Characterization of spinal somatosensory neurons having receptive fields in lumbar tissues of cats

85

Pain, 54 (1993) 85-98 0 1993 EIsevier Science Publishers B.V. All rights resewed 0304-3959/93/$06.00

PAIN 02313

Characterization

of spinal somatosensory neurons having receptive fields in lumbar tissues of cats

Richard

G. Gillette

b, Ronald

C. Kramis

a and William J. Roberts

a

’ R.S. Lkw Neurological Sciences, institute Good Samaritan Hospitul and Med~aI Center, PortIand, OR 97209 @JSA) and b Division of Basic Science, Western States Chiropractic College, Portland, OR 97230 (USA) (Received 19 October 1992, revision received 25 January 1993, accepted 1 February 1993)

Summary In pentobarbital anesthetized cats, extracellular unitary recordings were made from neurons in the extreme lateral dorsal horn of spinal segments LA-5 All 118 units reported had receptive fields in deep somatic tissues and/or skin of the lumbar region, hip and/or proximal leg. Neurons were functionally characterized according to their responses to non-noxious and noxious mechanical stimuli and to injections of algogens. Most neurons (92%) were either wide-dynamic range (WDR) or nociceptive specific (NS), and most of these had very large nociceptive receptive fields in the back/hip/leg that included both skin and deep somatic tissues innervated through both the dorsal (back/hip) and ventral (leg/ventral spine) rami. Most (72%) were ‘hyperconvergent’ in that they were responsive to stimulation of many different somatic tissues including skin, muscles, facet joint capsules, ligaments, and periosteum. Some units were tested and found also to be activated by noxious stimulation of spinal dura and ventral annulus fibrosis and ventral longitudinal ligament. Twelve of 22 neurons tested were found to have ascending axons extending beyond ThlO. The nocireceptive neurons (NS and WDR) in the population tested are suitable for processing information about tissue damage in deep somatic tissues in the back, hip and proximal leg. The apparent relative paucity of such neurons and their very large hyperconvergent receptive fields suggest that sensations served by these neurons, such as’low back and referred leg pain, would be neither well localized nor attributable to pathology in a specific tissue. These deductions, based on physiological characteristics in cats, are consistent with clinical reports from humans who experience pain as a consequence of spinal or paraspinal injuries. Key words: Vertebral joint; Muscle; Dura; Algogen; Nociception;

Introduction LOWback pains are a major medical problem in the industrialized societies (Loeser et al. 1990). The diagnosis and treatment of pains localized in the low back are complicated by the existence of multiple mechanisms, all of which may contribute to pain, and by the lack of consensus regarding differential diagnostic criteria and therapeutic procedures, as reviewed earlier (Kellgren 1977; Bogduk 1980, 1983; Mooney 1987).

Co~es~ndence io: William J. Roberts, Ph.D., R.S. Dow Neurological Sciences, Institute Good Samaritan Hospital and Medical Center, 1120 N.W. 20th Avenue, Portland, OR 97209, USA. Tel.: (503) 229-7502; FAX: (503) 229-7229.

Back pain

Surprisingly, there seem to have been no systematic studies done of spinal sensory neurons serving this region (Janig 1985), and the likelihood that persistent changes in central nervous system function play a substantial role in low back pain has to date received little emphasis (Frymoyer and Gordon 1989a). The study reported here was an effort to characterize spinal somatosensory neurons subserving acute nociception in lumbar tissues. Sensory input from deep tissues in the lumbar region was a specific focus in this study, as low back pain is most commonly referred to deep somatic tissues, not skin (Kellgren 1977; O’Brien 1984). Anatomical investigations of the primary afferent systems serving noncutaneous tissues around the lumbar spine include studies in humans (Hirsch et al. 1963; Jackson et al.

86

1966; Groen et al. 1988; Groen et al. 1990; Coppes et al. 19901, rodents (Kojima et al. 19901, carnivores (Bogduk 1976) and non-human primates (Stillwell 1956). Other studies of afferent fiber projections from lumbar paraspinal tissues have indicated that small caliber somatosenso~ fibers inne~ating one locus project to the spinal cord over multiple segmental nerves (Forsythe and Ghoshal 1984) and diverge widely within the spinal grey matter (Roberts et al. 1989). This anatomical work has been supplemented by a few single-fiber electrophysiological studies that describe the nociceptive response properties of deep paraspinal afferents with somatosensory receptive fields restricted to the anterior spinal column (Bahns et al. 19861, root pia (J&rig and Koltzenburg 1991) and posterior spinal structures (Cavanaugh et al. 1989; Yamashita et al. 1990). In addition, a small number of lumbar somatosensory neurons have been shown to have superficial receptive fields involving the hip or low back in the cat (Devor and Wall 1976; Fields et al. 1977; Light and Durkovic 1984; Ammons 1987, 1988; Collins 1987; Saito et al. 19901, rat (Ness and Gebhart 1988; Knuepfer et al. 1988), and monkey (Applebaum et al. 1975; Mime et al. 1981; ~mons 1989). We know of no studies that have elucidated the characteristics of central neurons responsive to noxious stimulation of deep lumbar paravertebral structures. It is not known, for example, whether nociceptive sensory processing mechanisms that serve deep somatic tissues around the lumbar spine are similar to processing mechanisms described previously for deep somatic tissues of the extremities (Foreman et al. 1977; Meyers and Snow 1982; Thies 1985; Schaible et al. 1987; Hoheisel and Mense 1989, 1990; Yu and Mense 1990) or for deep tissues served by the trigeminal system (Amano et al. 1986; Davis and Dostrovsky 1986; Sessle et al. 1986; Broton et a1. 1988; Sessle and Hu 1991). Studies in humans have shown that noxious mechanical and chemical stimulation of diverse paraspinal tissues produces diffuse low back and hip pain which often radiates into the proximal leg; tissues stimulated included lumbar spinal joints (Hirsch et al. 1963; Mooney and Robertson 1976; McCall et al. 1979; Fairbank et al. 1981; Kuslich and Ulstrom 1991) ligaments and periosteum (Kellgren 1939; Inman and Saunders 1944; Kuslich and Ulstrom 1991); paraspinal muscle (Simons and Travel1 1983; Kuslich and Ulstrom 1991); intervertebral disc (Fernstrom 1960; Hirsch et al. 1963; Park et al. 1979; Kuslich and Ulstrom 1991); and spinal dura @myth and Wright 1958; El Mahdi et al. 1981; Kuslich and Ulstrom 1991). The uniformity of symptoms arising from provocative stim~ation of each of these structures has suggested to some that there must be an underlying population of spinal neurons that receive convergent input from all of the clinically relevant deep spinal sources (Kellgren 1977; J&rig 1985) -

not unlike the convergent-projection mechanism postulated for viscera-somatic pain and referral by Ruth (1946). These issues prompted us to undertake experiments in cats to determine the response properties of spinal somatosenso~ neurons activated by noxious mechanical and chemical stimulation of deep lumbar paraspinal structures including joints, periosteum, muscles, ligaments, dura and intervertebral discs. Preliminary reports of this work have been published (Gillette et al. 1990, 1991)

