Since the first reports of a lobular subdivision of the liver in the 17th century,1, 2 various organizational concepts have been proposed,3–15 and two conflicting models are still being used: the lobular unit,3, 4 in which blood is supplied to the surface and drained from the center, and portal and acinar units,5–7 in which the supplying vessels are at the center and draining vessels on the surface. In rat liver, the demonstration of “lobular” units using glucose-6-phosphatase as a marker16–18 led to the reconstruction of a secondary unit consisting of 14 primary units.19 Recently, we reported on considerable degrees of variability in the shape and size of primary units as well as in the number of primary units integrated into secondary units in rat liver and used the term “modular microarchitecture” to describe this variability.20 In the human liver,9, 21 three-dimensional reconstruction has provided detailed information on the lobular microvasculature, particularly on the vascular septum,3, 22–24 which connects portal venules and forms a continuous vascular surface from which sinusoids originate. Structural differences occur among sinusoids depending on their site of origin. In the rat liver, these structural differences are correlated with the metabolic heterogeneity of hepatocytes,16–18 leading us to differentiate between “portal” and “septal” sinusoids. Matsumoto and colleagues9, 21 also reported on the subdivision of lobular “secondary units” into “primary units.” Without a suitable marker, however, the borders of the primary units could be determined only on the surface of the secondary unit. It was not possible to assess the shapes or sizes of the primary units or the internal subdivision of the secondary unit. Therefore, in the present study, alkaline phosphatase activity was used as a marker for three-dimensional reconstruction to investigate the modular microarchitecture of the human liver.
The morphological homogeneity of the liver parenchyma has represented a major obstacle in finding an acceptable definition of the structural/functional units of the liver. Concepts such as the “lobule,” the “portal unit” and the “acinus” remain debatable. This study investigates the modular microarchitecture on the basis of the lobular concept. Using alkaline phosphatase activity as a histochemical marker, modules could be recognized clearly. In autopsy specimens of human liver, modules were traced through sequential cryosections, and a “secondary” module having a height of 1.9 mm, a surface of 14.7 mm2, and a volume of 5.1 mm3 was reconstructed three-dimensionally. It was subdivided into 14 “primary” modules by portal tracts and vascular septa and by a common draining central venular tree. Primary modules were polyhedral, with seven to nine facets, having heights from 0.3 to 0.9 mm, surface areas from 1.7 to 5.0 mm2, and volumes from 0.1 to 0.9 mm3. Such variation in shape and size is considered an important part of the modular organization of the human liver. In conclusion, the findings on the three-dimensionality and microcirculation of liver modules support and extend the lobular concept and, at the same time, make apparent the shortcomings of the concepts of acinar and portal units. The results of this study should permit a better interpretation of histological sections of normal and pathological liver and provide a basis for understanding the metabolic heterogeneity of liver cells and their functional integration into parenchymal units. (HEPATOLOGY 2005.)
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
Materials and Methods
Preparation of Materials.
Human liver specimens were obtained from the Department of Pathology of the University of Ulm. Tissue was collected during routine autopsy and after surgical removal of solitary tumor metastases of the liver. Liver tissue was frozen and stored in liquid nitrogen. Tissue samples were trimmed and marked so that the outlines of the sections could be used for the vertical orientation of the section planes. The study protocol conformed to the ethical guidelines of the 1975 Declaration of Helsinki, as reflected by the a priori approval (no. 18/2001) of the ethics commission of the medical faculty of the University of Ulm.
Histochemical Demonstration of Alkaline Phosphatase, E.C.184.108.40.206.
Serial sections (14 μm) were cut parallel to the organ surface at −20°C, air-dried for 5 minutes at 37°C, and fixed at 4°C for 1 minute in a solution containing HEPES-NaOH buffer (100 mmol/L; pH 7.4), glutaraldehyde (1.25%), and saccharose (1%). Enzyme activity was demonstrated according to Burstone25 using a staining kit (SK 5200, Vectastain ABC-AP; Vector Laboratories, Burlingame, CA). Cell nuclei were stained with Mayer's hemalum.26 Incubation of sections in a medium containing 0.5 mmol/L levamisole resulted in inhibition of most of the enzyme activity.
