THE AMERICAN JOURNAL OF ANATOMY 192:293-306 (1991)

Sensory Innervation of t h e Canine Esophagus, Stomach, a n d Duodenum RAMESH K. KHURANA AND J.M. PETRAS Department of Neurology, University of Maryland School of Medicine, Baltimore, Maryland 21201 (R.K.K.); and Department of Medical Neurosciences, Division of Neuropsychiatry, Walter Reed Army Institute of Research, Washington, D.C. 20307-5100 (J.M.P.)

ABSTRACT The sensory innervation of the postpharyngeal foregut was investigated by injecting the enzyme horseradish peroxidase (HRP) into the walls of the esophagus, stomach, or duodenum. The transported HRP was identified histochemically, labeled neurons in the spinal and vagal ganglia were counted, and the results were plotted using an SAS statistical program. The spinal sensory fields of each viscus were defined using three determinations: craniocaudal extent, principal innervation field, and peak innervation field. The data revealed that innervation fields are craniocaudally extensive, the sensory field of each viscus overlaps significantly with its neighbor, yet each viscus can be characterized by a field of peak innervation density. Craniocaudal innervation of the esophagus spans as many as 22-23 paired spinal ganglia (ClL2). There are two peak innervation fields for the cervical (CzC6 and T,-T,) and for the thoracic (TzT4 and TrT12) sectors of the esophagus. The sensory innervation of the stomach extends craniocaudally over as many as 25 paired spinal ganglia (C,-L,). The peak innervation field of the stomach spans a large area comprising the cranial, middle, and the immediately adjoining caudal thoracic ganglia (T2-Tlo).The duodenum is innervated craniocaudally by as many as 15 paired thoracolumbar ganglia (T,L,). Peak innervation originates in the middle and caudal thoracic ganglia and cranial lumbar (T,L,) ganglia. There is a recognizable viscerotopic organization in the sensory innervation of the postpharyngeal foregut; successively more caudal sectors of this region of the alimentary canal are supplied with sensory fibers from successively more caudal spinal dorsal root ganglia. Vagal afferent innervation of the esophagus, stomach, and duodenum is bilateral and originates predominantly, but not exclusively, from vast numbers of neurons in the nodose (distal) ganglia. The esophagus is innervated bilaterally and more abundantly by jugular (proximal) ganglia neurons than is either the stomach or duodenum. The physiological significance of the findings C

1991 WILEY-LISS, INC.

are discussed in relation to the phenomena of visceral pain and referred pain. INTRODUCTION

Afferent signals from the pharyngoesophageal and gastroesophageal sphincters and the pyloric sphincter function in the reflex regulation of these organs, in the coordinated functions of deglutition, and in gastric and duodenal filling and transport (Code and Schlegel, 1968; Doty, 1968; Thomas and Baldwin, 1968). These signals are conveyed from esophageal, gastric, and duodenal receptors in the epithelium, longitudinal, and circular muscular layers, blood vessels, serosa, and mesenteries. Historically, the literature demonstrates wide discrepancies in the anatomical, physiological, and clinical data on sensory innervation fields of the esophagus, stomach, and duodenum. According to Ross (188'71, Head (1893), Gaza (1924), Gage1 (1931), and White (1943) the human esophagus, stomach, and duodenum appeared to receive sensory innervation from a small sensory field of no more than 2-4 thoracic spinal segments (T,-T,; T,-T,). Human sensory visceral fields were determined through the mapping of the affected dermatomes over which gastric and duodenal pain was referred. An estimated 3- t o 4-fold increase in the segmental sensory field size was documented for the esophagus, stomach, and duodenum using experimental surgical and physiological techniques in animals (Balchum and Weaver, 1943; McSwinney and Suffolk, 1938; Hazarika et al., 1964). For example, Hazarika and co-workers (1964) demonstrated t h a t the sensory innervation of the feline esophagus, stomach, and duodenum originated from nearly all thoracic segments (T3--Tl2).Using similar techniques, large thoracic sensory fields were reported by Lebedenko and Brjussowa (1930) and McSwinney and Suffolk (1938) for the canine and feline stomachs, respectively. Watanabe (1954) made comparable observations for the feline duodenum. Smaller thoracic sensory fields of some 8-9 segments were described for the duodenum by Lebedenko and Brjussowa (1930) and McSwinney and Suffolk (1938). The anatomical axonal degeneration stud-

Received January 3, 1991. Accepted June 7, 1991. Address reprint requests to Dr. J.M. Petras, Department of Medical Neurosciences, Division of Neuropsychiatry, Walter Reed Army Institute of Research, Washington, DC 20307-5100.

294

R.K. KHURANA AND J.M. PETRAS

ies of Kimura (1966) implicated as many a s 9, 13, and 12 spinal ganglia for the innervation of the canine esophagus (T3-Tll), stomach (T3-L2), and duodenum (T3-Ll), respectively. These very substantial differences in sensory visceral innervation field size between the clinical and experimental research have not been resolved. Ellison and Clark (1975), using Cauia porcellus, the guinea pig, injected horseradish peroxidase (HRP) into the walls of either the stomach, or both the stomach and ileum. This resulted in the bilateral labeling of the cell bodies of the nodose ganglion. Similarly, HRP injections under the epicardium on the ventral aspect of the left and right atria and ventricles of the feline heart resulted in labeled cell bodies in the spinal ganglia of cranial thoracic segments. This early demonstration of the value of the HRP technique emphasized the utility of centripetal transport as a histochemical marker for the elucidation of visceral sensory pathways. Since then, several authors have utilized this method, or the transport of fluorescent dyes, for the study of visceral afferents (Neuhuber and Niederle, 1979; Elfvin and Lindh, 1982; Khurana and Petras, 1982, 1989; Clerc, 1983; Sharkey et al., 1984; Hudson and Cummings, 1985). The modern data provided some indication of the innervation density stemming from selected spinal ganglia. Hino et al. (1979), in their study of the feline stomach, described 14 spinal ganglia innervating the stomach, but also reported that the area of maximal innervation was derived from a single caudal thoracic segment, TI,. Clerc (1983) injected HRP into the feline lower esophageal sphincter and described a n innervation field of 15 spinal segments, namely, the Tl-L2 dorsal root ganglia. Hudson and Cummings (1985) found that the cervical and thoracic sectors of the esophagus were innervated by the C,-T, and the c8-T, spinal ganglia, respectively. Neuhuber and Niederle (1979) found labeled dorsal root ganglion neurons in the T,-L, spinal ganglia following HRP injections of the rat’s stomach. Three segments appeared to contain the greatest number of labeled neurons; the T,-TI, ganglia. Elfvin and Lindh (1982) injected HRP into the guinea pig pyloric sphincter and found sensory innervation arising from the dorsal root ganglia of all thoracic segments and two cranial lumbar segments, T,-L,. The sphincter appeared to receive its maximal innervation from six thoracic segments, the T,-TI, spinal ganglia. Injection of the fluorescent dye true blue into the ventral wall of the rat stomach was described by Sharkey et al. (1984). This region of the gut was supplied by afferent neurons originating in nine thoracolumbar ganglia, T,-L,. Thus, most authors have investigated the sensory innervation of a single gastrointestinal sphincter, single organ, or organ sector. Limits have also been placed on the number of spinal ganglia studied. The current approach has been to investigate a much larger area of the alimentary tract, together with a larger number of the spinal ganglia ((2,-L,) which may be presumed to innervate the postpharyngeal foregut. The objective was to ascertain some of the more general properties of visceral sensory anatomy such as (1)craniocaudal limits, (2) field overlapping, and (3) innervation densities. These characteristics were heretofore unexplored or beyond the technical capability of the early historical pe-

riod. Modern histochemical methods perhaps provided the promise of ascertaining visceral innervation density. We believe that the current results have achieved these ends through the use of extensive injections of horseradish peroxidase into the walls of the esophagus, stomach, or duodenum and by the subsequent counting and mapping of labeled neurons within 26 paired spinal dorsal root ganglia (Cl-L5), and the paired proximal and distal vagal ganglia. MATERIALS AND METHODS Surgical Procedures

