Int. J . Cancer: 46, 452455 (1990) 0 1990 Wiley-Liss, Inc.
Publication of the International Union Against Cancer Publication de I’Union lnternationaie Contre le Cancer
CONTRIBUTIONS OF VASCULARIZED LYMPH-NODE METASTASES TO HEMATOGENOUS METASTASIS IN A RAT MAMMARY CARCINOMA Leonard WEISSand Pamela M. WARD Department of Experimental Pathology, Roswell Park Cancer Institute, Buffalo, NY 14263, USA. Rats received hind-foot-web (FWI) injections of MT- 100-TC mammary carcinoma cells; the resultant tumor metastasized first to the popliteal lymph nodes. Over the course of 4 weeks, in association with increases in tumor weight, the blood-flow to the popliteal nodes increased 18-fold, and their vascular densities increased 2-fold. In spite of this vascularization, cancer cells were detected in only 3 of 648 blood vessels associated with involved, ipsilateral lymph nodes compared with intravascular cells in 82 of 314 vessels associated with “primary” foot-pad lesions. The presence of tumorigenic cancer cells in the right ventricular blood of animals bearing these tumors is, therefore, considered to result from their direct entry into blood vessels from the “primary” lesions, and/or from extra-nodal invasion of vessels in tissues to which nodal tumors were adherent, as distinct from passage via lymphatico-venous communications between tumors and nodal blood-vessels. The reconstructed events occurring in the rat model, with effective restriction of regional node metastases to the nodes themselves for a time, could possibly account for the long-term survival of some patients with breast cancer and regional-node metastases, following surgery.
In common with metastases growing in other sites, lymphnode metastases become vascularized as an essential requirement for future growth. The neovasculature also provides a potential pathway for the subsequent blood-borne dissemination of cancer cells from nodal metastases, originally seeded by lymphatic delivery from primary cancers. In order to assess the contributions to metastasis of this potential lymphatico-venous pathway, measurements were made of the blood flow and vascularization of popliteal node metastases from MT- 100-TC mammary carcinomas growing in rat foot-pads, over periods of up to 4 weeks. In addition, intravasation and exit of cancer cells from nodes in the bloodvessels surrounding them was assessed, together with the development of lung metastases. MATERIAL ANDMETHODS
Animals Wistar-Furth female rats, weighing 150 to 175 g (HarlanSprague-Dawley, Anderson, IN) were used throughout. Cancer cells The MT-100-TC rat mammary carcinoma (Ghosh et al., 1983) was maintained by S.C. passage. Single-cell suspensions were prepared from tumor fragments by pronase/DNase digestion (Ward and Weiss, 1989a). Foot-web injections (FWI) Ether-anesthetized rats were injected through 25-gauge needles, into the right foot-web between the second and third metatarsals, with 5 X lo5 cancer cells suspended in 0.1 ml phosphate-buffered saline (PBS). Back-flow from the injection site was prevented by application of light pressure as the needle was withdrawn. Determination of blood flow Blood flow to lymph nodes, at different times after FWI, was expressed as normalized gamma counts (cpm), made on lymph nodes removed within 2 min after left ventricular injec-
tions of ‘251-labelled,carbonized plastic tracer microspheres (15 2 3 bm diameter: 3M, St. Paul, MN). Rats were anesthetized with intraperitoneal injections of nembutal (40 mg/kg). The intercostal location of the left ventricular pulse was determined by palpation; lo5 microspheres (=250,000 cpm) suspended in 0.2 ml PBS plus 1% fetal bovine serum, were injected into the left ventricle through the intercostal space, through a 25-gauge needle, attached to a 1-ml syringe by a 10-cm length of 0.38 mm (internal diameter) polyethylene tubing. The position of the needle in the left ventricular cavity was confirmed by the appearance of bright red (oxygenated) blood in the tubing, before and after injection. Two minutes after injection, the popliteal lymph nodes and lungs were removed and placed in 70% ethanol, then 1min gamma counts were obtained. Lymph-node histology