From several series of sections from tissue samples of four different livers, one series of a total of 559 sections (autopsy sample 34310/92 from the right lobe of the liver of a 59-year-old man) was used for reconstruction. Because of the complexity of three-dimensional reconstruction, the present study was restricted to a single secondary module. The images were photographed at a magnification of ×12.5, scanned at high resolution into Photoshop 7.0 (Adobe Systems Inc., San Jose, CA), and oriented vertically using the section outlines. Drawings of the images were imported into “Surfdriver” (David Moody and Scott Lozanoff, 1999; http://www.surfdriver.com) for reconstruction. Topographically, the secondary module extended from 4.5 mm to 6.4 mm below the organ surface. Physical models were built at a magnification of 75:1, as previously described.19
The drawings of alternating section planes were imported into Axiovision 3.1 (Zeiss, Oberkochen, Germany) for morphometric measurements.
Results and Discussion
Distribution of Alkaline Phosphatase Activity.
The enzyme is known to occur in variable activity in the endothelium of portal and central veins, hepatic arteries, sinusoids, and bile canaliculi.27, 28 For reconstruction, we chose a tissue specimen in which the endothelium of portal and central veins and sinusoids was clearly marked, so that the relevant structural elements of the modular microarchitecture could be identified. Fig. 1 shows a cross-section of a secondary module, which is composed of primary modules (Ia, IIb-VIb). At the surface of the modules, areas of high enzyme activity mark endothelial cells of portal veins and of the initial tortuous segment of portal sinusoids that originate directly from and in the vicinity of the portal veins (Fig. 1A-B; red arrows in module Ia and IIb). Between neighboring portal veins, where blood is distributed along interconnecting vascular septa, enzyme activity decreased from both sides and was lowest where the flow-fronts meet and septal sinusoids originate (Fig. 1B; blue arrows in module IIb and IIIb). The areas of high enzyme activity at the surface of the modules (Fig. 1B; areas shaded in dark gray) correspond to the “sickle zones,”9, 21 the lateral ends of which meet where septal sinusoids originate. In direction of blood flow—that is, from the transition of the tortuous into the straight segment of portal sinusoids and from the origin of the straight septal sinusoids—enzyme activity first decreased and then increased toward the end of the sinusoids at the central veins (Fig. 1A-B; central areas shaded in light gray). Although little is known about the specific function of hepatic alkaline phosphatase,29 the heterogeneity of sinusoidal endothelial cells correlates with structural differences among sinusoids of different origin9 and thus supports a differentiation of “portal“ and “septal“ sinusoids in human liver.16–18
General Organization of the Parenchyma.
Reconstruction revealed a group of 14 primary modules integrated into a secondary module (Fig. 2). Integration resulted from a common drainage by the branches of a central venular tree (Figs. 3, 4A) and from the arrangement of portal venular branches and vascular septa, which form a continuous vascular surface over the entire module and separate it from adjacent modules (Figs. 2, 3). The secondary module had a height of 1.862 mm, a surface area of 14,666 mm2, and a volume of 5.146 mm3. The maximum length and width were 3.463 and 3.355 mm, respectively. With the exception of modules Vc and Vd, primary modules were arranged in two horizontal layers (Figs. 3, 5): a top layer comprising modules Ia to VIa and a bottom layer composed of modules Ib to VIb. Additionally, the modules were arranged in six vertical groups (Figs. 3 [right column], 4A, 6, 7): 5 groups of two modules each (Ia/b, IIa/b, IIa/b, IVa/b, and VIa/b), and one group of 4 modules (Va-d). Group VI was located next to the draining sublobular vein (Fig. 3, right column), followed by groups I and V, in an opposing position, and by groups IV and II, also in an opposing position, adjacent to group III. This assembly is quite different from that of rat liver,19, 20 in which modules are stacked along a vertical axis.
The Central Venular Tree.