Seventeen neonatal dogs (Canis familiaris) of either sex and ranging between the ages of 9 and 21 days were used. Atropine (0.08 mg/kg) was administered intramuscularly as a n anticholinergic agent. Anesthesia was induced with intravenous thiopental sodium (Pentothal, 6.0 mg/kg) followed by intubation. Effective surgical anesthesia was maintained by the delivery of a Halothane, nitrous oxide, and oxygen mixture. The viscera were surgically exposed and freshly prepared aqueous solutions of 30% horseradish peroxidase (Sigma type VI) were injected into the walls of either the esophagus, stomach, or duodenum using a multidelivery device containing a Hamilton microliter syringe fitted with a 30-gauge needle. The target areas were injected bilaterally using numerous closely spaced penetrations of the viscus wall in order to fill the organ or organ sector as completely as possible. Some diffusion of the HRP solution was visible within the wall of the viscus and beyond the injection site. Extreme care was taken to ensure that the HRP remained subserosally. The esophagus was injected in a total of 6 dogs. The cervical esophagus, the thoracic esophagus, and both the cervical and thoracic sectors of the esophagus were injected in two cases each. The cervical esophagus was injected from the level of the pharynx to the thoracic inlet. Injection of the thoracic esophagus began cranially a t the thoracic inlet and extended caudally to the esophageal opening in the diaphragm. The retrocardiac sector of the thoracic esophagus was not injected. The stomach was injected in 7 dogs. The cardia and the pylorus were injected in 2 cases each, while the cardia, fundus, body, and pylorus were injected in 3 cases. The duodenum was injected in 4 dogs a s follows: one case each of the pars cranialis, pars descendens, and pars ascendens, and all three sectors in another case. The pars caudalis was not injected. Laparotomy cloths were used to protect the surgical field outside the immediate target area. This precaution was used to prevent contact of HRP with areas other than the target site. The injection needle, furthermore, was left in situ for a few minutes before its withdrawal to eliminate the reflux of the HRP from the injection site. The surgical incision was closed in layers following HRP injections. The animals were monitored postsurgically until they fully recovered from the anesthetic. They were treated prophylactically with penicillin and carefully followed thereafter. Euthanasia, Fixation, and Dissection

On the third postoperative day, dogs were surgically anesthetized by intravenous pentobarbital sodium (6.0 mg/kg administration). Thoracotomy exposed the peri-

SENSORY INNERVATION O F POSTPHARYNGEAL FOREGUT

cardium. This was opened to permit simultaneous exsanguination from the right atrium and transcardial perfusion through the left ventricle using a heparinized O.1M phosphate buffer (pH 7.4) solution followed by a freshly prepared fixative containing a mixture of 0.5% paraformaldehyde and 2.5% glutaraldehyde. The central nervous system remained in situ for 2 h r before dissection was begun. This was done to avoid the development of neuronal hyperchromatosis (the “dark neurons” of Cammermeyer, 1960, 1961, 1962, 1978). The abdominal viscera, neck, or thoracic viscera were examined after completing the perfusion. The HRP-injected viscus was scrutinized for evidence of postoperative extravasation of HRP (1)beyond the target site and (2) beyond the viscus under study. We did not find evidence in any of the cases reported here that HRP entered any adjacent viscera, the abdominal cavity, the thoracic cavity, or the adjoining structures of the neck in esophageal cases. The brains and spinal cords were dissected in toto and immersed overnight in fresh fixative. They were then transferred to a phosphate-buffered fixative-sucrose solution (0.5% paraformaldehyde and 2.5% glutaraldehyde and 30% sucrose) before sectioning. Spinal dorsal root ganglia were dissected bilaterally from the C, through the L, segments in all dogs. The proximal vagal ganglia and the distal vagal ganglia (Nomina Anatomica Veterinaria), also known as the jugular and nodose ganglia, respectively, were removed bilaterally in 9 animals; 4 esophageal cases, 4 stomach cases, and 1 duodenal case (see Table 3). The spinal cords, brainstems, prevertebral, and paravertebral ganglia were dissected and removed in 4 esophageal cases (DGI-8, DGI-17, DGI-22, DGI-27), 5 gastric cases (DGI-7, DGI11, DGI-16, DGI-30, DGI-31), and two duodenal cases (DGI-20, DGI-24). All spinal, cranial, and sympathetic ganglia were immersed overnight in sucrose-phosphate buffer (pH 7.4) a t 4°C. Histochemical and Histologjcal Methods

The spinal cords were divided into their respective cervical, thoracic, and lumbar segments for subsequent microtomy in the transverse or horizontal planes. The brainstems were blocked and cut transversely. Frozen sections of the spinal and cranial ganglia, spinal cords, brainstems, and sympathetic ganglia were cut a t 30 Fm on a sliding microtome and all sections were collected serially. The method of de Olmos et al. (1978) was used. Two stock solutions (A and B) were prepared. The gelatin-buffer solution (solution A) contains absolute ethyl alcohol (100 ml), dimethyl sulfoxide (10 ml), 0.05M acetate buffer, pH 4.3 (20 ml), gelatin (5 gm), and deionized water (870 ml). The tetramethyl benzidine (TMB) solution (solution B) contains 0.36 gm of TMB in 270 ml of absolute ethyl alcohol. Sections were rinsed in deionized water, washed in 5% nickel ammonium sulfate for 5 min, and rinsed again in deionized water before incubation a t 0°C for periods varying between 20 min to 1 hr. The incubation medium is a mixture of (1) gelatin-buffer solution (76 ml), (2) tetramethyl benzidine solution (4 ml), (3) 5% sodium nitroprusside (1.35 ml), and (4) 0.75% hydrogen peroxide (1.0 ml). The hydrogen peroxide is added at 20-min intervals while microscopically checking the staining

295

process. A 20-min incubation period was sufficient, in most instances, for full development of the reaction product together with the avoidance of extracellular crystal formation. An ice-cold deionized water rinse followed. Sections were mounted on chrome alum-coated slides and were subsequently counterstained with a 0.5% neutral red-0.1% Safranin 0 solution for the demonstration of nerve cell bodies. Analytical Methods

