The lymphatics (i.e., lymph nodes and lymphatic vessels) in the human abdominopelvic region are generally believed to accompany arteries and veins (Haagensen,1972; Gabella,1995). This rule, perhaps even a dogma, has been applied to cancer surgery in which lymphadenectomy should be conducted after treatment or ligation of the artery and vein (Japanese Research Society for Gastric Cancer,1995; Japanese Pancreas Society,1996). Does the fetal lymphatic vessel develop along the artery and vein? Actually, previous dye-injection studies using fresh full-term fetuses found that thin lymphatic vessels run along both arteries and veins, and these studies might provide the basis for the present knowledge of the adult lymphatic morphology (Inoue,1936; Hiraki,1958). However, to the best of our knowledge, no adult lymphatic vessels have been described as running along the portal vein tributaries and joining at the venous confluence. According to Sabin (1909,1912), in the early stages of gestation, for example, at 9 weeks, the thoracic duct generates buds for the future intestinal lymph trunks, and the retroperitoneal lymph sac (i.e., cisterna chyli) appears at the root of the mesentery of the midgut. Thus, it seems likely that the fetal upper abdominal lymphatic vessels develop along the superior mesenteric artery (SMA) and celiac trunk (CT). However, Jin et al. (2010) described that fetal visceral lymphatic vessels do not usually run along the artery and vein.
In this context, we would like to elucidate the lymphatic morphology “later than that Sabin described” because of no or few references on the topographical anatomy of the abdominopelvic lymphatics developing from the early buds. Moreover, in adults, the upper abdominal lymphatic vessels have a character different from the lower abdominal and pelvic vessels: the former should descend to the level of the left renal vein and drain into the thoracic duct (Deki and Sato,1988; Ito and Mishima,1994; Hirai et al.,2001a,b). Does the developing autonomic nerve plexus disrupt the outgrowth of the lymphatic vessels to provide the descending course? We hypothesized that specific morphologies that would promote budding of the lymphatic vessels should be present around the left renal vein, to allow easy connection between the upper abdominal viscera and the thoracic duct. Consequently, the first aim of this study was to elucidate how and why the upper abdominal lymphatic vessels descend to the lower level and connect with the thoracic duct. We do not believe that the topographical anatomy of fetal lymphatics simply corresponds to a miniversion of the adult morphology. Therefore, with a close relationship to the first aim, the second aim of this study was to find and discuss the critical differences between the adult and fetal lymphatic morphologies, according to observations in midterm fetuses.
Abbreviations used: AO = aorta; CHA = common hepatic artery; crus = crus of the diaphragm; CT = celiac trunk; D1, D2, D3, and D4 = the first, second, third, or fourth portion of the duodenum, respectively; GL = autonomic ganglion; IVC = inferior vena cava; JA = upper jejunal artery; LGA = left gastric artery; LF = lymph follicle; LRV = left renal vein; panc = pancreas; PC = peritoneal cavity; PV = portal vein; RGEA = right gastroepiploic artery; SMA = superior mesenteric artery; SMV = superior mesenteric vein; SPA = splenic artery; t.colon = transverse colon; TD = thoracic duct; VP = Vater's papilla.
MATERIALS AND METHODS
According to the provisions of the Declaration of Helsinki (1995, revised Edinburgh 2000), we examined 19 early stage and late stage fetuses: eight fetuses between 10 and 15 weeks of gestation (four males and four females; craniocaudal length [CRL] 50–115 mm) and 11 fetuses between 18 and 30 weeks (six males and five females, CRL 152–280 mm). With the agreement of the families, the early stage group was donated to the Department of Anatomy, Chonbuk National University, Korea, and their use for research was approved by the University's ethics committee. These specimens were fixed in 10% formalin solution and embedded in paraffin for histology. The late-stage group had been fixed in 10% formalin solution and preserved as a collection for medical education and research at the Medical Museum of Sapporo Medical University, Japan, for more than 30 years. Because there was no possibility of contacting the families, and the fetuses were part of the museum collection, the project using these late-stage fetuses did not include a specific protocol that was examined and approved by a suitably constituted institutional ethics committee.