After gamma counting, the lymph nodes were weighed, formalin-fixed, wax-embedded, cut into 5-bm sections and stained with hematoxylin and eosin. The extent of nodal involvement was recorded at 1 , 2, 3 and 4 weeks after FWI. The numbers of blood vessels containing MT-100-TC cells were counted. The vessels were small veins (mean diameters 23.3 2 3.2 bm; median 17.8 bm, range 9.5 to 143 km) identified on the basis of thin basement membranes and incomplete or no smooth muscle, endothelium and occasional intravascular erythrocytes and leukocytes. The observed vessels were either within the hilar regions of lymph nodes or in peripheral to large nodal metastases. Individual MT-100-TC cells were identified on the basis of their large nuclei (12 -C 3 bm diameter) and distinct nucleoli. Although small clumps and single cancer cells may be difficult or impossible to identify within solid tissues in routinely stained sections, they were comparatively easily recognized lying within small blood-vessels. Lymph-node vascular densities Sections of popliteal lymph nodes taken 1, 2 , 3 and 4 weeks after FWI were examined at X400 magnification, using a grid eyepiece of 0.0625-mm2 projected area to make counts of blood vessels within lymph nodes. “Primary” foot-pad tumors In these and other experiments (Ward and Weiss, 1989a), 4 weeks after FWI, 14/14 animals had pulmonary metastases. Using the same criteria as described above, counts were made of the proportions of tumor-associated blood vessels containing cancer cells, in sections from 8 of these “primary” foot-pad lesions. Lung histology Sections were prepared from lungs to detect the possible presence of micrometastases, in the absence of overt lesions. RESULTS
Bloodflow to lymph nodes This is expressed as normalized gamma counts on lymph
Received: February 20, 1990 and in revised form April 24, 1990.
453
LYMPHOGENOUS A N D HEMATOGENOUS METASTASIS
nodes, following left ventricular injections of radio-labelled microspheres. The results summarized in Table I show an increased blood flow to the ipsilateral popliteal nodes (i.e., increased microsphere retention) over the 1- to 4-week period after FWI.Over this period, no significant counts above background (105 cpm) were obtained in the contralateral popliteal nodes. Lymph-node histology Popliteal lymph nodes removed from animals 1 to 4 weeks after FWI were examined histologically, and the results are summarized in Table 11. Cancer cells were first observed in the ipsilateral popliteal nodes 2 weeks after FWI.After 3 weeks, although all ipsilateral nodes were largely replaced by cancer cells, the node capsules were generally intact. Infiltration of cancer cells had occurred through the hilar regions, into the perinodal fat and along the fascia1 planes of muscle. At 4 weeks, normal nodal tissues were entirely replaced by tumor, with mean cortical width of 0.5 ? 0.07 cm (n = l l ) , containing morphologically intact cancer cells. There were central, amorphous necrotic cores, measuring 1.53 2 0.01 cm (n = 9) in diameter. In the cortices, the cancer cells were arranged in loose strands, interspersed with extravasated erythrocytes. In most cases, the metastatic tumors had extended into the adjacent fat and muscle, to completely surround the blood vessels in these sites. The contralateral popliteal nodes were seen to contain scattered cancer cells 3 weeks after FWI in 1/12 animals, and by 4 weeks, 319 animals had their contralateral popliteal nodes partially replaced by cancer cells. Intravascular small clumps of cancer cells were detected only in animals 3 weeks after FWI in 3/316 identified vessels (Table 11). In these cases, the blood vessels were observed lying within the fatty tissues. Lymph-node weights As summarized in Table 111, there is a consistent, non-linear increase in the weights of the ipsilateral nodes in contrast to the contralateral nodes, over the 4-week period of observation. Approximately 20% of the volume of the 4-week ipsilateral nodes is due to central, necrotic material. After 3 and 4 weeks, paired t-tests reveal that the ipsilateral nodes weigh significantly more than the contralateral nodes (p = 0.05 and GO.001 respectively).