All histochemically prepared and counterstained sections were studied microscopically using bright-field and dark-field illumination. Horseradish peroxidaselabeled neurons were counted for each of the 26 paired spinal ganglia collected from every case (a total of 884 spinal ganglia). The HRP technique does not lend itself easily to quantitative analysis. Cell counts made of neurons with visible nucleoli would render analysis more accurate (Ebbesson and Tang, 1965). Seldom was this possible, however, since the density of the HRP label obscured nucleoli and even the cell nucleus. We counted, therefore, all labeled neurons and then calculated the percentage of labeled neurons in each spinal dorsal root ganglion from the entire population of labeled dorsal root ganglia cells. This approach also addressed the problem created by the volume related variability in number of ganglion cells labeled. An estimate of the actual number of visceral sensory neurons innervating the esophagus, stomach, and duodenum was not possible. Numerical data from all spinal ganglia were stored on a disc for subsequent analysis in a computer (VAX) database. The number of labeled neurons at each of the paired segmental ganglia were counted and expressed as a percentage of the total number of labeled neurons ( 2 6 ~ 2 = 5 2 )The . results were plotted as a function of location (within the ganglia) using a n SAS statistical program. The data were plotted in each dog. We also plotted mean values by pooling the data derived from dogs with similar injections; either esophageal, gastric, or duodenal. The ganglia of “principal innervation” were, after surveying the data, arbitrarily defined a s those ganglia containing 2-7.9% of the total labeled cell population. The term “peak innervation” was applied to those dorsal root ganglia containing 8% or more of the labeled population of neurons. In a few instances, intervening segments were included even if the number of labeled neurons fell below the arbitrary limit. The inclusion of these ganglia was done following review of all the data and was applied because experience indicated that these rare instances of variation represented an injection artifact. Use of Animals

At the time that these experiments were performed, the research was conducted in compliance with the Animal Welfare Act, and other Federal statutes and regulations relating to animals and experiments involving animals, and adhered to principles stated in the 1978 Guide for the Care and Use of Laboratory Animals. The views of the authors do not purport to reflect the position of the Department of the Army or the Department of Defense (para 4-3, AR 360-5).

296

R.K. KHURANA AND J.M. PETRAS

TABLE 1. Spinal afferent innervation of the esophagus, stomach, and duodenum’

Organ Esophagus, cervical and thoracic DGI-8 DGI-27 Esophagus, cervical DGI-22 DGI-17 Esophagus, thoracic DGI-28 DGI-9 Stomach: cardia, fundus, body, and pylorus DGI-7 DGI-30 DGI-31 Stomach, cardia and fundus DGI-11 DGI-16 Stomach, pylorus

Volume of HRP injected

Craniocaudal

(IJ.1)

extent

Principal innervation

Peak innervation

~

DGI-1

DGI-15 Duodenum, pars cranialis DGI-24 Duodenum, pars descendens DGI-20 Duodenum, pars ascendens DGI-26 Duodenum, pars cranialis, pars descendens, pars ascendens TXI-5

225 320

C,-C4, TZ-T, (6) C3-C,, T,-T, (6)

157 75

164 100 250

676 700 100 78

Not recorded 110

172 190

197 150

‘These case summaries list the spinal segments of origin for each of the viscera studied. The cranial and caudal limits of labeled dorsal root neurons for each organ are also illustrated in the graphs of Figure 1. Spinal segments supplying the largest number of afferent neurons are termed ganglia of “peak innervation.” The term “principal innervation” is defined as 2-7.9% of the total number of labeled dorsal root neurons.

RESULTS Spinal Afferent innervation of the Esophagus, Stomach, and Duodenum

The esophagus Injections of the cervical and thoracic sectors of the esophagus labeled neurons in 23 paired spinal dorsal root ganglia; C1-L, (Table 1, DGI-08 and DGI-27). The ganglia of principal innervation were located in two regions, the C2-C, segments and the T2-Tll segments (Table 1). Peak innervation density originated from cranial cervical and cranial thoracic segments in both cases, that is, C2-C4 and T,-T, in DGI-08 compared with C3-C, and T3-T, in DGI-27. Far fewer labeled neurons were present in the C1, cg, TI, and the Tlz-L2 ganglia. Cervical esophageal injections labeled neurons craniocaudally from the C1 through the T, segments (Fig, l a ; Table 1). Principal innervation occurred in two regions, the C1-C, ganglia and the TZTT5 ganglia. Peak innervation arose in 4-6 rostra1 cervical ganglia and was limited to a single thoracic segment. Thoracic esophageal injections, by contrast, labeled neurons craniocaudally from the C4-TI, ganglia (Fig. l a , Table 1).A few labeled cervical dorsal root ganglia neurons (1-3 cells per ganglia; DGI-28) were identified at the C4-C, segmental levels. The area of principal

innervation included the Tl-Tl, ganglia. Peak innervation stemmed from two regions, the T,-T4 and the T8-T9 ganglia. In summary, the cervical esophagus was innervated by cervical and cranial thoracic segments whereas the thoracic esophagus received its principal innervation from cranial and middle thoracic segments. The stomach Of three animals with combined injections of the cardia, fundus, body, and pylorus of the stomach, two were selected for data presentation because of their differing results. Following the injection of 676 pl of HRP into the cardia, fundus, body, and pylorus in case DGI-30 (Table 21, the location of labeled spinal afferent neurons extended craniocaudally from the C,-L5 ganglia (Table 1, Fig. lb). Whereas labeled cells were present and plentiful in every section through these more cranially located ganglia, the ganglia of segments cg and TI did not contain labeled neurons. The area of principal innervation included the ganglia of segments CG, and T4-L,; peak labeling occurred in the T,L1 ganglia. In the second case (DGI-31),which received 700 pl of HRP injected into the cardia, fundus, body, and pylorus, labeled neurons were identified in 19 paired ganglia extending craniocaudally from the CTLl seg-

297

SENSORY INNERVATION O F POSTPHARYNGEAL FOREGUT

TABLE 2. Counts of labeled dorsal root ganglion neurons following injections of horseradish peroxidase (HRP) into the walls of the esophagus, stomach, or duodenum'

Organs Esophagus, cervical and thoracic DGI-8 DGI-27

Volume of HRP injected ( ~ 1 )

Number of labeled spinal dorsal root ganglion neurons

225 320

1,383 4,464

157 75

2,732 1,764

164 100

2,058 884

250 676 700

1,532 13,941 10,744

100 78

1,669 1,426

Not recorded 110

799 1,678

Esophagus, cervical DGI-22 DGI-17

Esophagus, thoracic DGI-28 DGI-9

Stomach: cardia, fundus, body, and pylorus DGI-7 DGI-30 DGI-31

Stomach, cardia and fundus DGI-11 DGI-16

Stomach, pylorus DGI-1 DGI-15

Duodenum, pars cranialis DGI-24

Duodenum, pars descendens DGI-20

Duodenum, pars ascendens DGI-26

172

729

190

220

197

90

150

557

Duodenum, pars cranialis, pars descendens, pars ascendens DGI-5 'Counts were made of paired ganglia

ments (Table 1).The ganglionic neurons of segments C,, C,, and C, were not labeled. The area of principal innervation extended from the C8-TlO ganglia, whereas peak innervation was located in the T,-T, ganglia. We compared two cases each of injections limited to either the cardia and fundus, or pylorus. The results were strikingly similar (Table 1, Fig. 2). Cardia and fundus were innervated craniocaudally from the TZTL, ganglia while the pylorus received afferent innervation from the T,-L, ganglia in one case and the C,-T,, ganglia in the other. The cardia and fundus received their principal innervation from the T,-L, ganglia in DGI-11 and from the T4-Tl2 ganglia in DGI-16. The pylorus received its principal innervation from the T,TI2 ganglia. Peak innervation for the cardia and fundus arose in the T2 and T,-T12 ganglia in DGI-11 and in the T,-T,, ganglia in DGI-16. For the pylorus, peak innervation arose from the T,-T, ganglia (Table 1). 'The duodenum

DGI-24) arose in the T,-L, ganglia, while principal and peak innervation stemmed from the T,-T,, and the T6-T12 ganglia, respectively. Innervation of the pars descendens (Table 1, DGI-20) appeared comparable to that of the pars cranialis, t h a t is, labeled ganglionic neurons extended craniocaudally from the T4-L, segments while principal and peak innervation arose in the T,-L, and the T,,-L, segments, respectively. The pars ascendens (Table 1, DGI-26) was innervated by neurons in the T,-L, ganglia. Principal and peak innervations were strikingly similar and arose from the T9-T13 ganglia (Fig. 2 ) . Axons of the dorsal root ganglia (Figs. 3-71, the mixed spinal nerves, and the immediately adjoining portion of the dorsal roots were labeled in all segments containing labeled spinal ganglion somata. Many neurons with labeled axon hillocks, and their proximal and distal axonal processes were identified. These findings were made in all esophageal, gastric, and duodenal cases. Sensory visceral neurons innervating the esophagus, stomach, and duodenum varied in size between small, medium, and large neurons in all cases (see Figs. 8-16).

