The “sacral parasympathetic”: ontogeny and anatomy of a myth

At a time when the autonomic nervous system was still considered to be entirely “sympathetic”, foreshadows of the “sacral parasympathetic” outflow can be spotted in the parallel drawn by Walter Gaskell between the cranial and sacral visceral nerves. He first described [10] that the nerves from the CNS to autonomic ganglia form a continuous series at thoracic and upper lumbar levels, framed by two gaps at brachial and lower lumbar levels and then resuming below at sacral levels as projections to the pelvic ganglia and above at cranial levels as projections of the vagus nerve to “distal” ganglia (that Gaskell changed his mind about: “ganglia of the trunci vagi and ganglion petrosum” in 1885 [10] (p. 11) give way to “excitor neurons of the main viscera” in 1920 [11]). Thus, the cranial and sacral outflows symmetrically bracket, at a distance, the thoracolumbar one. Gaskell acknowledges that the parallel between the brachial and lumbar gaps is “rough” [11] (p 27). This parallel is actually  false: the caudal gap coincides with the lumbar plexus (and might be explained by an ontogenetic depletion of preganglionic neurons in favor of somatic motor neurons for the hindlimb, both born from the same pool of progenitors [12]). In contrast, the rostral gap extends beyond the brachial plexus to include the entire cervical spinal cord.

A second “cranio-sacral” parallel suggested by Gaskell was that, above the brachial and below the lumbar gaps, projections resume only to distal or “collateral” ganglia, not to the paravertebral sympathetic chain; in other words, there are no white rami communicantes [10] (Figs. 1a, 2). Again, the parallel is superficial: there is no paravertebral chain at the cranial level, so that the absence of a white ramus communicans there is trivial; in contrast, there are sacral paravertebral ganglia, however diminutive, so that the absence of this connection at sacral levels is a genuine oddity. The pelvic ganglion is connected to branches of the spinal nerves that bypass the paravertebral chain, the nervi erigentes of Eckhard [13], which can appear to the anatomist as unrelated to anything at thoracolumbar levels (Fig. 3). (According to Langley [14] (p. 234), these nerves had already been divorced from the “great sympathetic” by J.-B. Winslow in the 1730s). However, Gaskell was keen, at first, to emphasize the kinship of the nervi erigentes to the “abdominal splanchnics” (that connect the thoracolumbar preganglionics with the prevertebral—coeliac and mesenteric—ganglia “without having anything to do with the [paravertebral] chain”) [11] (p. 24)—even though they join it at gross anatomical level (Fig. 1a); to mark this similarity he originally named the nervi erigentes “pelvic splanchnic” [10], but he eventually backtracked and validated Langley’s “pelvic nerves” [11] (p. 25). The fact that sacral paravertebral ganglia receive their spinal input not from the nearby sacral cord but from the lumbar cord via the sympathetic chain needs an explanation, which might reside in a temporal or spatial idiosyncrasy in the development of this caudal region.

An often-cited differential feature of sympathetic and parasympathetic ganglia is their distance to their targets, long and short, respectively. This rule, however, is soft and implicitly excludes prevertebral ganglia (sympathetic yet close to the gut, kidneys, etc.). The pelvic ganglia, which are supposedly mixed, inevitably conflict with this criterion: they are so close to the bladder and internal genital organs as to have spawned the category of “short noradrenergic [i.e. sympathetic] neurons” [15] and so distant from external genitals as to create the unusual danger of accidental or surgical disruption of nerves between a supposedly parasympathetic ganglion and its target [16].

Conversely, two anatomical features could have inspired long ago a grouping, as we propose, of all spinal preganglionic neurons and their distinction from cranial ones:

A more recent anatomical counterargument to the sacral “parasympathetic” label is that a number of individual pelvic ganglionic neurons receive dual lumbosacral innervation (which undermines the hodological definition of sympathetic versus parasympathetic ganglionic neurons), as inferior mesenteric ganglionic neurons do (which undermines the singularity of the pelvic ganglion) [6].

In sum, there are more anatomical arguments against than in favor of a parasympathetic identity of the sacral autonomic outflow.

