Sarah Wuest, DC, Bruce Symons, DC, Timothy Leonard, MSc, Walter Herzog, PhD
Human Performance Laboratory,
University of Calgary,
Calgary, Alberta, Canada.
OBJECTIVE: The purpose of this study was to measure strains in the human vertebral artery (VA) within the cervical transverse foramina and report the first results on the mechanical loading of segments of the VA during spinal manipulation of the cervical spine.
METHODS: Eight piezoelectric ultrasound crystals of 0.5-mm diameter were sutured into the lumen of the left and right VA of one cadaver. Four hundred-nanosecond ultrasound pulses were sent between the crystals to measure the instantaneous lengths of the VA segments (total segments n = 14) at a frequency of 200 Hz. Vertebral artery engineering strains were then calculated from the instantaneous lengths during cervical spinal range of motion testing, chiropractic cervical spinal manipulation adjustments, and vertebrobasilar insufficiency testing.
RESULTS: The results of this study suggest complex and nonintuitive strain patterns of the VA within the cervical transverse foramina. Consistent (for 2 chiropractors) and repeatable (for 3 repeat measurements for each chiropractor) elongation and shortening of adjacent VA segments were observed simultaneously and could not be explained with a simple model of neck movement. We hypothesized that they were caused by variations in the location and stiffness of the VA fascial attachments to the vertebral foramina and by coupled movements of the cervical vertebrae. However, in agreement with previous work on VA strains proximal and distal to the cervical transverse foramina, strains for cervical spinal manipulations were consistently lower than those obtained for cervical rotation.
CONCLUSIONS: Although general conclusions should not be drawn from these preliminary results, the findings of this study suggest that textbook mechanics of the VA may not hold, that VA strains may not be predictable from neck movements alone, and that fascial connections within the transverse foramina and coupled vertebra movements may play a crucial role in VA mechanics during neck manipulation. Furthermore, the engineering strains during cervical spinal manipulations were lower than those obtained during range of motion testing, suggesting that neck manipulations impart stretches on the VA that are well within the normal physiologic range of neck motion.
From the Full-Text Article:
The topic of vertebral artery dissection (VAD) has been discussed and studied in the literature for many years. Although the actual risk remains unknown, the chiropractic clinical practice guidelines on evidence-based treatment of adult neck pain not due to whiplash reported the risk to be approximately 1 per million cervical spinal manipulative treatments (cSMTs) ; and the general consensus in the literature is similar. [2-6] Nevertheless, the mechanism responsible for the relationship between cSMT and VAD remains unresolved for a lack of biomechanical data. The stresses and strains experienced by the VA during normal everyday movements are currently not known.
Recently, Cassidy et al  reported that strokes resulting from injury to the VA are no more common after chiropractic visits than visits to medical doctors.  This finding contradicts some researchers in the health care community who report that chiropractic treatment is a risk factor for this type of stroke. Although these epidemiologic studies may answer the question of the statistical risk of VAD associated with cSMT, they add little to our understanding of the biomechanics of the VA during spinal manipulation.
Although much has been written on vertebrobasilar accidents associated with chiropractic manipulation, few studies have investigated the biomechanics of the VA during neck manipulation; and little is known about what happens to the VA during movements of the cervical spine. The majority of research into vertebrobasilar injury has focused on blood flow through the VA. Thus, the biomechanics of the tissues of the VA itself remain poorly understood.
To our knowledge, no studies have investigated the mechanics of VA segments C1-C6 during movements of the cervical spine. Nevertheless, it has been suggested clinically for many decades that these movements are predictable from the anatomy. For instance, it has been tacitly assumed that, during right lateral flexion, the right VA segments will shorten whereas the left segments will elongate. This understanding of VA movement has been used to support claims that chiropractic cSMT may cause damage to the VA.
We have previously reported the strains experienced by the VA during range of motion (ROM) testing, vertebrobasilar insufficiency testing, and chiropractic cSMT.  However, for technical reasons, this study was limited to the most proximal and distal ends of the VA, thereby leaving the clinically and biomechanically relevant sections between C1-C6 unexplored.
The purpose of this study was to investigate the biomechanics of these VA segments within the cervical transverse foramina. Specifically, the strains (elongations from the neutral position) of the VA segments were measured at high temporal and spatial resolution for clinically relevant examinations and treatment procedures.
Consistent with our earlier findings on the distal and proximal segments of the VA,  the strains during cSMT in the midsegments of the VA spanning C1-C6 were substantially lower than the strains obtained during diagnostic and/or passive ROM testing. These findings support our previous conclusions that the VA strains experienced during cSMT are substantially less than the strain in VA segments C1-C6 experienced during normal neck rotation.
However, and in complete contrast to our previous findings on the distal and proximal VA segments,  the results for VA segments C1-C6 are highly complex in nature, as stretching and shortening of the VA occurred simultaneously in adjacent segments during ROM testing, cervical spinal manipulative therapy, and vertebrobasilar insufficiency testing (Fig 3, Fig 4). Therefore, we conclude that the VA does not act in a predictable fashion (ie, the VA segments elongate on the convex side of positive neck curvature and shorten on the concave side of negative curvature). Our data suggest that additional factors play a role influencing the strains placed on the VA within the cervical transverse foramina.
