Trunk muscle activation in low-back pain patients, an analysis of the literature

[문형철]

통증의 두가지 가설

1. pain adaptation model 

   - 통증이 발생하면 주동근 underactivity, 길항근 overactivity되면서 악화된다는 모델

   - 1942년 Travell에 의해 제안된 이론

2. pain spasm pain model 

  - 통증이 발생하면 반사성 근수축이 발생하여 muscle spasm이 발생하여 통증이 악화된다는 모델

  - 1991년 Lund에 의해 제안된 이론

두 이론은 충돌한다. 그래서 대안이 필요하다. spinal stability를 완성하여 primary pain이 secondary pain으로 악화되지 않도록

Trunk muscle activation in low-back pain patients, an analy.pdf

Abstract

This paper provides an analysis of the literature on trunk muscle recruitment in low-back pain patients. Two models proposed in the literature, the pain–spasm–pain model and the pain adaptation model, yield conflicting predictions on how low- back pain would affect trunk muscle recruitment in various activities. 

The two models are outlined and evidence for the two from neurophsysiological studies is reviewed. Subsequently, specific predictions with respect to changes in activation of the lumbar extensor musculature 

are derived from both models. These predictions are compared to the results from 30 clinical studies and three induced pain studies retrieved in a comprehensive literature search. Neither of the two models is unequivocally supported by the literature. 

These data and further data on timing of muscle activity and load sharing between muscles suggest an alternative model to explain the alterations of trunk muscle recruitment due to low-back pain. It is proposed that motor control changes in patients are functional in that they enhance spinal stability.

1. Introduction

Low-back pain (LBP) is one of the most preval‎ent and costly health problems in western society [7]. In spite of

extensive research efforts, the causes of LBP are still elusive and treatment effects are unsatisfactory. It is often suggested that the occurrence of LBP should be accepted as a fact of life and efforts of researchers and clinicians should focus on preventing LBP from becoming chronic rather than at prevention of first-time occurrence [7].

To design a rational program for secondary prevention, it needs to be established whether behavioral responses displayed by patients should be considered adaptive and supportive of recovery or as adverse and contributing to a vicious cycle leading to chronicity. An important debate in the literature on LBP in this respect focuses on the interpretation of changes in trunk muscle activity in LBP patients. On the one hand, these changes are interpreted in the context of the pain–spasm–pain model, postulating that pain results in increased muscle activity, which in turn will cause pain [110,129]. In contrast, the pain adaptation model postulates that pain reduces activation of muscles when active as agonists and increases activation of muscles when active as antagonists. This will reduce movement velocity and range of motion, which would prevent mechanical provocation of pain in damaged tissues and further damage of these tissues [78].

It is striking that proponents of both models have found evidence supporting their interpretation in reviews of the literature on electromyographically recorded back muscle activation in LBP patients [78,110]. 

The aim of  the present paper therefore is to more systematically review the literature on effects of clinical and experimentally induced LBP on trunk muscle activation in terms of levels of activation, the timing of activation, and load sharing between muscles. 

First, the two models will be outlined and specific predictions on trunk muscle activation will be derived for each model. Second, studies on trunk muscle activation in LBP patients will be reviewed. The results from the studies reviewed will be

compared to the model predictions. Finally, an alternative interpretation for the experimental results based on

spinal instability as an important component of LBP will be proposed.

2. Pain–spasm–pain model

Pain occurring immediately after acute trauma is often accompanied by intense contractions of muscles surrounding

the injured structures. This is thought to be functional, since it will prevent motion of the injured structures. In clinical practice, it is often assumed that similar muscular reactions occur following non-traumatic pain. In these cases, it is not the functional, adaptive nature of such a response that is emphasized but rather its possible adverse consequences. The pain–spasm–pain model was first formally proposed by Travell [129]. 

Travell suggested that pain would lead to muscular hyperactivity referred to as spasm, which in turn will cause pain. Treatment modalities based on this model involve relaxation and stretching of muscles. Two distinct neural pathways have been proposed to form the basis of a vicious pain–spasm–pain cycle (Fig.1). 

Most tissues in the low back, including, muscle, tendon, ligament, bone, endplate, and annulus fibrosus, are

innervated by thin myelinated and unmyelinated fibers with free nerve endings. These afferents, which are considered

to be nociceptors, enter the posterior horn, from where they project onto higher centers giving rise to the perception of pain. In addition, these fibers project via interneurons onto alpha motorneurons at the segmental level. Wyke [138] has suggested that these projections form the basis of a reflexive hyperactivity of back muscles in response to pain. 

