Anatomical studies of the neurovascular bundle and cavernosal nerves


Anthony J. Costello, Department of Urology, Division of Surgery, The Royal Melbourne Hospital, Parkville, Victoria, Australia.



To accurately define the anatomy of the neurovascular bundle (NVB) in men.


The NVB was microdissected in detail bilaterally in 12 fixed human male adult cadavers. The anatomy of the NVB and its relationship to surrounding pelvic structures was ascertained in each specimen.


Previous reports of the anatomy of the NVB have not mentioned its levator ani and anterior rectal component. The anatomy of the cavernosal nerves is such that accurate graft anastomosis to proximal and distal cavernosal nerve segments is extremely difficult.


The current anatomical description of the cavernosal nerve and NVB is inaccurate.


radical prostatectomy


neurovascular bundle.


The major long-term morbidity from radical prostatectomy (RP) remains sexual dysfunction, despite advances in surgical technique. The development of ‘nerve-sparing’ RP has lead to improved potency rates, with studies reporting the recovery of erectile function in 16–76% of men in whom both neurovascular bundles (NVBs) were preserved and 0–56% in those with one NVB preserved [1,2]. Deliberate bilateral excision of the cavernosal nerves during prostate surgery to achieve adequate cancer resection margins results in erectile dysfunction in almost all men [3]. For men with intentional unilateral or bilateral cavernosal nerve resection, nerve reconstruction using an interposition sural nerve graft has been proposed as a means of restoring neurological control of erectile function. Studies assessing the ability of sural nerve grafts to enhance cavernosal nerve regeneration have been conducted in both rats and humans [4–6]. Despite some promising preliminary data in these models, the technique is viewed with a degree of scepticism. Criticism of the technique in humans focuses on the ability to accurately interpose the sural nerve graft.

Thus in the present study we describe the anatomy of the NVB in detail, and outline potential technical difficulties in sural nerve grafting.


The pelves of 12 fixed male human adult cadavers (age at death 56–74 years) were dissected in detail. Before dissection the cadavers had undergone the standard embalming process of the anatomy and cell biology department at the authors’ institution (a mixture of 40% formaldehyde, glycerine, ethanol, phenol and saline). Of the 12 specimens, one was known and noted during dissection to have metastatic prostate cancer. For this cadaveric specimen only the findings of dissection proximal to the prostate were included, i.e. the anatomy of the pelvic splanchnic nerves, hypogastric nerve and pelvic plexus. Other specimens with prostate, bladder and/or rectal pathology considered to be a contributing cause of death or that was grossly apparent when starting the dissection were excluded from the study. The NVB was dissected bilaterally in each cadaver, and its anatomy and relationship to surrounding pelvic structures documented photographically. The dissection sequence varied with two different approaches adopted, i.e. hemisection or en bloc pelvic resection.

Four of the 12 cadaver specimens were hemisected. The fascial sheaths overlying the sacral nerve trunks were excised. The parasympathetic pelvic splanchnic nerves were identified branching from the anterior sacral nerve trunks S2–S4 and dissected as they descended anteriorly to pass beneath the lateral border of the rectum. The rectum was transected at the level of S1–S2 and its anterior surface reflected from the posterior aspect of the prostate. Extreme care was taken when freeing the anterior surface of the rectum from the prostate, as dispersed throughout the separating fascial layers (collectively known as Denonvilliers’ fascia) and adipose tissue lies the neurovascular tissue under investigation. Both layers of Denonvilliers’ fascia (prostatic and rectal) were left on the posterior surface of prostate to ensure an adequate safety margin for preserving neurovascular tissue. The mobilized rectum was displaced posteriorly, revealing the underlying pelvic splanchnic nerves entering the pelvic plexus. The pelvic plexus was dissected in detail with the hypogastric nerve (the major sympathetic input) identified descending from the lower abdomen. The branches of the pelvic plexus were meticulously dissected under magnification (×6). The constituents of the NVB were traced to their target organs, with the findings documented. The relationship of the NVB to surrounding pelvic structures was recorded.