TABLE 11- LYMPH-NODE [NVOLVEMENTAND INCIDENCE OF INTRAVASCULAR CANCER CELLS AT SPECIFIC TIMES AlTER FOOT-WEB INJECTIONS OF MT-100-TC CARCINOMA CELLS Weeks after foot-web injections
1 2 3 4
Intravascula? cancer cells
Nodal involvement’ Ipsilateral
Contralateral
(incidence)‘
- (0112) (17117) (16/16) (12/12)
++ +++
++++
bsilateral Contralateral -
- (0/7) - (0/10) (1/12) (3/9)
+
++
0/196 0/171 3/316 01161
0/134 0/85 0/122 0/215
+
+
++ +
TABLE 111 - POPLITEAL SODE-WEIGHTS AT SPECIFIC TlMhS AFTER FOOT-WEB INJECTIONS OF MT-100-TC CARCINOMA CELLS Popliteal node weights Weeks after gm ? SE (n) foot-web injection bsilateral Contralateral
Controls 1 2 3 4
0.006 0.013 0.019 0.09 4.66
k
0.002 (4)
+. 0.001 (7)
*k 0.005 (6) 0.04 (9)
k
0.5
(4)
0.004 0.01 0.01 0.006 0.01
f 0.001 (4) f 0.002 (7)
*f 0.003 (6) 0.001 (9) f 0.002 (4)
TABLE IV - VASCULAR DENSITIES IN IPSILATERAL AND CONTRALATERAL POPLITEAL LYMPH NODES/TUMORS AT SPECIFIC TIMES AFTER FOOT-WEIB INJECTIONS OF MT-100-TC CARCINOMA CELLS Weeks after foot-web injection
Controls 1 2 3
4
Vascular densities’ -C SE (n)’
t-tests Ipsilateral
Ipsilateral
*
0.89 0.08 0.45 0.05 (144;5) 0.64 k 0.07 ( 183;6) 1.04 2 0.09 (187;16) 1.34 0.18 (205:12)
* *
vs.
Contralateral
contralateral p -
(223;15) 0.60 2 0.09 (97;4) 0.61 2 0.07 (142;6) 1.15 k 0.10 (164; 12) 1.02 2 0.11 (168: 10)
-
0.117 0.765 0.423 0.142 ~
Lymph-node vascular densities The vessel counts per unit areas of the lymph nodes at different times after FWI are summarized in Table IV. One week after FWI the mean vascular densities in the ipsilateral nodes showed a statistically significant decrease over those in the untreated controls @ < 0.0001). The mean densities showed statistically significant increases between 1 and 2 weeks (p = 0.036) and 2 and 3 weeks @ = O.OOOS), but not between 3 and 4 weeks (p = 0.15). The contralateral nodes TABLE I - BLOOD-FLOW TO LYMPH NODES’ Weeks after foot-web iniection
Ipsilateral popliteal node gamma counts corn 2 SE (n)*
Controls3 1 2
approx. bkgd4 (4) 216 67 (7) 233 f 94 (6) 1112 2 304 (9)
3 4
*
3832 2 1971 (4)
‘Gamma counts in the ipsilateral popliteal nodes, made 2 min after left ventricular injection of radiolabelled microspheres (approx. 250,000 cpm), in tumor-bearing animals at different times after foot-web injections.-’Minus background (105 cpm). -31.e., non-tumor-bearing animal~.-~cpmnot significantly above background.
-
‘Extent of nodal involvement: - no cancer cells present; scattered cancer cells present in node; + + normal nodal tissues partially replaced by cancer cells; + + normal nodal tissues largely replaced by cancer cells; + normal nodal tissues completely replaced by cancer cells.- Incidence of involved n~des.-~Incidenceof intravascular clumps of cancer cells in peripheral and hilar blood vessels.