Primary modules were drained by a single branch of the central venular tree (Fig. 4A), which was located in the center and oriented along the vertical axis of the modules, as well as by occasional additional smaller branches (not shown). The diameter of the branches varied between 11 ± 2 (±SEM) μm at the distal beginning and 57 ± 4 μm at the central end where branches fused. These diameters are smaller than those seen in rat liver, which ranged from 62 ± 14 μm to 216 ± 9 μm,19 as was the number of branching steps (2 in human liver and 4 in rat liver). Tracing the central venular tree and its branches19 revealed which of the primary modules were part of a particular secondary module and where individual modules ended (i.e., the fusion of central veins). The central areas of high enzyme activity (Figs. 1, 2, and 3 [right column]) were also used for tracing individual modules, because they extended throughout a larger portion of the modules. Following the direction of blood flow (Fig. 4A, arrowheads), the central veins of modules IIIa/b, IVa/b, IIa/b and Va/b joined and connected in order from left to right to the adjacent central veins. The fusion of most of the central veins occurred in a narrow horizontal plane (planes 43-49): the central veins of modules Va and Vb joined and connected to those of modules VIa and VIb, which emptied into the sublobular vein. The fusion of the central veins of modules Vc and Vb occurred in section plane 74. The central veins of modules Ia and Ib were connected to that of module Vb (planes 80-83). The central venular tree had a horizontal main axis and two perpendicular axes (Fig. 3 [right column]), resulting in a shape that was quite different from that reconstructed in rat liver,19 where the branches and the branching steps were oriented preferentially along a vertical axis.
Portal Venous Trees.
The secondary module was supplied by six portal veins (Figs. 3 [left column], 4B). According to the branching pattern, three types of branches could be distinguished (Fig. 4B): 2 type 1 vessels (Fig. 4B [2, 3]) with an average diameter of 55 ± 3 (±SEM) μm; 8 type 2 branches with a diameter of 47 ± 2 μm (Fig. 4B [0.6, 0.5, 2.2, 2.1, 0.4, 0.1, 3.1, and 3.2]); and 24 type 3 branches with a diameter of 26 ± 1 μm. Of a total of 34 portal venular branches, 31 were located on the surface of the secondary module, and only 3 supplied blood to the “internal” surfaces of the primary modules (Fig. 4B [0.6.2, 0.5.3, 0.4.2]). Matsumoto and colleagues9, 21 have described a similar three-step branching. Third-step septal branches, in their course, lose their connective tissue sheets, their wall structure resembles that of the sinusoids, and they can then be recognized only by their “interlobular” location and larger lumina. In our study, the “proximal” part of the septal branches could be traced directly, while the “distal” sinusoid-like part was identified by areas of high alkaline phosphatase activity along the vascular septa (Figs. 6 and 7, red lines).
Organization of the Secondary Module.
A total of six groups of primary modules were integrated into a secondary module. Five groups consisted of two primary modules, and one group (group V) consisted of 4 primary modules each (Figs. 3, 5, 6, and 7). In the following discussion, group I and modules Ia and Ib are described in detail. Information on the other groups and the primary modules is presented in Figs. 2, 3, 5, 6, and 7 and in Table 1.
|Group||Blood Supply (Portal Veins)||Module||Facets||Height (μm)||Volume (mm3)||Surface Area (mm2)||Areas of Vascular Septa, mm2 (Number of Septa)||Length of Portal Sinusoids (μm), ± SEM*||Length of Septal Sinusoids (μm), ± SEM*|
