Although stellate ganglion block is used to diagnose and treat complex regional pain syndromes and vasospastic disorders of the upper extremity , detailed information on the outflow pathway of sympathetic preganglionic vasoconstrictor fibers to the upper extremity in experimental animals, comparable to that available for the lower extremity [2,3], is lacking. Animal studies [4-7] in which changes in volume or skin color of the forelimb in response to electrical stimulation of the thoracic spinal nerves were measured suggest that the most caudal thoracic ramus contributing to the forelimb vascular innervation is T9 or T10, but they disagree regarding the most rostral ramus, which ranges from T1 to T4. Although the measured variables are indirectly related to forelimb vascular resistance, it is difficult to assess from these studies the relative contribution of individual thoracic rami because the responses to spinal nerve stimulation were not compared quantitatively.
Studies in humans concerned with identifying the spinal segments that give rise to sympathetic preganglionic axons innervating the upper extremity have also yielded variable results. Plethysmographic measurements of the changes in upper extremity volume evoked by electrical stimulation of ventral roots suggest that these preganglionic axons arise from T3-6 . A wider range, T2-10, was suggested when changes in skin resistance were measured . However, as in the animal studies, responses were not compared quantitatively; therefore, the relative contribution of each thoracic ramus cannot be evaluated. Studies assessing the effect of interruption of nerve activity in the sympathetic preganglionic chain by surgical or chemical means on temperature, resistance, or moisture of the skin of the upper extremity suggest an upper limit of T1-3 . In attempting to determine experimentally the most rostral ramus that carries preganglionic axons subserving vasoconstrictor control of the upper extremity, a potentially complicating factor is the presence of postganglionic axons in the same ramus. Anatomical studies in cats  and humans [11,12] demonstrate that intrathoracic branches of the second thoracic nerve may join the first, and it has been assumed that these branches carry postganglionic axons based on the presence of unmyelinated fibers. Although it has been assumed that these postganglionic axons innervate the vasculature of the upper extremity, this has never been demonstrated physiologically.
Knowledge of the rostral limit of the spinal cord segments supplying preganglionic innervation, as well as the course of the postganglionic innervation of the upper extremity, has important practical significance because it determines the level of sympathectomy required to abolish sympathetic discharge to the upper extremity . We performed the present study to delineate the organization of sympathetic preganglionic input to, and postganglionic output from, the stellate ganglion, which mediates vasomotor control of the forelimb of the cat, with particular attention to the contribution of the rostral thoracic rami. For comparison, heart rate was measured using the cardiac pacemaker cell as another cardiovascular effector that receives sympathetic input via thoracic rami projecting to the stellate ganglion [13-17].
This study was approved by the McGill University Animal Care Committee. Cats (2.0-4.0 kg, n = 28) were initially anesthetized with pentobarbital (35 mg/kg intraperitoneally). To avoid subsequent administration of anesthetics, which could transiently influence measurement of the physiological variables being studied (forelimb perfusion pressure and heart rate), cats were decerebrated anemically by occlusion of the common carotid and vertebral arteries, Disappearance of the corneal reflex and the presence of a fixed and dilated pupil were considered evidence of effective decerebration. This resulted in a preparation with stable cardiovascular parameters, such that surges in systemic arterial pressure and heart rate in response to surgical stimulation were not observed. During the course of the experiment, baseline heart rate did not change (0.7 +/- 2.5 bpm), whereas systemic arterial pressure actually decreased (12 +/- 2.0 mm Hg; P < 0.0001). After tracheal cannulation, the cats were paralyzed with pancuronium (initial dose 0.1 mg/kg, followed by 0.02 mg/kg every 2 h), and their lungs were artificially ventilated with 100% oxygen. Endtidal CO (2) was monitored and maintained at 30-35 mm Hg. Core temperature was maintained at 37[degree sign]C by using a thermistor-controlled heating blanket. A continuous infusion of 0.9% NaCl solution (10 mL [center dot] kg-1 [center dot] h-1) and drugs were administered via a catheter in a femoral vein. Systemic arterial blood pressure was recorded using a transducer connected to a catheter in a femoral artery. The arterial pressure pulse triggered a cardiotachometer for continuous recording of heart rate. The vagus nerves were sectioned bilaterally in the neck to avoid any heart rate effects of reflex activation of the cranial parasympathetic nervous system. The stellate ganglion, sympathetic chain, vertebral nerve, and thoracic rami T1-8 on the right side were identified after exposure by a ventral approach after removal of the overlying ribs by subperiosteal resection, leaving the pleura intact . Unless otherwise indicated, the stellate ganglion was isolated from the central nervous system by sectioning the T1-8 rami and the sympathetic chain just caudal to the most caudal ramus exposed in any given experiment (Figure 1), thus blocking tonic efferent sympathetic activity, which could affect the evoked changes in heart rate or forelimb perfusion pressure. In three experiments, the T1 ramus was not sectioned (see Results). To prevent drying, the stellate ganglion, sympathetic chain, and rami were covered with warm mineral oil.