~~
‘Mean number blood vessels per 0.0625 mm2.-*n = (total number of fields examined; total number of nodes examined).
also showed a decreased vascular density compared with controls @ = 0.03), after 1 week; no change was observed between 1 and 2 weeks @ = 0.93); there was a significant increase between 2 and 3 weeks (p < O.OOOl), and no significant change between 3 and 4 weeks (p = 0.38). As shown in Table IV, no statistically significant differences were observed between the mean vessel densities in the tiniepaired ipsi- and contra-lateral nodes. “Primary” foot-pad tumors In sections of foot-pad tumors from 8 animals bearing pulmonary metastases, 82 of 3 14 identifiable blood-vessels contained cancer cells. Lung histology Pulmonary micrometastases were detected in 4/4 rats only after 4 weeks following FWI. DISCUSSION
Initially, cancer cells enter the bloodstream and lymphatic
454
WEISS AND WARD
system by independent processes. Subsequently, cells in the lymphatic system may enter the bloodstream mainly via the thoracic, subclavian and right lymphatic ducts, which terminate in the venous system at the junctions of the internal jugular and subclavian veins (Haagensen, 1972). Conversely, cells in the bloodstream may seed lymph nodes via the nodal arteries. Although various attempts have been made to demonstrate lymphatico-venous communications, the demonstration of lymph-to-blood passage of cancer cells has been much less convincing than blood-to-lymph passage (Weiss, 1980). Metastases may themselves generate further metastases (Weiss, 1985a), which disseminate via the blood-stream, the lymphatic system and other routes (Weiss, 19856). In transplanted tumors, which in some ways are analogous to metastases with respect to neovascularization, cancer cells enter blood-vessels in the tumor periphery (Warren et al., 1978). In S.C. T241 tumors in mice, a tumor neovasculature appeared after 4 days, and cancer cells were detected in vessel perfusates 1 day later; a linear relationship was observed between the densities of perfused vessels and the concentrations of effluent cells (Liotta et al., 1974). The extent of cancer-cell release into the bloodstream may be considerable. Thus, MTW9 mammary carcinomas growing in subcutaneous pouches in the rat released more than lo6 cancer cells per gram of tumor per 24 hr (Butler and Gullino, 1975); similar results were obtained with 3LL and B16 melanoma in mice (Glaves, 1983). The specific question addressed here is whether or not vascularized metastases in lymph nodes also release sufficient numbers of cancer cells into the bloodstream to affect the progression of hematogenous metastasis. Jirtle (1981) estimated the blood flow to lymphatic metastases from the MT mammary carcinoma by arrest of 25-pm diameter, radioactive microspheres injected into the left ventricle of rats. In this and the MTW-9B rat mammary carcinoma, relative blood flow was reported as inversely proportional to tumor weight. In the present studies, which were made with a different tumor and smaller (15 pm) microspheres, the results are different in some respects from those reported by Jirtle. Thus, we also observed a progressive increase in lymph-node blood flow (Table I) associated with increasing total lymph-node weight (Table HI). However, the relationship between blood flow and tumor weight was not linear; when the logarithm of tumor weight was regressed against blood flow (cpm), a highly significant correlation coefficient was obtained (r = 0.99; p = 0.003), indicating a complex relationship between tumor volume and vascularity. In spite of the complexity of the relationship, increases in blood flow to the lymph-node metastases over time were associated with increased nodal weight which, to a major extent, reflected increased tumor burden. The increased blood flow did not appear to depend to a major extent on increased vascular density because, on the one hand, no statistically significant increases in vascular density were observed between 3 and 4 weeks (Table IV) although, during this time, there was an approximately 3-fold increase in blood flow (Table I). On the other hand, the approximate 2-fold increase in vascular density occumng between 2 and 4 weeks corresponded to an approximately 16-fold increase in blood flow. As the lymph-node metastases increased in size, many more blood vessels were present. Based on the modest increases in blood-vessel density, the probability of an individual cancer cell coming into contact with a vessel closely associated with a tumor is expected to be increased by only a small amount; however, in terms of the whole cancer, more cancer cells are expected to come into contact with the tumor-associated vasculature as a whole, because more vessels and cancer cells are present.
