0022-34 I7/80/23630287$2.00

L Y M P H A T I C M E T A S T A S I S : I N V A S I O N OF L Y M P H A T I C V E S S E L S A N D E F F L U X OF T U M O U R C E L L S I N T H E

A F F E R E N T P O P L I T E A L L Y M P H A S S E E N I N T H E W A L K E R R A T C A R C I N O M A

JEAN CARR, IAN CARR, BARBARA DREHER AND KEN BETTS Departments of Physiology and Pathology, University of Saskatchewan, Saskatoon,

Saskatchewan, S7N 0 WO Canada

L Y M P H A T I C metastasis is an important mode of spread of many human tumours, but until recently relatively little investigation had been carried out on animal experimental models (see van de Velde and Carr, 1977; Carr and Carr, 1980; Weiss, 1980 for reviews). A useful and quantifiable model of lymphatic metastasis is obtained by injecting tumour cells into the footpad and studying the spread to the draining popliteal node (Carr and McGinty, 1974). Using an anaplastic chemically induced transplantable tumour, the Rd3, in this footpad model it was shown that tumour cells migrate into the lymphatic vessels in the footpad by migrating between endothelial cells, a process described as reverse diapedesis (Carr, McGinty and Norris, 1976).

The footpad model has become more useful with the observation that the prepopliteal lymphatic ducts can be cannulated and tumour cells in them directly observed and counted (Carr et al., 1979). The metastatic behaviour of the Walker rat carcinoma has been explored at the EM level with particular relation to the mode of escape of tumour cells from the circulation (Jones, Wallace and Fraser, 1971, summarised in Chew, Josephson and Wallace, 1976) but lymphatic metastasis of this tumour and in particular the mode of entry of tumour cells into the lymphatics are unexplored. It has been shown that Walker rat carcinoma cells reach an extravascular position in haematogenous metastasis, not by diapedesis, but by destroying the vessel (Chew el d.). We have investigated lymphatic metastasis of the Walker rat carcinoma to determine whether this difference is due to the fact that different tumours penetrate vessels in different ways, or because tumour cells enter lymphatic vessels in a manner different from that by which they leave blood vessels.

Before an ultrastructural study of the relationship of tumour cells to lymphatic vessels can be held to be significant in relation to lymphatic metastasis it must be shown that consistent metastasis occurs in the model used. The present report describes ( I ) experiments establishing and quantifying a footpad model of lymphatic metastasis using the Walker rat carcinoma and (2) a description of the ultrastructure of penetration of lymphatic vessels in the footpad in this model.

Accepledforpublication 1 1 Feb. 1980

287

288 J. C A R R . I . C A R R , B . DREHER AND K . BETTS

MATERIALS AND METHODS

The basic metastasis experimenr The tumour used was the Walker rat carcinoma, carried by transplantation of 0.2 ml of a

concentrated cell suspension every 10-14 days in adult male white Wistar rats (150-250 g). For experimental purposes a tumour cell suspension was prepared as described by Carr and

McGinty. Cell viability was judged by trypan blue exclusion. Tumour cells with greater than 9 0 per cent. viability were injected into the left footpad in a dose of 5 million cells to weanling rats and 20 million cells to adult male rats (150-250 g). Animals were killed sequentially in pairs at intervals thereafter (0 hr, 6 hr, 1, 2, 5. 7, 10, 15 and 20 days) and the left footpads fixed in Bouins solution, lymph nodes in alcohol-formalin-acetic solution, and liver and lungs in 10 per cent. buffered formalin, processed to paraffin wax, sectioned and stained with haematoxylin and eosin. Nodes at 0 and 6 hr were serially sectioned and all sections stained and examined. The adult animals were selected for further study because their lymphatics were big enough to cannulate.

Lymphotic cannulation In this tumour model there is greater than 90 per cent. metastasis to the draining popliteal

node, further metastasis occurs to the para-aortic nodes, and occasionally by aberrant lymphatic drainage to the inguinal node. Lymphatic cannulation of the afferent popliteal lymphatics is carried out following exposure of the short saphenous vein under Nembutal/ether anaesthesia. The lymphatics are demonstrated by the injection of 20 pI of Evans blue (BDH) (30 pg/ml in 0.45 per cent. saline solution) into the web between the second and third toes followed by passive movement of the foot, holding the toes and malleoli to avoid compression of the footpad itself. Normally the lymphatics are visible within 10-30 s. In animals with tumour present for 6 or more days the main trunks are usually not visualised until a second injection of Evans blue is made into the lateral aspect of the heel. This is presumably due to lymphatic blockage by the tumour.

