Electrical stimulation o f the digital nerves o f the index finger at a strength 2.5 times the threshold for perception produced a reflex modulation o f the ongoing average rectified EM G recorded from the I D l hand muscles. In healthy subjects, this CMR response was usually triphasic consisting o f an initial increase, E l, in average rectified EM G followed by a decrease, 11, and then a second increase, E2. Earlier studies o f cutaneomuscular reflexes in ID l following stimulation o f the digital nerves o f the index finger were carried out by G arnett & Stephens (1980). They dem onstrated that during a sustained voluntary contraction the first increase in average rectified EM G occurred at approximately 30ms after the stimulus and termed this the E l com ponent, and also noted that this latency was similar to that o f the earliest reflex response obtained following a mechanical tap o f the muscle belly.
Origin of the E1 component
Using the same experimental paradigm, Jenner & Stephens (1982) carried out additional investigations which included measuring the latencies o f F and M waves in order to calculate the peripheral afferent and efferent conduction times and to estimate central
delay. Based on the latencies o f these F and M responses, and the extremely short central delay (4.4ms), they concluded that the E l com ponents m ust be generated in the spinal cord. The presence o f a short latency spinal E l com ponent present between 30-40ms following digital nerve stimulation is now widely accepted, based on several studies in the
upper limb o f both healthy adults and stroke patients (Evans et al 1989; Bruce et al 1991;
Chen & Ashby 1993; Chen et al 1998a&b). A n elegant study recently published has
furtherm ore clearly dem onstrated reflex coupling between single cutaneous afferents and
m otoneurones in humans (McNulty et al 1999). In these authors’ study, activity in single
cutaneous afferents o f the median nerve was recorded using tungsten electrodes while delivering cutaneous stimulation by stroking the skin o f the index finger, and at the same time subjects produced a weak voluntary muscle contraction o f ID I. Spike triggered averaging showed changes in EM G time-locked to the afferent discharge, and stimulation o f single afferents produced a short latency reflex response in ID I, thereby confirming that spinal reflex com ponents are produced in ongoing voluntary EM G in response to impulses in cutaneous afferents.
O ther studies have also reported modulation o f ongoing voluntary EM G in response to electrical stimulation. U pton and colleagues (1971) directly stimulated the nerves at the wrist and recorded a biphasic reflex modulation o f ongoing EM G in various hand muscles which occurred at a spinal latency and also a response 25ms later. Subsequently, Caccia and colleagues reported that electrical stimulation o f the digital nerves o f various digits o f the hand produced a polyphasic modulation o f the ongoing average rectified EM G recorded from abductor poUicis brevis (Caccia et al 1973). In their study they did not identify an early spinal response but observed two increases in average rectified EM G, the first
occurring at an onset latency o f 75ms with each increase preceded by a decrease (Caccia et
al 1973).
Reflex com ponents obtained following electrical stimulation are n o t solely confined to the upper Hmb. In healthy adults, electrical stimulation o f the second toe produces reflex com ponents in voluntary EM G recorded from extensor digitorum brevis in the foot
T he main advances relative to the study by Jenner & Stephens (1982) are that by investigating CMR responses in a hand muscle over a prolonged period o f time we have been able to relate some o f these results to the patient’s function. W e have shown that exaggerated E l spinal responses early post-stroke may predict a poor functional outcome. In a patient who showed slow, progressive recovery to a moderate level we have also shown a relationship between the size o f the E2 cortical com ponent with these im provements in function.
Using physiological forms o f cutaneous stimulation has also shown modulation o f ongoing voluntary EM G . Using air pu ff stimulation delivered to the tip o f the digits or the naü beds in the hand produced reflex responses in the ongoing EM G recorded from ID I, abductor
poUicis brevis and abductor digiti rnmirni (Deuschl et al. 1995). Mechanical stimulation
using a small m etal disc to indent the skin o f the Up vermiUon produced reflex responses in the peri-oral muscles (Wohlert 1996).
Johansson & Westling (1987) also investigated reflex responses following cutaneous stimulation during functional activities. If a subject picked up an object between the index finger and thum b and a perturbation was deUvered which caused the object to sUp out o f the subject’s grasp, reflex com ponents were recorded from ID I at a latency o f 60-80ms. N o early shorter latency responses were shown prior to this time.
