The vaginal support structures comprise a series of connective tissues that are well known to gynecologists, i.e., levels I, II and II described by DeLancey [1-3]. However, few studies have attempted to demonstrate the configuration of composite fibers in these tissues. In order to consider the transfer of tensile stress from the vagina to the pelvic floor and wall, we believe that a detailed knowledge of the morphology of the composite fibers is necessary. In particular, since the initial report by Oelrich [4], there have been very few histological demonstrations of level III support. Soga et al. [5] demonstrated a distinct connective tissue mass in the lateral side of the distal vagina, which they termed a difficult name, the lateral extension of the perineal body. They considered that the connective tissue, comprising mainly smooth muscles alongside the vestibule, is located adjacent to and connects with both the external and internal anal sphincters, thus suggesting functional cooperation between the vaginal support and the anal sphincter.

Oelrich [4] considered the female perineal smooth muscles to be a membranous structure corresponding to the male urogenital diaphragm. In a histological section covering a small area, Hudson et al. [6] identified several fiber bundles as the perineal membrane; however, this was beyond the scope of a “topohistological” demonstration. Using histology and macroscopic slices, Stein and DeLancey [7] found that both the compressor urethrae and urethrovaginal sphincter are closely “associated” with the female perineal membrane. However, in their photos, it was difficult to identify the composite fibers of the membrane. Unlike our interpretation, they considered the perineal membrane to connect with the endopelvic fascia (fascia pelvis parietalis) covering the internal aspect of the levator ani muscle. In the same year, Kato et al. [8] clearly demonstrated that elastic fibers between the urethral rhabdosphincter muscle fibers join together to form the female perineal membrane. Thus, an elastic fiber cage for the urethral rhabdosphincter, which is evident in males [9], seems to be a common feature in both genders. However, the topographical relationship between the female perineal membrane and the perineal smooth muscles has not been demonstrated. Consequently, with an aid of immunohistochemistry, the present review attempts to demonstrate the topohistology of the supportive tissues of the vagina, especially the most distal part. In addition, we focused on nerves passing in and along the vaginal supportive tissues, as nerves in the lower paracolpium seem to correspond to the distal part of the pelvic autonomic nerve plexus [10-13].

This study was performed in accordance with the provisions of the Declaration of Helsinki 1995 (as revised in Edinburgh 2000). The histological sections shown in this review were obtained from 10 donated female cadavers ranging in age from 78 to 96 years, with a mean age of 88 years. The cause of death had been ischemic heart failure or intracranial bleeding, and we confirmed that none of the individuals had undergone surgery by reference to medical documentation as well as macroscopic observation after opening the abdominopelvic cavity. These cadavers had been donated to Tokyo Dental College for research and education on human anatomy, and their use for research had been approved by the university ethics committee. The cadavers had been fixed by arterial perfusion with 10% v/v formalin solution and stored in 50% v/v ethanol solution for more than 3 months.

After routine procedures for paraffin-embedded histology, most sections were subjected to hematoxylin and eosin (HE), azan, Masson trichrome or silver staining, and some were used for immunohistochemistry as well as elastica-Masson staining (a variation of Masson-Goldner staining) or aldehyde fuchsin staining for elastic fibers. The primary antibodies used for nerve immunohistochemistry were 1) mouse monoclonal anti-human S100 protein (1:200 dilution; Dako Z0311; Dako, Glostrup, Denmark), 2) rabbit polyclonal anti-human neuronal nitric oxide synthase (nNOS) (1:200; Cell Signaling Technology, Beverly, MA), 3) mouse monoclonal anti-human vasoactive intestinal polypeptide (H-6 or VIP) (1:100 dilution; Santa Cruz sc25347; Santa Cruz, CA) and 4) rabbit polyclonal anti-human tyrosine hydroxylase (TH) (1:100; Millipore-Chemicon ab152, Temecula, CA). In addition, for fibrous structures, we used mouse monoclonal anti-human alpha smooth muscle actin (1:100; Dako M0851, Glostrup, Denmark) and mouse monoclonal anti-human desmin (dilution, 1:50; Dako N1526). The secondary antibody was labeled with horseradish peroxidase (HRP), and antigen-antibody reactions were detected by the HRP-catalyzed reaction with diaminobenzidine. Counterstaining with hematoxylin was performed on the same samples. A negative control without a primary antibody was set up for each of the specimens. Observations and photography were usually performed with a Nikon Eclipse 80, but photos at ultra-low magnification (objective lens less than ×2) were taken using a high-grade flat scanner with translucent illumination (Epson scanner GTX970).