|I||2, 2.1, 2.1.1, 2.1.2, 2.2, 2.2.1||Ia||8||896||0.890||5.013||0.492; 0.556; 0.808; 0.755; 0.803 (5)||654 ± 20 (104)||557 ± 11 (100)|
|2.2.3, 0.5, 0.5.3, 0.6, 0.6.0||Ib||8||630||0.659||4.907||0.680; 0.337; 0.264; 0.267; 0.505; 0.515; 0.191 (7)||666 ± 23 (89)||545 ± 17 (89)|
|II||0.1.2, 0.1.3, 0.4, 0.4.1, 0.4.2||IIa||7||630||0.263||2.576||0.518; 0.619; 0.572 (3)||632 ± 43 (39)||393 ± 57 (31)|
|0.4.3, 0.3.2||IIb||7||686||0.309||2.680||0.229; 0.622; 0.398; 0.119; 0.382 (5)||525 ± 27 (49)||368 ± 24 (49)|
|III||3, 3.1, 3.1.1, 3.2, 3.2.1||IIIa||7||364||0.348||3.352||0.167; 0.114; 0.239; 0.053; 0.147; 0.286; 0.286 (7)||721 ± 26 (40)||593 ± 27 (39)|
|IIIb||9||658||0.282||2.572||0.305; 0.274; 0.124; 0.546; 0.525 (5)||573 ± 32 (57)||428 ± 37 (53)|
|IV||2.1.2, 3.1, 0.5.3||IVa||8||252||0.132||1.708||0.157; 0.265; 0.403 (3)||623 ± 55 (18)||543 ± 58 (15)|
|IVb||8||434||0.259||2.305||0.272; 0.279; 0.450; 0.251; 0.091 (5)||525 ± 22 (46)||424 ± 21 (46)|
|V||0.1, 0.1.1, 0.1.2, 2.2.3, 3.2,||Va||7||672||0.368||2.849||0.263; 0.372; 0.571; 0.340; 0.495 (5)||576 ± 24 (61)||400 ± 19 (61)|
|0.4, 0.4.1, 0.4.2, 0.4.3, 0.4.4||Vb||8||518||0.372||3.314||0.465; 0.385; 0.402; 0.326; 0.196; 0.158 (6)||578 ± 22 (59)||430 ± 19 (59)|
|0.5, 0.5.1, 0.5.2, 0.5.3||Vc||9||588||0.415||3.062||0.241; 0.449; 0.396; 0.358; 0.561 (5)||590 ± 19 (70)||474 ± 20 (70)|
|0.6.1, 0.6.3, 0.6.4, 0.6.5||Vd||8||588||0.388||3.118||0.180; 0.705; 0.219; 0.305; 0.604 (5)||594 ± 17 (62)||450 ± 22 (62)|
|VI||2.2, 2.2.2, 2.2.3, 0.6.5||VIa||9||546||0.237||2.728||0.265; 0.383; 0.943; 0.326; 0.163 (5)||525 ± 21 (47)||364 ± 29 (45)|
|VIb||8||602||0.224||2.149||0.610; 0.345; 0.309; 0.244 (4)||447 ± 20 (53)||331 ± 19 (53)|
|I-VI||Total height||1862||588 ± 19 (14)||450 ± 22 (14)|
Group I (Figs. 5, 6) consisted of two primary modules (Ia and Ib) that opposed each other with respect to the direction of central venular blood flow. This group was located on the left side of the secondary module. The blood supply (Fig. 6 [left column]) came from portal veins 2, 0.5, 0.6 and their branches (Fig. 4B). Distal portions of septal branches (Fig. 6 [left column], red lines) were identified based on the distribution of areas of high alkaline phosphatase activity on the vascular surface of the modules (Figs. 1A [IB], 2). This is indicated in Fig. 6 by darker shading of the surface. Analysis of the blood supply by portal veins and septal branches revealed the different vascular septa, the direction of blood flow (Fig. 6, red arrows on the surface of the modules) and where along the individual septa flow-fronts from neighboring inflow sources meet (violet lines) and septal sinusoids originate. This location corresponds to the border between adjacent “sickle-zones.”9, 21
Module Ia (Figs. 5, 6) had a total of 8 facets. It had a height of 896 μm, a volume of 0.890 mm3, and a surface area of 5.013 mm2 (Table 1). Its surface was subdivided into 5 vascular septa with areas from 0.492 to 0.808 mm2. The apparent length of portal sinusoids was 654 ± 20 (±SEM) μm; and the apparent length of septal sinusoids was 557 ± 11 μm. The module was drained by one central vein (and a smaller side branch), which fused with that of module Ib in planes 88/89 and emptied here into the central veins of modules Vb and Vd (Fig. 6 [90° view]).
Module Ib had 8 facets. It was 630 μm high and had a volume of 0.659 mm3 and a surface area of 4.907 mm2. The surface was subdivided into 7 septa, with areas between 0.191 and 0.680 mm2. The apparent length of portal sinusoids was 666 ± 23 (±SEM) μm; the apparent length of septal sinusoids was 545 ± 17 μm. Module Ib was drained by one central vein and a smaller side branch (Fig. 5). The central vein fused with that of module Ia and emptied into the joined central veins of modules Vb and Vd.