The right axillary artery was ligated, and the distal portion was cannulated for perfusion. Heparin (500 U/kg) was administered approximately 3 min before perfusion, followed by 100 U/kg every 2 h. The right common carotid artery was ligated, and the proximal portion was cannulated. Blood for perfusion of the forelimb was withdrawn from the proximal portion of the right common carotid artery and pumped at a constant flow (6-12 mL/min) into the distal portion of the right axillary artery using a precalibrated roller pump and tubing (inner diameter 1.65 mm, outer diameter 4.93 mm). Pump flow was constant over this range of flows and with outflow pressure resistances <or=to350 mm Hg . Perfusion pressure of the axillary artery was measured 2-3 cm distal to the site of cannulation using a transducer. At the start of each study, the blood flow into the axillary artery was adjusted so that perfusion pressure (110.5 +/- 4.3 mm Hg, n = 28) was similar to the mean systemic arterial blood pressure (108.6 +/- 4.7 mm Hg, n = 28). When the pump was stopped for 30 s, perfusion pressure decreased to 7.6 +/- 2.2 mm Hg (n = 10), which suggests minimal collateral arterial flow into the perfused arterial bed. At constant flow, an increase or decrease in perfusion pressure indicates vasoconstriction or vasodilatation, respectively. Systemic arterial pressure, forelimb perfusion pressure, and heart rate were continuously recorded.
To stimulate sympathetic preganglionic axons, bipolar silver hook electrodes were placed on the sympathetic chain and on the cut distal portion of individual thoracic rami. To stimulate postganglionic axons, electrodes were placed on the vertebral nerve and the proximal portion of the cut T1 ramus (Figure 1). Electrical stimulation of pre- or postganglionic axons began 300 +/- 16 min after decerebration. Electrodes were connected via a stimulus isolation unit to a stimulator. Parameters of stimulation were 10- to 40-s trains at 2-20 Hz using supramaximal intensity (10 V, 0.5 ms). In preliminary studies, it was determined that maximal cardiac and vascular responses to stimulation of the sympathetic chain were evoked with 2.8 +/- 0.5 V (n = 5). Each train was repeated at least twice at an interval of approximately 5 min to assess the stability of the response. In each animal, the change in perfusion pressure and heart rate evoked by electrical stimulation of a given ramus was normalized by assigning the value of 100% to the response produced by supramaximal stimulation of the sympathetic chain just caudal to the stellate ganglion. Stimulation at this site produced an increase in perfusion pressure of 68 +/- 6 mm Hg and an increase in heart rate of 66 +/- 7 bpm (from a baseline heart rate 123 +/- 3 bpm; n = 13). Comparisons of means were made by using a paired or unpaired Student's t-test, depending on whether a comparison was within the same animal or between groups of animals, respectively. P <or=to 0.05 was considered significant. Data are expressed as mean +/- SEM.
The effects of electrical stimulation (10-s trains of 10 V, 0.5 ms at 2 Hz) on perfusion pressure and heart rate of the distal portion of sectioned individual rami from T1-8, and of the sympathetic chain immediately caudal to T8, are shown in Figure 2. The most cranial ramus that produced a significant, albeit small, response in perfusion pressure was T2 (4.5% +/- 1.5% of maximum). The response increased in magnitude as successively more caudal thoracic rami were stimulated. The largest response was obtained from T7 (55 +/- 2.3 mm Hg, 73% +/- 6.6% of maximal response, n = 4). Although the most caudal thoracic ramus capable of producing an increase in forelimb perfusion pressure was not determined, it must be >or=toT9 because marked vasoconstrictor responses were evoked by stimulation of the sympathetic chain caudal to T8 (44 +/- 15 mm Hg, 54% +/- 14% of maximal response, n = 3).