Johansson and Sojka [59] have proposed an alternative neural pathway to form the basis of a pain–spasm–pain loop. They proposed that nociceptors affect the output of muscle spindles via direct excitatory projections on the gamma motorneurons. The increased muscle spindle output will cause hyperexcitability of the alpha motorneuron pool. The

common denominator in the two theories is the hyperexcitability of the alpha motorneuron pool, which could

form the basis for more sustained and more intense muscle activity in pain patients as compared to controls. 

This sustained activity is in turn expected to cause pain due to the accumulation in the muscles of for example arachidonic acid, bradykinin, potassium and lactate. These substances have been shown to have strong excitatory

effects on nociceptors in muscle tissue [66,91]. 

Several basic neurophysiological studies have been performed to verify the existence of the pathways proposed

to underlie the pain–spasm–pain cycle. Evidence of hyperexcitability of the alpha motorneuron pool due to nociceptive stimuli was found in cats after mechanical or chemical noxious stimulation of facet joints, fascia, ligaments, periosteum, and muscle. The hyperexcitability was shown in the form of short erratic bursts of paraspinal muscle activity [56,101,118,119,123,137]. In pigs, paraspinal muscle activity was found to increase during electrical stimulation of the intervertebral disc and facet joints [54]. In rat, the flexion reflex was increased following induction of muscle pain [134].

Likewise resting EMG levels in rat masticatory muscles were found to be increased after induction of pain. This effect did however last only up to 10 min. Experimental results in humans are limited. 

The amplitude of the Hofmann’s reflex, which is an indicator of alpha motorneuron excitability, was not increased after injection of hypertonic saline in the calf muscles [88]. However, resting EMG levels were increased in human masticatory

muscles [124], although the effect was only short-lived. Part of the neurophysiological studies inspired by the pain–spasm–pain model focused on the pathway proposed by Johansson and Sojka [59], by looking at the effects of nociceptive afference on muscle spindle activity. 

An increased output of muscle spindles in cat hind leg and neck muscles, as a consequence of nociceptive stimuli, was found in a range of experiments [30,76,102,103]. However, an experiment in which myositis was induced in the hind leg of the cat showed a decreased activity of gamma motorneurons [92]. Also in cat back muscles no increased gamma motorneuron activity was found after injection of bradykinin and capsaicin [65]. Inflammation of the knee joint produced

mixed excitatory and inhibitory effects on gamma motorneurons in cats [42]. In human calf and masticatory muscles, the stretch reflex, which is mediated by muscle spindle afferents, was found to be enhanced after injection of hypertonic saline [88,125].

 However, in back muscles no enhancement of stretch reflexes was found after hypertonic saline injection [141]. In addition, during walking the stretch reflex in the calf muscles was unaffected by induced pain [87]. In conclusion, the studies reviewed do not unequivocally support nor clearly reject the mechanisms, which have been proposed to underlie the pain–spasm–pain model. In addition, most of this work involved animal experiments with anesthetized or decerebrate animals, greatly reducing or excluding any influence from the brain. 

The pain-adaptation model (see below) states that the effects of peripheral projections of nociceptors onto

alpha motorneurons are modulated by central nervous influences related to voluntary activity. Furthermore, it has been suggested that higher levels of muscle activation can occur as part of a response to pain as a psychological stressor [57]. Therefore, testing the model with data obtained in studies of induced or clinical pain in awake humans may provide further information. 

The basic tenet in the pain–spasm–pain model is that pain will result in more sustained and increased muscle activation. The model predicts that, in every submaximal task as well as in rest, muscle activation will be higher in patients as compared to healthy individuals. By mechanical necessity this implies that cocontraction of agonist and antagonist muscles occurs in the patients. During maximum efforts, patients are predicted to attain the same level of muscle activation as controls, with however a lower moment production due to the antagonistic cocontraction. These predictions (Table 1) will be tested below using data from clinical studies.

3. Pain-adaptation model

The pain-adaptation model was proposed by Lund et al. [78] to account for clinical findings on muscle activity in pain syndromes. The model states that pain decreases the activation of muscles when active as agonists and increases it when the muscle is active as antagonist [78]. The effects of such a control strategy would be that movement velocity is reduced and movement excursions are limited. These kinematic effects are believed to prevent pain provocation.