Methods Subjects and preparation Experiments were conducted on 27 adult cats of either sex weighing 2.2-4.0 kg and anesthetized with sodium pentobarbital (35-40 mg/kg, ipf. Anesthetic supplements (2.5 mg) were given i.v. whenever toe pinch elicited a flexion reflex or pressor response. Ringer’s solution was continuously infused through a cephalic vein at 20 ml/h. Cannulae were placed in the right common carotid artery to monitor arterial pressure and in the bladder to collect urine. The experiment was terminated if the mean arterial blood pressure fell below 80 mm Hg. The animals were ventilated without paralysis to maintain end-tidal CO, between 3.5 and 4.5%. Rectal temperature was maintained at 38 + O.S”C. A series of 2-4 microlaminectomies, l-2 mm in diameter, were made in the L4 vertebra for access to the lateral dorsal horn. A very small incision was made in the dura just prior to recording to permit m~~roelectrode entry while minimizing loss of cerebrospinaf fluid. This ‘micro’ technique provided better spinal cord stability and minimized the development of spinal bleeding and edema that occurs with large dural incisions during long experiments (24-56 h). A small hole was made in the pia with forceps prior to each electrode penetration to minimize dimpling of the spinal cord. The spinal pool was covered with warm mineral oil maintained at 38°C with a heat lamp. Another laminectomy was performed at ThlO in 11 animals to permit transection under visual control of the right dorsolateral funiculus (DLF) to block conduction in most descending modulatory fibers (Hoheisel and Mense 1990) and to allow implantation of pairs of insulated tungsten microelectrodes (1 mm tip exposure) in the left ventrolateral funiculus (VLF) and the right DLF. These stimulating electrodes were used to identify neurons having long ascending axons (0.1 msec pulses, lo-30O/sec, 0.1-0.6 mA). Standard criteria were applied to identify antidromic responding (Lipski 1981). To irreversibly block paraspinal muscle contractions in response to spinal stimulation at ThlO, 95% ethanol was injected bilaterally into paravertebral muscles at multiple sites in Th9-11 (3 ml total). Systemic neuromuscular blocking agents were not used, as they would be expected to attenuate synaptic transmission in sympathetic ganglia (Bell et al. 19851, this being needed for tests of s~pathet~~ trunk stimulation not reported here. The animal was mounted in a spinal frame after first injecting deep tissues and skin with long-acting local anesthetic at the sites to be used for hip pins and laminectomies. This was done to minimize the ‘sensitization’ of spinal neurons due to nociceptor activation. Supplementary injections were not made.

Recording techniques Extracellular single-unit recordings were made through glass micropipettes filled with either 3 M NaCl or Pontamine Sky Blue (Z%,

87 Mult!f

@-

1(‘

.4

Disc Stim

Fig. 1. Experimental setup. Shown collapsed onto a single transverse plane are: the lateral location of the recording pipette in LA-5; the suture used to stimulate the disc (Disc Stiml including both the annulus fibrosis (AF) and ventral longitudinal ligament (VLL); the lower lumbar injection sites for stimulating multifidus muscles (Multif) and facet joints (Facet) innervated by the dorsal ramus (DR); and the ventral ramus (VR). Also shown are the positions of the antidromic stimulating electrodes in the contralateral VLF and ipsilatera1 DLF in thoracic segment TlO.

in 0.5 M sodium acetate) and impedances of 2-8 Ma. Amplified voltages from the pipette were filtered (l-10 kHz1 and displayed using conventional recording equipment. Action potentials from isolated single units were converted to pulses with a window discriminator and recorded with a special purpose computer (CED 1401).

Search strategy The recording pipette was advanced in 6 pm steps through the lateral dorsal horn as illustrated in Fig. 1 while the lumbar back, hips and hind limbs were manually stimulated, first at non-noxious and then just noxious intensities (brief pinch and/or sustained pressure onto skin and deep tissues). A recorded unit was presumed to be a neuron, not a fiber in passage, if the recorded impulses had a predominant negative component of less than a few millivolts and if stimulus-response characteristics and receptive field properties of the unit differed from those of known classes of superficial and deep primary afferents (Bahns et al. 1986; Mense 1986; Willis and Coggeshall 1991b). Indeed, we rarely encountered units with large-magnitude negative spikes in our investigation. Prior to receptive field mapping and the application of prolonged noxious stimuli, background activity of each neuron was assessed from a 60-see epoch taken from the peristimulus time histogram.

Unit stimulus-response

classification

The non-noxious cutaneous receptive field (if existent) of each unit was mapped using brief non-noxious mechanical stimuli delivered with blunt probes or cotton swabs. Subsequently, noxious mechanical stimuli were briefly (l-5 secl delivered by lifting and pinching small folds of skin with serrated forceps, using force just sufficient to produce pain when applied to the dorsal hairy skin of the experimenter’s hand. The non-noxious and noxious cutaneous receptive fields of each unit were mapped separately onto a drawing. Each isolated neuron was characterized as low-threshold (LT), nociceptive-specific/high-threshold (NS), or wide-dynamic-range/multire-

ceptive (WDR) according to conventional criteria (Mendell 1966; Dubner and Bennett 1983). The superficial receptive fields of isolated neurons were divided into three groups according to relative size. The boundaries of ‘small’ receptive field units were restricted to small areas of the low back, hip or proximal leg. However, these small receptive field areas were not as small as those found in the distal leg (Meyers and Snow 1982; Wilson et al. 1986). ‘Intermediate’ sized cutaneous fields encompassed either the hip and proximal leg or the back and hip region. ‘Very large’ unit receptive fields included large portions of the caudal back, hip and proximal leg. A search was then made for deep receptive fields for each unit, initially by mechanically stimulating skin plus deep tissues and comparing the unit’s responses to those evoked by stimulation of skin alone. Additionally, deep structures were, in many cases, stimulated directly with mechanical probes and von Frey filaments, the overlying tissues having been surgically mobilized hours earlier. Care was taken to protect all exposed tissues from drying. In the initial search, non-noxious forces (as judged by application to the experimenter’s forearm) were used. Subsequently, sustuined noxious pressure (5-20 set) was applied; such pressure activated many units not activated by transient noxious pressure ( < 5 secl applied to deep tissues. Dorsolateral dura was stimulated mechanically by pinching with fine forceps through a microlaminectomy hole. The anterior longitudinal ligament/annulus fibrosis was stimulated at L5 by sustained traction applied with a 5-O suture inserted through these tissues and out through the flank. An additional search for deep-tissue receptive fields was made, as illustrated in Fig. 1, with injections of algogens (100 ~1 of: 6% saline, or 100 pg/ml bradykinin, or 0.01% capsaicin, with 7.5% Tween-80 in normal saline). Deep-tissue injections of these substances have been shown in previous studies to produce pain or nocifensive responses in conscious humans and other animals (Hirsch et al. 1963; Mooney and Robertson 1976; McCall et al. 1979; El Mahdi et al. 1981) and vigorous spike activity in deep-tissue nociceptive afferents (Kanaka et al. 1985; Mense 1986). The spread of material from such injections into deep tissues was estimated in 4 animals by making postmortem injections of Pontamine Sky Blue dye (100 ~11 into deep paraspinal tissues. Dissections revealed that injections into the multifidus muscle were confined to single portions of a muscle slip while facet joint injections led to dye accumulations throughout the fibrous capsule and onto the adjacent periosteum. Consequently, joint algogen injections were presumed to activate both articular and periosteal small-caliber nociceptive afferents (Kanaka et al. 1985; Cavanaugh et al. 1991). No attempt was made to map the entire deep-tissue field for each neuron. Only certain deep areas were routinely tested (low back, hip and hind limbs both ipsi and contralateral to the side of recording) and not all areas with all stimuli. Therefore, the characterization of neurons in this study provides only a minimum estimate of the deep-tissue receptive field of each cell.

Localization of recording sites and histology Recording electrode penetrations were directed through a narrow ( < 0.3 mm) sagittal strip immediately medial to the root entry zone of spinal segments L4 and L5, to explore areas previously shown to have neurons with proximal receptive fields (Devor and Wall 1976; Light and Durkovic 1984; Wilson et al. 1986). The depth of the recording electrode tip relative to the cord surface was recorded from the stepper drive for each neuron tested. In addition, the position of the electrode tip was marked subsequent to recording for 20 neurons by iontophoretic deposit of Pontamine Sky Blue (2 PA, 10 min) (Hellon 19711.At the completion of the experiment, the lumbar spinal cord was fiied by transcardial infusion of 10% neutral formalin, removed, and stored at 4°C in 30% sucrose/O.1 M phosphate buffer solution. Dye marks were later localized by light mi-

88 ‘d

O-

‘sd

l1

I 2-

: .

Fig. 2. Distributions of recording electrode depths for neurons having receptive fields in skin only f‘s’), deep-tissues only Cd’), or both skin and deep tissues f‘sd’). All cells were isolated in the lateral gray matter and had receptive fields that included the low back, hip and/or proximal hind leg.

croscopy in 50 F-m cryostat sections, counterstained

with 1% Neutral

Red to identify Rexed’s laminae (Rexed 1954).