A cannula consisting of the shaft of a disposable 30 g needle connected to a 15-20 cm piece of PE 10 polythene tubing (Clay Adams) is inserted by direct vision into one of the two draining lymphatics about 1 cm proximal to the ankle joint, facing distally. Lymph flows down the cannula after passive manipulation of the foot. The first drops of lymph are collected on a clean glass slide and gently smeared. The terminal 1 cm of the cannula (volume 0.628 mm’) is ther. cut off and the lymph expelled onto a clean glass slide, and gently smeared. At I0 min. intervals further I cm sections are cut off and similarly treated. The remainder of the lymph is collected in a graduated microtube to determine the flow rate. An anti-coagulant was not used. All smears are ringed by wax pencil for orientation, air dried, fixed in 10 per cent. neutral formalin and routinely stained with H & E. Collections of lymph were continued for up to 90 min.

Lymphatics were cannulated in pairs of animals at 24 hr intervals for 9 days. At day 0 the cannula was established before the injection of the tumour cells into the footpad to determine whether there might be a direct intra-lymphatic injection of the tumour cells. At each period a control animal was killed at the same time as the cannulated animals to determine whether the terminal manipulation of the foot resulted in a change in the number of tumour cells found in the node. Four additional animals were cannulated over the period 7-9 days after injection of tumour when tumour cells were very numerous in the lymph. In all animals the popliteal node and where relevant the para-aortic inguinal and axillary nodes were examined histologically, to verify the presence of metastasis. The presence of turnour cells in the cannulated lymphatic was verified at the end of two experiments by serial histological examination of the afferent lymphatic after ligation at both ends. A further lymphatic from a non-cannulated rat was similarly examined.

Differential counts of tumour cells and non-neoplastic cells were performed by scanning the entire ringed area of the slides. Cells were identified as tumour cells by their large size and dense nuclear staining; only unequivocally neoplastic cells were accepted as such. The tumour cell counts may therefore be marginally low. Small lymphocytes, polymorphs and macrophages were recognised by their characteristic morphologic features. The only area where identification was occasionally difficult was in discrimination between macrophages and large lymphocytes.

LYMPHATIC METASTASIS 289

The cells in the lymph of 6 animals, 8 days after injection were counted for periods up to 1; hr. Duplicate counts were carried out on adjacent 1 cm pieces of tubing in some cases, and a maximum variation of & 12 per cent. was seen in these. Cells did not appear to sediment in the tubing as low counts followed high counts successively. Lymph Row rates were calculated for each 10 min. period between cell samples.

Counting turnour cells in a ! m p h node The popliteal lymph node was removed with minimal trauma, dissected clean, placed in cold

phosphate buffered saline (PBS) and weighed on filter paper. The node was then dispersed in PBS and the cell suspension centrifuged. The cell pellet was resuspended in PBS to make a stock cell suspension. This was diluted 1 : 500 in PBS and counted in duplicate on a Coulter Model B Counter with a lower threshhold of 12. I f duplicate samples were not within i5 per cent. the counts were repeated until agreement was obtained. The stock cell suspension was centrifuged to obtain a concentrate which was smeared on a glass slide, fixed in 10 per cent. formalin and stained with H & E. Two hundred cells were counted. The absolute number of tumour cells present was obtained from the total cell count and the percentage of turnour cells in the differential count. This was correlated with the weight of the lymph node. After a preliminary incomplete experiment to establish the technique. this experiment was carried out on pairs of animals at daily intervals of 0 to 7 days.

Ultrastructural examination Seven days and 9 days after injection of tumour cells into the footpad animals were perfusion

fixed with 3 per cent. glutaraldehyde in cacodylate buffer and thin slices of foodpad tumour immersed in 3 per cent. glutaraldehyde for I5 min. to harden. chopped into small blocks and held in cacodylate buffer. Some of these animals received 1 ml intravenous lmferon (Fisons) while the heart was beating and prior to injection of fixative. This marked the blood vessels to aid discrimination from lymphatics. Parallel blocks of footpad were processed through paraffin and H & E sections were examined to check that areas of apparent lymphatic invasion could be seen. The draining lymph nodes were similarly checked histologically for the presence of metastasis. The footpad blocks were then stained with uranyl acetate, post osmicated in 2 per cent. osmium tetroxide and embedded in epon-araldite. Over one thousand I p m sections stained with toluidine blue were examined by light microscopy to identify the few areas where lymphatic invasion was occurring. Thin sections of one hundred of these were then cut for transmission electron microscopy and double stained with lead citrate and uranyl acetate.