In the pelvic floor, the strongest striated (or skeletal) muscle contraction is generated by the levator ani muscle. It is perhaps pertinent to begin this review by considering the specificity of the levator ani from the viewpoint of the composite fibers in and around the striated muscle (Figure 1). In the human body, striated muscle cells or fibers are surrounded by the endomysium, which comprises type IV and other collagens, whereas intramuscular and extramuscular tendons are composed of type I collagen [14-17] (Figure 1(A) insert). Typical striated muscle fibers carry a series of specific collagenous structures (a form of enthesis [18]) to conduct the force of contraction to a bone or ligament via a tendon. In contrast, the levator ani insertion is formed by an elastic fiber-smooth muscle complex (Figures 1(C) and (D)) classically termed the “conjoint longitudinal muscle coat” [19].

The term “conjoint” was given to the muscle insertion because it is formed by joining between 1) the longitudinal smooth muscle layer of the rectum and 2) the other smooth muscles in the endopelvic fascia covering the internal aspect of the levator: Arakawa et al. [20] and

Hieda et al. [21] clearly demonstrated the conjoint or combined connective tissue fibers. However, as the histology has not been well clarified, its context in pelvic anatomy has often been ignored or misunderstood (for details, see the final paragraph in the subsection entitled “Perineal smooth muscles”). In this context, a new term that does not involve the concept of joining fibers has recently been adopted, i.e., the “longitudinal anal muscle” or LAM [22]. We have chosen the latter term in this review for the convenience of readers. The LAM seems to provide a “fibrous skeleton” for the anal sphincters to minimize any damage or tears resulting from sudden and strong contraction of the levator ani. The major part extends through the intersphincteric space toward the anal skin and minor parts penetrate the subcutaneous part of external sphincters. Figure 1(A) displays a poorly developed LAM without the minor parts dividing the external anal sphincter as intramuscular septa.

Collagen fibers (type I collagen fibers) possess very little elasticity, whereas elastic fibers absorb tensile stress and recover their length. Tendons of skeletal muscles, which are composed of mostly type I collagen fibers, require a small proportion of elastic fibers in order to recover their length after muscle contraction [23,24]. However, as typically seen in extremities, the covering fascia of a striated muscle contains few or no elastic fibers, because its elastic nature is due largely to the mesh structure of its constituent collagen fibers [25] and partly to elasticity of the muscle cells thmselves [26]. Fasciae that are exceptionally elastic fiber-rich cover or bundle striated muscle fibers in the extra-ocular muscles of the eye [27], the intrinsic lingual muscles along the tongue surface [28], and also the levator ani muscle. Fasciae of the levator ani, including the endopelvic fascia, are rich in both elastic fibers and smooth muscles (Figures 1(C), (D), 2 and 3), especially at the inferomedial edge facing the urethral rhabdosphincter [29].

Other than striated muscles, elastic fiber-mediated connections have been described in the larynx [30,31], the middle ear [32] and the urethra [9] (Figure 1(E)).

According to Hinata et al. [9], the urethra in males contains a much greater amount of elastic fibers than the female counterpart. At these three sites, in combinationwith hyaluronic acid as a lubricant, elastic fibers seem to avoid damage at their connection point when subjected to vibration or drastic changes in shape. Notably, the female endopelvic fascia is characterized by a rich content of elastic fibers [33], possibly resulting from the difference in hormonal backgrounds, as estrogen is known to increase the formation of elastic fibers [34].

The most striking difference between the levator ani and other striated muscles lies in the relationship between the direction of muscle action and that of the muscle fibers: in skeletal muscle, the muscle fibers, tendon and fibrous tissue connecting these are consistently arranged “in series” along an almost straight line (Figure 4(A)),

whereas the levator muscle fibers are not directed to the urethra and vagina (Figures 4(C) and (D)). A mesh-like

fascial structure lying between the levator ani and vagina (Figures 2(E) and 3(A)) would also seem to play a role in 1) stabilizing structures in the event of elevation force, and 2) regulating and distributing tensile stress from the levator ani. However, rather than a mechanical role, the levator-vagina interface acts as a major nerve pathway to

the rectum (see the subsection below). Therefore, any direct transfer of force from the levator to the vagina would seem to be limited.