A neural pathway suggested to account for the recruitment changes in the pain-adaptation model consists of the nociceptors that project on the alpha motorneuron via both inhibitory and excitatory interneurons (Fig. 2).

The excitability of these interneurons is controlled by the central nervous system, in such a way that depending on the instantaneous motor command inhibition or excitation of the alpha motorneuron dominates. The pain-adaptation model has motivated several neurophysiological studies addressing the effects of induced pain. Schwartz and Lund [113] found noxious pressure to cause changes in EMG and kinematics during cyclic jaw movements in decerebrate rabbits, which were consistent with the pain adaptation model. In an induced pain condition postural activity of the jaw muscles was

increased with respect to baseline. 

However, it was not different from that in a sham pain condition [121]. Thus, in line with the model no consistent evidence for an effect of pain on resting activity of these muscles was found. Jaw kinematics were affected by pain in accordance

with the model’s predictions [120]. Clear evidence

in support of the pain-adaptation model was found in a

study by Graven-Nielsen et al. [41]. Induced pain in

human gastrocnemius muscle was found to decrease the

activity of this muscle during gait, while the activity of

the tibialis anterior, its antagonist, increased. The reverse

was found when pain was induced in the tibialis anterior

muscle. Furthermore, motor unit firing rates were found

to decrease in human forearm muscles after injection of

hypertonic saline [14].

The pain-adaptation model predicts muscle activation

to be reduced in agonists and increased in antagonists,

irrespective of the level of exercise. The theory is somewhat

aspecific as to what is meant by antagonists and

agonists. In our interpretation, Lund et al. [78] used the

term antagonists to refer to muscles that are lengthening

and agonists to refer to muscles that are shortening. This

interpretation is based on the predicted effects on kinematics.

An alternative definition of antagonist and

agonist roles is based on the sign of the moment produced

by the muscle relative to the sign of the net

moment [53]. This definition does clearly define agonists

and antagonists in isometric as well as dynamic contractions.

The implications of this definition may clearly

deviate from the definition based on shortening versus

lengthening. The fact that the originators of the model

interpreted unchanged activity in static tasks as supportive

of the model [121], is in line with our interpretation.

Following this interpretation, the model predictions are

given in Table 2.

중간 생략

6. Discussion

The literature reviewed here reveals that neither one

of the two models adequately predicts the effects of back

pain on trunk muscle activation. In some cases evidence

for reduced activation is found in line with the painadaptation

model and in conflict with the pain–spasm–

pain model. These changes appear to be adaptive in that

heavy exertion of painful muscles and high accelerations

that may impose a risk of pain provocation are avoided.

In line with this Marras et al. [84,85] reported that LBP

patients generally perform movements relatively slow.

However, in other cases increased activation is found,

conflicting with the prediction of the pain-adaptation

model. Moreover, the wide variance in results within and

between subjects appears irreconcilable with the hardwired

neural pathway proposed in the pain-adaptation

model. In addition, some of the data reviewed indicate

that pain may in some cases lead to disturbances of

motor control, which are not likely to be adaptive, such

as for instance left/right asymmetry of LES activation

and the erratic temporal pattern of LES activation

reported by Grabiner et al. [39]. Also the delayed responses to perturbations of trunk equilibrium might

fall into this category. As we will argue below, a loss of

control may actually be one of the reasons why muscle

activation patterns are adapted.

We suggest that instances of increased activation may

be adaptive in nature. The nature of these adaptations

however appears much more complicated than suggested

by the pain-adaptation model. First the changes may be

tuned to the individual problem probably through learning,

as is suggested by the substantial variance between

subjects. In support of this Arena et al. [8] found erector

spinae activity to differ between groups diagnosed to

have LBP of different origins. However, too few studies

have attempted to discriminate between different diagnostic

groups to allow more definitive conclusions. It is

also conceivable that between-subject variation may

occur due to differences between patients in the developmental

stage of their low-back disorder. However, most

of the studies reviewed have included only patients with

chronic LBP. Second, the changes appear tuned to the

mechanical circumstances, or in other words to the task

at hand. Both this and the former point are illustrated in

a study by Sherman [115]. In this study each LBP patient

was found to show increased LES activation during at

least one of the five experimental tasks, but when analyzing

the data per task no differences between patients and

controls were found.