RE s ULTS

The model In 95 per cent. of newborn rats the injection of 5 million tumour cells resulted

in metastasis to the draining popliteal node. In adult rats the injection of 20 million cells produced greater than 90 per cent metastasis. Detailed histological evaluation showed a progression of metastasis. Complete serial section of lymph nodes immediately after, and 6 hr after injection of tumour cells into the footpad, demonstrated that no tumour cells were present in the lymph node. Significant direct intralymphatic injection of tumour cells therefore did not occur. In the popliteal node cells were seen initially only in the subcapsular sinuses in which they accumulated and from which they spread down the radial sinuses and penetrated into the pulp of the node. At 2 days tumour cells were seen in the efferent lymphatic vessels in the hilum of the node and at 4 days or after occasionally in the blood vessels. By 5 or 6 days the node was often totally

290 LYMPHATIC M E T A S T A S I S

FIG. I.-A lymphatic vessel in a Walker rat carcinoma footpad tumour 9 days after implantation of 20 million cells into the footpad. Turnour cells bulge the endothelium and at one point apparently pass into the lumen. Toluidine Blue plastic section. x450.

replaced by tumour which established its own characteristic framework and blood supply.

The appearances in the para-aortic nodes are similar but occur one to two days later. A significant number of regressions did occur in adult rats, and degree of metastasis at any given time varied considerably. This may be related to the fact that the tumour is not syngeneic. The overall histologic appearances are similar to those illustrated earlier in the case of the Rd/3 tumour (Carr and

FIG. 2.--Lymphatic trunks from similar tumour-bearing rat. This is the trunk efferent from the footpad and afferent to the popliteal node and is the vessel cannulated. It contains a number of large tumour cells and some small reactive cells. Haematoxylin and Eosin paraffin section. x43.

J . C A R R , I . C A R R , B . D R E H E R A N D K . B E T T S 29 1

McGinty). A lymphatic vessel as seen in a 1 micron thick plastic section stained with toluidine blue is shown in fig. 1. Tumour cells bulge the endothelium and at one point are apparently passing into the lumen. Serial section of the lymphatic trunks showed numerous tumour cells in the lumen; often several were caught in the pocket behind a valve (fig. 2).

Lymphatic Cannulation Tumour cells were readily identified in lymph because they were large and

had grossly pleomorphic and hyperchromatic nuclei (fig. 3). The total number of tumour cells exported per min. at varying times after tumour implantation and at the outset of each cannulation is indicated in fig. 4. Apart from two specimens from one animal, with particularly high counts (1 300 and 4300 tumour cells/cm cannula) it was unusual to see tumour cells in groups. About 1 per cent of tumour cells were multinucleate and 1 per cent. dividing. As has been shown previously, export of tumour cells is relatively constant over a period of cannulation in any given animal (Carr et a/.). With time there is a progressive rise in the number of tumour cells leaving the footpad. Initially few cells travel in the afferent lymph; the number increases sharply after the third day. This is later checked markedly presumably due to lymphatic blockade and rises again thereafter. In general the later the stage of metastasis, the more tumour cells are found in the afferent lymph and the larger the metastatic deposit in the popliteal node. Those showing gross oedema and extension of the tumour to the heel exported more cells than those where there was no obvious oedema. In all cases there were at least two visible lymphatic trunks draining the area. In the footpads which were most grossly enlarged, injection of Evans blue into the usual site between the third and fourth toes did not demonstrate the lymphatic

FIG. 3.-Cells from cannulated lymphatic trunk from similar turnour-bearing rat, allowed to settle on a glass slide. The tumour cells are large and have densely staining nuclei. Haematoxylin and Eosin. x 360.

292 J. C A R R , 1. C A R R , B . D R E H E R A N D K . BETTS

* O I 18 i LL

Days FIG. 4.-Cell output from lymphatic trunk during the progression of metastasis. Each point represents the

mean of two animals. All the animals indicated had histologically confirmed metastasis in the popliteal node. LL = large lymphocytes. SL = small lymphocytes, M = macrophages. TC = tumour cells, P = polymorphs.