The distribution of the thoracic rami that, when stimulated, produced changes in heart rate was shifted to the left of that of the rami that produced changes in perfusion pressure. The most cranial thoracic ramus that produced a cardioacceleratory response was T1 (28.3% +/- 13% of maximal response, n = 9). The greatest increase in heart rate was evoked in response to activation of the T3 ramus (55.2 +/- 8 bpm, 77.6% +/- 7% of maximal response, n = 12). Caudal to T5, stimulation produced insignificant changes in heart rate. The perfusion pressure and heart rate responses were produced by activation of preganglionic axons, as the responses to stimulation of the sympathetic chain just caudal to the stellate ganglion were greatly attenuated after block of ganglionic cholinergic transmission (hexamethonium 15 mg/kg and scopolamine 0.5 mg/kg): perfusion pressure 56.9 +/- 5.3 and 6.6 +/- 2.5 mm Hg before and after cholinergic block, respectively (P < 0.005); heart rate 53.3 +/- 5.1 and 1.1 +/- 0.7 bpm before and after cholinergic block, respectively (P < 0.005, n = 8).
The vertebral nerve contains sympathetic postganglionic axons that mediate vasoconstrictor responses of the upper extremity in the cat . In three experiments with the T1 ramus intact, sectioning the vertebral nerve during continuous stimulation of the sympathetic chain just caudal to the stellate ganglion only partially abolished the increase in forelimb perfusion pressure (by 62% +/- 8%). However, the perfusion pressure response was completely abolished after the additional section of the T1 ramus. In five other experiments, we compared the increases in forelimb perfusion pressure evoked by stimulation of the vertebral nerve and of the proximal portion of the sectioned T1 ramus (20-s trains of 10 V, 0.5 ms at 20 Hz). Stimulation of the vertebral nerve produced an increase in forelimb of perfusion pressure of 69 +/- 9 mm Hg, which was greater than the increase (34 +/- 14 mm Hg) produced by stimulation of the T1 ramus (P = 0.01) (Figure 3). Responses to electrical stimulation of the vertebral nerve or T1 ramus were produced by activation of postganglionic axons, as they were unaffected by block of ganglionic cholinergic transmission with hexamethonium bromide 15 mg/kg and scopolamine 0.5 mg/kg (precholinergic block 42 +/- 3 mm Hg; postcholinergic block 40 +/- 5 mm Hg [n = 3]; vertebral nerve precholinergic block 35 mm Hg; postcholinergic block 40 mm Hg [n = 1, T1 ramus]) and were blocked by the alpha-adrenergic antagonist phentolamine 10 mg/kg (pre-alpha-adrenergic block 48 +/- 10 mm Hg; post-alpha-adrenergic block 6 +/- 5 mm Hg [n = 4, vertebral nerve]; P = 0.005).
Our findings indicate that the sympathetic preganglionic axons that mediate vasoconstriction in the cat forelimb are contained within a wide range of thoracic rami. The results suggest that relatively few of these axons are contained in T1-2; most are carried in more caudal thoracic rami. The more rostral distribution of cardioacceleratory fibers found in the present study agrees with that described previously [14-17]. It is well established that preganglionic neurons projecting to the stellate ganglion originate over a wide range of spinal segments (T1-9) [20,21]. Anatomical studies have shown that the cell bodies of the sympathetic preganglionic neurons that project axons to a given thoracic ramus are confined to a single segment at the corresponding spinal level. Thus, the axons originate at the same spinal segment as the ramus and travel extraspinally in the sympathetic chain, rather than intraspinally, before synapsing in the stellate ganglion [21,22].
Previous studies in the cat [4-6], dog [4,7], and monkey  have addressed the same question as the present study by measuring changes in forelimb volume or changes in color of the skin of the forelimb produced by electrical stimulation of the distal portion of sectioned thoracic nerves (see Introduction). However, in these previous studies, responses were not compared quantitatively; thus, it is difficult to assess the relative contribution of individual thoracic rami. These previous findings are also complicated because the forelimb vasculature was not isolated and muscle paralysis was not used; therefore, some of the responses may have been influenced by changes in cardiac output or by contraction of the forelimb muscles because of unavoidable stimulation of sympathetic cardioaccelerator fibers and somatic efferent axons, respectively. A study in which the evoked compound action potential was recorded from the median, ulnar, and radial nerves in response to electrical stimulation of the distal portion of sectioned thoracic ventral roots in the monkey suggests that the sympathetic outflow to the upper limb is carried in thoracic rami T4-8 . A similar study in which the compound action potential was recorded from the vertebral nerve in the cat shows that responses can be obtained from stimulation of thoracic rami caudal to T3 . Although these approaches circumvent the problem of stimulation of sympathetic cardioaccelerator fibers and somatic efferent axons (see above), the function subserved by the preganglionic axons being stimulated (e.g., vasomotor, sudomotor, piloerector) cannot be determined. Qualitative studies concerned with the distribution of sympathetic preganglionic fibers that mediate sweating of the upper extremity in the cat [25,26] and monkey  suggest that the preganglionic axons are contained in thoracic rami T4-10. The present study is the first quantitative description of the organization of the sympathetic preganglionic input to the stellate ganglion that mediates vasomotor control of the forelimb. Because, in the present study, the forelimb vasculature was isolated and perfused at constant flow, and the animals received a muscle relaxant, the responses observed must have been caused by stimulation of the sympathetic fibers that innervate forelimb resistance vessels. This was confirmed by the effect of drugs blocking transmission at the ganglion or neuroeffector junction.