Statistical analyses Group differences were evaluated with 4 distribution-free fnonparametric) tests: the Wilcoxin matched-pairs signed-ranks test for repeated measures, chi-square test, Mann-Whitney U test and the Newman-Keuls multiple-range test. P values equal to or less than the 5% level (2-tailed) were considered to be significant.

Results The results to be presented here are based upon recordings from 118 neurons isolated in the extreme lateral dorsal horn of spinal segments L4-5 in 29 cats

of both sexes. All neurons had receptive fields in the low back, hip and/or proximal leg. The results from preparations with (n = 11) and without (n = 18) ipsilatera1 DLF Iesions at ThlO are combined because the neuronal characteristics observed in both populations of animals were similar in terms of the percentages of units having deep-tissue input and the percentages of neurons categorized as LT, WDR or NS. Neurons with conuergent parasp~nal inputs The most distinctive feature of the population of units sampled was the complexity of convergent afferent input onto these cells. Most neurons in our sample (72% of 118 cells) received excitatory input from skin and from multiple deep tissues Csd’ units) of the low

Rl?SS

Brush

Pinch

~.._--LI1 Brush

PWISS

Pinch

Fig. 3. Responses to skin stimulation of a representative ‘sd’ neuron. This unit was recorded from the dye-marked site indicated in the camera lucida drawing. A: histogram showing responses to brush and pinch applied within the black area in the drawing. B: histogram showing responses to similar stimuli applied to the surrounding nociceptive field (shaded area). Note the very large, bilateral receptive field. Responses of this unit to deep-tissue stimuli are shown in Fig. 4. The histogram bin width is 1 set in this and subsequent figures.

89 TABLE I OF SOMATOSENSORY

‘s’ cells

‘d’ cells

‘sd’ cells

LT (n = 10) NS (n = 261 WDR (n = 82)

10 4 13

_ 5 1

_ 17 68

Totals: 118

27 (23%)

6 (5%)

85 (72%)

Input modality

back/hip/proximal leg muscles and tendons. The ‘sd’ units (n = 85) were also responsive to stimulation of skin (cf., Figs. 3 and 6). Figs. 3 and 4 illustrate responses from one representative ‘sd’ neuron recorded at the laminae II-III border of the lateral dorsal horn in W. This unit had superficial and deep excitatory receptive fields involving the low back, hip and proximal leg. It was driven by both innocuous and noxious somatic stimulation of tissues of both the dorsal compartment (dorsal skin and muscles, facet joints and periosteum) and the ventral compartment (proximal leg skin and hamstring muscle) and was antidromically driven from the ipsilatera1 DLF. Brush, press and pinch of the skin revealed characteristic WDR responding as shown in Fig. 3A and B. The afterdischarge seen subsequent to brush stimulation in this particular unit was, however, not typical of this neuron population. Note that the receptive field crossed the midline. The most effective excitatory stimuli for this cell were injections of algesic substances into deep paraspinal muscles and joints (Fig. 4). Injection of bradykinin (10 pg in 100 PL) into the L6-7 multifidus muscle produced a slowly increasing discharge that peaked in 50 set and then declined over the next 5 min (Fig. 4A). Injection of 6% NaCl into the L6-7 facet joint capsule led to a vigorous discharge which peaked at 20 set and then steadily decreased back to baseline over the next 4 min (Fig. 4B). This cell was also activated, although less strongly, by injection of hypertonic saline into the biceps femoris muscle of the ipsilateral hind limb (Fig. 40. Algogen low

NEURON TYPES AND MODALITIES INPUT

back and hip and often from the thigh (Table I). All of the ‘sd’ units were nocireceptive, responding to either noxious input alone (17 NS cells) or to both noxious and non-noxious stimulation (68 WDR cells). These units were found widely distributed in depth throughout the lateral gray matter, 0.6-2.8 mm below the dorsal surface with the greatest numbers recorded between 1.0 and 1.6 mm (Fig. 2). Some neurons (23% of 118) had receptive fields restricted to just skin (Ls’ cells), and a very few (5%) had fields only in deep somatic tissues (‘d’ cells) (Table I). From Fig. 2, it can be seen that the ‘d’ neurons tend to be deeper than the others, although the numbers are too small to be sure about this. No neuron type appeared to be associated with a particular depth in the dorsal horn (Fig. 2). Neurons with deep-tissue receptive fields All nocireceptive ‘d’ and ‘sd’ cells (n = 91) were activated by noxious stimulation of deep, paraspinal tissues in both the dorsal and ventral compartments including two or more of the following: facet joints, periosteum, ligaments, intervertebral disc, spinal dura,

Multifidus

I-

Needle.: Facet Joint

\I

V

\”

-

L

L‘lO

Needle Biceps Femoris

i 0

Fig. 4. Responses of the ‘sd’ neuron from Fig. 3 to needle insertions followed by algogen injections into deep tissues. This neuron was activated by: (A) bradykinin in the ipsilateral multifidus muscle at L6; (B) 6% saline in the L6-7 facet joint capsule; and (Cl 6% saline in the biceps femoris muscle. Muscles shown to be part of a receptive field are indicated by dashed lines in all relevant figures.

90

TABLE II CHARACTERISTICS

Facet joint 6% NaCl Bradykinin Capsaicin Multifidus 6% NaCl Bradykinin Capsaicin Dura mater Bradyki~in

OF NEURONAL

RESPONSES TO CHEMICAL INJECTIONS

Latency of response (set)

Time-to-peak firing (see)

Increase in firing (Hz)

Half-decay time (set)

(n = 46) (n = 41) (n = 3)

21.5 ( f 23.9)SD 27.3 (+ 24.3) 10.7 ( f 2.5)

47.3 (+ 41.9) 59.8 (_t 51.8) 15.0 (rt 6.5)

36.0 ( t 46.6) 23.5 (it 28.3) 42.3 (rt 12.0)

22.0 ( + 27.5) 30.7 (+ 61.4) 16.7 (+ 4.7)

(n = 25) (n = 27) (n = 2)

15.6 I+ 10.7) 19.4 (f 16.3) 18.0 ( f 0.0)

33.3 (It 23.5) 46.8 (+ 54.7) 24.0 (?r: 6.0)

24.1 (rir 16.5) 22.5 (rt 13.9) 35.0($:11.0)

23.0(+39.1) 24.6 ( + 43.6) 21.5 (rt 1.5)

fn = 1)

20.0

65.0

18.0

20.0

= 16) = 4) = 2)

21.6 (& 11.9) 22.6 (i 11.4) 10.0

47.6 ($27.7) 58.8 (rt51.61 20.0

21.3 (t 17.8) 12.4 (+ 7.2) 42.0

12.8 (+ 10.3) 5.2(+ 2.4) 6.0

= 12) = 7) = 2)

14.9 (+ 10.7) 23.7 ( f 7.9) 15.5 ( f 5.5)

28.5 ( f 14.9) 47.0 ( f 23.0) 35.0 (rt 15.0)

24.7 (+ 18.3) 11.4 ( j, 4.2) 14.5 (?r 7.5)

ll.S(k11.5) 10.6 (i 8.7) 17.5 (2 7.5)

= 2) = 2)

8.0 (& 5.0) 21.5 (+ 10.5)

14.0 (F 4.0) 39.5 (+ 19.5)

24.0 f F 5.0) 12.0 ( f 8.0)

37.5 (+ 22.5) 90.0 (k 85.0)

