J Manipulative Physiol Ther. 2001 (Jan); 24 (1): 2–11
Joel G. Pickar, DC, PhD, John D. Wheeler, DC
Kansas State University,
Department of Anatomy and Physiology,
Manhattan, KS, USA.
OBJECTIVE: The mechanisms underlying the benefits of spinal manipulation are not well understood. Neurophysiological mechanisms likely mediate its effects, at least in part, yet we know little about how the nervous system is affected by spinal manipulation. The purpose of the present study was to determine whether muscle spindles and Golgi tendon organs in paraspinal muscles respond to a mechanical load whose force-time profile is similar to that of a spinal manipulation.
METHODS: Experiments were performed on 10 anesthetized adult cats. The L6 dorsal root was isolated for electrophysiological recordings while the L6-L7 vertebrae and associated paraspinal tissues on one side of the vertebral column were left intact. Single unit recordings were obtained from 5 muscle spindles, 4 Golgi tendon organs, and 1 presumed Pacinian corpuscle afferent with receptive fields in paraspinal muscles. Loads were applied at the spinous process of the L6 vertebra through use of an electronic feedback control system. The load simulated the force-time profile of a spinal manipulation. Loads were applied in compressive and distractive directions and at 2 different angles (0 degrees and 45 degrees) with respect to the long axis of the vertebral column.
RESULTS: Golgi tendon organ afferent discharge frequency increased more to the impulse than to the preload during 13 of 15 spinal manipulations. Generally, the 4 Golgi tendon organ afferents became silent immediately at the end of each impulse. Similarly, muscle spindle discharge frequency increased more to the impulse than to the preload during 10 of 16 manipulations. Distractive manipulations loaded the spindles more effectively than compressive manipulations. After 7 of these 10 manipulations, muscle spindles became silent for 1.3 +/- 0.6 seconds (range, 0.1-4.3 seconds). Six of the 16 manipulations unloaded the muscle spindles. A presumed Pacinian corpuscle responded to the impulse of a manipulative-like load but not to loads with a slower force-time profile.
CONCLUSION: The data suggest that the high-velocity, short-duration load delivered during the impulse of a spinal manipulation can stimulate muscle spindles and Golgi tendon organs more than the preload. The physiologically relevant portion of the manipulation may relate to its ability to increase as well as decrease the discharge of muscle proprioceptors. In addition, the preload, even in the absence of the impulse, can change the discharge of paraspinal muscle spindles. Loading of the vertebral column during a sham manipulation may affect the discharge of paraspinal proprioceptors.
From the Full-Text Article:
This study demonstrates that muscle spindles and GTOs with receptive endings in the paraspinal muscles respond to vertebral loads whose force-time profiles are similar to that of a load delivered during spinal manipulation.  The spinal manipulative-like load in the present experiment was applied at the L6 spinous process along the long axis of the vertebral column in anesthetized cats. Muscle spindles discharged at rest, and they responded to the load preparatory to the impulse (preload) and to the impulse of the manipulation. Distraction of the L6-L7 facet loaded the muscle spindles more often than did compression of that facet. Discharge frequency was generally greater in magnitude during the impulse than during the preload. After the impulse, muscle spindles often paused for up to 4 seconds. GTOs were generally silent at rest and did not respond to the preload. The impulse increased the discharge frequency of GTO afferents. The largest increase in GTO afferent activity occurred when the manipulation distracted the L6-L7 facet. However, GTO afferents also responded when manipulation compressed the L6-L7 facet.
Most chiropractic theories underlying the reason for and explaining the effects of spinal manipulation hypothesize a role for the nervous system.  The chiropractic subluxation—a structural dysrelationship, typically between contiguous vertebrae—is thought to affect reflex neural activity. As a corollary, correction of the subluxation through the use of spinal manipulation is thought to affect the neuromusculoskeletal system. The unique effect of the impulse portion of a spinal manipulation on the responses of paraspinal proprioceptors may contribute, in part, to the effects of spinal manipulation.
The anatomy and reflex organization of spindles in paraspinal muscles have some unique aspects in comparison with those of the hindlimb. In the cat, hindlimb muscle spindles are described as single receptors located both deep in the muscle belly and close to the musculotendinous junction. [18-20] Spindle density ranges from 5 to 45 spindles/g of hindlimb muscle weight.  In the cervical spines of human beings [22-23] and cats, [24-25] however, muscle spindles are rarely seen as single entities, and their densities are greater than in the peripheral musculature. Richmond and Abrahams [15, 25] describe cervical spindle complexes wherein 2-6 spindles are in close contact with each other or share capsules and/or intrafusal fibers. Spindle density can be 2 to 8 times higher (47-107 spindles/g) in superficial cervical muscles  and 10 to 25 times higher (137-460 spindles/g) in deep cervical muscles  than in hindlimb muscles. Carlson  states that in the lumbar spine of the cat, muscle spindles are present in the longissimus, iliocostalis, sacrocaudalis, intertransversarii, multifidus, and interspinalis muscles; however, quantification and morphologic description of the spindles were not performed.