Concerning our observation that some postganglionic vasoconstrictor fibers course through the T1 ramus, previous anatomical studies in cat  and humans [11,12] may be mentioned. These studies demonstrate that an intrathoracic branch of the second thoracic nerve may join the first, and it has been assumed that the branch carries postganglionic axons based on the presence of unmyelinated fibers. However, this assumption may be questioned because it has been shown in cats that a significant number of preganglionic axons may be unmyelinated . The findings from the present study provide physiological evidence that the T1 ramus does, indeed, carry postganglionic axons subserving vasomotor control of the upper extremity. The distribution of the vasoconstrictor sympathetic preganglionic axons described above reflects only that portion of the forelimb perfusion pressure response transmitted via the vertebral nerve, not T1, because, in those experiments, the thoracic rami (including T1) were sectioned. Therefore, we cannot discount the possibility that a different distribution of vasoconstrictor sympathetic preganglionic axons would have been observed had the T1 ramus been left intact. The demonstration of vasoconstrictor postganglionic axons in the T1 ramus in the cat suggests a possible explanation for the observation in humans that, for complete sympathectomy of the upper extremity, it may be necessary to block conduction through the T1 ramus in addition to interrupting neuronal transmission in sympathetic preganglionic axons below the second ramus (or at the level of the stellate ganglion itself; see ).
Standard anesthesia textbooks concerned with the anatomical basis for regional anesthesia provide descriptions of the rostral-caudal limits of the spinal segmental distribution of sympathetic preganglionic neurons subserving vasomotor control of the upper extremity [28,29]. However, the relative segmental contribution of sympathetic preganglionic neurons subserving such a function has not been previously described, despite the availability of qualitative information for more than a century (see Introduction). This oversight may have had important implications concerning the understanding of how major conduction anesthesia differentially affects the autonomic and somatic nervous systems. For example, in patients undergoing spinal anesthesia, the spinal level of sympathetic blockade, as inferred from the loss of temperature discrimination, has been reported to be approximately two spinal segments rostral to that of sensory blockade . This sympathetic-sensory difference has been ascribed to the combination of a decreasing concentration gradient of drug from the site of injection and that sympathetic preganglionic fibers are more sensitive to the effects of local anesthetics and so are blocked by concentrations of drug too low to block somatic sensory fibers . In patients undergoing spinal anesthesia, it has been more recently shown that the spinal level of sympathetic blockade, as determined by changes in skin temperature of the upper extremity, may be as many as six segments rostral to that of sensory blockade . In an accompanying editorial on that study, Greene  stated, "such an extensive zone of differential sympathetic denervation is a finding so unexpected as to be mind boggling." The findings reported in the present study emphasize an alternative anatomical explanation to this conundrum, namely, that sympathetic preganglionic neurons subserving vasomotor tone of the upper extremity reside primarily in the middle and caudal thoracic segments.
In summary, the findings from the present study in cats suggest that forelimb resistance vessel control is subserved by sympathetic preganglionic neurons located mainly in the middle to caudal thoracic spinal segments. Some of the postganglionic axons subserving vasomotor function course through the T1 ramus, in addition to the vertebral nerve.
1. Breivik H, Cousins MJ, Lofstrom JB. Sympathetic neural blockade of upper and lower extremity. In: Cousins MJ, Bridenbaugh PO, eds. Neural blockade. 3rd ed. Philadelphia: Lippincott-Raven, 1998:411-47.
2. Donald DE, Ferguson DA. Study of the sympathetic vasoconstrictor nerves to the vessels of the dog hind limb. Circ Res 1970;26:171-84.
3. Sonnenschein RR, Weissman ML. Sympathetic vasomotor outflows to hindlimb muscles of the cat. Am J Physiol 1978;235:H482-7.