In the remaining eight cadaver specimens the dissection sequence was altered with an en bloc pelvic resection adopted. Pelvic blocks were excised from the remaining eight cadaveric specimens. Both lower limbs were amputated at the upper extremity of the thigh and the abdomen transected 1 cm above the iliac crest (with previous removal of lower abdominal viscera). The lateral and posterior aspects were sectioned using an electric band-saw, crudely exposing the levator ani musculature and the posterior wall of the rectum. Remnants of the bony pelvis and perirectal adipose tissue were removed, clearly exposing the lateral surface of the levator ani musculature and the posterior and lateral walls of the rectum, respectively. The rectum was reflected and displaced posteriorly, exposing Denonvilliers’ fascia with an investing layer of perirectal fat. When reflecting the rectum, care was taken to preserve its neurovascular supply. The perirectal fat was excised and Denonvilliers’ fascia carefully dissected, with the anatomy of Denonvilliers’ fascia and its relationship to the interposed and underlying neurovascular tissue noted. Loose connective tissue and adipose tissue surrounding the prostate, seminal vesicles and bladder was removed, rendering them clearly visible. After gross dissection, the NVB was meticulously dissected under magnification (×6) with interposed adipose tissue and fascial layers carefully removed. The relationship of the NVB to surrounding structures and the target organs of its constituents were documented. This dissection sequence enable the pelvic plexus to be dissected in full, and exposed the additional branches innervating the bladder, seminal vesicles, anterolateral prostate and levator ani musculature. Although innervation to these structures was noted, the focus was directed on documenting the nerve branches descending within the NVB. The levator ani musculature was excised to varying extents, and the lateral pelvic fascia resected to gain greater exposure of the NVB and anterolateral surface of the prostate. The prostatic apex, external urethral sphincter, perineal membrane and membranous urethra were exposed. The cavernosal nerves were traced to the corpora cavernosa.


The pelvic splanchnic nerves arise form the anterior sacral roots, with most branches originating from S4 and smaller contributions from S2 and S3. These parasympathetic fibres converge with sympathetic fibres from the hypogastric nerve to form the pelvic plexus (Fig. 1).

Figure 1.

The pelvic plexus, showing contributions of hypogastric nerve and S2–S4 segments.

The pelvic plexus is located retroperitoneally on the lateral surface of the rectum. A fascial layer (pararectal fascia) and 1–2 cm of perirectal adipose tissue separates the lateral surface of the rectum from the pelvic plexus (Fig. 2). The fenestrated pelvic plexus is situated in a sagittal plane, with moderate variations in its size and position between dissections. It extends as far as 1.5 cm posterior to the dorsal edge of the rectum and 1 cm superior to the rectovesical pouch (pouch of Douglas). Gauging pelvic plexus size is difficult, with borders between it and its branches hard to define. However, generally the pelvic plexus ranges from 3–5.5 cm long and 2.5–5 cm high. There is a quantitative relationship between the size and mass of neural tissue within the pelvic plexus and the number of nerve branches within its projections. The branches of the pelvic plexus form three major projections; (i) anterior, extending across the lateral surface of the seminal vesicle and the inferolateral surface of the bladder; (ii) antero-inferior, extending to the prostatovesical junction and obliquely along the lateral surface of the prostate; and (iii) inferior, running between the rectum and the posterolateral surface of the prostate, forming the neural constituents of the NVB.

Figure 2.

Fascial relationship of the NVB, showing the position of the NVB and its relationship to the prostate (P), rectum (R) and fascial layers. The widening Denonvilliers’ fascia (DF) laterally fuses with the lateral pelvic fascia (LPF) and pararectal fascia (PF). The posterior and lateral divisions of the NVB run within these fibrous leaves.

The pelvic plexus is closely associated with branches of the inferior vesical vein and artery. These large vessels are predominantly in a sagittal plane that is superimposed on the lateral surface of the pelvic plexus. On removing investing adipose and connective tissues, these vascular and neural (pelvic plexus) structures generally lay in distinct separable layers posteriorly, only to converge at the level of the pelvic plexus projections.