Leg Biceps femoris 6% NaCl (n Bradykinin (n Capsaiein (n Quadriceps femoris 6% NaCl (n Bradykinin (n Capsaicin (n Gastr~nemius 6% NaCl (n Bradykinin (n

injections into deep lumbar muscles and vertebral joints were more effective than injections in the proximal leg in producing intense, long-lasting activity in this unit and in most nocireceptive neurons recorded in the lateral dorsal horn, as shown quantitatively in Table II. The c-fiber excitant, capsaicin, was injected into deep paraspinal and leg tissues in a few experiments and was found to be an extremely effective excitatory stimulus in all 6 ‘sd’ neurons tested (not shown). Capsaicin produced large-magnitude, short-latency responses similar to those to the other two algogens (Table II). Some units with deep paraspinal receptive fields were tested by applying von Frey probes to multifidus muscle, facet joints or intervertebral ligaments after reflection of the overlying skin or skin and muscles. Thresholds for activation from the exposed tissues were mostly less than 100 mN, as shown in Table III. The responses to sustained pressure typically occurred with a delayed onset (up to 5 set) and a slowly increasing firing rate. Records from a representative WDR unit with a cutaneous receptive on the distal lateral thigh show responses to stimulation of thigh skin (Fig. 5A), to sustained pressure applied directly onto the multifidus muscle at L7 (Fig. 531, and to hypertonic saline injected into the L6-7 facet joint (Fig. 5C). Some units that were unresponsive or minimally responsive to brief or even sustained noxious pressure on skin and deep

paraspinal muscles were responsive to deep tissue algogen injections into the same region, suggestive of input from mechanically insensitive nociceptors. Nine of 25 ‘sd’ neurons (2 NS, 7 WDR) tested with mechanical stimulation of ventral paraspinal tissues could be driven by mechanical traction on the annulus of the intervertebral disc and the associated ventral longitudinal ligament. The responsive units were recorded throughout the superficial and deep dorsal horn (from laminae I through laminae VIII>. Fig. 4 illustrates the response to disc stimulation in one ‘sd’ neuron recorded at a depth of 920 pm. The extent of the low- and high-threshold superficial receptive fields are shown in the drawing. The histogram shows the responses to mechanical stimulation of the skin followed by firm traction on the annulus and ligament of TABLE III ~RESHOLDS FOR A~IVA~ON OF NEURONS BY MECHANICAL (Van Frey) STIMULATION OF SPECIFIC PARASPINAL TISSUES Tissues timulated

Median Range

Facet joints (n = 7)

Multifidus (n = 6)

Spinal ligaments (n = 2)

38 mN 12-285 mN

62 mN 12-87 mN

62 mN 38-87 mN

Pinch 20 Multifidus

4

20 Facet

Joint

Fig. 5. Responses of an ‘sd’ neuron to pressure applied to deep paraspinal muscle. The histograms illustrate that: (A> this is a WDR unit, based on its responses to stimulation of lateral thigh skin; (B) the unit gave slowly developing responses to sustained pressure (Pressure, 280 mN von Frey) applied to the ipsilateral multifidus muscle at L7 (skin reflected); (C) the unit also responded to injection of 6% saline into that L6-7 facet joint capsule. Peak responses truncated (open arrows).

the L5 disc via a suture emerging through the flank. This and all other units were unresponsive to movement of the suture relative to the muscle, fascia and skin between the vertebral column and the flank of the animal (not shown). This neuron had a typically large deep-tissue receptive field. It responded to algesic

Pig. 6. Responses of an ‘sd’ neuron to noxious mechanical stimulation of the L4-5 intervertebral disc.This WDR unit, with a cutaneous receptive field restricted to the hind leg as shown in the drawing, responded as shown in the histogram to skin stimulation and to sustained traction on the L5 disc/longitudinal ligament (Disc). It also responded (not shown) to 6% saline injections into: the hamstring muscle group, the ipsilateral L6-7 facet joint (01, and the L5 multifidus muscle. Other deep tissues were not tested.

stimulation of deep tissues in both the back and leg, including the ipsilateral hamstring muscle group, and the ipsilateral L6-7 facet joint at sites marked in the drawing (Fig. 6). Nine of 14 ‘sd’ neurons tested (4 NS, 5 WDR) were found to be responsive to noxious mechanical or chemical stimulation of spinal dura mater. Fig. 7 is derived from a representative WDR unit isolated 1402 pm below the cord surface. This neuron was driven by all of the following: mechanical stimulation of the skin, algesic (bradykinin) stimulation of the ipsilateral LS-6 facet joint and periarticular periosteum, and algesic stimulation of ipsilateral multifidus muscle at L6-7

Fig. 7. Responses of an ‘sd’ neuron to repeated noxious pinch of the spinal dura at L4 (Dura). This WDR neuron had a cutaneous receptive field localized as shown on the hip and hind leg and was responsive to 6% saline injections into the L6-7 multifidus muscle and L5-6 facet joint (records not shown).

92

(records not shown). The histogram in Fig. 7 shows the vigorous response of the cell to pinching of the dorsolateral spinal dura at LA. The unit was unresponsive to touching the dura with forceps at the same site. In all neurons responsive to dura pinch, the discharge began abruptly, persisted throughout the period of stimulation and then returned to baseline either immediately or rapidly over the next few seconds, as shown in Fig. 7. The 2 cells tested with topical application of bradykinin to the dura were responsive to it as well as to mechanical stimulation (not shown). The possibility of leakage of the bradykinin from the spinal dura to surrounding deep tissues could, however, not be precluded. Responses to bradykinin occurred with a latency of about 20 set, peaked rapidly, and decayed over the next 3-4 min (Table II) Receptive field sizes for nocireceptive neurons

The nociceptive cutaneous receptive fields of most of the 78 nocireceptive cells recorded from the lateral dorsal horn were ‘very large’, even when determined prior to prolonged noxious stimulation; these fields included large portions of the caudal back and hip and lateral, proximal leg (n = 56). A few cells had ‘intermediate’ sized cutaneous fields including only the hip and proximal leg (n = 6) or back and hip (n = 2); others had ‘small’ fields covering only part of the leg (n = 161, the hip (n = 2), or the low back (n = 1). Receptive fields rarely extended onto the ventral body surface or the medial aspect of the leg. It should be noted that the size of the receptive field for most units was markedly state dependent - 75% of 28 neurons examined in detail showed expanded receptive fields after noxious stimulation that persisted more than a few seconds (e.g., after injection of bradykinin into lumbar paravertebral tissues). Many ‘sd’ neurons (82% of 44) had receptive fields that crossed the midline to involve the skin on the contralateral back and hip (cf., Fig. 3); these units had bilateral deep-tissue input as well (8 units confirmed, not all tested). The contralateral field was, in every case, less extensive than the ipsiIatera1 field. A comparison of the depth-distribution of units with ‘very large’, ‘intermediate’, and ‘small’ receptive fields (Fig. 8) demonstrated no significant differences (Mann-Whitney U, 2-tailed). However, there was a tendency for units with ‘small’ fields to be isolated more superficially (Fig. 8). Cells with background discharge More than one-half of the recorded neurons (54% of 118) demonstrated background activity in the absence of recent noxious stimulation in the hindquarter; note, however, that the receptive fields of many units included the ipsilateral hip pin site. Background discharge ranged from 0.1 to 14 impulses/set with a

. c : .

-

Leg Hip

I

’

Hip/Leg

Sack

’

1

i

I

Sack/Hip

I

Receptive

1

L Very Large

intermediate

Small

Back/Hip/Leg

Field Size

Fig. 8. Distributions of recording depths for units with ‘small’, ‘intermediate’, and ‘very large’ cutaneous receptive fields.

median of 1.0 imp/set. Ongoing discharge was seen in all unit types; however, ‘sd’ neurons were significantly more likefy to display background activity than were the other unit types (P < 0.001, ,y* = 14.1, df = 1). The distribution in depth of units with background discharge was significantly different from that of units without background discharge, the latter being found more superficially (Fig. 9, P < 0.001, Mann-Whitney U, U = 1151.0, 2-tailed). Stimulus-induced suppression of activity

Ongoing activity was attenuated by noxious and/or non-noxious stimulation in 20% of the ‘sd’ population (17 of 85). For most of these units, noxious pressure caused a reduction in ongoing activity when applied to deep somatic tissues in a region of the low back or proximal Ieg, and this ‘inhibitory’ field was located

.

: Cell Type .s’

‘d’

‘sd’ Background

l

-s.

‘ci‘

‘sd’

Discharge

Fig. 9. Distributions of recording depths for units with (filled symbols) and without (open symbols) backgmund discharge.

