the artery and its associated structures. Image reconstruction techniques were used successfully to show the presence of vascular injury after either overinflation of an angiographic balloon (aneurysm) or manual translation of an inflated balloon within the vessel (dissection).
IVUS is a well-entrenched clinical technique used to visualize coronary artery pathologies and to help identify locations for the delivery of therapeutic agents (eg, stents, pharmaceuticals). [23–24] The application of this technology is limited only by the diameter of the ultrasound transducer (~1 mm) and the ability to place and operate the catheter from a remote access point. For these reasons, we supposed that IVUS would be an excellent technique to study the fine structures of the vertebral arteries in vivo. At present, IVUS is not approved for human use in vertebral arteries.
Most typically, human vertebral arteries arise from the subclavian artery in 90% of the population.  After exiting the subclavian artery, each vessel enters osseous foramina located on the transverse processes of the cervical vertebrae beginning typically at the sixth cervical vertebra. Vertebral arteries are unique in that they pass within rigid channels and are linked by connective tissue attachments to these channels at each of the vertebral segments they transcend. As a result, vertebral arteries are thought to undergo constant positional alterations from neck movement. [25–26] After entering the transverse foramina of the sixth cervical vertebra, the course of the artery is superior and linear until the third cervical vertebrae. After exiting the third transverse foramina, the vertebral artery courses more laterally, then sharply cephalad, creating a loop as it passes through the atlanto-occipital membrane and into the subarachnoid space.  The guidance wire of our catheter system can be seen at this level in Fig 2. After the vertebral artery passes through each foramen, it unites with its counterpart to form the basilar artery which, in normal circumstances, provides approximately 10% of the blood to the brain. 
The supply of blood from the vertebral artery to the brain may become interrupted through direct injury from a variety of mechanisms such as motor vehicle accidents.28 Because spinal manipulation is a therapeutic agent which is defined as the manual application of an external force over a short period resulting in movements of the vertebrae, some feel that this therapeutic modality is a potential form of trauma that may cause injury to the vertebral artery. [29–33] While evidence exists to support the use of cervical manipulation for providing relief of musculoskeletal complaints of the head and neck, [34–37] a number of cases have been reported which make a temporal link to the application of cervical manipulation and injury to the vertebral artery. [2–4, 9–11] Such an injury may in turn cause symptoms of downstream ischemia including nausea, nystagmus, ataxia, tinnitus, and anesthesia. [38–39] These symptoms range from being transient in nature to those that are permanently disabling or, in rare circumstances, fatal.
Unfortunately, these case studies are typically inadequate in their description of the performed manipulation and/or the evaluation of iatrogenic injury. Of approximately 115 cases of vertebral artery injury after manipulation reported in the English literature, only 2 cases exist in which investigators interviewed the practitioner to obtain an accurate description of the procedure used.  Therefore, only 2 reports exist which contain accurate information regarding the manipulative procedure.  Although these case studies make plausible the idea that a link between manipulation and cerebrovascular injury exists, the lack of information within these case reports is of little use toward establishing a cause-effect relation.
To further assess a potential relation between vertebral injury and cervical manipulation, other investigative approaches have been used. These include cadaveric, imaging, and population investigations. The limitations of these approaches are discussed below.
As a whole, cadaveric studies which perfuse the vertebral artery artificially have shown a diminished flow in the contralateral vertebral artery with ipsilateral neck rotation. [41–44] . Although some studies have shown the opposite of this generalization, [45–46] a popular notion has surfaced which contends that the rotational component of cSMT reduces blood flow to the brain which leads to cerebrovascular injury. This is an unlikely mechanism given the extremely short time the neck is rotated during cSMT [47–48] and the presence of other vessels to compensate for decreased cerebral flow.  A more plausible injury mechanism is that neck rotation during cSMT injures the vertebral artery which then creates a sustained interruption of blood flow to the posterior brain. To investigate this possibility, investigators have used cadaveric preparations to determine the strain on unperfused vertebral arteries during neck rotation. The resulting strain during cSMT was then expressed as a percentage of the strain required for the isolated vessel to fail when stretched along its long axis.  Although results from this study suggest that it is “very unlikely” that the vertebral artery is damaged during neck manipulation, there has been no recorded case that would suggest that this mode of failure (long-axis distraction) is physiologically relevant. More importantly, this approach, like all cadaveric approaches, cannot determine if biologic injuries occur in the vessel before the onset of overt damage (ie, microtearing, thrombus formation, vasospasm). Events such as these would require a biologically viable system, which is not the case in cadaveric preparations.