4. Kuntz A, Alexander WF, Furcolo CL. Complete sympathetic denervation of the upper extremity. Ann Surg 1938;107:25-31.
5. Kuntz A, Dillon JB. Preganglionic components of the first thoracic nerve. Arch Surg 1942;44:772-8.
6. Langley JN. Note on the connection with nerve-cells of the vaso-motor nerves for the feet. J Physiol 1891;12:375-7.
7. Bayliss WM, Bradford JR. The innervation of the vessels of the limbs. J Physiol 1894;16:10-22.
8. Foerster O. Operativ-experimentelle Erfahrungen beim Menschen uber den EinfluB des Nervensystems auf den Kreislauf. Z Ges Neurol Psychiatry 1939;167:439-61.
9. Ray BS, Hinsey JC, Geohegan WA. Observations on the distribution of the sympathetic nerves to the pupil and upper extremity as determined by stimulation of the anterior roots in man. Ann Surg 1943;118:647-55.
10. Haxton HA. The sympathetic nerve supply of the upper limb in relation to sympathectomy. Ann R Coll Surg Engl 1954;14:247-66.
11. Kuntz A. Distribution of the sympathetic rami to the brachial plexus. Arch Surg 1927;15:871-7.
12. Kirgus HD, Kuntz A. Inconstant sympathetic neural pathways. Arch Surg 1942;44:95-102.
13. Kollai M, Koizumi K. The mechanisms of differential control in the sympathetic system studied by hypothalamic stimulation. J Autonom Nerv Syst 1980;2:377-89.
14. Langley JN. On the origin from the spinal cord of the cervical and upper thoracic sympathetic fibres. Phil Trans R Soc Ser B 1892;183:85-124.
15. Faden AI, Jacobs T, Woods M. Cardioacceleratory sites in the zona intermedia of the cat spinal cord. Exp Neurol 1978;61:301-10.
16. Szulczyk A, Szulczyk P. Spinal segmental preganglionic outflow to cervical sympathetic trunk and postganglionic cardiac sympathetic nerves. Brain Res 1987;421:127-34.
17. Kamosinska B, Nowicki D, Szulczyk P. Control of the heart rate by sympathetic nerves in cats. J Autonom Nerv Syst 1989;26:241-9.
18. Phillips JG, Randall WC, Armour JA. Functional anatomy of the major cardiac nerves in cats. Anat Rec 1986;214:365-71.
19. Lioy F, Hanna BD, Polosa C. CO2-dependent
component of the neurogenic vascular tone in the cat. Pflugers Arch 1978;374:187-91.
20. Petras JM, Cummings JF. Autonomic neurons in the spinal cord of the rhesus monkey: a correlation of the findings of cytoarchitectonics and sympathectomy with fiber degeneration following dorsal rhizotomy. J Comp Neurol 1972;146:189-218.
21. Olfield BJ, McLachlan EM. An analysis of sympathetic preganglionic neurons projecting from the upper thoracic roots of the cat. J Comp Neurol 1981;196:329-45.
22. Rubin E, Purves D. Segmental organization of sympathetic preganglionic neurons in the mammalian spinal cord. J Comp Neurol 1980;192:163-74.
23. Sheehan D, Marrazzi AS. Sympathetic preganglionic outflow to limbs of monkeys J. Neurophysiol 1941;4:68-79.
24. Kamosinska B, Nowicki D, Szulczyk A, Szulczyk P. Spinal segmental sympathetic outflow to cervical sympathetic trunk, vertebral nerve, inferior cardiac nerve and sympathetic fibres in the thoracic vagus. J Autonom Nerv Syst 1991;32:199-204.
25. Geohegan WA, Wolf GA, Aidar OJ, et al. The spinal origin of the preganglionic fibers to the limbs in the cat and monkey. Am J Physiol 1942;135:324-9.
26. Langley JN. On the course and connections of the secretory fibres supplying the sweat glands of the feet of the cat. J Physiol 1891;12:347-74.
27. Coggeshall RE, Hancock MB, Applebaum ML. Categories of axons in mammalian rami communicantes. J Comp Neurol 1976;167:105-23.
28. Bonica JJ, Cailliet R. General considerations of pain in the neck and upper limb. In: Bonica JJ, ed. The management of pain. vol 1. 2nd ed. Malvern, PA: Lea & Febiger, 1990:812-47.
29. Winnie AP. Plexus anesthesia. vol 1. 3rd ed. Philadelphia: WB Saunders, 1993:42-5.
30. Greene NM. Area of differential block during spinal anesthesia with hyperbaric tetracaine. Anesthesiology 1958;19:45-50.
31. Greene NM. A new look at sympathetic denervation during spinal anesthesia. Anesthesiology 1986;65:137-8.
32. Chamberlain DP, Chamberlain BDL. Changes in skin temperature of the trunk and their relationship to sympathetic blockade during spinal anesthesia. Anesthesiology 1986;65:139-43.