If cancer cells in fact enter the bloodstream directly and frequently from lymph-node metastases, it might reasonably be expected that they would have been detected in the hilar and other lymph-node-associated vessels. However, it is extremely difficult to detect invasion of small vessels in tumor sections stained with hematoxylin and eosin (Martin et al., 1987). We have attempted to avoid this problem by confining our search for intravasated cancer cells to the easily defined hilar vessels of the lymph nodes and the peripheral vessels of advanced nodal metastases, since cancer cells entering less well defined vessels within the tumor itself must ultimately disseminate via these vessels. However, between 2 and 4 weeks, when large numbers of cancer cells were present in the ipsilateral nodes, intravascular cancer cells were observed in only 3 of a total of 648 vessels examined (Table 11); in these cases, observed 3 weeks after FWI, small clumps of cancer cells were observed lying within 3 vessels associated with fat, to which the “fixed” nodes were adherent and which cancer cells were invading. Using the same criteria for blood-vessel and cancer-cell identification, significantly more (82 of 314; p = 0.001) blood vessels associated with metastasizing, “primary” footpad lesions contained cancer cells than did similar vessels associated with involved lymph nodes. However, the presence of intravascular cancer cells is not necessarily predictive of hematogenous metastasis (Weiss, 1990). Bioassays previously made on right ventricular blood from rats bearing MT- 100-TC tumors in their foot-pads revealed tumorigenic cancer cells in 4/8 host animals after 3 and 4 weeks, and lung metastases were observed in 10/10 animals after 4 weeks. However, the lung metastases were of the diffuse type traditionally associated with lymphogenous metastasis, as distinct from the discrete nodular lesions associated with hematogenous metastasis (Ward and Weiss, 1989a). In the present study, pulmonary micrometastases were only detected after 4 weeks (4/4 animals) by microscopy. In contrast to this limited blood-borne dissemination, extensive lymphatic metastasis, extending up to the axillary nodes, was observed after 4 weeks. These observations are in accord with those previously described in detail elsewhere (Ward and Weiss, 19896). On the basis of the present experiments, the sequence of metastasis-related events occurring in the popliteal lymph nodes may be reconstructed. Tumors seeded in the nodes by lymphogenous metastasis from foot-pad tumors grow and become progressively more vascularized. However, in spite of dramatic increases in blood-flow to the involved lymph nodes, cancer cells are not detected within associated blood vessels, in “mobile” nodes which are not adherent to adjacent structures. Therefore, although in other experiments (Ward and Weiss, 1 9 8 9 ~ viable ) cancer cells were recovered from the venous blood draining foot-pad tumors, these cells must have entered the bloodstream via veins draining the “primary” tumors, as distinct from those draining the lymph-node metastases. The vast majority of cancer cells entering the bloodstream from many types of cancers are killed in the microvasculature by a variety of fast and slow mechanisms (Weiss et al., 1989), which are major contributors to “metastatic inefficiency” (Weiss, 1990). However, passage through the lymphatic system is expected to be much less traumatic (Weiss, 1980). This differential trauma may well account for minimal (micrometastatic) involvement of the lungs in the presence of massive lymph-node involvement. The results are also in accord with the view that, until the degree of lymph-node involvement progresses to lymph-node “fixation”, when cancer cells invade extranodal tissues and intravasate at these sites, hematogenous and lymphogenous metastasis remain independent processes, proceeding at different rates. Even when distant lymph-node metastasis occurs, and can-
LYMPHOGENOUS AND HEMATOGENOUS METASTASIS
cer cells are released into the major ducts terminating in the subclavian veins, most of the intravenous cells will be killed in the pulmonary microvasculature (Weiss, 1990). Although some might consider that a transplantable mammary carcinoma growing in the hind foot-pad of a rat is a less than ideal model for human breast cancer, our results nonetheless provide a possible explanation for some limited aspects of human disease. Thus, in a large series of breast cancer patients with regional node involvement, following surgery, 24% survived for 20 years (Axtell et al., 1976). These patients survived presumably because, at the time of operation, metastases were confined to the regional lymph nodes and, although cancer cells were probably entering the bloodstream, they did not
455
form hematogenous metastases. Analogous results were obtained in the rat model system following surgery (Ward and Weiss, 1989b). Fixation of involved lymph nodes adversely affects prognosis in humans and, as in the rats used in the present experiments, probably indicates an increased risk of hematogenous metastasis. ACKNOWLEDGEMENTS
We thank Dr. J.P. Harlos for help in statistical analyses, and Ms. G. Elkin and Mr. D. Graham for their technical assistance This work was partially supported by a grant to L. Weiss from the NCI, DHHS.