We hypothesized that the unexpected elongations and shortenings occurring simultaneously on VA segments were caused by coupled motions of the adjacent vertebrae, such that rotation of the neck may be associated with an approximation of 2 vertebrae comprising 1 motion segment, and divergence may occur between 2 vertebrae comprising an adjacent motion segment. Furthermore, fascial connections or tethering between the VA and the bony canal of the cervical transverse foramen might restrict motion of the VA and cause these results. It is also possible that the facet orientation, disk height, and disk properties might affect the lengthening or shortening of segments of the VA and thus might contribute to this nonintuitive behavior of the VA strain distributions. Finally, we cannot exclude the possibility that strains in the VA might change depending on the circumferential location. All of our measurements were made on the most lateral aspect of the VA, where the tensile strains would be expected to be greatest.
Only 2 VAs were tested in this study. Thus, these results need to be viewed and interpreted with caution; and generalizations of these findings in vivo are not warranted. Nevertheless, we felt it important to document these preliminary results for several reasons: first, they are consistent with those obtained from our previous study on 7 VAs obtained from 6 cadavers  in that the diagnostic and ROM procedures caused VA tensile strains greater than those observed during cSMT.
One of the limitations of the previous study  was that the piezoelectric ultrasound crystals could only be placed in 2 locations. In this study, newer and smaller crystals were used that overcame this limitation; and thus, we could instrument multiple VA segments simultaneously. Furthermore, Kawchuk et al  recently reviewed 2 large case series of VAD patients with a history of recent cSMT and reported that the V3 segment of the VA was the one most statistically implicated in these patients as compared with the V1, V2, and V4 segments. In addition, Sheth et al  used a 3-dimensional reconstruction technique from magnetic resonance angiography images to demonstrate that the V3 segment, specifically the C1-C2 segment, undergoes the greatest strain during rotation and is the segment most strongly associated with VAD. In this study, we placed 2 piezoelectric crystal pairs spanning the C0-C2 segments; thus, we are confident that we have the most important and representative region of interest of the VA instrumented.
Our previous study  was also criticized for the use of only one chiropractor and for the inability to determine if the cSMT delivered in situ was similar to one delivered in vivo in terms of its biomechanical characteristics. In this study, we have addressed these points by measuring the force-time histories of 2 different chiropractors while also measuring the strains in situ. The force-time histories measured here were qualitatively similar to those observed previously in this laboratory (for review, see Herzog and Symons ). However, the peak forces measured in this study of 200 to 273 N were higher than the 150-N peak force previously reported.  It is possible that the use of this newer Pedar pressure transducer pad system gave larger values than in previous studies. This is supported by the fact that unpublished observations in live subjects showed that both chiropractors in this study had peak forces during cervical manipulation of 150 to 250 N using this newer Pedar system.
However, the results from this study contradict our previous results that found perfectly predictable strain patterns in the nontethered proximal and distal sections of the VA. Second, these results are highly nonintuitive and cannot be predicted merely based on neck curvature and the unconstrained anatomy of the VA. Finally, the results were observed in both VA specimens, were perfectly repeatable during multiple tests for a given chiropractor, and were consistent across 2 chiropractors.
There are several limitations to the present study. First, a sample size of one precludes any statistical comparison of our results. Nevertheless, this is a preliminary experiment reporting the successful application of newer technology to our previous protocol; and the observation of these nonintuitive strain patterns prompted us to report these surprising results. Continuing experiments are now under way with 3 chiropractors. Second, the use of a 90-year-old cadaveric specimen has obvious limitations in terms of extending these results in vivo, particularly to the population younger than 45 years who are at highest risk for VAD. [15, 16] However, we approached this study from a biomechanical engineering viewpoint to quantify the material properties of the VA, rather than to model the biological effects of cSMT. Third, we inflated the VA in this study with ultrasound gel to better emulate the in vivo shape of the VA. However, we did not measure the pressure inside the VA; nor did we attempt to pressurize the VA. We did observe that the sonomicrometry signals tended to degrade somewhat with time, and this was rectified by refilling the VA with ultrasound gel. Although the VA was sutured shut, we presume that there was leakage of the gel both distally into the basilar artery and/or proximally back into the subclavian arteries. Fourth, one inherent limitation of using sonomicrometry is that this method uses sound waves to quantify distance and thus measures length in a straight line. As shown in Figure 1, the upper loop of the VA in the V3 segment is a curved structure and not conducive to a linear measurement method. We have partially overcome this limitation by spacing the crystals as closely as possible, and the Sonolab software used to analyze the signals is able to distinguish between rebounding sound signals. Nevertheless, measurement error in this region of the VA may be increased.
The strains experienced by VA segments between C1 and C6 may not be predictable solely based on neck deformation. The biomechanics of VA strains in situ appear to be highly nonintuitive, and the historical notion of VA behavior was not confirmed here. These results have clinical implications because they suggest that VA strains in individual patients may not only differ in magnitude, but may also differ in direction, thereby making any a priori biomechanical prediction impossible. Further study of this surprising result is essential to fully understand VA mechanics for everyday neck movements in general and chiropractic neck treatments specifically. However, as observed previously, the strains during neck manipulations were always substantially lower (less than 20%) than those observed during diagnostic and ROM testing.
Vertebral artery mechanics during cervical spinal manipulation are highly nonintuitive.
Vertebral artery strains during cervical spinal manipulation are much smaller than during ROM or diagnostic testing.
Cervical spinal manipulation does not appear to put any tensile stress beyond normal on VA segments C1-C6.
Funding Sources and Potential Conflicts of Interest
This research was funded by grants from the Canadian Chiropractic Research Foundation, The Canadian Chiropractic Protective Association, and the Alberta College and Association of Chiropractors. No conflicts of interest were reported for this study.