The tendinous arch of the pelvic fascia, or the so-called “white line”, has been well studied in terms of gross anatomy [35-38]. However, the histology does not exhibit a “linear” structure but a mesh-like complex of multiple fasciae [12,33,39] (Figures 2(A)-(E)). Along the tendinous arch, the rectovaginal septum and pubocervical fascia meet the endopelvic fascia [2]. Both are composed of elastic and collagenous fibers and they contain no or few smooth muscles [40] (Figures 2(F) and (G)). Although it is not commonly known, the tendinous arch and its associated loose connective tissue (i.e., the paracolpium) is one of the striking anatomical features in the female pelvic floor. Because of the bulky mass of the prostate, the periprostatic tissue along the endopelvic fascia is very narrow, without any definite tendinous arch from a histologic viewpoint [33,41]. The morphology of the tendinous arch varied between individuals (Figures 2(A)-(E)), possibly due to age-related degeneration and damage sustained during vaginal delivery [42]. Recently, celloidin-embedded large sections have been used for demonstration of the pelvic fasciae [43], but these may not be applicable for immunohistochemistry because of their thickness. Abundant nerves run antero-inferiorly along the tendinous arch and pass through small spaces divided by multiple fasciae continuous with either or both the rectovaginal septum and pubocervical fascia (Figure 2(H)). Thus, the nerves cross the composite fibers of the fasciae.

Nerves in the lower paracolpium provide the origins of nerves extending to the internal anal sphincter [13,21] as well as the cavernous nerves [42,44]. In view of the origins of these nerves, the posterior half of the lower paracolpium seems to correspond to a nerve-rich space at the posterolateral corner of the prostate in males. The rectovaginal septum corresponds to Denonvilliers’ fascia in females [43]. Thus, as in males [45,46], the rectovaginal septum provides the anterior boundary of a fatty tissue layer surrounding the rectum, i.e., the mesorectum [47]. However, in elderly individuals, because the septum is usually fragmented or unclear in the lateral area, the lower paracolpium becomes continuous with the mesorectum without any clear demarcation (Figure 3(A)).

Using three markers (nNOS and VIP for parasympathetic nerves; TH for sympathetic nerves; for abbreviations see Materials and Methods), Hieda et al. [48] found four combinations of autonomic motor nerve fibers in the lower paracolpium: [nNOS+, VIP+, TH+], [+, −, +], [−, −, +] and [−, −, −], not 8, which would logically be the largest number of combinations (i.e., 2 × 2 × 2). Nerves negative for all three markers, i.e., [−, −, −], were composed of thick, myelinated sensory fibers. The first and second patterns were predominant (Figures 3(E)-(G)). In contrast to the male pelvic floor, it has been shown that the VIP-dominant nerve supply is rich in the female pelvic floor, especially along the vagina [49,50]. However, VIP-positive fibers in the female pelvic floor may include sensory fibers [51,52].

When observed in sagittal sections, the upper half of the rectovaginal septum, or Denonvillers’ fascia in females, is so thin or weak as a target of surgical repair (Figure 5).

Inferiorly, the septum reaches the bilateral large masses of the perineal smooth muscles (PSMs) occupying a large subcutaneous space on the posterolateral and lateral sides of the distal vagina and vestibulum (Figures 5-10). Because their major interest was individual variations in perineal body morphology, Soga et al. [5] called PSMs

the “lateral extension of the perineal body”. In fact, the bilateral masses of the PSMs are connected by the perineal body (Figure 8), and together they are arranged in a catamaran-like configuration (i.e. a double-hulled boat). In fact, DeLancey [53] considered the perineal body and membrane to be critical factors for level III support. However, rather than the PSMs, the female perineal body is difficult to demonstrate because of variations in histology: it can be a round mass of irregularly arrayed smooth muscles (Figures 7(A) and 8(A)), a fatty tissue mass (Figure 9(A)), or part of the PSMs without a clear demarcation (Figure 6(A)). Nevertheless, irrespective of morphology, the perineal body as well as the PSMs receive the LAM at the inferior end (Figures 6(B), (C), 7(A) and 9(A)). Similarly to the perineal body [33,54], the PSMs appear to originated and migrate from the de-

veloping longitudinal muscle layer of the fetal rectum.