We surmise that the changes observed due to pain in

general are aimed at avoiding noxious tensile stresses in

injured structures. Our interpretation is in line with the

hypothesis on stability as a cause of LBP put forward

by Panjabi [98,99,99a]. In rest postures, stability of the

spine can be critical [19]. If the spine becomes unstable,

excessive rotations of segments will occur and pain may

be provoked [98,99]. The increased activation in rest

postures, which reflects increased cocontraction and the

changes in load sharing between lumbar and thoracic

erector spinae may be aimed at enhancing stability.

Absence of flexion-relaxation is associated with limited

segmental rotations [64]. It prevents excessive lengthening

of muscle, ligaments, and posterior annulus by supporting

the upper body against gravity through active

muscle force, rather than with elastic forces in the aforementioned

structures.

Three reasons could be put forward, as to why back

pain patients would need additional muscular stabilization

of their spines:

the passive stiffness of the spine is reduced as a

consequence of damage to disc or ligaments;

muscle force and consequently the capacity to correct

perturbations is reduced;

sensorimotor integration is disturbed, hampering corrective

responses.

Below, each of these possible reasons will be discussed.

LBP in many cases appears to be caused by injury to

the ligaments or discs as a consequence of mechanical

overloading [1,29,104]. These injuries and especially

those of the intervertebral discs have a substantial effect

on the mechanical behavior of the spine. In relation to

spinal stability the main effects are:

an increase of the neutral zone, the part of the movement

trajectory where stiffness is minimal [62,63].

a reduction of stiffness outside the neutral zone

[62,63,105].

an increase of the range of motion [62,63,105].

Although it is difficult to verify the existence of such

changes in living subjects, there is ample evidence that

in many LBP patients the stiffness of affected spinal

motion segments is decreased and the range of motion

increased [36,69,112,114]. These changes will lead to a

reduced spinal stability as evidenced by the fact that

injured spinal motion segments buckle under lower compression

forces as compared to intact specimens [136].

This buckling occurs at a very high velocity, which

might preclude muscular corrections. Consequently,

adaptations of muscular activation might be required to

compensate for the reduced stiffness. This could involve

increased cocontraction [18,20,40] and adaptations in

load sharing between extensor muscles [26]. With

respect to the increased LES activation found in patients

in full trunk flexion, it seems likely that this is explained

by the need to limit the segmental range of motion

[64,116]. It could limit the excursion of the vertebrae

with respect to each other where the passive stiffness is

insufficient to do so.

Many studies have shown a reduced trunk muscle

force in LBP patients as compared to healthy controls

[72,90,126,127]. This is not only caused by a lack of

maximal activation during the test, since wasting of the

extensor muscle mass [44,45] and a loss of type II fibers

[83] have been demonstrated. Furthermore, several studies

have shown that LBP patients are characterized by a

reduced endurance of the trunk extensor muscles

[13,79,89,95,111]. On the basis of these findings it is

likely that LBP patients will be less able to rapidly

develop trunk muscle force. This would limit their

capacity to correct perturbations of trunk equilibrium

and prevent spinal instability.

Injuries of spinal ligaments are likely to cause a disturbance

of the control of trunk equilibrium, since ligaments

have been shown to have an important sensory

function in feedback control of joint position [58,117].

This function probably also holds for injuries of the

annulus fibrosus, which is richly supplied with mechanoreceptors

[108,139]. A number of animal experiments

and limited experimentation in humans has indeed

shown a reflexive coupling between damage to the annulus

and ligaments on one hand and the activity of the multifidus and longissimus muscles on the other hand

[51,54,55,119,123,119a]. In addition, reduced proprioception

[15,37] and disturbances of postural control

[80] have been found in back pain patients. The same

study demonstrated increased reaction times in LBP

patients. Impaired postural control in back pain patients

is associated with a reduced capacity to react to perturbations

of trunk equilibrium [107].

In summary, all three subsystems that subserve spinal

stability (the passive system, the control system and the

muscular system [98]) appear to be affected in LBP,

though not necessarily all at the same time. The

reduction in passive spinal stiffness would increase the

risk of buckling under perturbations, whereas the

changes in the control system and muscular system

would impair the capacity to respond to perturbations.

Recent studies have indeed confirmed that LBP patients

respond less adequately to trunk perturbations [82,106].