The paraffin sections examined were 5 μm (early stage fetuses) or 10 μm thick (late-stage fetuses) and had been cut at 50 μm (early stage) or 100 μm intervals (late stage). Horizontal sections were prepared for 13 fetuses (five early stage and eight late-stage fetuses), and sagittal sections for six fetuses (three early stage and three late-stage fetuses). More than 100 sections were cut from each early stage fetus (more than 200 from each late-stage fetus). From three of the late-stage group (one male and two females; horizontal sections; Table 1), we obtained specimens containing the entire abdominopelvic region. However, the specimens from the other fetuses were limited to the upper abdomen. Most of the sections were stained with hematoxylin and eosin (HE), but some (5–6 sections per fetus) obtained from the early stage group were used for immunohistochemistry of the lymphatic vessels (see below).
Table 1. Distribution of the abdominal and pelvic lymph follicles: A sum of cross sectional areas (mm2) measured with 0.3 mm interval
Sites of lymph follicles
Specimens A, B, and C: a male (20 weeks, CRL 170 mm), a female (20 weeks, CRL 180 mm), and a female (18 weeks, CRL 152 mm) fetuses, respectively.
CA, celiac axis; IMA, inferior mesenteric artery; SMA, superior mesenteric artery.
Distal to the left renal vein; proximal to a site in the anterior side of the duodenum; along the peritoneum facing the duodenojejunal junction.
Plane in front of and along the retropancreatic fusion fascia or a remnant of the primitive mesoduodenum (Michels, 1955).
Right aspect of the mesentery containing hepatic artery and SMA; along the peritoneum facing the liver caudate lobe.
Along the obturator nerve; along the iliac vessels below the umbilical artery.
For D2-40 immunohistochemistry for lymphatic endothelium, according to the methods of Yonemura et al. (2006) and Yajin et al. (2009), the primary antibody (monoclonal anti-human podoplanin; Nichirei D2-40; Nichirei, Tokyo, Japan; 1:100 dilution) was used after immersion in a ligand activator (Histofine SAB-PO Kit; Nichirei, Tokyo, Japan) and autoclaving (105°C, 10 min). The second antibody (Dako Chem Mate Envison Kit; Dako, Glostrup, Denmark) was labeled with horseradish peroxidase (HRP), and antigen-antibody reactions were detected by the HRP-catalyzed reaction with diaminobenzidine, followed by counterstaining with hematoxylin. For smooth-muscle actin immunohistochemistry, the primary antibody was monoclonal anti-human alpha-1 smooth-muscle actin (Dako M85; Dako, Glostrup, Denmark), and the second antibody reaction was the same as for D2-40. According to Gerecht-Nir et al. (2004) and Hayashi et al. (2008), Dako M85 stains the endothelium of arteries and veins as well as vascular smooth muscles, but produces no reaction with lymphatic endothelium. Counterstaining was not performed for the smooth-muscle actin immunohistochemistry. In addition, lymph follicles (LFs; developing lymph nodes) were examined immunohistochemically using anti IgM, anti-CD3 and anti-CD68 antibodies (Dako, Glostrup, Denmark) as markers of B lymphocytes, T lymphocytes, and macrophages, respectively.
Smooth-muscle actin immunohistochemistry displayed a negative reaction to candidate lymphatic vessels (Fig. 1A), in contrast to positive immunoreactivity with the endothelium of thin arteries and veins as well as any visceral and vascular smooth muscle. D2-40 effectively stained thin lymphatic vessels, even when they were collapsed (Fig. 1B,C). However, as recently reported by Yajin et al. (2009), the staining was usually uneven. Thus, candidate lymphatic vessels were likely to show negative immunoreactivity, despite the fact that their lumens were directly attached to accumulations of lymphocytes in the follicles. Following these immunohistochemical observations, in the descriptions below, we identified lymphatic vessels as a dilated lumen enclosed by a very thin monolayer of cells using HE staining.