Additional animals killed at 7 and 9 days showed similar results.

trunks. In these animals the dye penetrated the lymphatics when injected into the heelpad, above the level of oedema. This suggests that the more distal lymphatics may have been totally or partially obstructed by the tumour. The normal (reactive) cells increase in number in parallel with the increase in number of tumour cells; there is no evident change in the proportion of different cell types. When a differential count (minimum 200 cells) is carried out at each time interval, tumour cells do not usually exceed 20 per cent. of the total cells present. The rate of flow of lymph varied widely from 0.3 pl/min. to 50 pl/min. (mean 5 pl/min.). At day 0 when the cannula was inserted before injection of tumour cells and the foot then manipulated a few tumour cells were recovered. This did not represent an intralymphatic injection of a large bolus of cells; there is no way of knowing whether any tumour cells at all could be recovered without manipulation of the foot. When tumour cells were injected into the footpad of two animals whose lymphatic trunks had previously been cannulated and then manipulated, the recovered lymph showed respectively two and three tumour cells per mm3 of lymph. In one of these a total of 87 tumour cells was recovered from the cannula in the first hour, while complete serial section of the lymph node showed nine tumour cells. Presumably the lower hydrostatic pressure in the cannulated lymphatic diverted lymph away from the other duct leading to the lymph node. It is possible, however, that even at this early stage tumour cells were passing directly through the node. In all five animals where concurrent

LYMPHATIC METASTASIS 293

blood samples were taken from veins draining the primary, tumour cells were identified. Single tumour cells were seen in pulmonary capillaries but the animals were all killed before massive pulmonary metastasis had time to develop.

The flow of tumour cells may have been increased by the prior increase in tissue pressure induced by the introduction of 20 pl of Evans blue. However, one serially sectioned lymphatic from the animal which did not receive Evans blue did not contain a different number of cells from the similarly sectioned lymphatics of animals injected with dye. This is probably not an important factor but it cannot be assessed. It is not possible to obtain significant amounts of lymph without delineation of the lymphatic with Evans blue and gentle manipulation of the foot to stimulate lymph flow.

Tumour cell counts in lymph nodes The results of this experiment with lymph node weights are expressed in figs.

5 and 6. There is a progressive increase in tumour cell content and lymph node weight throughout the experiment; when tumour cell count was plotted against lymph node weight there was a direct linear relationship.

Whole Cell Counts

(x1000/mm3) Weight (mq)

16

14

12

10

0

6

4

2

whole cell counts

weight

tumour cell counts / 160

140

120

100

80

60

40

20

0

Time (days)

FIG. 5.-CelI counts from left popliteal nodes of rats during the process of metastasis of Walker rat carcinoma from the left footpad.

294 LYMPHATIC METASTASIS Whole Cell Counts

Weight (mg)

FIG. 6.-Cell counts from para-aortic nodes of rats during the process of metastasis carcinoma from the left footpad.

of Walker rat

Ultrastructure The ultrastructural characteristics of lymphatic capillaries have been

described by others (reviewed by Casley-Smith, 1976). The lymphatic endothelium is very thin and protrudes cytoplasmic spikes externally, and the basal lamina is thin. It can nevertheless, be difficult to identify lymphatic vessels unequivocally. An intravenous marker was therefore used to mark blood vessels. Lymphatic capillaries are found only near the edge of the tumour.

J . C A R R , 1. C A R R , B . D R E H E R A N D K. B E T T S 295

FIG. 7.-Lymphatic vessel from the edge of a similar footpad tumour. The animal (and the others from which electron micrographs were obtained) was given an intravenous iron marker to mark the blood vessels. A tumour cell is protruding broad cytoplasmic processes from which arise fine cytoplasmic processes. These pass towards a lymphatic vessel. This and other EM material was fixed in glutaraldehyde and osmium and stained with lead citrate and uranyl acetate. ~ 6 3 0 0 .

The marker was absent from the vessels illustrated.