REFEREINCES
AXTELL,L.M., ASIRE,A.J. and MYERS,M.H., Cancer patient survival. Report 5 , p. 16, US Dept. Health Education and Welfare (DHEW h b l . No. (NIH) 77-992), Bethesda, MD (1976). BUTLER,T.P. and GULLINO,P.M., Quantitation of cell shedding into efferent blood of mammary adenocarcinoma. Cancer Res., 35, 512-516 (1975). GHOSH,S., ROHOLT,O.A. and KIM, U., Establishment of two nonmetastasizing and one metastasizing rat mammary carcinoma cell lines. In Vitro, 19, 919-928 (1983). GLAVES, D., Correlation between circulating cancer cells and incidence of metastases. Brit. J . Cancer, 48, 665-673 (1983). HAAGENSEN, C.D., General anatomy of the lymphatic system. In: C.D. Haagensen, C.R. Feind, F.P. Herter, C.C. Slanetz, and J.A. Weinberg, (eds.), The lymphatics in cancer, pp. 2 2 4 1 , Saunders, Philadelphia ( 1972). JIRTLE,R.L., Blood flow to lymphatic metastases in conscious rats. Europ. J. Cancer. 17, 53-60 (1981). LIOTTA,L.A., KLEINERMAN, J. and SAIDEL,G.M., Quantitative relationships of intravascular tumor cells, tumor vessels, and pulmonary metastases following tumor implantation. Cancer Res., 34,997-1004 (1974). MARTIN,S.A., PEREZ-REYES, N. and MENDELSOHN, G., Angioinvasion in breast carcinoma. Cancer, 59, 1918-1922 (1987).
WARD,P.M. and WEISS,L., The relationship between lymphogenous and hematogenous metastasis in rats bearing the MT-100-TC mammary carcinoma. Clin. exp. Metasi., 7, 253-264 (19890). WARD,P.M. and WEISS,L., Metachronous seeding of lymph node metastases in rats bearing the MT-100-TC mammary carcinoma: the effect of elective lymph-node dissection. Breast Cancer Res. Treat., 14, 315-3210 (19896). WARREN, B.A., SHUBIK,P. and FELDMAN, R., Metastasis via the blood stream: the method of intravasation of tumor cells in a transplantable melanoma of the hamster. Cancer Lett., 4, 245-251 (1978). WEISS,L., The pathophysiology of metastasis within the lymphatic system. In: L. Weiss, H.A. Gilbert and S.C. Ballon (eds.), Lymphatic system metastasis, pp. 2-40, G.K. Hall, Boston, MA (1980). WEISS,L., The metastasis of metastases. In: Principles ofmetastasis, pp. T (198%). 201-207, Academic Press, Orlando, l WEISS,L., Metastatic pathways. In: Principles of Metastasis, pp. 181194, Academic Press, Orlando, FL (19856). WEISS,L., Metastatic inefficiency. Advanc. Cancer Res., 54, 159-211 (1990). WEISS,L., ORR,F.W. and HONN,K.V., Interactions between cancercells and the microvasculature: a rate-regulator for metastasis. Clin. exp. Metast., 7, 127-167 (1989).