What happens to the E 1 componentsfollowing stroke'^
In the present study one o f the main results was that for 4 /5 patients w ho showed a poor recovery o f m o to r function, the sizes o f the E l com ponents recorded from ID I were exaggerated com pared with the non-stroke side and com pared with healthy control subjects, and the II and E2 com ponents were usually absent. N one o f the patients with either m oderate or good recovery showed exaggerated E l com ponents and the II and E2 com ponents were usually present.
Exaggerated E l com ponents with absent or smaller II and E2 com ponents have been previously reported in ID I following digital nerve stimulation for patients either with stroke or a dorsal column lesion (Jenner & Stephens 1982; Choa & Stephens 1982). However, a recent study o f CMRs recorded from ID I from patients w ho had suffered a lacunar stroke reported that E l com ponents were usually very small and difficult to detect (Chen et al. 1998a). Although this appears to contrast with our results, all the patients in their study were able to produce a relatively strong contraction o f ID I, at grade 4 /5 on the MRC scale (able to contract muscle against gravity and with some external resistance) at their initial assessment, within the first week after stroke. Patients followed up 1 m onth
com ponents recorded from the stroke with those recorded from the non-stroke side. In a study o f patients with pure sensory stroke due to a thalamic lesion, CMRs recorded from thenar muscles following stimulation o f the median nerve also failed to produce exaggerated E l responses but these patients had not shown any m otor deficits following their stroke (Chen et al. 1998b).
Other reflex studies of the central nervous system
The discovery that reflex responses are changed following damage to the central nervous system is n ot new. As long ago as 1840, Marshall Hall noted that stroke patients were more sensitive to cutaneous stimulation and Romberg (1853) reported that gently stroking the sole o f the affected foot produced a series o f spasms. Babinski showed that stimulating the lateral border o f the foot with a blunt instrum ent produced a reflex response which involved extension o f the great toe and fanning o f the other toes not normally observed in healthy adults. Tendon reflexes were independently described by E rb (1885) and W estphal (1885) w ho reported that these were also exaggerated in lesions o f the spinal cord but depressed in Tabes Dorsalis.
While Marshall HaU is credited as the first person to develop the concept o f a sensorimotor reflex arc it was BeU (in London) and Magendie (in France) w ho noted that m ovem ent o f the limb only occurred when the anterior roots but no t the posterior limbs were stimulated and thus they distinguished between m otor and sensory nerves (reviewed by Louis & Kaufman 1996). D uring the end o f the 19th and the early part o f the 20th centuries, Sherrington attem pted to clarify the organisation o f the spinal reflex arc and elucidate the mechanisms underlying changes in spinal reflex responses following damage to the CNS by studying the effects o f lesions to the spinal cord and brain perform ed on cats and dogs. He dem onstrated that the quadriceps tendon jerk in cats could be abolished by sectioning the dorsal roots and thus confirmed the reflex nature o f this response. Sectioning the brainstem between the superior and inferior coUicuh produced exaggerated stretch reflexes and intense stiffness (decerebrate rigidity) in the muscles, particularly those with an antigravity (extensor) action in the leg o f the cat. This stiffness was abolished if the dorsal roots were then cut and therefore he concluded that the increase in muscle stiffness was due to muscle activity resulting from ongoing unopposed activity in the muscle afferents. It was also noted by Sherrington that the effects on muscle tone depended on the site o f the lesion; if perform ed above the superior coUiculi, this rigidity did not occur.
In humans it is also well recognised that following an upper m otoneurone lesion, for example after a stroke, patients display abnormal postural tonus with abnormal co ordination o f m ovem ent patterns (Bernstein 1967). D uring the early stage after a stroke (days to weeks), muscle tone may be initially decreased with absent or decreased tendon jerks (Bobath 1990). Subsequently these clinical features are reversed and there is a velocity-dependent increase in muscle tone, spasticity (Lance 1980), and tendon reflex responses show an increased amplitude, have a lower threshold to response, and may be elicited from a wider distribution o f muscles (RothweU 1994).