Alterations of trunk muscle recruitment patterns might

be functional in LBP patients, since they would stiffen

the trunk, thus precluding the probability of perturbations

to which the patient could not adequately

respond.

The above rationale might also account for the disappearance

of anticipatory muscle activation and the

consequent anticipatory postural adjustments in back

pain patients preceding fast limb movements

[46,47,49,50]. Healthy subjects generally use specific

anticipatory control to counteract ensuing perturbations

of trunk equilibrium. For instance when raising the arms

forward, the back muscles are activated before the arm

muscles to counteract the flexing torque on the trunk

imposed by the arm movement [11,33,48,140]. This

anticipatory activation is specific with respect to the

magnitude and direction of the ensuing perturbation and

antagonistic cocontractions appear to be absent

[11,33,48,140]. When moving the arms, specific anticipation

is possible since the perturbation is entirely

internal. However, it turns out that this can be generalized

to situations in which the mechanical nature of

the perturbation is to a lesser extent under the control of

the subject. A good example of an everyday activity in

which the mechanical nature of the perturbation is a

priori unknown to the subject is the lifting of a load. The

mass and the center of mass position can only be known

when the object has already been lifted [60,61]. It

appears that also when lifting loads the trunk extensor

musculature is activated in an anticipatory fashion,

whereas no anticipatory activation of the antagonistic

abdominal muscles is found [27]. In addition, this anticipatory

activation was shown to be specific with respect

to load magnitude [77] and center of mass position [27].

With an incorrect expectation of the ensuing perturbation,

anticipatory actions will actually further disturb

equilibrium [128]. Therefore, in cases where the possibilities

for corrections are limited, such as appears to be

the case in LBP patients [81,106], specific anticipation

might impose a risk. For this reason, patients might prepare

by cocontracting muscles, as healthy subjects do

when specific anticipation is impossible [71]. Furthermore,

anticipatory postural adjustments might be suppressed

as was found in healthy subjects when the risk

associated with balance loss was high [2] and in patient

groups with problems in maintaining whole-body equilibrium

[70]. Finally it is conceivable that anticipatory

adjustments are less relevant in patients, since the muscle

activation patterns, such as elucidated in the present

study, provide them with adequate stability to deal with

ensuing perturbations [50a]. Increased coactivation

could even account for the delayed and reduced

responses after a perturbation [81,106], since prior activation

of muscles reduces the amplitude of responses in

these muscles to perturbations [122].

Besides the positive effect of pain-related changes in

muscular recruitment, some negative consequences are

likely to occur. Hyperactivity could cause pain in the

muscles themselves, as proposed in the pain–spasm–pain

model. In addition, increased cocontraction will increase

the forces acting on the spine [28,40]. However, the

increased activation reported in the literature review is

generally only small or moderate. Maybe more

importantly, both cocontraction and selective derecruitment

of muscles will limit functional abilities of patients.

It is possible that changes in recruitment of trunk musculature

remain present after their functional significance

has disappeared, when injured structures have recovered.

In chronic LBP, aspects of pain behavior in many cases

appear to remain, whereas the physiological cause may

no longer be present [34,132]. The fear of movement

and re-injury, which characterizes many LBP patients

[133], could underlie the remaining changes. This

assumption is supported by the relationship found

between hyperactivity and other aspects of pain behavior

[4] and by a study indicating that mental and hence

mechanically irrelevant stressors may trigger the

responses observed in muscle activation [35]. However,

these effects of mental stressors were not generally

observed [22,23,38]. Nevertheless, caution should be

exercised when rehabilitating LBP patients with the sole

purpose of restoring a normal muscle recruitment pattern.

The ‘abnormalities’ may represent compensation

mechanisms to stabilize the spine. The problem a clinician

is facing then, is to differentiate between adaptive

changes and unwanted residuals of former injury.

Further scientific research may provide the knowledge

and methods to support clinical decision-making. 

7. Conclusion

Findings on trunk muscle recruitment in LBP patients

fit neither the pain–spasm–pain model, nor the pain adaptation model. The changes observed are task-dependent,

related to the individual problem and hence highly

variable between and probably within individuals. We

propose that the alterations in trunk muscle recruitment

in patients are functional in that they reduce the probability

of noxious tissue stresses by limiting range of

motion and providing stabilization of the spine. This

explanation should be tested in future experiments

specifically designed to refute hypotheses derived from

it.