Injections into the pars cranialis, pars ascendens and pars descends of the duodenum labeled sensory neurons located in 15 paired spinal ganglia, T,-L, (Table I, DGI-05). Ganglia of principal and peak innervation oc- The spinal cord The spinal cord was studied in 12 animals: 5 esophcurred in the Ts-L2 and the T,,-L1 segments, respectively. Innervation of the pars cranialis alone (Table 1, ageal, 5 gastric, and 2 duodenal cases (see Table 1).

R.K. KHURANA AND J.M. PETRAS

A

Esophagus C e r v i c o l , DGl 1 I & DGI 22 Esophagus Thoracic. OGI 09 & DGI 2 8

J

30

2

B LL

0

Stomach: Cardia, Fundus, Body, & Pylorus DGI 0 7 , DGI 3 0 , & DGI 31

25

I-

$20

w W

a “15 Vl Z

0

w

10

3

w

7-\A-

Z

n Y

5

?

0

W

C C C C C C C C T T T T T T T T T T T T T L L L L L 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 1 1 1 1 1 2 3 4 5 0 1 2 3

SEGMENTS

Stomach: Cardia, Fundus, Body, & Pylorus DGI 0 7 , DGI 3 0 , & DGI 31

-

-

-

a.

CCCCCCCCTTTTTTTTTTTTTLLLLL

-

SEGMENTS

J

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 1 1 1 1 1 2 3 4 5 0 1 2 3

30

a +

B

-

A

Stomoch: Cordia, DGI 1 1 & DGI 1 6 Stomach: Pylorus, DGI 0 1 & DGI 15

L 2 5

n

0

+

5 20 0 LT W

a

“15 v, z 0

l x

2 10 Z

n w 1 5 W m

%

CCCCCCCCTTTTTTTTTTTTTLLLLL 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 1 1 1 1 1 2 3 4 5 0 1 2 3 SEGMENTS -30 J

Q c

2 LL

0

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

,

1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 1 1 1 1 1 2 3 4 5 0 1 2 3 SEGMENTS

Duodenum p a r s c r a n l a l l s , d e s c e n d e n s . et a s c e n d e n s DGI 0 5 , DGI 2 0 , DGI 24, & DGI 26

25

+

5 20 V LT

Fig. 2. Gastric sensory innervation. (a)Mean of data from three cases with widespread coinjection of HRP into the walls of the four major sectors of the stomach: cardia, fundus, body, and pylorus. The principal innervation field is located within the thoracic and the cranial lumbar spinal ganglia. (b) Findings resulting from injections of the cardia and fundus compared with injections of the pylorus. A remarkably close correspondence exists between the sensory segments of origin for the cardia and for the pylorus.

I\--

W

m

c% .s

b.

l

CCCCCCCCTTTTTTTTTTTTTLLLLL

CCCCCCCCTITTTTTTTTTTlLLLLL 1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 9 1 1 1 1 1 2 3 4 5 01 23 iEGMENTS Fig. 1.

Fig. 1 . The segmental origins of visceral sensory neurons supplying innervation to the esophagus (a), stomach (b), and duodenum ( c ) are summarized in these graphs. Sensory field determinations include craniocaudal extent, principal innervation, and peak innervation densities. The peak innervation field is indicated by the dashed horizontal line which marks 8%on the ordinate. Numerical data were plotted in each dog. The results presented in this figure are pooled (mean) data. Comparison of the three graphs demonstrates a craniocaudal shift in the sensory innervation of the alimentary tract as one moves caudally in the postpharyngeal foregut.

Figs. 3-7. Dark-field photomicrographs of the nodose (Figs. 3 , 6) and spinal ganglia (Figs. 4 , 5 , 7) following HRP injections of the stomach wall. Figure 3 is a photomontage of the nodose ganglion, while Figure 4 is a single photograph of the T, spinal ganglion. The far LTeater number and much higher density of vagal afferents stands in sharp contrast to the relatively low number and density of labeled spinal dorsal root ganglion neurons typical of all cases. The relative

density of labeled neurons can be appreciated further by comparing the higher power photomicrographs of the nodose ganglion (Fig. 6) with the density of spinal ganglion neurons (Figs. 5 , 7). The arrows in Figures 4 , 5 , and 7 identify labeled afferent fibers. Figures 3 and 4 are printed at the same magnification. Bar in Figure 4 = 100 pn for Figures 3,4; bar in Figure 5 = 40 km for Figures 5-7.

300

R.K. KHURANA AND J.M. PETRAS

Figs. 8-16. Spinal dorsal root ganglion neurons innervating the stomach. These high-power bright-field photomicrographs illustrate how some neurons are lightly labeled (Figs. 10, 12) while their neighbors are densely labeled. This dense labeling was found in all sizes of neurons whether the cells are described as small (Figs. 8, 10, 13),

medium (Figs. 8, 10, 12, 13-16), or large (Figs. 9, 11, 16). In some instances the cell nucleus could be identified only poorly (Figs. 8, 10, 15). The nucleolus was seldom seen. These results were typical of all cases. Figure 12 contains a large neuron with visible nucleus and nucleolus. Bar in Figure 8 = 10 Fm for all figures.

SENSORY INNERVATION O F POSTPHARYNGEAL FOREGUT

301

TABLE 3. Counts of labeled vagal sensory neurons in the jugular (proximal) and nodose (distal) ganglia following injections of horseradish peroxidase (HRP) into the walls of the esophagus, stomach, or duodenum' Volume of HRP Organ

injected ( ~ 1 )

Number of labeled vagal neurons Nodose ganglia

Jugular ganglia

Total

Esophagus, cervical and thoracic DGI-8 Esophagus, cervical DGI-22 DGI-17 Esophagus, thoracic DGI-9 Stomach: cardia, body, and pylorus DGISO DGI-31 Stomach, cardia DGI-11 DGI-16 Duodenum, pars cranialis DGI-24

225

286 (12%)

2,148 (88%)

2,434

157 75

324 (16%) (L61, R263) 800, Right (16%)

1,650 (84%) 4,360 (84%)

1,974 5,160

100

91, Right (6%)

1,411 (94%)

1,502

676 700

51 71 (1%)

Too numerous to count 8,750 (99%)

8,821

100 78

53 (2%) 100 (2%)

3,180 (98%) 4,960 (98%)

3,233 5,060

1,290 (99%)

1,305

172

15 (1%)

'Paired ganglia were counted except in the cases indicated where labeled neurons of the right ganglia were counted.