A lymphatic capillary devoid of intravascular marker is depicted in fig. 7. The lymphatic is lying in dense fibrous tissue at the edge of the neoplasm. Tumour cells adjacent to the lymphatic protrude broad cytoplasmic processes which spread along the exterior of the lymphatic, or give off fine processes which behave similarly. At higher magnification these processes contain micro- filaments 6-8 nm in diameter, presumably actin (fig. 8). The endothelium of the lymphatic is in places deeply indented. The tumour cell cytoplasmic processes may protrude through gaps between lymphatic endothelial cells, often appearing rather more electron dense than adjacent endothelial cell cytoplasm. It is notable that at other points in such lymphatics endothelial cells are still adherent by junctional complexes, so that there is no generalised patency of interendo- thelial cell junctions (fig. 9).

Migration of tumour cells through open gaps between adjacent endothelial cells was not identified. However, tumour cells indenting lymphatic en- dothelium seemed to produce degenerative changes in the endothelial cell cytoplasm. Such an indenting cell is seen in fig. 10. The tumour cell is elongated with nucleus at one end and a cytoplasmic tail at the other, appear- ances suggestive of cell movement. Numerous fine cytoplasmic processes of the

296 J . C A R R , r. C A R R , B . D R E H E R A N D K . BETTS

FIG. 8.-Detail of cytoplasmic processes from a similar tumour cell. Within the core of the processes lie fine filaments 6-8 nm in diameter usually in transverse section. but at some points (arrows) visible in longitudinal section for short distances. x 96,ooO.

tumour cell protrude towards the endothelium of the lymphatic vessel which has disintegrated (fig. 11). The junctional complex between the endothelial cell and its neighbour: is, however, intact. This is not a mechanical artefact because the closely related endothelium on the other side of the lymphatic vessel is not disrupted (fig. 10). Elsewhere tumour cell processes can be seen protruding through gaps between endothelial cells. The endothelial cells show marked cytoplasmic and perinuclear vacuolation indicative of cell degeneration (figs. 12 and 13). In the last stage of this degenerative process a segment of the wall of the lymphatic disappears leading to an apparently open-ended lymphatic. This allows direct access from tumour to lymphatic lumen (fig. 13). Finally, presumably due to spread down the vessel once access has been gained to it, large numbers of tumour cells can be seen within lymphatics whose wall at the level sectioned, is intact (fig. 14). Eight of the original 1000 blocks showed lymphatic invasion.

DISCUSSION The origin of the Walker rat carcinoma as a mammary tumour was

described by Earle (1935) and early accounts of its ultrastructure given by Fisher and Fisher (1961), and Mercer and Easty (1961). It is of course allogeneic. This for many purposes is far from ideal. However, the penetration of lymphatic vessels is a relatively rare phenomenon and therefore difficult to

LYMPHATIC M E T A S T A S I S 297

FIG. 9.-A similar lymphatic vessel. A tumour cell (T) extends towards a lymphatic vessel (V). Gaps are present in the endothelium of the lymphatic through one of which a tumour cell process protrudes. Numerous tumour cell cytoplasmic processes lie in the space immediately outside the lymphatic vessel. Elsewhere (asterisk) the interendothelial junctions are closed. x 7030 (montage).

298 J . C A R R , I . C A R R . B . D R E H E R A N D K. BETTS

FIG. 10.-A similar lymphatic vessel. A tumour cell (n deeply indents the endothelium of a lymphatic capillary. Endothelium on the vessel wall away from the tumour cell is intact but over the tumour cell shows discontinuity. A little of the " tail " of the tumour cell is visible indicating possible motility. x 5900.

investigate except in a rapidly growing and metastasising tumour. There is no reason to suppose that the allogeneic nature of the tumour interferes seriously with the interpretation of the present results. The present experiments show that when 20 million Walker rat carcinoma cells are introduced into the adult rat footpad, there is usually progressive metastasis to the popliteal lymph node and thence to the para-aortic node; the model is similar to that previously described (Carr and McGinty). It is less consistently successful, possibly because the model is not syngeneic; also the mitosis of sinus macrophages is less evident. Serial section of nodes at 0 hr and 6 hr did not show tumour cells in the popliteal nodes. Paradoxically, however, when lymphatic cannulation was established before tumour cells were injected, and the foot was then passively dorsiflexed and plantarflexed tumour cells were found in both the lymph and the draining

LYMPHATIC METASTASIS 299

FIG. 11.-Detail of cell shown in Fig. 10. but in deeper serial section. An interendothellal junction IS closed (arrow) but endothelial cytoplasm shows extensive degeneration and fragmentation in relation to tumour cell processes. x 20.000.

node during the first hours. This probably indicates that tumour cells pass into lymphatic vessels under the influence of increased lymph flow and possibly in relation to minor trauma to the lymphatic. The ultrastructural results to be described relate to animals whose feet were not passively moved and who would not therefore have this early passage of tumour cells into the lymph.