In the field of pelvic physiology, few topics are more confusing than the lumbar inhibitory pathway to the detrusor muscle of the bladder. Langley himself never used the bladder to make his case for the sympatho-parasympathetic duality of the lumbo-sacral outflow. And for good reason: he had explicitly excluded any inhibition of the bladder in his 1895 extensive monography on the subject [27]: “We give some additional—and we think conclusive—evidence that both the lumbar and the sacral nerves cause contraction of all the muscle fibers of the bladder […]. Inhibitory fibers of the bladder are few, if indeed they exist.”. Three years earlier, Sherrington had published data on monkey and cat [28] that “confirmed […] the existence of a motor outflow [to the bladder] in the upper lumbar anterior roots”, and, without mentioning any inhibition, plainly formulated the continuity of action between the lumbar and sacral levels: “we may suppose that a long nucleus for the bladder exists in the lumbosacral region, which has however a gap in its continuity […]. The outflow from the anterior roots above the gap is into the sympathetic system, from the anterior roots below the gap is direct by the sacral nerves. To a suggestion as to the wherefore of this gap I hope to return later” (which he apparently forgot to do). Elliott [29], following Stewart [30], contradicted Langley on the subject concerning the cat and monkey, yet failed to detect lumbar inhibition in all other species he tested: dog, rabbit, ferret, Indian mongoose and Indian civet—except, very tentatively, in the goat. Moreover, in his opinion, the human bladder failed to respond to being “painted” with noradrenaline. A feature of the bladder response to stimulation of the hypogastric nerves is that whenever a relaxation occurs, it is only apparent, without exception, tens of seconds after an initial contraction, comparable in timing to the contraction evoked by pelvic nerves (and possibly related to the contraction of the base and trigone of the bladder (discussed in [31], p. 295). Langley himself noted [27] that “a slight flaccidity follows the contraction brought about by the hypogastrics, but we never observed the flaccidity without the contraction, nor any considerable degree of flaccidity.” The disagreement seems to bear on the interpretation of this sluggish and inconstant secondary response (the main published traces we are aware of are reproduced in Fig. 4). Langworthy is possibly the second major provider of evidence for an inhibitory action of the lumbar pathway on the bladder, by demonstrating, in cats, a diminished vesical capacity after hypogastric nerve transection [33]. Yet, he later failed to find any sympathetic fibers in the detrusor muscle [34] (followed in this by many others [35‐38]) and eventually turned into the bluntest critic of the lumbosacral antagonism on the bladder [39]: “Section of the sympathetic fibers has no significant effect on the urinary bladder, which acts as well as ever. Section of the parasympathetic fibers paralyzes the bladder completely.” De Groat et al. [40] (also on cat) is a later notable contribution to the “lumbar inhibition” case. The paper, however, is introduced by a warning that “a considerable body of earlier data [indicates] that the sympathetic pathways [are] relatively unimportant in the regulation of bladder function” and closes on the concession that “we have been unable to demonstrate that […] depression of the detrusor […] is activated by naturally occurring sympathetic firing”. Intense firing of the hypogastric nerve occurs during micturition [41], an observation rarely if ever cited, possibly because it is more suggestive of synergy rather than antagonism of the lumbar and sacral pathways. Rare patients with a genetic deficiency for dopamine-β-hydroxylase, and thus unable to synthesize noradrenaline, have normal urinary function [42] and studies in humans with lesions in the spinal cord [43] make it “unlikely that spinal reflexes governing lower urinary tract function in man include the sympathetic [i.e. lumbar] nervous system”—i.e. either the proposed relaxation of the detrusor or the proposed contraction of the bladder outlet by the thoracolumbar outflow [44]. Yet all contemporary reviews and textbooks on the subject, particularly in their take-home schematics, portray a well-balanced antagonism between the lumbar pathway and the sacral one during bladder filling and micturition (e.g. [44, 45]).

In this monotonously discordant literature that covers more than a century, there is no sign of a final convergence of the reviews, textbooks and primary literature. We propose that this stalemate is caused by the double bind of the “sacral parasympathetic” dogma, with its implication of a lumbosacral antagonism, and the scarcity of evidence for the latter. The shift towards purely pharmacological experiments in the 1970s, highlighting alpha- and beta-adrenergic receptors in different parts of the bladder and under different physiopathological conditions, is built on the notion of a sympathetic lumbar inhibitory pathway without providing further evidence for its reality. Local non-neuronal sources of adrenalin have been recently proposed [38].