93

within a larger area in which noxious pressure had excitatory effects (not shown). The excitatory responses of these neurons to stimulation of other tissues were indistinguishable from the rest of the ‘sd’ population. Five of these cells ‘inhibited’ by local stimulation were also ‘inhibited’ by distant noxious stimulation, with a time course characteristic of ‘diffuse noxious inhibitory controls’ (DNIC) (Le Bars et al. 1979; Morton et al. 1988). Most neurons were not tested for the effects of distant stimulation. Neurons with ascending axons

Antidromically conducted action potentials were observed in 12 of 22 neurons tested. Conduction velocities for these ranged from 7.3 to 90 m/set with a mean of 36.5 m/set f 23 m/set SD. Eleven of the 12 antidromically activated neurons were ‘sd’ nocireceptive neurons (9 WDR, 2 NS cells). The remaining neuron was a superficially isolated (925 pm), slowly-conducting NS neuron with a small, skin-only receptive field. The 12 neurons were antidromically activated with stimuli localized in either the ipsilateral DLF (n = 6) or the contralateral VLF (n = 6).

Discussion Hyperconvergence

onto lateral dorsal horn neurons

A major finding of this investigation was that most of the nocireceptive (NS and WDR) neurons isolated in the lateral border of the cat lumbar dorsal horn receive somatosensory input from many different deep somatic tissues and from skin in the low back, hip and proximal leg - a sensory organization we will refer to as ‘hyperconvergence’. In this preparation we found some neurons (23%) that were responsive to stimulation of skin only, but all neurons responsive to noxious stimulation of a deep somatic tissue were responsive to stimulation of multiple tissues. Functional consequences of this hyperconvergence will be discussed at the end of this section. Previous studies of nocireceptive neurons serving the face and head, another axial region, showed analogous patterns of hyperconvergence. For example, Sessle and colleagues (Amano et al. 1986; Sessle et al. 1986; Sessle 1987; Broton et al. 1988; Sessle and Hu 1991) reported that a large proportion of nocireceptive neurons in trigeminal subnucleus caudalis receive both nociceptive and non-nociceptive inputs from many facial tissues including skin, perioral, tooth pulp, visceral, joint, and neck muscle structures in both the cat and rat. Those skin/deep neurons, like ours, tended to have larger superficial receptive fields than neurons unresponsive to deep somatic stimulation; such neurons were recorded throughout the depth of the medullary dorsal horn (Amano et al. 1986). Conver-

gence onto single spinal neurons from multiple deep tissues in the hind limb has been demonstrated by others (Schaible et al. 1986, 1987; Hoheisel and Mense 1990). No input from axial tissues was reported in those studies; however, stimulation was confined to the hind limb. The present study is the first, to our knowledge, in which lumbar somatosensory neurons have been shown to be responsive to stimulation of spinal dura and intervertebral discs, structures thought to be major contributors to low back pain in humans (El Mahdi et al. 1981; Mooney 1987; Kuslich et al. 1991). Note, however, that in our neuronal sample, all neurons responsive to dura and/or disc were also responsive to stimulation of other paraspinal tissues, reinforcing the concept of hyperconvergence. Our finding of responses to stimulation of lumbar dura . complements recent findings from subnucleus caudalis in cats which showed that many nocireceptive neurons were responsive to noxious stimulation of cranial meninges and associated vascular elements as well as to many other superficial and deep facial structures (Davis and Dostrovsky 1986; Dostrovsky et al. 1991). The percentage of nocireceptive neurons responsive to deep-tissue stimulation that were also responsive to stimulation of skin in the present study (93%, n = 91) is greater than the 71% reported by Hoheisel and Mense (1990) in comparable units with d&al receptive fields in cats. This difference may reflect an anatomical or functional difference in convergence of primary afferents onto neurons subserving somatosensation for proximal versus distal regions, although it may be due to differences in anesthetics, search strategies, or other technical differences between this and other studies (Hoheisel and Mense 1989; Mense 1991). Broton et al. (1988) reported a ratio of ‘sd’ to ‘d’ units that is similar to ours in a small population of nocireceptive subnucleus caudalis neurons in cats. In a study of deep nociceptive input from the hind limb of the rat, Yu and Mense (1990) reported a much lower incidence of ‘sd’ units (42%) and a much higher incidence of ‘d’ cells (58%) than was observed in our study. Interestingly, Foreman’s group (Hobbs et al. 1992) has presented data which indicates that spinothalamic neurons in the middle segments of the lumbar and cervical enlargements of the monkey are likely to have strictly cutaneous receptive fields, whereas SIT neurons in the adjacent segments (as in our own investigation) have predominantly deep and/visceral input. The percentage of neurons with background discharge in the present study (54%) is similar to that reported by Schaible et al. (1987) for cells with distal deep-tissue receptive fields. In the population of neurons reported here, all of those located deep in the gray matter (below 1.6 mm) had a background discharge.

94

Size and exfent of s~pe~~c~ai receptiue fiekis

The superficial receptive fields in the majority of our nocireceptive units were very large, even when determined prior to noxious conditioning stimulation. The fields typically included portions of the back, hip and posterolateral thigh to the knee (e.g., Fig. 31, including three or more adjacent lumbar dermatomes (Hekmatpanah 1961). This type of large, proximal field has been only infrequently reported in the literature on the cat somatosenso~ system (Brown and Fuchs 1975; Devor and Wall 1976; Light and Durkovic 1984; Wilson et al. 1986; Collins 1987). The relevant cells in these earlier studies were, like ours, isolated in the lateral portions of the lumbar dorsal horn from laminae I through VIII. Such fields are very large in comparison to the cutaneous fields of neurons with fields located distal to the knee (Wilson et al. 1986). Many of the units in our sample had bilateral cutaneous receptive fields on the low back and hip. This finding in lumbar tissues is consistent with an earlier anatomical study of deep-tissue afferents serving the tail of the cat (Mense et al. 1981) and with studies of cat spinal neurons having cutaneous receptive fields on the hip and back (Devor and Wall 1976; Collins 1987). Our present results are also consistent with the recent studies of c-fos expression in the lumbar dorsal horn after noxious stimulation of the hip (Bullitt 1991), and with our recent neuroanatomical finding of bilateral projections of primary afferent fibers innervating lumbar facet joints or multifidus muscles (Roberts et al. 1989). In the present study, however, contralateral receptive fields were in every case much smaller than the ipsilateral fields. Somatosen~~ neurons with receptive fields in the d&al hind limbs generally do not respond to contralateral stimulation (Wilson et al. 1986), although contralateral fields have been reported for units responsive to stimulation of deep tissues in the hind leg (Schaible et al. 1987). The sizes of the superficial receptive fields determined for units in the present study did not vary systematically with the depths of the recorded units. This result is consistent with the observations of Brown and Fuchs (1975) and Light and Durkovic (1984) but is not consistent with the reports of Wall (1967) and Sorkin et al. (1986). It should be noted that the convolution of laminae in the lateral dorsal horn (Rexed 1954) distorts the relationship between electrode depth and laminar location more than elsewhere in the superficial dorsal horn and that no attempt was made to mark the laminar recording site for most units in the present study. The majority of ‘sd’ and ‘d’ units isolated in the lateral dorsal horn were determined to be WDR neurons (75%, Table I). None of the ‘sd’ and ‘d’ units were LT specific and only 24% were NS. The percentage of units classified as WDR in our sample is greater than