To investigate the modular microarchitecture of hepatic parenchyma, the problem of morphological homogeneity, which conceals the liver's internal organization, had to be resolved. This was accomplished using a metabolic marker rather than a histological staining procedure.9 Alkaline phosphatase activity was chosen because of its unique distribution in endothelial cells of the human liver, which allows the relevant structural elements of the modules to be clearly identified. The unusual stability of this enzyme27, 28 is also of advantage, because it permits the use of autopsy samples.
Using this approach, we confirmed and extended the results of Matsumoto and Kawakami's study on the lobular organization of human liver.9 The secondary module we reconstructed is larger than the secondary unit they described and is composed of 14 rather than 7 primary units. While Matsumoto and Kawakami's primary units were not reconstructed but were assumed to be cone-shaped, our reconstructed modules were polyhedral, with 7 to 9 facets, which are either plane, convex, or concave. In addition to variable shapes, the primary modules also varied in size (i.e., height, surface area, volume, number and area of vascular septa). Such morphogenetic plasticity is considered an important part of the modular microarchitecture of the liver.
Our findings also allowed for a critical reappraisal of Matsumoto and Kawakami's conclusion that their secondary unit correlates with a “classical lobule.”9 Primarily because of the size of the secondary unit and the branching of the central venular tree (see Figs. 12 and 13 in Matsumoto and Kawakami's paper9) we conclude that it is not the secondary but rather the primary units that correlate with “classical lobules.” Such assembly of primary modules into secondary modules then becomes comparable to that seen in rat19, 20 and pig liver.30, 31
Primary modules of human and rat liver9, 19, 20 share similar structural elements that are responsible for directing and timing of blood flow: (1) portal veins and their septal branches (and hepatic arteries); (2) vascular septa, which connect portal veins and septal branches to a continuous supplying surface and act as a “watershed” between adjacent primary modules9; (3) long portal sinusoids, which originate directly from and in the vicinity of the portal vessels, with an initial tortuous segment and a subsequent straight radially oriented segment; (4) short septal sinusoids, which are straight radial and lacking the initial tortuous segment that originate from the vascular septum, where inflow-fronts from neighboring portal vessels meet; and (5) a central venular branch located in the center of the primary module, draining the sinusoids.
There are also differences between the human and rat liver19: the shape of the primary modules is more complex in human than in rat liver; primary modules are arranged differently within the secondary module; portal and septal sinusoids in average are longer by 205 and 124 μm, respectively; the branching pattern of the central venular tree is different; and the diameters of the branches of the central venular tree are considerably smaller in human liver compared with rat liver.
Whereas the vascular septum in human and in rat liver9, 19 has been shown to be essential for the distribution of blood to the entire surface of the primary modules, it is not compatible with the concepts of either the acinus6–8 or the portal unit.5 Blood flow through the acinar and portal units is very different, because both units are arranged around a central axis of blood supply, from which sinusoids are thought to diverge in all directions. From this it follows that sinusoidal blood has to be collected from a large surface into a small number of draining vessels.19 Blood flow of the vascular septa, therefore, would cut through the middle of the acinar unit and separate the portal unit into three parts. In both concepts, sinusoidal blood supply is restricted to the terminal branches of the portal vein. This is not the case in either human9 or rat liver19 in which the portions of the portal venular tree supplying the sinusoids are much larger and comprise three generations of branches. Acinar or portal units, therefore, do not fit into the complex three-dimensionality of the portal venular tree.
In conclusion, knowledge of the three-dimensionality and microcirculation of primary and secondary modules can improve the interpretation of (two-dimensional) tissue sections from normal and, especially, pathologically altered liver tissue (e.g., chronic congestion and cirrhosis.9, 19, 21, 32 It provides a new basis for progress in the understanding of the metabolic heterogeneity of liver cells16–19 and its regulation.
The author is grateful to T. L. Kyander-Teutsch for the building of physical models of liver modules and morphometric measurements, to E. Groezinger for technical assistance with sectioning and histochemical staining, and to J. Baron, Ph.D., for assistance with the manuscript.