The follicles contained IgM-carrying cells (B lymphocytes), CD-3-positive T lymphocytes, and CD-68-positive macrophages (figures not shown). At all abdominopelvic sites, the T lymphocytes were segregated in the central part of the node-like structure, whereas the B lymphocytes and macrophages appeared to be distributed at random. No subcapsular sinus macrophages were observed. At all the fetal stages examined, the “nodes” had not yet differentiated into structures such as the cortex and medulla. Thus, the term “lymph follicle” (LF or LFs) is used in the present description instead of the usual term “lymph node.” The LFs accompanied, and even contained, the developing lymphatic vessels (Figs. 1, 2, 5, 6, 7, and 10). Often, the lymphatic vessels did not accompany an artery or vein (Figs. 2B, 4, 7, and 10). In the late-stage fetuses (Figs. 3, 5, 8, and 9B), the LFs were full of cellular components and embedded in a relatively tight connective tissue, and thus the lymphatic vessels were difficult to identify. Notably, the portal vein and its tributaries did not accompany thick nerves, in contrast to branches of the CT and SMA, which were surrounded by a thick autonomic nerve plexus (Figs. 1, 3–8). The inferior vena cava was also accompanied by autonomic nerves, but much less than the aorta (Figs. 2, 3, 5, 6, 8, and 9). Likewise, the pelvic arteries and veins were accompanied by no or only a few nerves (Fig. 10).
The thoracic duct and its thick tributaries ran along the posterior aspect of the aorta. However, at levels superior to the left renal vein, the bulky crura of the diaphragm interfered with the connection between the thoracic duct and the paracaval LFs (Figs. 2, 3, 5, and 6). However, the thick tributaries of the thoracic duct were able to pass through a gap between the crura of the diaphragm (Fig. 2). Paraaortic and paracaval LFs were found in all horizontal sections taken from levels between the origins of the SMA and the inferior mesenteric artery. However, the thick autonomic nerve plexus limited the anterior extension of these LFs (Figs. 3 and 6). The paraaortic and paracaval LFs tended to be larger when they occurred on the left anterior side of the aorta, as well as at a site between the aorta and the vena cava. A single, slender LF sometimes extended between the origin of the SMA and the aortic bifurcation. In contrast, the density of LFs was much lower on the superior side of the origin of the SMA (Table 1); in this region, the LFs were located on the lateral and posterior sides of the autonomic nerve elements. Notably, at levels immediately inferior to the left renal vein, thick lymphatic vessels were seen draining from the paraaortic LFs into the thoracic duct (Fig. 2). The cross-sectional areas of the paraaortic and paracaval LFs were usually up to 1.5-times greater than those of the other, peripherally located LFs (see below), but were often almost the same in both areas (Table 1).
Upper Abdominal Lymphatics
The most striking observation in this study was found in the sagittal sections of the upper abdomen (Figs. 1B, 3, and 4): LFs were densely distributed in a single retroperitoneal layer or plate, which extended along the cranio-caudal axis (Figs. 3C,D and 4A). Notably, most parts of the LF-rich layer were located along and immediately outside the peritoneum. The layer extended superiorly along the posterior wall of the omental bursa toward the gastric cardia, and inferiorly into the jejunal and ileal mesentery. The superior and inferior parts of the LF-rich layer were connected by the retropancreatic chain of LFs (Figs. 3D and 4A). This retropancreatic connection was not present in the subperitoneal area but ran along the anterior aspect of the retroperitoneal fusion fascia (Figs. 7 and 8). The retropancreatic LFs developed later than, and originated from, the subperitoneal LFs lying along the posterior wall of the omental bursa behind the liver (Fig. 4A). Figures 5A,B show an atypical specimen without the retropancreatic chain of LFs, but in which LFs developed well behind the retropancreatic fusion fascia (cf. Fig. 8 at the same level as in Fig. 5). The LF-rich layer issued a posterior “rim,” which ran along the hepatopancreatic peritoneal fold and communicated with the paracaval LFs to provide posterior drainage (Figs. 3A,B and 4A).