It is well recognized that in the cat hindlimb, the monosynaptic stretch reflex is elicited by excitation of muscle spindles, which in turn activates α motoneurons to the same muscle in which the spindle is located (homonymous α motoneurons). [27-29] The stretch reflex arc uses a single excitatory synapse to homonymous α motoneurons. [28, 30] The afferent arm of the reflex is comprised of group Ia and group II afferents. [29, 31] Each group Ia afferent from a given hind-limb muscle makes functional, monosynaptic connections with 50% to 100% of the homonymous α motoneurons. [32-33] Thus, stimulation of muscle spindles from a given hindlimb muscle evokes a monosynaptic excitatory postsynaptic potential (EPSP) in all α motoneurons to the same muscle. [34-35]
In the cervical spine, monosynaptic reflex connections to homonymous α motoneurons are weaker. EPSPs are smaller in amplitude, and group Ia afferents make functional connections with only 10% of the homonymous α motoneurons. [36-37] This probably contributes to the absence or weakness of monosynaptic reflexes in cervical muscle.  In the lumbar spine of the cat, stretch reflexes can be elicited from the longissimus muscle but not from the iliocostalis muscle. The existence of stretch reflexes from the multifidus muscle in the cat is unknown. Conduction delays suggest that the reflex arc, unlike that in the hindlimb, is not monosynaptic. [28, 30, 39] The presence of monosynaptic stretch reflexes from the deeper lumbar muscles has not been determined. In human beings, indirect evidence for the presence of muscle spindles and muscle spindle reflexes in lumbar paraspinal muscles was obtained by measuring evoked cerebral potentials in response to vibration of the lumbar paraspinal muscles,  which selectively stimulates muscle spindles. 
Paraspinal spindle reflexes could contribute to the short-lasting electromyographic (EMG) responses recorded from paraspinal skeletal muscle during spinal manipulation. [42-43] An activator thrust to a transverse process elicits paraspinal EMG activity at the same segmental level within 2-3 ms. However, if the EMG activity was initiated by muscle spindles, it was probably not produced by a monosynaptic reflex. If one assumes an average conduction velocity of 60 m/s and a one-way conduction distance of 3 cm between paraspinal muscles and the ipsilateral dorsal horn at the same segmental level, then the calculated conduction time would be 1 ms. Assuming synaptic delay to be approximately 0.5 ms, the reflex pathway would have to contain more than one synapse. Herzog et al  showed that spinal manipulation evoked paraspinal EMG activity in a pattern related to the region of the spine that was manipulated. The EMG responses from paraspinal muscles both near and distant from the site of manipulation occurred 50-200 ms after initiation of the manipulative thrust (impulse). These latencies are too slow for the reflex to be monosynaptic from muscle spindles. Our data support the possibility that muscle spindle reflexes can be initiated from a spinal manipulative impulse. Although the apparent reflex pathway from muscle spindles is likely oligosynaptic, other classes of afferents stimulated by vertebral movement [44-45] may contribute to reflex EMG activity.
Spinal manipulation is often applied to patients on the basis of clinical findings related to reduced segmental range of motion and to palpatory findings indicative of muscle hypertonicity. Spinal manipulation is thought to increase the range of motion and normalize muscle activity. Herzog et al  recently demonstrated that in a patient with spontaneous paraspinal muscle EMG activity, spinal manipulation initiated a reflex increase in EMG activity followed by a reduction in the spontaneous EMG activity. On the basis of findings in the present study, we speculate that combined activation of GTO afferents and silencing of muscle spindle afferents during the spinal manipulation can decrease spontaneous EMG activity by reflex inhibition or disfacilitation of α motoneurons.
Korr  proposed a neurophysiological mechanism to explain abnormal segmental function associated with the subluxation and the effects of spinal manipulation on segmental function. Abnormal segmental function may arise from altered paraspinal muscle activity. Korr suggests that paraspinal muscles of subluxated segments are responding to increased γ motoneuron discharge. The increased γ bias increases the sensitivity of paraspinal muscle spindles to stretch with consequent activation or facilitation of amotoneurons to paraspinal muscles. Korr proposes that spinal manipulation resets the γ bias by producing a high-frequency discharge in muscle spindle and GTO afferents. Although this hypothesis remains speculative because changes in γ bias associated with segments displaying characteristics of a subluxation have never been shown, the results of the present experiments suggest that spinal manipulation does bombard the central nervous system with sensory input from muscle proprioceptors and that this is followed by a prolonged silence. The central effects of these responses to impulse loads is not known, but it is interesting that input from muscle spindle and GTO afferents can converge on common reflex pathways in the central nervous system. 
Clinical studies involving sham manipulations should consider the types of sensory inputs they are trying to either mimic or exclude. Sham manipulations that provide a preload but not an impulse may still activate paraspinal muscles afferents. The present study has demonstrated this potential. Further investigation is necessary to clarify how the central nervous system integrates proprioceptive information evoked during each phase of the manipulation.
Our data suggest that short-duration, high-amplitude load delivered during a spinal manipulation can stimulate muscle spindles and GTOs. Some mechanoreceptors (eg, the presumed Pacinian corpuscle) may be stimulated only by the impulse and not by the preload. The results of these experiments confirm the speculation that muscle spindles and GTOs can be coactivated by spinal manipulative loads.11 However, the physiologically relevant portion of the manipulation may relate to its ability to increase as well as decrease the muscle proprioceptor discharge. In addition, the preload, even in the absence of the impulse, can alter the discharge of paraspinal muscle spindles. We speculate that loading of the vertebral column during a sham manipulation may affect the discharge of paraspinal proprioceptors.