Mechanisms which maj account for the exagération of the H1 components
Based on this earlier w ork in the 19"^^ and early 20*^^ centuries it became accepted that activity within spinal reflex pathways was normally under descending inhibitory control from inputs which originated from supraspinal or “higher centres” in the brainstem and the brain. T o discuss how this relates to cutaneous reflexes we will first consider w hat is known about the spinal cord circuitry mediating these effects.
Cutaneous reflex pathways in the spinal cord
In order to study the reflex effects o f cutaneous afferent inputs on motoneurones, Burke
et al (1970) electrically stimulated the sural nerve o f the hind limb (at 1.2 and 5 times
threshold) in neurologically intact anaesthetised cats. Post-synaptic potentials were recorded from spinal cord lumbar m otoneurones which innervate triceps surae muscles and these PSPs consisted o f both excitatory (EPSPs) and inhibitory com ponents (IPSPs). M ore recently, Wada et al (1999) also showed that electrical stimulation o f the sural, tibial and superior peroneus cutaneous nerves o f the hindhm b evoked b oth EPSPs and IPSPs in m otoneurones which innervate the spinal and abdominal muscles in decerebrate and in spinal cats. In their study, stimulation was perform ed at 2 and 5 times threshold thereby activating low threshold mechanoreceptors. Based on the segmental latencies o f responses to cutaneous stimulation being between 1.2-1.5ms, it is generally agreed that the shortest neuronal circuit from cutaneous afferents innervating m otoneurones in the hindhmb o f cats is disynaptic (Hongo et al 1969; lllert et al 1975; Fleshman et al 1988). Thus, in the
M ore recently, studies investigating spinal cord circuitry have been performed in macaque monkeys while they produced isometric flexion/extension torques around the wrist (Maier
et al 1998; Perlm utter et al 1998). Activity from C6-T^ intemeurones was recorded and
spike triggered averages o f E M G recorded from the forearm muscles were performed. Results showed that m ost o f the pre-m otor intem eurones produced post-spike effects in only one target muscle and the changes in post-spike EM G were consistent with m ono synaptic or oligosynaptic connections to motoneurones. It was suggested that because pre m otor intem eurones were twice as likely to influence the activity o f flexor muscles compared with extensor muscles that the flexors may be controlled m ore directly by
intem eurones whereas extensors could have m ore direct supraspinal inputs (Perlmutter et
al 1998). W ith respect to the response firing patterns, m ost intemeurones showed phasic-
tonic activity followed by purely tonic and purely phasic activity, and m ost were active
during both flexion and extension (Maier et al 1998). The differences between
corticom otoneuronal cells and spinal pre-m otor intem eurones in terms o f voluntary m ovem ent is summarised by Fetz et al (1999). CM cells usually fire with either flexion or extension, have larger muscle fields often producing co-ordinated activity in groups o f muscles and involve reciprocal inhibition o f antagonists. Pre-m otor intemeurones usually suppress or facilitate or suppress specific muscles but are bi-directionally active. Therefore, a more explicit pattern o f co-ordinated muscle activity is shown by CM cells whereas spinal intem eurones affect particular muscles (Fetz et al 1999).
relative amounts o f excitaton^ (EPSPs) and inhibitor)^ (IPSPs) inputs via intemeurones excited by afferent input.
While there may be “private” chains o f intemeurones mediating reflex responses produced by cutaneous afferent input, available evidence in cats suggests that this is not the case. In fact there is considerable convergence of different t\ pes o f peripheral afferent inputs onto the same intemeurones (reviewed by Lundberg 1975; Lundberg 1979; Jankowska 1992). For example, in addition to inputs from cutaneous afferents, lb inhibitory intemeurones in the spinal cord also receive inputs from golgi tendon organ afferents and joint afferents (Hultborn 1972). These lb inhibitor}' intemeurones also receive inputs from descending pathways including the mbrospinal and pyramidal pathways in cats (Lundberg 1979) and from the corticospinal tract in primates (Jankowska & Tanaka 1974; Lundberg & Voorhoeve 1962). The convergence o f inputs to the lb inhibitor}^ interneurone is represented diagrammatically below in Figure 1-45