Centrally projecting sensory axons labeled with the vical esophageal injection (DGI-22) resulted in the laHRP reaction product could not be traced into the spi- beling of 324 jugular ganglion cells compared with nal cord. This lack of transganglionic transport of the 1,650 nodose ganglion cells. When cardia, fundus, HRP label was evident in all cervical, thoracic, and body, and pylorus of the stomach were injected together lumbar segments. Study of the spinal cord gray matter (DGI-311, it resulted in labeling of 71 neurons in the a t caudal cervical (C5-C,), thoracic (Tl-T13), lumbar jugular ganglia whereas 8,750 neurons were labeled in (Ll-L6), and sacral (S,-S,) levels revealed that HRP the nodose ganglia. Similarly, the injection of pars cralabeling was not present in any of the neurons of the nialis (DGI-24) of the duodenum produced 15 labeled dorsal horns, zona intermedia, or the ventral horns. jugular ganglia neurons and 1,290 labeled nodose ganSpecial attention directed toward an examination of glion neurons. Jugular ganglia appeared to supply the thoracic and lumbar (T1-L4) preganglionic sympa- more afferent fibers to the esophagus than to the stomthetic neurons, including the nucleus intermediolater- ach or duodenum. The labeled visceral afferent neurons of the jugular alis pars principalis, n. intercalatus spinalis, or the n. intercalatus pars paraependymalis (Petras and Cum- ganglia were studied in order to ascertain topographic mings, 1972; Petras and Faden, 1978), produced simi- (viscerotopic) differences in labeled afferent neurons following HRP injection of different sectors of a single larly negative results. Paravertebral ganglia of the thoracic and lumbar viscus and between viscera. No evidence was found in sympathetic chains and the mesenteric ganglia (celiac, either the nodose or jugular ganglia to indicate a sencranial mesenteric, and caudal mesenteric) also were sory viscerotopic origin for afferents to the esophagus, studied in these cases. While the findings are beyond stomach, and duodenum. On the contrary, a striking the scope of this study it is, nevertheless, important t o uniformity of sensory neuronal labeling was found mention that, unlike our spinal cord findings, labeled when comparing the various sectors of a single viscus ganglion cells were identified bilaterally in the para- or between the esophagus, stomach, and duodenum. vertebral ganglia, and the mesenteric ganglia of esophLabeled vagal neurons varied in size as did splanchageal, gastric, and duodenal cases. nic sensory neurons innervating the postpharyngeal foregut. Large, medium, and small neurons were laVagal Afferent Innervation of the Postpharyngeal Foregut beled (Figs. 17-21). Axons in the vagal ganglia, the vagus nerve distally, A comparison of esophageal, gastric, and duodenal cases in which the paired (9 animals) jugular and no- and the dorsal rootlets were labeled bilaterally in all dosal vagal ganglia were collected and stained, re- cases. Some vagal neurons also contained the label in vealed large differences in the number of labeled affer- their axon hillocks, and their proximal and distal axent neurons found in the jugular and nodose ganglia onal processes. (Table 3). Similarly, the number and density of labeled neurons in the nodose ganglia far exceeded the number The brainstem Vagal afferents were traced bilaterally into the tracand density of labeled spinal ganglion neurons (compare Figs. 3 and 6 with Figs. 4,5, and 7). The nodose tus solitarius and solitary nuclei in 4 esophageal, 4 ganglia were, without question, the principal source of gastric, and 3 duodenal cases. Neurons of the solitary afferent innervation to these organs. For example, in nuclei were not transsynaptically labeled. Labeled DGI-08 where two sectors of esophagus were injected preganglionic neurons were located in the following (Table 3), 286 jugular ganglion cells were labeled, medullary structures: (1) the dorsal motor nucleus of whereas 2,148 nodose ganglion cells were labeled. Cer- the vagus, (2) n. ambiguus (occasionalcell labels), (3)n.

302

R.K. KHURANA AND J.M. PETRAS

Figs. 17-21. Nodose ganglion neurons. These bright-field photomicrographs, like the sensory spinal neurons seen earlier, illustrate the differing density of the HRP label and the occurrence of differing sized neurons described as small (Fig. 171, medium (Figs. 17-20), and large (Figs. 17, 18). The nuclei and nucleoli of vagal ganglion cells were

usually obscured by the high density of the HRP reaction product. Axon hillocks (Fig. 21) and labeled axons were commonly present. Bar in Figure 20 = 40 pm; bar in Figure 19 = 10 pm for Figures 17-19 and 21.

SENSORY INNERVATION O F POSTPHARYNGEAL FOREGUT

retroambiguus, and (4) the n. retrofacialis. The labeling of these brainstem neurons indicates a direct innervation of the viscera studied. By contrast, the absence of labeled spinal preganglionic neurons further illustrates what is already well known, the presence of a ganglionic relay in the autonomic pathway (prevertebral or paravertebral neurons) to the viscera. It also demonstrates that retrograde transsynaptic transport did not occur in these cases. DISCUSSION The Spinal Ganglia

303

tions have focused on the stomach (or a part thereof) or the esophagus. Only Kimura studied duodenal innervation. We have dissected the C,-L5 spinal ganglia bilaterally in every case and stained for the HRP reaction product. Microscopic study reveals a much larger craniocaudal extent of cervical and thoracic innervation for all esophageal sectors than heretofore demonstrated. The present data also demonstrate that the esophagus is innervated by 23 paired spinal dorsal root ganglia (Cl-Lz), the stomach by 22 (C2-Lz), and the duodenum by 15 (Tz-L3). The physiological and anatomical data of all authors indicated a variable degree of overlap in the sensory fields (ganglia) which supply the esophagus, stomach, and duodenum. Hazarika and co-workers (1964) reported an overlap of 100% (10 segments, T3-TiZ) for the esophagus, stomach, and duodenum. Kappis (1913) found an overlap of 3 segments (T,T,) for the stomach and duodenum. Lebedenko and Brjussowa (1930) described an overlap of eight segments (T5-TiZ) for the stomach and duodenum, and McSwinney and Suffolk (1938) observed an overlap between the stomach and duodenum of eight segments (T6-T13). Only one study compared the sensory innervation of the esophagus, stomach, and duodenum (Hazarika et al., 1964).