At the rectovaginal interface, the longitudinal smooth muscle layer of the rectum continues to the LAM without any contribution by the endopelvic fascia (Figures 5(A),

(B) and 7(A)). Thus, on the anterior side of the lower rectum, there is no “conjoint muscle coat of the rectum” in the true sense (see the subsection “Levator ani muscle, collagenous fibers and elastic fibers”). However, irrespective of whether joining of the endopelvic fascia is present or absent, some researchers may consider the inferior

continuation of the longitudinal muscle layer of the rectum to be the LAM. The LAM ends at the PSMs, and the

same PSMs provide the major anterior insertion to the external anal sphincter (Figures 6(B), 7(A) and 9(A)). Conversely, the strong striated sphincter seems to play a dynamic role in supporting the rectovaginal interface to maintain a suitable antero-posterior length of the urogenital hiatus against intra-abdominal pressure. The anterior attachment of the anal sphincter formed by the PSMs seems to be adapted more effectively to mechanical stress than the static posterior attachment formed by elastic fibers, i.e., the anococcygeal ligament and raphe [55].

At its posterior end, the female perineal membrane ends at the PSMs (Figure 9(A)). The typical multilayered perineal membrane is rarely found in elderly Japanese women [8] (Figures 9(F) and (G)) because the PSMs have enlarged, destroying the membrane from the superficial and posterior sides. Sometimes, most parts of the perineal membrane appear to be integrated into the bulky PSMs. Accordingly, the mass of the PSMs occupies the center of the fibrous skeleton in the female perineum, and integrates both a group of vertical structures (vaginal wall smooth muscles and the LAM) and a group of structures extending horizontally and/or frontally (the perineal membrane and perineal body). We propose a schematic representation showing the critical position of the PSMs (Figure 11; see the final paragraph of this subsection). In the vertical structures, the urethra is tightly connected with the vagina [56,57] because of vaginal descent with marked reconstruction of the urethral posterior wall [58]. The female urethral smooth muscles are characterized by their thick longitudinal layer[59]. The PSMs in females correspond to the recto-urethralis muscle in males: the male LAM also insert into the recto-urethralis muscle [60,61]. However, in contrast to the smooth muscle mass in males, PSMs are unlikely to contain the cavernous nerves in females.

As typically seen in arterial walls, smooth muscles and elastic fibers usually coexist because elastic fibers are necessary for maintaining the 3-dimesnional configuration of smooth muscle fibers [62]. As well as the vascular wall, in the pelvic floor, smooth muscle and elastic fibers often coexist (Figures 1(C), (D), 3(B), (C), 8(E)- (H) and 9(B), (E)). Do fibers of PSMs exert synchronized action with the levator ani under the control of nerve impulses? Smooth muscles in the pelvic floor connective tissue are not arrayed regularly and are directed at random (Figures 6(D) and 8(E)-(H)). PSMs contain abundant nerve terminals (Figures 7(C), (D) and 10(C)), but their density is much lower than in the circular and longitudinal smooth muscle layers of the rectum (Figure 1(B)). The PSMs are likely to differ in function from the regularly arrayed smooth muscle structures typically seen in the intestine. The latter type of structure shows organized contraction or peristalsis under the control of nerves and hormones, but smooth muscles in connective tissue may not, in view of their random arrangement. This is somewhat reminiscent of the nature of smooth muscle cells or fibers in the walls of arteries, which can act against blood pressure without nerve or hormonal control (i.e., Bayliss’ rule [63,64]).

Connective tissue composed of smooth muscle would seem to function as an ideal barrier, septum or protector against mechanical stress because, even without innervation, smooth muscle fibers resist (not absorb) pressure in accordance with Bayliss’ rule. This function seems to be much stronger than the passive action of elastic fibers. However, at the full term of pregnancy, PSMs are most likely to become highly relaxed under various hormonal influences to assist delivery. In contrast, extracellular collagenous and elastic fibers would be torn or elongated by excess tension during vaginal delivery. In addition, the venous plexus located around the vagina is well developed in multiparous women [5,42]. The venous plexus also appears to have a role in relaxation of the rectovaginal interface during delivery. Nevertheless, any dynamic function of structures other than PSMs would appear to be unlikely.