that reported in previous studies of neurons having deep-tissue receptive fields in the hind limbs of cats (31%, Hoheisel and Mense 1990; 32%, Fields et al. 1977). This discrepancy may be due to greater afferent convergence of LT afferents onto neurons with proximal fields; however, differences in anesthetics and other experimental conditions and procedures cannot be ruled out as factors. We were unable to demonstrate a statistically significant segregation in depth of neurons between these three categories in the dorsal horn; others have reported spatial segregation in cat (Sorkin et al. 1986) and monkey (Price and Mayer 1975). Respo~es to ~ec~a~~caI s~i~~ta~io~ of deep t~sues Sensory input from deep lumbar and/or hip or proximal leg tissues was demonstrated in 77% of the 118 neurons tested in the lateral dorsal horn in this study. To our knowledge, no previous investigators have reported input from deep somatic tissues of the lumbar region onto somatosensory neurons. Most cells in the present study that had deep-tissue receptive fields responded only to clearly noxious mechanical or algesic chemical stimulation of deep tissues, not to moderate pressures. However, the responses of a few cells to mechanical stimuli applied directly onto deep tissues was continuously graded over a wide range of stimulus intensities from innocuous to noxious, as judged by application to the experimenter’s skin. The range of thresholds for activation from spinal joints/periosteum, muscle and ligaments with von Frey filaments in this study (12-285 mN) is about the same as that reported from studies of group-III and -IV primary nociceptor afferents with receptive fields in lumbar facet joints, multifidus muscle, and posterior ligaments (Yamashita et al. 1990) as well as intervertebral disc, anterior ligaments and periosteum (Bahns et al. 1986). Although care was taken to protect tissues from drying, prior mobilization of the overlying tissues may have resulted in sensitization of nociceptive afferents. Resides to jeep-tissue algogen injec~~ns The most effective way to demonstrate input from the array of non-cutaneous paraspinal structures in the current study was through local injections of algesic substances. Indeed, some neurons that were unresponsive to sustained noxious pressure applied to both skin and deep paraspinal tissues were responsive to algogen injections (cf., Amano et al. 1986). Algogen injections into spinal joints, periosteum and related fascia tended to produce greater discharge rates than did injections into adjacent muscle bellies. Injection of 6% saline generally evoked larger responses that peaked earlier than did an equal volume of bradykinin (10 pg in 100 ,A>, as shown quantitatively in Table II.

95

The onset latencies and durations of responses of our nocireceptive neuron population to 6% saline and bradykinin injections were quite similar to those in studies of small caliber (groups III and IV>, highthreshold afferents from skeletal muscle (Franz and Mense 1975; Kumazawa and Mizamura 1976; Foreman et al. 1979; Kaufman et al. 1982; Abrahams et al. 1984; Mense and Meyer 19851, joints (Kanaka et al. 1985; Cavanaugh et al. 1991), and paraspinal structures (Bahns et al. 1986; J&rig and Koltzenburg 1991). Neurons with rostrally projecting axons

A small proportion of our units were tested for axonal projections, and a majority of those tested (12 of 20) were found to have ascending axons projecting through either the contralateral VLF or ipsilateral DLF; all of the projection cells except one were nocireceptive ‘sd’ neurons. The extensive range of conduction velocities of our projection cells encompasses the ranges of known nociceptive pathway systems traversing the VLF and DLF, including the spinothalamic, spinoreticular, spinomesencephalic and spinocervical systems (cf., Willis and Coggeshall 1991a). It is likely that our findings underestimate the percentage of neurons that are projection neurons, as no effort was made to ensure that all fibers were activated. Measures were not taken to ensure that only axons within the VLF and DLF were activated. Functional considerations

Our findings that almost all nocireceptive neurons in the lateral lumbar dorsal horn with receptive fields in the low back, hip and/or proximal leg have very large receptive fields and are hyperconvergent, receiving excitatory input from many different deep, paraspinal tissues and from skin, suggest that these neurons are better suited to subserve sensations of diffuse, poorly localized pain than pain that is precisely localized. Similarly, these neurons are poorly suited for discrimination of the tissue(s) from which a nociceptive input is being generated. Our findings are consistent with the fact that in humans, pain localization is poor in this region (Fernstrom 1960; Kellgren 1977; O’Brien 1984) and diagnostic determination of specific origin(s) of low back pain is difficult (Kellgren 1977; Fairbank et al. 1981; Mooney 1987). The discharge profiles and durations of hypertonic saline-induced spike activity in the ‘d’ and ‘sd’ nocireceptive neurons (Table II) showed a remarkable temporal correspondence with the subjective pain reported by human subjects receiving injections of this same algogen into deep paravertebral structures including lumbar spinal ligaments, joints, periosteum, muscle, dura and intervertebral disc (Feinstein et al. 1954; Hirsch et al. 1963; Mooney and Robertson 1976; McCall et al. 1979; Bogduk 1980; El Mahdi et al. 1981). It

therefore seems likely that such neurons subserve subjective intensity for pain resulting from noxious events in spinal and paraspinal tissues. However, speculation of a functional relationship based on a correlation between pain in humans and neuronal activity in cats must be tempered by uncertainties related to probable anesthetic effects (Collins and Ren 19871, possible species differences, and supraspinal transformations of sensory input. The hyperconvergence that is characteristic of these neurons may also explain the common clinical finding that pain associated with a focal spinal injury (e.g., a facet joint injury) appears to be subsequently exacerbated by input from other tissues (e.g., by spasms in paraspinal muscles that ‘splint’ the injured region or by inflammatory effects on nociceptive afferents) (Frymoyer and Gordon 1989b). Hyperconvergence leads to summation of excitatory inputs from many different tissues and presumably to increased pain, making it more difficult to identify a specific tissue pathology as the essential or original pain-inducing dysfunction. Other clinical phenomena that may be explained by the neuronal characteristics revealed in this study are spontaneous pain and tenderness in the hip and/or leg subsequent to lumbar injuries (Steindler and Luck 1938; Travel1 and Rinzler 1952; Fairbank et al. 1981). Many units in our sample that had cutaneous and deep-tissue receptive fields centered in the hip and proximal leg were also activated by noxious stimulation of deep paraspinal tissues. Because the receptive fields of these nocireceptive neurons are centered in the hip or proximal thigh, they are likely to evoke sensations of pain in the hip and/or leg, yet they are responsive to lumbar paraspinal stimulation. In summary, the functional characteristics determined in the present study for somatosensory neurons serving the low back region differ substantially from the characteristics of neurons that have been most widely studied and reported, namely neurons with receptive fields in the distal limbs - the most conspicuous difference being the predominance of sensory input from multiple deep tissues in the region of the low back, hip, and proximal leg. The neurons examined in this study appear to have the characteristics predicted by Kellgren (1977) on the basis of clinical findings characteristics that appear to explain some of the difficulties associated with the diagnosis and treatment of back pain.

Acknowledgements

We gratefully acknowledge the financial support for this research by the National Institutes of Health (NS 13447) and Good Samaritan Hospital and Medical Center. R.G.G. was the recipient of a Tarter Trust

96

Research Fellowship from Foundation of Oregon.