In the jejunal and ileal mesentery, LFs were not distributed along the artery or vein, but along the peritoneum (Figs. 1B, 4B, and 5). Thus, no or only a few LFs were found between the SMA and the superior mesenteric vein. The LFs in the mesentery developed in a proximal to distal direction (Fig. 4). Likewise, in the hepatoduodenal ligament (Fig. 6), LFs were distributed along the surface peritoneum. In contrast, the artery and its surrounding nerves created a “core” in the mesentery, although collapsed lymphatic vessels were identified between the tightly packed nerves (Fig. 1B,C). The inferior vena cava was located near the subperitoneal LFs around the left gastric artery, as well as along the right aspect of the hepatoduodenal ligament (Figs. 3 and 6); this close relationship corresponded to one of the eventual posterior drainage routes. In contrast, the LFs lying along the left aspect of the ligament continued into the retropancreatic chain of LFs (Fig. 6B,C). Moreover, the retropancreatic chain of LFs extended leftward and anteriorly along the left side of the proximal SMA (Fig. 7). In this area, near the duodenojejunal flexture, the LFs faced toward the peritoneum (Fig. 8) and were connected to the LFs in the jejunal and ileal mesentery (Figs. 4 and 5). The retropancreatic chain of LFs was located far inferior to the transverse course of the splenic artery because the width of the pancreatic head was much smaller than the cranio-caudal length. In the early stage fetuses, the retropancreatic chain of LFs appeared to be separated from the left renal vein by a fusion fascia (Fig. 7); however, each of the LFs was embedded in the thick fascia in the late-stage fetuses (Fig. 8). The retropancreatic area was adjacent to another posterior drainage route at the level of the left renal vein (Fig. 2).
The LFs along the left aspect of the proximal SMA also connected to a small group of LFs in the anterior aspect of the pancreatic head (Figs. 5 and 7). The LFs in the hepatoduodenal ligament also communicated with this anterior group (Fig. 6). The anterior group of LFs was located along or near the right gastroepiploic artery and middle colic vein. In addition, developing lymph vessels created a plexus near the third portion of the duodenum (Fig. 7A). However, this plexus appeared not to communicate with the LFs lying along the left aspect of the proximal SMA or with the retropancreatic chain of LFs. The LFs located in the hepatoduodenal ligament, along the left aspect of the proximal SMA and/or in the retropancreatic chain, consistently occupied larger areas in the horizontal sections than the other upper abdominal LFs (Table 1).
Overall, the lymphatic connections from the upper abdominal viscera to the paraaortic and paracaval areas could be classified as following two routes: (1) from the jejunal and ileal mesentery, via LFs on the left aspect of the proximal SMA which connected with the retropancreatic chain of LFs, to the paraaortic LFs near the left renal vein; (2) from sites along the splenic and left gastric arteries and from the hepatoduodenal ligament to the paracaval LFs. These two routes communicated mutually on the left aspect of the hepatoduodenal ligament and at the right end of the retropancreatic chain of LFs. In sagittal sections, most of those LFs were included in a LF-rich plate, which extended along the cranio-caudal axis.
Lower Abdominal and Pelvic Lymphatics
At levels inferior to the origin of the inferior mesenteric artery, the paraaortic and paracaval LFs were reduced in both size and number (Fig. 9, Table 1). Traveling along a similar course to the thoracic duct, thick lymphatic vessels ran along the vertebral column. The lower paraaortic and paracaval LFs, as well as the common iliac LFs, were connected to the deep vessels at several sites above the origin of the umbilical artery. The LFs lying along the inferior mesenteric artery (Fig. 9) were much smaller and fewer than those in the jejunal and ileal mesentery (Figs. 4 and 5). Details of this lower abdominal topographical anatomy have been published in a separate paper, in which both the present specimens and other midterm fetuses are described (Kinugasa et al.,2008; Matsubara et al.,2009).