The experiments to determine the movement of tumour cells out of the footpad at different stages of metastasis were all done in precisely the same way and so are comparable. It is not possible to collect significant amounts of lymph without passive movement of the foot. The progression of metastasis in the node as seen morphologically is similar to that seen when tumour cells are directly infused into the lymphatic vessel (Ludwig and Titus, 1967). The parallel and synchronous increase in tumour cell numbers in the popliteal and para-aortic nodes suggests that there is little significant hold-up of tumour cells in the popliteal node.

Two new findings emerge from these experiments. The output of cells from immune reactions had been studied in elegant detail (Hall et al., 1967) and it has been shown that immunoblasts are present in large numbers in the thoracic duct lymph of animals with implanted syngeneic tumours which did not show lymphatic metastasis (Delorme et af., 1969; Alexander et al., 1969). Franchi et al. (1972) describe the recovery of tumour cells from post nodal lymph after implantation of Ehrlich carcinoma into the tibia of mice. There is, however, little information in the literature on the cells exported in the prenodal lymph efferent from experimental tumours which spread by lymphatic metastasis. In the present experiments the majority of the cells in the lymph at all stages are not tumour cells but macrophages and lymphocytes. They presumably are related in part to an inflammatory response to dead or dying tumour cells, and in part to an immune response to the antigens of an allogeneic tumour. It is now well known (see Carr and Underwood, 1974 for review) that tumours contain numerous lymphoreticular cells and it has been suggested (see Alexander, 1976) that the ability of tumours to metastasise is inversely related to the number of

300 J . C A R R . I . C A R R , B. D R E H E R A N D K . BETTS

FIG. I2.-A similar lymphatic vessel. A tumour cell (T) extends towards a lymphatic vessel (V). Gaps The are present in the lymphatic vessel wall through which tumour cell processes protrude.

endothelial cells (E) show extensive degenerative vacuolation. x 8 100.

LYMPHATIC METASTASIS 30 1

FIG. 13.-A similar lymphatic vessel surrounded by tumour cells. The endothelium is intact to the points arrowed beyond which i t is deficient and could not be traced in this or deeper sections. Tumour cells therefore have direct access to what is functionally an open ended lymphatic. x 2500.

macrophages they contain. It is clear from the present experiments that these cells also leave the tumour in large numbers. The progressive nature of metastasis in this experiment makes it obvious that the lymphoreticular cells are

302 J . C A R R , I . C A R R , B . D R E H E R A N D K . B E T T S

FIG. 14.-A similar lymphatic vessel which elsewhere had a deficient wall. In this section the endothelium is complete and the lymphatic packed with cells, mostly tumour cells but including at least one macrophage. x 1900.

not effective in inhibiting metastasis. The tumour cells in the lymph usually lie singly and only occasionally show mitosis. Metastasis is so rapid, and the tumour so malignant that it is probably not a profitable model in which to seek peculiarities in the cells which are successfully metastatic. This requires a more

LYMPHATIC M E T A S T A S I S 303

slowly metastasising syngeneic tumour in which selection of successfully metastatic variants has time to occur. It is clear from the progressively rising number of tumour cells in the lymph that continuous recruitment of tumour cells from the simulated primary occurs. The temporary check in the increase in number of tumour cells in the lymph is associated with delay in passage of Evans blue from footpad to lymphatic trunk and probably therefore with lymphatic blockade by the tumour. What is of most importance is that passage into the lymphatics is a continuous and progressive phenomenon.

Since it is possible that tumour cells could be present in lymph without establishing successful metastasis in draining lymph nodes it was necessary to evaluate the extent of metastasis by counting cells in the draining popliteal and para-aortic lymph nodes. Any serious attempt to study the effects of therapeutic manipulation on lymph node metastasis implies quantitation of the extent of metastasis. A previous attempt to do this by histological means was laborious and not very accurate (Wood and Carr, 1974). Individual counts of cell suspensions from individual lymph node metastases were consistent to + 5 per cent. The variation in cell count from one animal to another at the same time interval after injection of tumour cells into the footpad is wider than this, due to individual variations in tumour growth or metastasis rates. This may be related to the fact that the tumour is allogeneic. It is clear from the increase in number of non-neoplastic cells in the nodes, rising in parallel with the neoplastic cells, that a reactive process is occurring. Any attempt to quantitate lymph node metastasis by weight or measurement of bulk DNA synthesis is bound to be inaccurate and a satisfactory method must involve visual identification of the tumour cells. The present technique shows clearly that metastasis to the para-aortic nodes occurs early, i.e., that almost certainly many tumour cells go straight through the node. The technique is dependent on the non-cohesive nature of the tumour under study and will require modification to deal with tumours which form cellular aggregates.