The earliest anatomical studies defined the innervation of the viscera through dissection of the paravertebra1 (sympathetic chain) and prevertebral ganglia (coeliac; superior and inferior mesenteric, and hypogastric), and their splanchnic branches (see Mitchell, 1953; Kuntz, 1934, 1953). Experimental research attempted to ascertain the composition of the splanchnic nerves and the nervi communicantes by defining the presence of myelinated and unmyelinated nerves (e.g., Langley, 1903; Warrington and Griffith, 1904; Ranson and Billingsley, 1918; Rossi, 1922; Kuntz and Farnsworth, 1931). Kimura’s (1966) work was perhaps the first experimental anatomical effort to ascertain the segmental origin of sensory fibers to the gastrointestiTopology of the Sensory Innervation Fields nal tract. He transected spinal nerves distal to the dorA peak innervation field appeared characteristic for sal root ganglia. By staining the esophagus, stomach, and duodenum for the presence of degenerated axons, a each organ studied. There are two peak innervation distinction could not be made between degenerated vis- fields each for the cervical and for the thoracic sectors ceral motor or sensory fibers. He reported, neverthe- of the esophagus. For the cervical esophagus, the first less, that the sensory innervation of the esophagus peak occurs in the cervical ganglia (C2-cG) and the spanned 9 thoracic segments (T3-Ti1) and that of the second in the cranial thoracic (T2-T,) spinal ganglia. stomach spanned 13 spinal segments (11thoracic and 2 For the thoracic sector, the first peak occurs in cranial lumbar; T3-L,). He also noted an overlap of 9 spinal thoracic ganglia (T,-T,) while the second peak occurs segments in the sensory supply to these organs (T3- in the mid- t o caudal thoracic (T8-Tlz) spinal ganglia. Tll). The duodenum was innervated by 12 spinal seg- The peak innervation field of the stomach spans a large ments (T3-Ll). Curiously, these anatomical data are area comprising the cranial, middle, and the immedinearly identical to the physiological results of Haza- ately adjoining caudal thoracic ganglia (Tz-Tlo). Comparison of the combined innervation fields of the cardia rika et al. (1964). Over the last two decades, improved neuroanatomi- and fundus on the one hand, with those of the pylorus cal labeling methods have made it easier t o define the on the other, reveals a remarkable overlap in craniosensory innervation of the viscera. Direct injection of a caudal extent and in their principal and peak innervaviscus with horseradish peroxidase, lectin, or toxin con- tion fields, despite the geographic interposition of the jugated HRP markers, or fluorescent dyes results in corpus between these sectors. For the duodenum, peak retrograde labeling of the somata of spinal ganglion innervation originates in the middle and caudal thoneurons. These methods provide perhaps the best racic ganglia and cranial lumbar (T,L,) ganglia (Fig. means of anatomically demonstrating the segmental lc). There is a recognizable viscerotopic organization in origin of sensory neurons to the viscera. Three such the sensory innervation of the postpharyngeal foregut; studies of the esophagus suggested innervation by 13 successively more caudal sectors of this region of the (Hudson and Cummings, 1985), 15 (Clerc, 1983),or 23 alimentary canal are supplied with sensory fibers from (Khurana and Petras, 1989) spinal segments. Simi- successively more caudal spinal dorsal root ganglia. Important differences exist regarding the segmental larly, studies of the stomach and the pylorus suggested innervation by 12 (Neuhuber and Niederle, 1979; rat), level of origin between splanchnic sensory and splanch15 (Elfvin and Lindh, 1982; guinea pig), or 22 nic motor neurons. Splanchnic preganglionic neurons (Khurana and Petras, 1982, dog) segments. Cottrell originate bilaterally from nuclear groups of the canine and Greenhorn (1987) found labeled neurons in 11 spi- zona intermedia located in thoracic and lumbar segnal ganglia (T6-L3) of the sheep following injections of ments (Petras and Cummings, 1972, 1978; Petras and cholera toxin-HRP in the gastroduodenal junction. No Faden, 1978). Splanchnic afferent neurons innervating injection study to date, however, has examined collec- the postpharyngeal foregut originate not only from the tively all cervical, thoracic, and lumbar spinal dorsal same segments but also originate from the cervical reroot ganglia (Kimura, 1966; Hino et al., 1979; Neuhu- gion as well. This vast craniocaudal distance constiber and Niederle, 1979; Elfvin and Lindh, 1982; Clerc, tutes between 61 and 68% of the spinal dorsal root 1983; Sharkey et al., 1984; Cottrell and Greenhorn, ganglia. A bilateral discontinuity is present in the sen1987; Hudson and Cummings, 1985). Most investiga- sory fields of the esophagus and stomach at the level of ,

R.K. KHURANA AND .J.M. PETRAS

304

the C8 and TI spinal ganglia. Labeled splanchnic afferent cell bodies were sparse in these ganglia following esophageal and gastric injections. Compared with other spinal ganglia, the C8 and T, ganglia would appear to make a minor contribution to foregut innervation. The Vagal Ganglia

A review of relatively recent data demonstrating vagal afferent innervation of the esophagus, stomach, and duodenum revealed that the proximal vagal (jugular) ganglia have been studied infrequently compared with the distal vagal (nodose) ganglia (Fryscak et al., 1984; Ellison and Clark, 1975; Gwyn et al., 1979; Scharoun et al., 1984; Sato and Koyano, 1987; Lundberg et al., 1978). When studied together, clearly the overwhelming number of vagal afferent neurons supplying the postpharyngeal foregut arises in the nodose ganglia (Hudson and Cummings, 1985; Elfvin and Lindh, 1982; Lindh et al., 1983). Some 84-99% of labeled vagal afferent neurons in our study arose in the nodose ganglia (Table 3). Jugular ganglion neurons were more frequently labeled following esophageal injections compared with gastric or duodenal injections. The regional distribution of vagal afferents to the gastrointestinal system has been discussed by only a few authors (Hudson and Cummings, 1985; Elfvin and Lindh, 1982; Lindh et al., 1983).In summary, we have not found evidence for a viscerotopic segregation of afferent cell bodies of the nodose or jugular ganglia innervating the esophagus, stomach, or duodenum (see Fig. 3). This is in agreement with the esophageal study of Hudson and Cummings (1985).

principalis (Petras and Cummings, 1972; Petras and Faden, 1978). The Medulla Oblongata

Vagal afferents labeled with the HRP reaction product, however, were traced into the solitary tract and the solitary nuclei. Solitary neurons were not labeled in any of the present cases. This also suggests that the labeled vagal ganglion neurons were labeled through the process of uptake a t their innervation sites in the viscus. Volume Effects

The segmental distribution for principal innervation and peak innervation fields varied somewhat despite attempts to control the volume and location of the injected HRP. The segments of principal innervation for the stomach in two cases reflect these results. Neurons of the C,, T,-L, ganglia (14 segments) and the C,-T,, ganglia (11 segments) were identified in DGI-30 and DGI-31, respectively. This variability occurred despite a relatively small percentage difference (3.6%) in the injected volume. Variability in the peak innervation density also appeared in the distributions of labeled neurons in the T,-L, ganglia (7 segments) in DGI-30 and in the TI-T, ganglia (9 segments) in DGI-31. The volume of the HRP injected affected the craniocaudal extent of the ganglionic labeling. A small injection, for example, of 250 pl of HRP, labeled cells in 16 spinal ganglia (T,-L,) in animal DGI-7. A larger volume of injected HRP (700 pl) labeled neurons in 22 paired ganglia in animal DGI-31. The injected volume also influenced the number of labeled neurons in each ganglion but had minimal effect upon the peak innervation density. For example, injection of 250 pl HRP in The Spinal Cord DGI-7 labeled 1,522 cells in all segments, whereas inCentrally projecting axons could not be traced into jection of 700 pl of HRP in DGI-31 labeled 10,774 cells the spinal cord. This lack of transganglionic transport (Table 2). Peak innervation density, however, spanned of the HRP label was evident in all segments studied 8 and 9 segments, respectively. These results demon(C,-L,) in all 17 cases. Transganglionic transport was strate clearly that larger injections labeled more senreported for the feline spinal cord by Cervero and Con- sory neurons or more segments. The number of labeled nell (1984). They treated the cut proximal ends of (I) neurons is not directly proportional, however, to the the dorsal and ventral primary rami of the left inter- volume of the injection. costal nerve (T,) and (2) the right greater splanchnic Clinical lmplications nerve and found labeled visceral afferents in the spinal dorsal horns along with labeled preganglionic neurons. Despite the marked prevalence of visceral pain and Their data differ significantly from our results, but so its significance in clinical diagnosis, the anatomical do our respective treatment techniques. We injected foundations of visceral sensory innervation have not visceral organs directly rather than applying HRP to been studied sufficiently. Visceral pain has been clasthe cut proximal ends of peripheral nerves (greater sified into two components: (1)true visceral pain, and splanchnic and mixed spinal nerves (Cervero and Con- (2) referred or transferred pain (Procaci and Zoppi, nell, 1984). Directly injecting the abdominal viscera, in 1986; Procaci et al., 1986; Janig, 1987). True visceral our cases, resulted in the centripetal transport of HRP pain has been characterized as dull, deep, and diffuse. to selected neurons of the dorsal root ganglia, prever- Its poorly defined localization and distribution over tebral ganglia, and paravertebral ganglia. Although larger areas is typical. It may be accompanied by a we found abundant labeling of peripheral axons, trans- sense of malaise, diffuse sweating, vasomotor reganglionic transport was not demonstrable. We did not sponses, pallor, and changes in arterial pressure and find evidence for retrograde transneuronal transport heart rate. Referred pain is far more localized and may either. Consequently, autonomic interneuronal cells of be felt within a somite. It is characterized by somatic the dorsal horns and zona intermedia (e.g.,nucleus in- and autonomic manifestations. The somatic manifestatermediomedialis, were not labeled. Similarly negative tions include cutaneous hyperalgesia, allodynia, skin results were obtained upon studying the preganglionic tenderness, white dermographism, muscle tenderness, cell groups of thoracic and cranial lumbar segments: n. and even muscle contraction. Hyperhidrosis, piloerecintercalatus spinalis, n. intercalatus pars paraependy- tion, vasoconstriction, and vasodilatation may occur in malis, and n. intermediolateralis thoracolumbalis pars the referred cutaneous zone.