The inferior projection of the pelvic plexus unites with several vessels to form a prominent NVB. The NVB descends along the posterolateral border of the prostate (Fig. 3). It extends laterally to the junction of the lateral pelvic fascia and pararectal fascia, and posteriorly to the dorsal layer of Denonvilliers’ fascia, which forms a thick fibrous sheath separating the prostatic capsule from the rectum. Laterally it becomes continuous with the pararectal fascia posteriorly and lateral pelvic fascia anteriorly. The pararectal fascia extends along the lateral surface of the rectum, while the lateral pelvic fascia separates the levator ani musculature from the lateral surface of the prostate (Fig. 4). At the prostatic midline, Denonvilliers’ fascia exists as a single sheet, only to significantly widen laterally. At the junction of the three fasciae there are numerous leaves of fibrous tissue. The posterior and lateral aspects of the NVB run through these leaves. Denonvilliers’ and pararectal fasciae are separated from the anterior and lateral surfaces of the rectum by variable amounts of perirectal adipose tissue.

Figure 3.

Left lateral view of an en bloc dissection; the levator ani musculature and lateral pelvic fascia have been excised, and the rectum displaced posteriorly.

Figure 4.

Posterior view of the prostate and seminal vesicles with the anterior wall of the rectum reflected and Denonvilliers’ fascia removed. The lateral pelvic fascia and its relationship to the NVBs can be seen. The posterior aspects of the NVBs are exposed. Superiorly at the level of the seminal vesicles the posterior-most fibres have been sectioned. The anterior rectal wall is innervated by the posterior-most nerves within the NVB, and by additional medial fibres.

In all 24 dissections, the plexus of nerves running within the NVB branch from the postero-inferior aspect of the pelvic plexus are 0.5– 2 cm inferior to the level of the tip of the seminal vesicle (Fig. 5). The number of macroscopic nerves present varies, with 6–16 noted. On branching from the pelvic plexus these nerves are spread significantly, with up to 3 cm separating the anterior- and posterior-most nerves. The nerves located most anteriorly are intimately associated with the seminal vesicle, coursing along the posterolateral surface, while the nerves located posteriorly run dorsal to the posterolateral verge of the seminal vesicle.

Figure 5.

Right posterolateral view; the levator ani musculature has been excised and the lateral pelvic fascia removed, revealing the relationship of the left and right NVBs to the seminal vesicles and prostate. The pelvic plexus and upper rectum have also been excised, and the posterior two-thirds of the NVBs resected 1–2 cm distal to their branching point. Note the width of the NVB, particularly the neural constituents proximally, and the many neural structures descending with the NVB.

Generally, most of the NVB descends posteriorly to the seminal vesicle. The nerves converge en route to the mid-prostatic level, forming a more condense NVB, only to diverge once again when approaching the prostatic apex.

The nerves of the NVB are intimately associated with vessels branching from the inferior vesical vein and artery. As these vessels course distally toward the prostatic apex numerous terminal branches are given off which, in most cases, mimic the course of the nerves.

The nerves running in the NVB innervate the corpora cavernosa, rectum, prostate and levator ani musculature. The last three also receive a vascular supply from vessels coursing in the NVB. In 20 of the 24 dissections, a large vein drained the rectum, piercing the pararectal fascia and entering the rectal musculature on its anterolateral surface at a variable level, ranging from mid-prostatic to prostatic apex (Fig. 6). Artery and nerve branches supply the anterolateral wall of the rectum from the prostatic apex to mid-prostate level. Nerves running in the NVB pass through slit-like openings in the lateral pelvic fascia to innervate the superior and middle sections of the levator ani musculature. Many nerve and vascular branches pierce the lateral pelvic fascia distally to supply the inferior portion. The nerves innervating the posterior aspect of the prostate are intimately associated with capsular arteries and veins of the prostate. These structures penetrate the prostatic capsule along its base, mid-portion and apex.

Figure 6.

Posterior view; the anterior wall of the rectum is reflected inferiorly. The large rectal veins (RVs) can be seen exiting the NVBs just proximal to the level of the prostatic apex, coursing posteriorly to drain the rectum. A capsular vessel and nerve of the prostate can be seen descending along the posterior aspect of the prostate in the midline.

The cavernosal nerves and several small vessels pierce the urogenital diaphragm posterolateral to the prostatic apex (Fig. 7A–C). At this level the clearly visible cavernosal nerves divide into numerous small branches that descend along the posterolateral aspect of the membranous urethra, before penetrating the posterior aspect of corpora cavernosa.

Figure 7.