The “integrated pelvic floor theory” [65] attributes a key role to the LAM in the static and dynamic support of the pelvic viscera, being involved in the closure and opening of the urethra and anal canal. Petros’ system resembles the line drawings published by Courtney [66], and according to both of them, the levator ani gives off “striated” muscle fibers extending longitudinally and inferiorly along the rectum. However, except for Macchi et al. [22], no other researchers have supported the existence of such striated muscles (e.g., Arakawa et al. [20]). Because Courtney [66] based his conclusions on gross anatomy and because the histological demonstration by Petros [65] was limited to the pubo-urethral ligament, their overemphasis of the morphology of the levator ani seemed to result from a lack of histological observations. We agree that, in the female pelvic floor, there is a complex of smooth muscles, especially vertically running muscles, including the LAM. However, smooth muscles are unlikely to play an active traction role without the cooperation of striated muscles. Moreover, the real LAM is located distantly from the female urethra or its rhabdosphincter, but it ends more posteriorly around the vagina and vaginal vestibule [13]. In accordance with Bayliss’ rule, rather than traction or elevation, the PSMs seem to act as an ideal barrier, septum or protector against mechanical stress such as that resulting from intra-abdominal pressure. In addition, Petros’s “posterior levator plate” (the anococcygeal ligament and raphe) is elastic [55,67] and unlikely to provide any strong posterior traction force. Another review [68] has focused on the limited effect of the levator ani on urethral function in both genders. Consequently, to demonstrate the key role of the perineal smooth muscles and perineal body for integrated function of the LAM, anal sphincters and perineal membrane, we have proposed our “catamaran” model (Figure 11/a>).

Nakajima et al. [69] reported that the deep transverse perineal muscle is attached to Cowper’s gland in males (Figure 10(G)), and is continuous with the urethral rhabdosphincter. With regard to the situation in females, Fritsch et al. [70] reported that the muscle was absent. Despite his excellent diagrams, which included the mem-

branous structure at the external genitalia, Oelrich [4,71] had a negative, rather than positive, opinion, because he had no opportunity to observe striated muscles in the membranous structures of the female. Oh and Kark [72] successfully demonstrated the transverse muscle in the lateral side of the perineal body in females, but their findings were not subsequently cited by others. We believe that the muscle is present, although vestigial, adjacent to Bartholin’s gland (Figure 10(B)). Thus, in both genders, the muscle is located inferiorly or superficially relative to the perineal membrane, and adjacent to the major perineal gland. Nakajima et al. [69] considered that, because the deep transverse perineal muscle is not sheet-like but a 3-dimensional pillar continuous with the rhabdosphincter, previous researchers had found it difficult to identify, especially in histology preparations.

On the basis of magnetic resonance imaging studies, the strictest argument against the existence of the urogenital diaphragm was provided by Myers [73], who established a safe treatment for the retropubic veins in radical prostatectomy. He stated that “there is not a hint of what might be called Henle’s artifact, his diaphragma urogenitale.” Likewise, Mirilas and Skanadalakis [74] had a strongly negative opinion regarding the existence of the diaphragm. The deep transverse perineal muscle has long been considered the core of the urogenital diaphragm. However, in our cadaver dissections, when we approached the urogenital hiatus from the ischiorectal fossa lateral to the rectum and extending to the lateral side of the urethra or vagina, we were able to palpate a diaphragm-like structure containing 1) the urethral and urethrovaginal sphincters, 2) the mass of the PSMs, 3) the perineal membrane, 4) Bartholin’s gland and its associated striated muscle fibers (the vestigial deep transverse perineal muscle) and/or 5) the vestibular bulb and its associated bulbospongiosus muscle. Likewise, for measurement of thickness using clinical imaging, Betschart et al. [75] considered the perineal membrane as a mass that includes striated muscle, smooth muscle and connective tissues. Because such a gross entity including multiple structures is a useful descriptor, we do not rule out the concept of the urogenital diaphragm.

Usually in the body, not only muscle contraction force for action but also passive mechanical stress can be transferred through a series of collagenous tissues of different types. However, in the pelvic floor, both are passed from collagen fibers to an elastic fiber-smooth muscle complex. When they are coexisted, elastic fibers bundle and connect with smooth muscles to maintain the configuration of the latter. PSMs provide the major anterior insertion to the external anal sphincter and, conversely, the strong striated sphincter seems to play a dynamic role in supporting the rectovaginal interface to maintain the antero-posterior length of the urogenital hiatus against prolapse. The bilateral large mass of the PSMs constitutes a pair of bases or cores of a smoothmuscle skeleton that includes both a group of vertical structures (the vaginal wall smooth muscles and the LAM) and a group of structures extending horizontally and/or frontally (the perineal membrane and perineal body). Randomly arrayed smooth muscle fibers in the PSMs are likely to act as a barrier, septum or protector against mechanical stress because, even without innervation, they would resist pressure in accordance with Bayliss’ rule.

We are grateful to the individuals who donated their bodies after death to Tokyo Dental College for research and education on human anatomy without any socio-economic benefit. We also thank their families for agreeing to the donation, as well as their patience in waiting for return of their remains after study.