At levels inferior to the umbilical artery (almost corresponding to the levels, which included the internal and external iliac arteries), the LFs became separated from the arteries and veins (Fig. 10). Thus, the LFs existed independently in the loose connective tissue between the obturator internus fascia and the rectal adventitia (Fritsch,1993), the latter of which contained autonomic nerves. The largest LF in the pelvis was consistently located along the obturator nerve (i.e., the obturator LF), while the second largest was found in the inguinal area between the gubernaculum and the external iliac vein (i.e., the inguinal LF). In this study, LFs lying outside the body wall were not counted as part of the inguinal LF (Table 1). The obturator LF accompanied a bundle of developing lymphatic vessels extending from the common iliac LFs to the inguinal LFs, which was separate from the external iliac artery and vein (Fig. 10). However, some lymphatic vessels accompanied arteries and veins near the pelvic viscera (Fig. 1A). LFs directly associated with the pelvic viscera were limited to a small number in the rectal adventitia (Table 1). Rarely, an LF was found near the urethra. In addition, a few LFs were consistently located near or along the sciatic nerve root. No intergender difference was found in the distribution, size or number of the pelvic LFs. Details of this pelvic topographical anatomy have been published in a separate article, in which both the present specimens and other midterm fetuses are described (Niikura et al.,2008,2010).
This study revealed that, in midterm fetuses, lymphatic connections from the upper abdominal viscera to the paraaortic and paracaval areas appear as chains of LFs or LF-rich plate-like tissue rather than as a bundle of lymphatic vessels. These chains could be classified as following two routes: (1) from the jejunal and ileal mesentery, via LFs on the left aspect of the proximal SMA, which connected with the retropancreatic chain of LFs, to the paraaortic LFs near the left renal vein; (2) from sites along the splenic and left gastric arteries and from the hepatoduodenal ligament to the paracaval LFs. Because of the specific courses, these two routes are most likely to correspond to the intestinal lymph trunks on the left and right sides of the SMA in adults (Hirai et al.,2001b). Notably, in midterm fetuses, the intestinal lymph trunks were not “vessels” but LF-rich plate-like tissues. Moreover, possibly due to the still-unfinished rotation of the duodenum (Kanagasuntheram,1960; Yu et al.,2010), most of these plates were located along the cranio-caudal axis, not along the dorsoventral axis. Because of the LF-rich plates, we were led to suspect that the classical theory that the “lymphatic vessels develop first, while the nodes develop second” (Sabin,1909,1912) might be erroneous. However, in the early stage specimens, the LFs had not yet been filled by cellular components but did contain vessel-like or reticular tissues. Thus, we confirmed the aforementioned theory.
When clinical terminology is used (Japanese Research Society for Gastric Cancer,1995), it is likely that only node groups Nos. 1, 2, 5–9, 11–14, 17, and 18 are involved in forming a single plate during ontogeny. In the LF-rich plate, the upper subperitoneal part was located along the posterior wall of the omental bursa behind the liver and at the root of the hepatoduodenal ligament, while the lower subperitoneal part extended into the jejunal and ileal mesentery. Notably, both parts were connected by the retropancreatic chain of LFs, which was distributed along the immediately anterior side of the retropancreatic fusion fascia or a remnant of the primitive mesoduodenum (Michels,1955; Cho et al.,2009). As Jeong et al. (2009) suggested, developing lymphatic vessels seem to pass through or run along a fusion site of the peritoneum. Overall, the upper abdominal LFs seem to develop at the root of the common mesentery containing the CT and SMA or the “mesenteric trunk” by Michels (1955). In contrast, outside the mesenteric trunk, such as near the kidneys and lateral body wall, no or only a few LFs were found; these were located along the parietal peritoneum. Even the mesentery for the inferior mesenteric artery did not contain so many LFs as in the upper abdomen. Moreover, in contrast to adults, the paraaortic and paracaval LFs were not much larger or greater in number than the upper abdominal visceral LFs; rather, the numbers at both sites seemed to be similar. Therefore, conversely, the mesenteric trunk seems to have a specific feature which promotes the development and growth of LFs. A limited reference to a hypothetical positive action of the peritoneum on lymphatic development was made by Solvason and Kearney (1992), although they did not discuss lymphatic vessel development. Likewise, previous research on vascular endothelial growth factors has not focused on abdominal topohistology (Karkkainen et al.,2003; Scavelli et al.,2004; Karpanen et al.,2006).