the

Medical

Research

References Abrahams, V.C., Lynn, B. and Richmond, F.J.R., Organization and sensory properties of small myelinated fibres in the dorsal cervical rami of the cat, J. Physiol. (Lend.), 347 (1984) 177-187. Amano, N., Hu, J.W. and Sessle, B.J., Responses of neurons in feline trigeminal subnucleus caudalis tmedullary dorsal horn) to cutaneous, intra-oral, and muscle afferent input, J. Neurophysiol., 55 (1986) 227-243. Ammons, W.S., Characteristics of spinoreticular and spinothalamic neurons with renal input, J. Neurophysiol., 58 (1987) 480-495. Ammons, W.S., Renal and somatic input to spinal neurons antidromically activated from the ventrolateral medulla, J. Neurophysiol., 60 (1988) 1967-1981. Ammons, W.S., Electrophysiological characteristics of primate spinothalamic neurons with renal and somatic inputs, J. Neurophysiol., 61 (1989) 1121-1130. Applebaum, A.E., Beall, J.E., Foreman, R.D. and Willis, W.D., Organization and receptive fields of primate spinothalamic tract neurons, J. Neurophysiol., 38 (1975) 572-586. Bahns, E., Ernsberger, U., Jlnig, W. and Nelke, A., Discharge properties of mechanosensitive afferents supplying the retroperitoneal space, Pfliiger’s Arch., 407 (1986) 519-525. Bell, C., Janig, W., Kummel, H. and Xu, H., Differentiation of vasodilator and sudomotor responses in the cat paw pad to preganglionic sympathetic stimulation, J. Physiol. (Land.), 364 (1985) 93-104. Bogduk, N., The lumbosacral dorsal rami of the cat, J. Anat., 122 (1976) 653-662. Bogduk, N., Lumbar dorsal ramus syndrome, Med. J. Aust., 2 (1980) 537-541. Bogduk, N., The innervation of the lumbar spine, Spine, 8 (1983) 286-293. Broton, J.G.. Hu, J.W. and Sessle, B.J., Effects of tem~romandibular joint stimulation on nociceptive and nonnociceptive neurons of the cat’s trigeminal caudalis tmedullary dorsal horn), J. Neurophysiol., 59 (1988) 1575-1589. Brown, P.B. and Fuchs, J.L., Somatotopic representation of hindlimb skin in cat dorsal horn, J. Neurophysiol., 38 (1975) l-9. Bullitt, E., Somatotopy of spinal nociceptive processing, J. Comp. Neural., 312 (1991) 279-290. Cavanaugh, J.M., El-Bohy, A., Hardy, W.N., Getchell, T.V., Getchell. M.L. and King, AI., Sensory innervation of soft tissues of the lumbar spine in the rat. J. Orthop. Res., 7 (1989) 378-388. Cavanaugh, J.. Avramov, A. and Ozakaty, A.C., Mechanosensitive and chemosensitive units of the lumbar spine: an in vitro study, Sot. Neurosci. Abst., 17 (1991) 1368. Collins, J.G., A descriptive study of spinal dorsal horn neurons in the physiologically intact, awake, drug-free cat, Brain Res., 416 (1987) 34-42. Collins, J.G. and Ren, K., WDR response profiles of spinal dorsal horn neurons may be unmasked by barbiturate anesthesia, Pain, 28 (1987) 369-378. Davis, K.D. and Dostrovsky. J.O., Activation of trigeminal brain-stem nociceptive neurons by dural artery stimulation, Pain, 25 (1986) 395-401. Devor, M. and Wall, P.D., Dorsal horn cells with proximal cutaneous receptive fields, Brain Res., 118 (1976) 325-328. Dostrovsky, J.O., Davis, K.D. and Kawakita, K., Central mechanisms of vascular headaches, Can. J. Physiol. Pharmacol., 69 (1991) 652-658.

Dubner, R. and Bennett, G.J., Spinal and trigeminal mechanisms of nociception, Ann. Rev. Neurosci., 6 (1983) 381-418. El Mahdi, M.A., Latif, U.A. and Janko, M., The spinal nerve root innervation and a new concept of the clinicopathological interrelations in back pain and sciatica, Neurochirurgia, 24 (1981) 137141. Fairbank, J.C.T., Park, W.M., McCall, I.W. and O’Brien, J.P., Apophyseal injections of local anesthetic as a diagnostic aid in primary low-back pain syndromes, Spine, 6 (1981) 598-605. Feinstein, B., Langton, J.N.K., Jameson, R.M. and Schiller, F., Experiments on pain referred from deep somatic tissues. J. Bone Joint Surg. (Am), 35A (1954) 981-997. Fernstrom, U., A discographical study of ruptured lumbar intervertebral discs, Acta Chir. Scandinav., Suppl. 258 (1960) l-60. Fields, H.D., Clanton, C.H. and Anderson, S.D., Somatosensory properties of spinoreticular neurons in cat, Brain Res.. 120 (1977) 49-66. Foreman, R.D., Schmidt, R.F. and Willis, W.D., Convergence of muscle and cutaneous input onto primate spinothalamic tract neurons, Brain Res., 124 (1977) 555-560. Foreman, R.D., Schmidt, R.F. and Willis, W.D., Effects of mechanical and chemical stimulation of fine muscle afferents upon primate spinothalamic tract cells, J. Physiol. &ond.f, 286 (1979) 215-231. Forsythe, W.B. and Ghoshal, N.G., Innervation of the canine thoracolumhar vertebral column, Anat. Rec., 208 (1984) 57-63. Franz, M. and Mense, S., Muscle receptors with group IV afferent fibres responding to application of bradykinin, Brain Res.. 92 (1975) 369-383. Frymoyer, J.W. and Gordon, S.L., New Perspectives on Low Back Pain, American Academy of Orthopaedic Surgeons, Park Ridge, 1989a, pp. l-406. Frymoyer, J.W. and Gordon, S.L., New Perspectives on Low Back Pain, American Academy of Orthopaedic Surgeons, Park Ridge, 1989b, pp. 291-380. Gillette, R.G., Kramis, R.C. and Roberts, W.J., Spinal neurons likely to mediate low back and referred leg pain, Sot. Neurosci. Abstr., 16 (1990) 704. Gillette, R.G., Kramis, R.C. and Roberts, W.J., Convergent input onto spinal neurons likely to mediate low back pain, 3rd IBRO World Cong. Neurosci., Abst. (1991) 188. Groen, G.J.. Baljet, B. and Drukker, J.. The innervation of the spinal dura mater: anatomy and clinical implications, Acta Neurochir. (Wien), 92 (1988) 39-46. Croen, G.J., Baljet, B. and Drukker, J., Nerves and nerve plexuses of the human vertebral column, Am. J. Anat., 188 (1990) 282-296. Hekmatpanah, J., Organization of tactile dermatomes Cl through L4, in cat, J. Neurophysiol., 24 (1961) 129-140. Hellon, R.F., The marking of electrode tip positions in nervous tissue, J. Physiol. (Land.), 214 (1971) 12P. Hirsch, C., Ingelmark, B.E. and Miller, M., The anatomical basis for low back pain, Acta Orthop. Stand., 33 (1963) l-17. Hobbs, S.F., Chandler, M.J., Bolser, D.C. and Foreman, R.D., Segmental organization of visceral and somatic input onto C3-T6 spinothalamic tract cells of the monkey, J. Neurophysiol., 68 (1992) 157551588. Hoheisel, U. and Mense, S., Long-term changes in discharge behavior of cat dorsal horn neurones following noxious stimulation of deep tissues, Pain, 36 (1989) 239-247. Hoheisel, U. and Mense, S., Response behavior of cat dorsal horn neurones receiving input from skeletal muscle and other deep somatic tissues, J. Physiol. tLond.f, 426 (1990) 265-280. Inman, V.T. and Saunders, J.B.D., Referred pain from skeletal structures, J. Nerv. Ment. Dis., 99 (1944) 660-667. Jackson, H.C., Winkelmann, R.K. and Bickel, W.H., Nerve endings in the human lumbar spinal column and related structures, J. Bone Joint Surg. (Am). 48A (1966) 1272-1281.