The second new finding relates to the way in which the tumour cells enter lymphatics. Here again there is little information in the literature. Luncksen and Strauli (1975) using scanning electron microscopy to study an ascites tumour showed tumour cells apparently crawling through the open lacunae of the diaphragmatic lymphatics. In a model similar to the present one, Carr et al. showed tumour cells migrating between the endothelial cells of relatively intact lymphatics (reverse diapedesis). Probe-like processes appeared to find their way along the outer surface of the lymphatic and towards the interendothelial junctions. These were not generally patent; it seemed therefore likely either that the tumour cells found the few open gaps, or that they induced the endothelial cells to contract and separate. Macrophages enter lymphatic vessels in a broadly similar way (Carr, 1977). However, Walker rat carcinoma cells have been shown to leave blood vessels by destroying the blood vessel wall (Chew et al., 1976).

The findings in the present series of experiments are different from those seen in lymphatic metastasis of the Rd/3 tumour. The tumour cells still show ultra-structural appearances which suggest that they are actively migrating

304 J . C A R R , I . C A R R , B. D R E H E R A N D K . BETTS

toward the lymphatic, perhaps as the line of least resistance in a mass of rapidly dividing cells. Cytoplasmic processes were seen between endothelial cells, but whole tumour cells were not seen to migrate through. The lymphatic endothelial cells, however, show extensive degeneration thus allowing ingress of tumour cells. In places massive defects appear in the lymphatic vessel wall making it in effect a channel with an open end, into which tumour cells pass. While the sampling error in ultrastructural examination of heterogeneous tissues is great, it seems likely that had reverse diapedesis been common in the Walker rat carcinoma it would have been identified. A reasonable inference is that non-cohesive turnours like the Rd/3 and the Walker rat carcinoma invade lymphatic vessels either by reverse diapedesis or by destruction of the lymphatic endothelium. It is likely that these mechanisms operate to varying degree in different tumours. The way in which tumour cells induce death of lymphatic endothelium is unknown, but induction of anoxia, production of short range diffusible toxins, and mechanical pressure are obvious possibilities. In a previous report it was estimated that lymphatic invasion was occurring about once in 5 mm3 of tumour tissue (Carr et d.). In the present material lymphatic invasion was similarly difficult to identify and presumably similarly rare, but because of technical problems was not estimated.

SUMMARY When twenty million Walker rat carcinoma cells are injected into the footpad

of albino outbred rats, there is progressive metastasis to the draining popliteal and thence para-aortic lymph nodes. The lymphatic duct efferent from the footpad and afferent to the popliteal node has been cannulated; it has been shown that there is a continuous and progressively increasing output of tumour cells, small and large lymphocytes, macrophages and polymorphs from the footpad. The number of tumour cells in the popliteal and para-aortic nodes has been counted using a Coulter counter and subsequent differential counting of stained smears; the nodes contain a progressively increasing number of both tumour cells and lymphoreticular cells. The early accumulation of tumour cells in the para-aortic node makes it evident that tumour cells pass rapidly through the primary node. Examination of the simulated primary tumour by transmission electron microscopy suggests that tumour cells move actively toward lymphatics and protrude cytoplasmic processes through gaps in the endothelium. The endothelial cell then degenerates in close proximity to tumour cell processes. This leaves gaps through which tumour cells may pass and ultimately results in lymphatics with large defects in their walls.

About 20 per cent of the cells are tumour cells.

This work was supported by a grant from the National Cancer Institute of Canada and a Development Grant from the Medical Research Council of Canada. Mr Ken Betts and Mr Ian Etches participated as summer students supported by the Dean of Medicine’s Fund. We are grateful to Mrs W. Kao and to Mr I. Etches for technical help and to the Photographic Section, Department of Pathology, University of Saskatchewan for illustrations.