(A,B) Lateral views (R) of the NVB; (C) Lateral view (L) of the NVB; the levator ani and lateral pelvic fascia have been excised. The NVB can be seen coursing over the posterolateral surface of the prostate, extending posteriorly to the anterolateral surface of the rectum. The coalescence of the NVB on approaching mid-prostatic level and its divergence to supply neural branches to the levator ani and cavernosal nerves is apparent. The rectum has been displaced posteriorly to varying degrees, exaggerating the distance between the rectum and the prostate. The general functional organization of the NVB constituents can be seen with the cavernosal nerves located in the anterior aspect, medial to the neurovascular supply of the rectum and levator ani at the level of the prostatic apex.

The constituents of the NVB are organized into three functional compartments (Fig. 8 and 9). The neurovascular supply to the rectum is generally in the posterior and posterolateral sections of the NVB, running within the leaves of Denonvilliers’ and pararectal fasciae. The levator ani neurovascular supply is in the lateral section of the NVB, descending along and within the lateral pelvic fascia. The cavernosal nerves and the prostatic neurovascular supply descend along the posterolateral surface of the prostate, with the prostatic neurovascular supply most anterior. Part of this anterior compartment runs ventral to Denonvilliers’ fascia. The functional organization of the NVB is not absolute, and is less pronounced proximally at the levels of the seminal vesicles and the prostatic base.

Figure 8.

Pelvic plexus; showing the multiple contributions of the pelvic plexus: rectum (R), levator ani, bladder (B), seminal vesicles (SV), prostate (P) and corpus cavernosum.

Figure 9.

Functional organization of the NVB; RNV, neurovascular supply to the rectum; DF, Denonvilliers’ fascia; PF, pararectal fascia; LPF, lateral pelvic fascia; LANV, neurovascular supply to levator ani; PNV, neurovascular supply to the prostate; CN, cavernosal nerves.

In addition to the nerves descending within the NVB, a scattering of nerves extends from the medial margin of the NVB to the prostatic midline (Fig. 10). The deepest nerves (from an anterior aspect) innervate the anterior surface of the rectum at the level of the prostatic apex. The more superficial nerves descend posterior to the prostatic apex and merge laterally with the NVB.

Figure 10.

Posterior view of the NVB and prostate; the anterior wall of the rectum has been reflected caudally. Note the slit-like openings in the lateral pelvic fascia through which the nerves to the levator ani musculature leave the NVB.


Before the studies by Walsh and Donker [7] on fetal and neonatal specimens, the cause of erectile dysfunction after RP was not well understood. By tracing the autonomic innervation of the corpora cavernosa, Walsh and Donker proposed and later showed that erectile dysfunction occurred secondary to injury of these nerves (termed the cavernosal nerves). These nerves were identified branching from the pelvic plexus (formed by the union of the sympathetic hypogastric nerve and the parasympathetic pelvic splanchnic nerves) and running as a plexus of small nerves within a prominent NVB on the posterolateral border of the prostate, before piercing the urogenital diaphragm and descending along the lateral aspect of the urethra. They are intimately associated with capsular vessels of the prostate and course outside the prostatic capsule. These initial findings have since been supported by additional anatomical studies, which have further characterized the anatomy of the NVB. Detailed histological studies have also revealed the cross-sectional profile of the NVB and shown it to run through leaves of the lateral pelvic fascia [8–11].

Interest in the precise anatomy of the NVB has increased since the advent of telerobotic laparoscopic prostatectomy, where the magnification may allow easier preservation of these nerves.

We present new information on the rectal and levator ani elements of the NVB. Although the initial hemipelvic sections enabled the constituents of the pelvic plexus to be traced from their origins, the pelvic splanchnic nerves and hypogastric nerve, exposure and visibility of the NVB and the pelvic plexus (in particular its anterior portion) were hindered by the bladder and prostate. Thus, for the remaining eight cadaver specimens the dissection sequence was altered to en bloc pelvic resection.

Previously it was reported that the cavernosal nerves and branches to the prostate were the only neural structures, and the capsular vessels of the prostate the only vascular structures to course through the NVB [1,7,9,10]. When assessing previous publications it is apparent that the terms NVB and cavernosal nerve are often used synonymously [12]; such reference is no longer appropriate.