The autonomic nerve element showed a strong tendency to accompany arteries, especially the branches of the CT and SMA, possibly because of vascular-derived neurotrophic factors (Honma et al.,2002). Actually, Isogai et al. (2010) recently demonstrated, using rat embryos, the tyrosine hydroxylase-positive paraaortic ridge contributes much to the development of retroperitoneal visceral arteries. Conversely, the nerve appeared to “avoid” association with tributaries of the portal vein. We could easily see that the developing nerves were likely to interfere with lymphatic development along arteries. However, we were unable to explain why LFs never developed along the tributaries of the portal vein. Moreover, according to van der Putte and van Limborgh (1980), budding of the intestinal trunks occurs in a gap between two large obstacles in front of the aorta, that is, the adrenal cortex and the definite kidney, but not between nerve elements. The nerve tissues are still loose at the 8–10 week stage, during which the intestinal trunks bud from the thoracic duct according to the histology described by Sabin (1909). Indeed, even in our midterm fetuses, the periarterial nerve plexus contained abundant lymphatic vessels according to D2-40 staining. Thus, it seems reasonable that the nerves and lymphatic vessels can “share” the paraaortic space for their development in the early stages. Likewise, the crura of the diaphragm are likely to be a great obstacle to budding from the thoracic duct. However, thick lymphatic vessels often ran through a narrow gap between the crura to connect the thoracic duct and the paraaortic LFs. Furthermore, Yamada et al. (2003) reported that, in adults, the thoracic duct tributaries pass through the diaphragmatic crus.
We found two posterior drainage routes to the thoracic duct in fetuses: (1) around and near the left renal vein and (2) in front of the vena cava, near the hepatoduodenal ligament. In the former site, the retropancreatic fusion fascia was attached to the left renal vein, and lymphatic vessels penetrated the fascia. In the latter site, the peritoneum was attached to the vena cava at the base of the mesentery. At these sites, developing lymphatic vessels, which appeared to be budding, were often seen in the early stage specimens. Previous macroscopic studies (Deki and Sato,1988; Ito and Mishima,1994; Hirai et al.,2001b) have demonstrated that the intestinal lymph trunks drain into a group of paraaortic nodes on the immediately inferior side of the left renal vein [No. 16b1 according to the Japanese Research Society for Gastric Cancer (1995)]. However, Murakami and Taniguchi (2004) provided an illustration showing a thick drainage vessel running from the hepatoduodenal ligament to node 16b1 along the anterior surface of the vena cava. These vessels in adults seem to correspond with the fetal posterior drainage routes. In fetuses as well as adults, the thoracic duct and its thick tributaries ran along the posterior aspect of the aorta and vena cava. Thus, along most of the course, they were located away from the peritoneum. However, we hypothesize that a specific portion of the peritoneum produces some kind of growth factor, and that abdominal lymphatic vessels bud from the thoracic duct at sites in which a specific area of the peritoneum or mesentery is located close to the great vessels (and the thoracic duct).
Other than the aforementioned hypothetical induction by the mesenteric trunk, however, the development of the paraaortic, paracaval, and pelvic LFs seemed to follow the simple rule that “LFs develop along the major drainage vessels.” Thick arteries and/or veins did not appear to be required for lymphatic development because most of the pelvic LFs did not accompany arteries or veins. In contrast to the upper abdomen, the peritoneum surrounding the rectovaginal and rectovesical pouches seemed to have little or no ability to accelerate the development of LFs. The large obturator LFs were located along the major route from the lower extremities, via the femoral canal and into the common iliac LFs. This observation is not consistent with adult morphology because the major route from the lower extremities does not run along the obturator nerve but along the external iliac vein (Haagensen,1972). Does the major lymphatic route “switch” from the obturator nerve to the iliac veins? Although the nerve and the LFs were not mutually interdigitated (indeed, they were clearly separated), we hypothesize that, instead of the lymphatic course switching, the topographical relation among the obturator nerve, and femoral canal alters drastically at stages later than 30 weeks, as well as during the postnatal period: they are included in a single horizontal section of fetuses, but not in adults. In this context, we reiterate that the upper abdominal LFs were not distributed along the arteries and veins, but along the peritoneum, in specific mesenteries. Consequently, topographical changes in the peritoneum or mesenteries during the later stages of gestation are likely to cause significant repositioning of LFs.