LYMPHATIC METASTASIS 305

REFERENCES ALEXANDER, P. 1976. The function of the macrophage in malignant disease. Ann. Rev. Med.,

27,207. ALEXANDER, P., BENSTED, J., DELORME, E. J. , HALL, J . G., AND HODGEIT, J. 1969. The

cellular immune response to primary sarcomata in rats. 11. Abnormal responses of nodes draining the tumour. Proc. Roy. SOC. Lond. B, 174,237.

CARR, I., AND MCGINTY, F. 1974. Lymphatic metastasis and its inhibition; an experimental model. J. Path., 113,85.

CARR, I., AND UNDERWOOD, J. C. E. 1974. The ultrastructure of the local cellular reaction to neoplasia. Int. Rev. Cytol., 37,329.

CARR, I., MCGINTY, F., AND NORRIS, P. 1976. The fine structure of neoplastic invasion: Invasion of liver, skeletal muscle and lymphatic vessels by the Rd/3 tumour. J. Path., 118, 91.

CARR, I. 1977. The passage of macrophage across lymphatic walls by reverse diapedesis: An ultrastructural study. J. Reticuloendothel SOC., 6,397.

CARR, I., AND CARR, J . 1980. Experimental lymphatic invasion and metastasis. I n Lymphatic metastasis, edited by L. Weiss and H. Gilbert, G. K. Hall, Boston. (in press).

CARR, J. , CARR, I., DREHER, B., AND FRANKS, C. R. 1979. Lymphatic metastasis of tumour; persistent transport of cells. Experientia, 35,825.

CASLEY-SMITH, J. R. 1976. The functioning and inter-relationships of blood capillaries and lymphatics. Experientia, 32, 1 .

CHEW, E. C., JOSEPHSON, R. L., AND WALLACE, A. C. 1976. Morphologic aspects of the arrest of circulating cancer cells. In Fundamental aspects of metastasis, edited by L. Weiss, North Holland/Elsevier, Amsferdam, p. 12 I .

DELORME, E. J., HODGETT, J., HALL, J . G., AND ALEXANDER, P. 1969. The cehlar immune response to primary sarcomata in rats. I. The significance of large basophilic cells in the thoracic duct lymph following antigenic challenge. Proc. Roy. SOC. Lond. B, 174,229.

EARLE, W. R. 1935. A study of the Walker rat mammary carcinoma 256, In Vivo and In Vitro. Amer. J. Cancer, 24,556.

FISHER, E. R., AND FISHER, B. 1961. Electron microscopic, histologic and histochemical features of the Walker rat carcinoma. Cancer Res., 21,527.

OTTAVIANI, G . 1972.. Lymphatic dissemination of cancer cells in mice transplanted intratibially with Ehrlich carcinoma. Lymphology, 5,3 1.

HALL, J. G., MORRIS, B., MORENO, G. D., AND BESSIS, M. C. 1967. The ultrastructure and function of cells in lymph following antigenic stimulation. J . Exp. Med., 125.91.

JONES, D. S. , WALLACE, A. C., AND FRASER, E. E. 1971. Sequence of events in experimental metastases of Walker 256 tumour: Light, immunofluorescent, and electron microscopic observations. J. Nut. Cancer Inst., 46,493.

LUNCKSEN, CH., AND ~TRAULI, P. 1975. Penetration of an ascitic reticulum cell sarcoma of the golden hamster into the body wall and through the diaphragm. Virch. Arch. I3 Cell. Path., 17,247.

LUDWIG, J., AND TITUS, J. L. 1967. Experimental tumour cell emboli in lymph nodes. Arch. Path., 84,304.

MERCER, E. H., AND EASTY, G. C. 1961. The fine cytology of the Walker tumour. Cancer Res., 21,52.

VAN DE VELDE, C., AND CARR, I. 1977. Lymphatic invasion and metastasis. Experientia, 33, 837.

WEISS, L. 1980. The pathophysiology of metastasis within the lymphatic system. In Lymphatic metastasis, edited by L. Weiss and H. Gilbert, G. K. Hall, Boston (in press).

WOOD, P., AND CARR, I. 1974. The quantitation of experimental lymph node metastasis. J. Path., 114,85.

FRANCHI, G., REYER-DEGLI INNOCENTI, I., GARAITINI, S., ALFIERJ, O., CADEMARTIRI, G., AND