91 Janig, W., Physiologie von Schmerz und Schmerzprojektion im Bereich der Brustwirbelsaule, Neuroorthopadie, 3 (1985) 84-102. Jinig, W. and Koltzenburg, M., Receptive properties of pial afferents, Pain, 45 (1991) 77-85. Kanaka, R., Schaible, H.-G. and Schmidt, R.F., Activation of fine articular afferent units by bradykinin, Brain Res., 327 (19851 81-90. Kaufman, M.P., Iwamoto, G.A., Longhurst, J.C. and Mitchell, J.H., Effects of capsaicin and bradykinin on afferent fibers with endings in skeletal muscle, Circ. Res., 50 (1982) 133-139. Kellgren, J.H., On the distribution of pain arising from deep somatic structures with charts of segmental pain areas, Clin. Sci., 4 (1939) 35-46. Kellgren, J.H., The anatomical source of back pain, Rheumatol. Rehabil., 16 (1977) 3-12. Knuepfer, M.M., Akeyson, E.W. and Schramm, L.P., Spinal projections of renal afferent nerves in the rat, Brain Res., 446 (1988) 17-25. Kojima, Y., Maeda, T., Arai, R. and Schickikawa, K., Nerve supply to the posterior longitudinal ligament and the intervertebral disc of the rat vertebral column as studied by acetylcholinesterase histochemistry. 1. Distribution in the lumbar region, J. Anat., 169 (1990) 237-246. Kumazawa, T. and Mizamura, K., The polymodal C-fiber receptor in the muscle of the dog, Brain Res., 101 (1976) 589-593. Kuslich, S.D., Ulstrom, C.L., Michael, E.J., The tissue origin of low back pain and sciatica: a report of pain response to tissue stimulation during operations on the lumbar spine using local anesthesia, Orthop. Clin. North Am., 22 (1991) 181-187. Le Bars, D., Dickenson, A. and Besson, J.M., Diffuse noxious inhibitory controls (DNIC). I. Effects on dorsal horn convergent neurones in the rat, Pain, 6 (1979) 283-304. Light, A.R. and Durkovic, R.G., Features of laminar and somatotopic organization of lumbar spinal cord units receiving cutaneous inputs from hindlimb receptive fields, J. Neurophysiol., 52 (1984) 449-458. Lipski, J., Antidromic activation of neurones as an analytic tool in the study of the central nervous system, J. Neurosci. Meth., 4 (1981) l-32. Loeser, J.D., Bigos, S.J., Fordyce, W.E. and Volinn, E.P., Low back pain. In: J.J. Bonica (Ed.), The Management of Pain, Vol. 2, Lea and Febiger, Philadelphia, PA, 1990, pp. 1148-1483. McCall, I.W., Park, W.M. and O’Brien, J.P., Induced pain referral from posterior lumbar elements in normal subjects, Spine, 4 (1979) 441-446. Mendell, L.M., Physiological properties of unmyelinated fiber projection to the spinal cord, Exp. Nemo]., 16 (1966) 316-332. Mense, S., Light, A.R. and Per], E.R., Spinal terminations of subcutaneous high-threshold mechanoreceptors. In: A.G. Brown and M. Rethelyi (Eds.), Spinal Cord Sensation, Scottish Academic Press, Edinburgh, 1981, pp. 79-86. Mense, S., Slowly conducting afferent fibers from deep tissues: neurobiological properties and central nervous actions, Prog. Sens. Physiol., 6 (1986) 139-219. Mense, S., Considerations concerning the neurobiological basis of muscle pain, Can. J. Physiol. Pharmacol., 69 (1991) 610-616. Mense, S. and Meyer, H., Different types of slowly conducting afferent units in cat skeletal muscle and tendon, J. Physiol. (Lond.1, 363 (1985) 403-417. Meyers, D.E.R. and Snow, P.J., The responses to somatic stimuli of deep spinothalamic tract cells in the lumbar spinal cord of the cat, J. Physiol. (Lond.1, 329 (1982) 35.5-371. Milne, R.J., Foreman, R.D., Giesler, G.J., Jr. and Willis, W.D., Convergence of cutaneous and pelvic visceral nociceptive inputs on to primate spinothalamic neurons, Pain, 11 (1981) 163-183. Mooney, V., Where is the pain coming from, Spine, 12 (1987) 754-759.

Mooney, V. and Robertson, J., The facet syndrome, Clin. Orthop., 115 (1976) 149-156. Morton, C.R., Du, J.-J., Xiao, H.-M., Maisch, B. and Zimmerman, M., Inhibition of nociceptive responses of lumbar dorsal horn neurones by remote noxious afferent stimulation in the cat, Pain, 34 (1988) 75-83. Ness, T.J. and Gebhart, G.F., Characterization of neurons responsive to noxious colorectal distension in the T13-L2 spinal cord of the rat, J. Neurophysiol., 60 (1988) 1419-1438. O’Brien, J.P., Mechanisms of spinal pain. In: P.D. Wall and R. Melzack (Eds.), Textbook of Pain, Churchill-Livingstone, London, 1984, pp. 240-251. Park, W.M., McCall, I.W., O’Brien, J.P. and Webb, J.K., Fissuring of the posterior annulus fibrosus in the lumbar spine, Br. J. Radio]., 52 (1979) 382-387. Price, D.D. and Mayer, D.J., Neurophysiological characterization of the anterolateral quadrant neurons subserving pain in M. mulatta, Pain, 1 (19751 59-72. Rexed, B., The cytoarchitectonic atlas of the spinal cord in the cat, J. Comp. Neurol., 100 (1954) 297-380. Roberts, W.J., Gillette, R.G. and Kramis, R.C., Somatosensory input from lumbar paraspinal tissues: anatomical terminations and neuronal responses to mechanical and sympathetic stimuli, Sot. Neurosci. Abst., 15 (1989) 755. Ruth, T.C., Visceral sensation and referred pain. In: J.F. Fulton (Ed.), Howell’s Textbook of Physiology, Vol. 15, Saunders, Philadelphia, PA, 1946, pp. 385-401. Saito, Y., Collins, J.G. and Iwasaki, H., Tonic 5-HT modulation of spinal dorsal horn neuron activity evoked by both noxious and non-noxious stimuli: a source of neuronal plasticity, Pain, 40 (1990) 205-219. Schaible, H.G., Schmidt, R.F. and Willis, W.D., Responses of spinal cord neurones to stimulation of articular afferent fibres in the cat, J. Physiol. (Lond.1, 372 (1986) 575-593. Schaible, H.G., Schmidt, R.F. and Willis, W.D., Convergent inputs from articular, cutaneous and muscle receptors onto ascending tract cells in the cat spinal cord, Exp. Brain Res., 66 (1987) 479-488. Sessle, B.J., Hu, J.W., Amano, N. and Zhong, G., Convergence of cutaneous, tooth pulp, visceral, neck and muscle afferents onto nociceptive and non-nociceptive neurones in trigeminal subnucleus caudalis (medullary dorsal horn) and its implications for referred pain, Pain, 27 (1986) 219-235. Sessle, B.J., The neurobiology of facial and dental pain: present knowledge, future directions, J. Dent. Res., 66 (1987) 962-981. Sessle, B.J. and Hu, J.W., Mechanisms of pain arising from articular tissues, Can. J. Physiol. Pharmacol., 69 (1991) 617-626. Simons, D.G. and Travel], J., Common myofascial origins of low back pain, Postgrad. Med., 73 (19831 55-108. Smyth, M.J. and Wright, V., Sciatica and the intervertebral disc, J. Bone Joint Surg. (Am), 40A (1958) 1401-1418. Sorkin, L.S., Ferrington, D.G. and Willis, W.D., Jr., Somatotopic organization and response characteristics of dorsal horn neurons in the cervical spinal cord of the cat, Somatosens. Res., 3 (1986) 323-338. Steindler, A. and Luck, J.V., Differential diagnosis of pain low in the back: allocation of the source of pain by procaine hydrochloride method, J. Am. Med. Assoc., 110 (1938) 106-113. Stillwell, D.L., Jr., The nerve supply of the vertebral column and the associated structures in the monkey, Anat. Rec., 125 (1956) 139-162. Thies, R., Activation of lumbar spinoreticular neurons by stimulation of muscle, cutaneous, and sympathetic afferents, Brain Res., 333 (1985) 151-155. Travell, J. and Rinzler, S.H., The myofascial genesis of pain, Postgrad. Med., 11 (1952) 425-434.

98 Wall, P.D., The laminar organization of dorsal horn and effects of descending impulses, J. Physiol. (fond.), 188 (1967) 403-423. Willis, W.D. and Coggeshall, R.E. (Eds.), Sensory Mechanism of the Spinal Cord, Plenum Press, New York, 1991a, pp. 307-400. Willis, W.D. and Coggeshall, R.E. (Eds.), Sensory Mechanisms of the Spinal Cord, Plenum Press, New York, 1991b, pp. 13-45. Wilson, P., Meyers, D.E.R. and Snow, P.J., The detailed somatotopic organization of the dorsal horn in the lumbosacral enlargement of the cat spinal cord, J. Neurophysiol., 55 (1986) 604-617.

Yamashita, T., Cavanaugh, J.M., El-Bohy, A.A., Getchell, T.V. and Ring, AL, Mechanosensitive afferent units in the lumbar facet joint, J. Bone Joint Surg. (Am), 72-A (1990) 865-870. Yu, X.-M. and Mense, S., Response properties and descending control of rat dorsal horn neurons with deep receptive fields, Neuroscience, 39 (1990) 823-831.