SENSORY INNERVATION O F POSTPHARYNGEAL FOREGUT

To our knowledge, the literature on primates does not contain information on the spinal sensory innervation of the postpharyngeal foregut derived from modern retrograde-labeling techniques. The following comparisons and interpretations of our data, as they apply toward understanding the functions of splanchnic afferents in man, must be based on the current canine data and t h a t of other mammals. The present anatomical study reveals a broad craniocaudal innervation and marked overlap of adjacent visceral sensory fields. This may explain the vague and diffuse nature of visceral pain, and the accompanying generalized autonomic responses, as well as a large overlap in pain distribution from adjacent organs. Pain of the referred type with distinct segmental localization may be anatomically related to peak visceral innervation. Neurectomy, rhizotomy, and cordotomy have been employed to relieve visceral pain in patients with a short life expectancy (Patrick and Sanford, 1977; Dubuisson, 1989; Lipton, 1989). The return of pain is not uncommon following these procedures. A craniocaudal innervation field which spans 15-22 segments indicates that complete surgical deafferentation of a n organ may be impossible through rhizotomy. Transcutaneous electrical nerve stimulation (TENS) has been effective in certain types of visceral pain, e.g., angina pectoris and dysmenorrhea, while it has failed in patients with other types of visceral pain (Long, 1983; Millea, 1983; Woolf, 1989) thus correlating with our anatomical findings of widespread sensory innervation. To be effective, TENS may have to extend over numerous segments. Pharmacotherapy may be considered a more rational approach in the management of visceral pain. ACKNOWLEDGMENTS

The authors thank Michie A. Vane and Cyril P. Wingfield for their assistance with all neuroanatomical techniques; D. Hinkle and Richard Vigue who provided photographic assistance; and J. Bailey, Raoul Vargas, and John Callahan for their experienced help with anesthesia and operating room support. We also thank Dr. Timothy F. Elsmore for his help and use of the SAS statistical program applied to our numerical data. A portion of this research was accomplished while R.K.K. was a National Research Council Associate a t the Walter Reed Army Institute of Research, on leave from the Department of Neurology, University of Maryland School of Medicine, Baltimore, MD. R.K.K. wishes to thank Dr. Erland R. Nelson for providing encouragement and support during his tenure as a n NRC Research Associate. LITERATURE CITED Balchum, O.A., and H.M. Weaver 1943 Pathways for pain from the stomach of the dog. Arch. Neurol. Psychiat., 49t739-753. Cammermeyer, J. 1960 The postmortem origin and mechanism of neuronal hyperchromatosis and nuclear pyknosis. Exp. Neurol., 2:379-405. Cammermeyer, J. 1961 The importance of avoiding dark neurons in experimental neuropathology. Acta Neuropathol., 1:245-270. Cammermeyer, J . 1962 An evaluation of the significance of the “dark’ neuron. Ergebn. Anat. Entwickl-Gesch., 36:l-61. Cammermeyer, J. 1978 Is the solitary dark neuron a manifestation of postmortem trauma to the brain inadequately fixed by perfusion? Histochemistry, 56:97-115. Cervero, F., and L.A. Connell 1984 Distribution of somatic and vis-