Like cadaveric studies, various imaging techniques have been used to address how cervical manipulation may disrupt vertebral artery flow. These include Doppler ultrasound [50–51] and magnetic resonance angiography. [52, 53] Unfortunately, each of these studies possesses limitations that inhibit study of the vertebral artery in vivo. Doppler ultrasound is capable of visualizing only gross alterations in vessel size or flow in small portions of the artery because of the size and shape of the transducer and the inability for ultrasonic waves to pass through the osseous transverse foramen.  Magnetic resonance angiography is limited in that its resolution can only describe significant alterations in vessel contour. [52–53] Although these limitations provide support for investigators who have concluded that these imaging techniques [55–57] and clinical tests of forced rotation  are of little value in predicting persons at risk for cSMT-induced injury, they do not provide the resolution or coverage necessary to determine if small-magnitude injuries or biologic responses to those injuries occur after manipulation.
Whereas cadaveric and imaging studies attempt to observe direct injury to the vertebral artery after cSMT, other investigators have used a population-based approach to estimate the risk of cSMT-induced injury in relation to cerebrovascular accidents. These population studies have generated risk estimates that range in value from 1 in 1.3 million to 1 in 400000. [13, 40] Unfortunately, this approach has severe limitations—a reflection of the wide range of published risk values. Specifically, population studies of risk tend to be disadvantaged by insufficient sample sizes because of the general infrequency of the condition. In addition, they may be corrupted because of underreporting and/or misdiagnosis of the condition.12 More importantly, population studies are limited in that they describe only the temporal probability between cSMT and cerebrovascular symptoms—these studies cannot define causation. As noted by the authors of one population-based study which created a risk estimate, “Statements about attributable rates imply a causal association and assume that observed relative risk values are unaffected by bias. This study design does not permit us to estimate the number of cases that are truly the result of trauma sustained during manipulation.” 
Creation of an animal-model toward the study of manipulation/vascular relations would solve many problems of the experimental approaches listed above and was the impetus for this project. Through use of IVUS, we have shown that the internal structures of the canine artery can be observed in vivo. This possibility provides an opportunity to apply mechanical forces to the model and determine the models' biologic and mechanical responses. Assuming that more than 1 animal can be tested, this technique can solve the sample size limitations of case studies. The high-resolution and indwelling nature of IVUS means that under physiologic conditions, the model can be probed for small structural aberrations and biologic responses to injury to manipulation—an impossibility in cadaveric studies. The ability to know the condition of the system before the application of a manipulative force provides the possibility that this model can delineate cause and effect relations between vascular injury and manipulation. Furthermore, the ability to create arterial injuries (aneurysm, dissection) in the model additionally provides an opportunity to determine how force application to the model may extend or aggravate preexisting vascular injuries.
It is the aggravation of these preexisting injuries or premorbid conditions by manipulation which is presently thought by many to be a significant mechanism by which cervical manipulation causes cerebrovascular injury. Specifically, given the large number of uneventful cSMT applications every year and the large number of vertebral artery strokes which do not involve cSMT, it has been suggested that persons who experience cSMT-induced stroke may have a preexisting vascular injury, [6, 59] genetic susceptibility, [60–62] or altered biochemistry. [19, 63–66] These conditions may in turn create symptoms (ie, headache, neck pain) that cause care-seeking behavior. Should that care provider deliver a manipulation, it may be conceivable that the cSMT application (1) would aggravate or extend the preexisting injury or (2) is independent of the lesion's progress, making treatment coincidental, but temporally linked to the onset of cerebrovascular symptoms. In either case, current investigative techniques based on cadaveric, imaging, or population-based studies are ill-suited to investigate this possibility.
Although advantageous in many respects, the animal model described here has limitations. Techniques used to synchronize the collection of IVUS images with the cardiac cycle and to register the IVUS images with the local anatomy must be used. In addition, we do not understand presently how congruent a canine model is to specific aspects of human vertebral artery physiology and the responses of the system to applied manipulation. This includes potential differences in the vascular stiffness of the model when compared with human beings who exhibit hypertension and atherosclerosis. Although no animal model is a perfect facsimile for human investigation, animal models provide a unique opportunity to increase our knowledge beyond the status quo when human investigation is not possible.
The 2 goals of this project were met. The canine model developed in this project is capable of direct study of its vertebral arteries (via IVUS) and can accommodate creation of controlled interventions resulting in arterial injury (dissection, aneurysm).
The long-term impact of this project will be the availability of an experimental platform capable of providing investigators of all backgrounds with the means to quantify the biologic outcomes of cervical manipulation. Information from these studies is needed desperately to evaluate current treatment efficacies, therapeutic dosage, and safety. As a result, regulatory agencies will possess the data needed to make meaningful recommendations regarding cSMT safety in the face of increasing public use. Without adequate tools for investigation, the cause-effect relation between cSMT and vascular injury will remain speculative.