Studies of the innervation of the levator ani musculature have yielded conflicting results. Several investigators reported that the innervation is derived from two sets of branches of the pudendal nerve, one arising close to the origin of the anal sphincteric branches and the other after the pudendal nerve exits the pudendal canal [13,14]. However, Walsh and Donker [7] reported levator ani innervation to arise from branches of the pelvic plexus that run laterally to the lateral pelvic fascia. Whilst the present study cannot exclude innervation by branches of the pudendal nerve, no branches were detected running laterally to the lateral pelvic fascia. Levator ani innervation is instead derived from branches of the pelvic plexus that run within the NVB, generally in its lateral section.

Innervation to the rectum and levator ani contained within the NVB contributes to the decreasing nerve density of the NVB as it courses distally along the prostate. Previously, this phenomenon was solely attributed to the autonomic innervation of the prostate [10].

The present findings have important implications for the surgical technique of interpositional sural nerve grafting during retropubic RP. The current practice of anastomosing sural nerve grafts to any intraoperatively identified nerve fibre within the NVB, assuming that it is part of the cavernosal nerve plexus, is now complicated by the discovery of additional innervation to the rectum and levator ani. To maximize the accuracy of graft anastomosis to the cavernosal nerves, the branches to the corpora cavernosa need to be distinguished from branches to the prostate, rectum and levator ani. The delineation of a topographic map of the neural constituents of the NVB by the present study aids in identifying distal cavernosal nerves. The cavernosal nerves from the mid-prostatic level to the prostatic apex are generally in the anterior section of the NVB. They are positioned posterior to the capsular vessels and nerves of the prostate, and medial to the nerves and vessels of the rectum and levator ani. Identifying the cavernosal nerves proximally is difficult, especially above the level of the prostatic base, where there is no reliable functional organization of the neural constituents [15]. Thus, to which severed proximal nerves should the nerve graft be anastomosed?

Attaching the sural nerve graft to all of the severed nerve ends proximally, in an attempt to overcome problems with identifying cavernosal nerves, is not feasible, as the width of cavernosal nerve plexus exceeds the graft diameter in some men. Although the NVB was relatively compact in most of the dissections at the mid-prostatic level (0.5–1.5 cm), when approaching the prostatic apex and base the neural structures are extensively spread. Branches to the rectum and levator ani are responsible for the widening of the NVB distally. As these nerves branch proximal to the level of the prostatic apex they can generally be distinguished from the cavernosal nerves. At the level of the prostatic base and seminal vesicles, the anterior- and posterior-most nerves of the NVB are separated by up to 3 cm. As the cavernosal nerves cannot reliably be distinguished from the neural supply to the prostate, rectum and levator ani at this level, graft anastomosis would have to encompass all severed nerve endings to guarantee accurate graft interposition. A single sural nerve graft will not always be capable of achieving this. As the nerves within the NVB converge en route to the mid-prostatic level, inaccuracies in graft anastomosis will inevitably be more profound in men who have a high proximal resection of the NVB (i.e. with prostate cancer and seminal vesicle infiltration).

We recognize that the width of the NVBs reported here may be criticised as being overstated, as a result of the dissection process. Additional measurements of NVBs before meticulous dissection showed that dissection had a negligible effect. Furthermore, similar findings of NVB divergence at the prostatic apex and base are evident in several previous histological studies [9–11].

Injuries to the facial nerve, and nerves of the upper and lower limbs are routinely repaired by re-anastomosis of severed ends or by graft interposition (for larger deficits). In both cases, accurate alignment is meticulously sought, even to the level of fascicular arrangement. This level of accuracy cannot be achieved when grafting into resected cavernosal nerves, because of difficulties in identifying them, and insufficient graft diameter. Accurate graft interposition is critical to achieving a good functional outcome. Inaccurate anastomosis not only decreases the probability of regenerating cavernosal nerves entering the graft and distal cavernosal nerve segment, but also exposes them to scar tissue stemming from the radical resection of the prostate and damage to the NVB and surrounding tissues.

In conclusion, the NVB is more complicated than previously thought and there are implications for sural nerve grafting. Providing the cavernosal nerves are capable of regenerating, improvements in the accuracy of graft interposition may enhance potency after RP. Better identification during RP of the cavernosal nerves and the use of conduits with wider diameters (e.g. dual nerve grafts) will increase the probability of attaining an accurate graft anastomosis. We hope that the present improved description and photographic documentation of the NVB and cavernosal nerves will enhance nerve preservation during radical prostate surgery.


None declared.