305

ceral primary afferent fibers within the thoracic spinal cord of the cat. J. Comp. Neurol., 230:88-98. Clerc, N. 1983 Afferent innervation of the lower oesophageal sphincter of the cat. An HRP study. J. Auton. Nerv. Syst., 9523-636. Code, C.F., and J.F. Schlegel 1968 Motor action of the esophagus and its sphincters. In: Handbook of Physiology, Sect. VI, Alimentary Canal, Vol. IV, Motility. C.F. Code, ed. American Physiological Society, Washington, DC, pp. 1821-1839. Cottrell, D.F., and J.G. Greenhorn 1987 The vagal and spinal innervation of the gastroduodenal junction of sheep. Q.J. Exp. Physiol., 72513-524. de Olmos, J., H. Hardy, and L. Heimer 1978 The afferent connections of the main and the accessory olfactory bulb formations in the rat: An experimental HRP-study. J . Comp. Neurol., 181t213-244. Doty, R.F. 1968 Neural organization of deglutition. In: Handbook of Physiology, Sect. VI, Alimentary Canal, Vol. IV, Motility. C.F. Code, ed. American Physiological Society, Washington, DC, pp. 186-1902. Dubuisson, D. 1989 Root surgery. In: Textbook of Pain. P.D. Wall and R. Melzack, eds. Churchill Livingstone, Edinburgh, pp. 784-794. Ebbesson, S.O.E., and D. Tang 1965 A method for estimating the number of cells in histological sections. J.R. Micros. SOC., 84t449464. Elfvin, L.G., and B. Lindh 1982 A study of the extrinsic innervation of the guinea pig pylorus with the horseradish peroxidase tracing technique. J. Comp. Neurol., 208t317-324. Ellison, J.P., and G.M. Clark 1975 Retrograde axonal transport of horseradish peroxidase in peripheral autonomic nerves. J . Comp. Neurol., I61 r103-114. Fryscak, T., W. Zenker, and D. Kanter 1984 Afferent and efferent innervation of the rat esophagus. A tracing study with horseradish peroxidase and nuclear yellow. Anat. Embryol., I70t63-70. Gagel, 0.1931 Zur Frage der vegetativen Zentren im Ruckenmark. 2. ges. Neurol. Psychiat., 134:499-511. Gaza, W. von 1924 Die resektion der Paravertebralen Nerven und die isolierte Durchschneidung des ramus communicans. Arch. Klin. Chir., 133:479-500. Gwyn, D.G., R.A. Leslie, and D.A. Hopkins 1979 Gastric afferents to the nucleus of the solitary tract in the cat. Neurosci. Lettr., 14: 13-17. Hazarika, N.H., J . Coote, and C.B.B. Downman 1964 Gastrointestinal dorsal root viscerotomes in the cat. J . Neurophysiol., 27,107-116. Head, H. 1893 On disturbances of sensation with special reference to the pain of visceral disease. Brain, 16.1-133. Hino, O., K. Kanafusa, and K. Tsunekawa 1979 Quantitative histological study of spinal afferent innervation on the ventral surface of the cat stomach by horseradish peroxidase (HRP) method. Experientia, 35t379-380. Hudson, L.C., and J.F. Cummings 1985 The origins of innervation of the esophagus of the dog. Brain Res., 326t125-136. Janig, W. 1987 Neuronal mechanisms of pain with special emphasis on visceral and deep somatic pain. Acta Neurochir. Suppl., 38: 16-32. Kappis, M. 1913 Beitrage zur Frage der Sensibilitat der Bauchhoele. Mitt. Grenzgeb. Med. Chir., 26t493-530. Khurana, R.K., and J.M. Petras 1982 Anatomic demonstration of the afferent innervation of the dogs stomach A possible implication for visceral pain in humans. Neurology, 32:A71 (abstr.). Khurana, R.K., and J.M. Petras 1989 Anatomical demonstration of the afferent innervation of the dogs esophagus: Possible clinical implications in humans. Neurology, 39 (Suppl. 1): 137 (abstr). Kimura, C. 1966 Visceral sensation, clinical manifestation and experimental bases. Neurovegetativa, 28t405-436. Kuntz, A. 1934 The Autonomic Nervous System. Lea & Febiger, Philadelphia. Kuntz, A. 1953 The Autonomic Nervous System, 4th ed. Lea & Febiger, Philadelphia. Kuntz, A., and D. Farnsworth 1931 Distribution of afferent fibers via the sympathetic trunks and gray communicating rami to the brachial and lumbosacral plexuses. J. Comp. Neurol., 53,389-399. Langley, J.N. 1903 The autonomic nervous system. Brain, 26:l-26. Lebedenko, W., and S. Brjussowa 1930 Beitraege zur Frage der Bahnen der Schmerzimpulse. 111. Die Schmerzernpfindlichkeit der Bauchhoehlenorgane. 2. gesamte Exp. Med., 71:198-211. Lindh, B., C-J. Dalsgarrd, L-G. Elfvin, T. Hokfelt, and A.C. Cuello 1983 Evidence of substance P immunoreactive neurons in dorsal root ganglia and vagal ganglia projecting to the guinea pig pylorus. Brain Res., 269t365-369. Lipton, S. 1989 Percutaneous cordotomy. In: Textbook of Pain. P.D. Wall and R. Melzack, eds. Churchill Livingstone, Edinburgh, pp. 832-839.

306

R.K. KHURANA AND J.M. PETRAS

Long, D.M. 1983 Stimulation of the peripheral nervous system for pain control. Clin. Neurosurg., 31:323-343. Lundberg, J.M., A. Dahlstrom, I. Larsson, G. Pettersson, H. Ahlman, and J . Kewemter 1978 Efferent innervation of the small intestine by adrenergic neurons from the cervical sympathetic and stellate ganglia, studied by retrograde transport of peroxidase. Acta Physiol. Scand., 104:33-42. McSwinney, B.A., and S.F. Suffolk 1938 Segmental distribution of certain visceral afferent neurons of the pupillo-dilator reflex in the cat. J . Physiol. (London),93:104-116. Millea, T.P. 1983 Transcutaneous electrical nerve stimulation in the management of non-operative intra-abdominal pain. Phys. Ther., 63t1280-1282. Mitchell, G.A.G. 1953 Anatomy of the Autonomic Nervous System. Livingstone Ltd., Edinburgh. Neuhuber, W.L., and B. Niederle 1979 Spinal ganglion cells innervating the stomach of the rat as demonstrated by somatopetal transport of horseradish peroxidase (HRP). Anat. Embryol., 155: 355-362. Patrick, B.S., and R.A. Sanford 1977 Management of pain of malignancy. 111. Neuroectomy and rhizotomy. J. Miss. State Med. AsSOC., 18:276-278. Petras, J.M., and J.F. Cummings 1972 Autonomic neurons in the spinal cord of the Rhesus monkey: A correlation of the findings of &oarchitectonics and sympathectomy with fiber degeneration following dorsal rhizotomy. J . Comp. Neurol., I46t189-218. Petras, J.M., and J.F. Cummings 1978 Sympathetic and parasympathetic innervation of the urinary bladder and urethra. Brain Res., 153t363-369. Petras, J.M., and A.I. Faden 1978 The origin of sympathetic preganglionic neurons in the dog. Brain Res., 144;353-357. Procaci, P., and J.J. Zoppi 1983 Pathophysiology and clinical aspects of visceral and referred pain. In: Advances in Pain Research and Therapy, Vol. 5. J.J. Bonica, ed. Raven Press, New York, pp. 643-658. Procaci, P., M. Zoppi, and M. Maresca 1986 Clinical approach to visceral sensation. In: Visceral Sensation. F. Cerverro and J.F.B. Morrison, eds. Elsevier Science Publ., Amsterdam, Prog. Brain Res., 6 7 2 - 2 8 .

Ranson, S.W., and P.R. Billingsley 1918 An experimental analysis of the sympathetic trunk and greater splanchnic nerve in the cat. J . Comp. Neurol., 129t441-456. Ross, J. 1887 On the segmental distribution of sensory disorders. Brain, 10:333-361. Rossi, 0. 1922 On the afferent paths of the sympathetic nervous system, with special reference t o nerve cells of spinal ganglia sending their peripheral processes into the rami communicantes. J . Comp. Neurol., 34:493-505. Sato, M., and H. Koyano 1987 Autoradiographic study on the distribution of vagal afferent nerve fibers in the gastroduodenal wall of the rabbit. Brain Res., 4OO:lOl-109. Scharoun, S.L., F.C. Barone, M.J. Wayner, and S.M. Jones 1984 Vagal and gastric connections to the central nervous system determined by the transport of horseradish peroxidase. Brain Res. Bull., 13: 573-583. Sharkey, K.A., R.G. Williams, and G.J. Dockray 1984 Sensory substance P innervation of the stomach and pancreas. Gastroenterology, 87t914-921. Thomas, J.E., and M.V. Baldwin 1968 Pathways and mechanisms of regulation of gastric motility. In: Handbook of Physiology, Sect. IV, Alimentary Canal, Vol. IV, Motility. C.F. Code, ed. American Physiological Society, Washington, DC, pp. 1937-1968. Warrington, W.B., and F. Grifith 1904 On the cells of the spinal ganglia and on the relationship of their histological structure to the axonal distribution. Brain, 27t297-326. Watanabe, Y. 1954 Studies on the viscerogenic reflexes. Arch. Jpn. Chir., 23.38-57. White, J.C. 1943 Sensory innervation of the viscera; studies on visceral afferent neurons in man based on neurosurgical procedures for relief of intractable pain. Assoc. Res. Nerv. Ment. Dis., 23; 373-390. Woolf, C.J. 1989 “Segmental afferent fibre-induced analgesia: transcutaneous electrical nerve stimulation (TENS) and vibration. In: Textbook of Pain. P.D. Wall and R. Melzack, eds. Churchill Livingstone, Edinburgh, pp. 884-896.