If bladder physiology did little, originally, to help theorizing the sacral parasympathetic outflow, blood vessels were the crux of the matter. In his 1899 Presidential Address to the Physiological Society [2], credited by Langley himself [21] as his first published statement on “the sacral and cranial pathways as one system”, he puts forward “one reason” to support this view: “if regions above and below [the middle portion of the spinal cord] were mere separated parts of the sympathetic region we should expect that, when some of these regions and the sympathetic region send nerves to the same spots, the effects produced by both sets of nerves would be the same in kind though it might differ in extent. But it is often not the case. Thus, certain blood vessels may receive nerve fibers from four spinal nerves in the sympathetic region and three spinal nerves in the sacral region, all the former cause contraction of the blood vessels, all the latter cause dilatation”. Prominent among the “certain blood vessels” are those to the external genitals, analyzed by Langley and Anderson [27], that cause erection by vasodilation and detumescence by vasoconstriction. On this issue, however, Langley was and remained somewhat of an outsider. Twenty years earlier, Eckhard had found [13] that “in the rabbit, the mechanism of erection is provided by nerves via a second pathway, that is by the nerve trunk that anatomically corresponds to the superior hypogastric plexus in man [i.e. the hypogastric nerve]”.¹ In 1895, the same year as Langley and Anderson’s paper, François-Franck announced that he “obtained a penile vasodilation from the descending branches of the inferior mesenteric ganglion with the same clarity as with the common sacral erector nerve”² [46]. He also found an occasional vasoconstrictor effect of the anterior pelvic nerve and concluded that “with the exception of the posterior erector nerve of Eckhard one finds associated in all nerves vasoconstrictor and vasodilator fibers”, “whose combined effect depended on the “nature form, intensity and rhythm of the excitations”. Half a century later, a similar explanation (“a masking effect produced by concurrent stimulation of vasoconstrictor fibers”) was offered by Root and Bard [47] for the failure of Langley to see “the suprasacral vasodilator pathway to the penis [that] runs through sympathetic channels”, that they deduced from a series of 154 tests on 21 cats with precise spinal lesions. Bessou and Laporte [48] concluded their own study on cat by a similar observation: “The hypogastric nerve, belonging to the orthosympathetic system, contain, in the cat, cholinergic vasodilatating fibers whose prolonged and repetitive stimulation can provoke an erection, despite the presence of fibers with an opposing action”.³ Human patients with complete destruction of the lower lumbar or sacral cord, but not the upper lumbar or thoracic one, experience psychogenic erections mediated by the lumbar pathway [49, 50]. In 1979, Sjöstrand and Klinge found that the pelvic and hypogastric outflows are synergistically vasodilatory in rabbit [51] and commented, with a hint of annoyance: “We feel that now, more than a hundred years after their original description (referring to [13]), it is time to generally accept the existence of the sympathetic hypogastric erectile fibers.” However, 40 years later, the time for “general acceptance” has yet to come and, here again, we incriminate the conceptual sway of the sympatho-parasympathetic antagonism on the field: contemporary reviews span the range from more or less explicit acknowledgments that thoracolumbar sympathetic neurons are “involved in erection” {[45] (p 357), [52] (p 29)}, to elaborate dismissals of this pathway as a plasticity phenomenon induced by lesions of the spinal cord [50] or protests that “under normal conditions, stimulation of pelvic splanchnic nerves (PSN; nervi erigentes, parasympathetic) and of the hypogastric nerve (sympathetic) have different effects on erectile tissues” [53]. Somehow, the lumbosacral synergy is less controversial concerning the emission phase of ejaculation (“nicely integrated” [50] since “both sympathetic and parasympathetic tones act in a synergistic fashion to initiate seminal emission” [54] and [45] (p. 357) and references therein)—and the ejaculation center is lumbar [55, 56], so that the command of the male sexual act has come to be understood as shifting seamlessly, and oddly, from start to finish, from “parasympathetic” to sympathetic neurons [50].

In other words, and barring the residual controversy about erection, the notion of antagonism, which inspired, in the first place, the partition of the spinal cord into sympathetic and parasympathetic [2] has all but evaporated. In conclusion, we can only concur with Jänig that “functions assumed to be primarily associated with sacral […] systems are well duplicated by thoracolumbar […] pathways” and that the “division of the spinal autonomic systems into sympathetic and parasympathetic with respect to sexual function is questionable” [45] (p. 357).