Publication of the International Union Against Cancer Publication de I’Union lnternationaie Contre le Cancer
CONTRIBUTIONS OF VASCULARIZED LYMPH-NODE METASTASES TO HEMATOGENOUS METASTASIS IN A RAT MAMMARY CARCINOMA Leonard WEISSand Pamela M. WARD Department of Experimental Pathology, Roswell Park Cancer Institute, Buffalo, NY 14263, USA. Rats received hind-foot-web (FWI) injections of MT- 100-TC mammary carcinoma cells; the resultant tumor metastasized first to the popliteal lymph nodes. Over the course of 4 weeks, in association with increases in tumor weight, the blood-flow to the popliteal nodes increased 18-fold, and their vascular densities increased 2-fold. In spite of this vascularization, cancer cells were detected in only 3 of 648 blood vessels associated with involved, ipsilateral lymph nodes compared with intravascular cells in 82 of 314 vessels associated with “primary” foot-pad lesions. The presence of tumorigenic cancer cells in the right ventricular blood of animals bearing these tumors is, therefore, considered to result from their direct entry into blood vessels from the “primary” lesions, and/or from extra-nodal invasion of vessels in tissues to which nodal tumors were adherent, as distinct from passage via lymphatico-venous communications between tumors and nodal blood-vessels. The reconstructed events occurring in the rat model, with effective restriction of regional node metastases to the nodes themselves for a time, could possibly account for the long-term survival of some patients with breast cancer and regional-node metastases, following surgery.
In common with metastases growing in other sites, lymphnode metastases become vascularized as an essential requirement for future growth. The neovasculature also provides a potential pathway for the subsequent blood-borne dissemination of cancer cells from nodal metastases, originally seeded by lymphatic delivery from primary cancers. In order to assess the contributions to metastasis of this potential lymphatico-venous pathway, measurements were made of the blood flow and vascularization of popliteal node metastases from MT- 100-TC mammary carcinomas growing in rat foot-pads, over periods of up to 4 weeks. In addition, intravasation and exit of cancer cells from nodes in the bloodvessels surrounding them was assessed, together with the development of lung metastases. MATERIAL ANDMETHODS
Animals Wistar-Furth female rats, weighing 150 to 175 g (HarlanSprague-Dawley, Anderson, IN) were used throughout. Cancer cells The MT-100-TC rat mammary carcinoma (Ghosh et al., 1983) was maintained by S.C. passage. Single-cell suspensions were prepared from tumor fragments by pronase/DNase digestion (Ward and Weiss, 1989a). Foot-web injections (FWI) Ether-anesthetized rats were injected through 25-gauge needles, into the right foot-web between the second and third metatarsals, with 5 X lo5 cancer cells suspended in 0.1 ml phosphate-buffered saline (PBS). Back-flow from the injection site was prevented by application of light pressure as the needle was withdrawn. Determination of blood flow Blood flow to lymph nodes, at different times after FWI, was expressed as normalized gamma counts (cpm), made on lymph nodes removed within 2 min after left ventricular injec-
tions of ‘251-labelled,carbonized plastic tracer microspheres (15 2 3 bm diameter: 3M, St. Paul, MN). Rats were anesthetized with intraperitoneal injections of nembutal (40 mg/kg). The intercostal location of the left ventricular pulse was determined by palpation; lo5 microspheres (=250,000 cpm) suspended in 0.2 ml PBS plus 1% fetal bovine serum, were injected into the left ventricle through the intercostal space, through a 25-gauge needle, attached to a 1-ml syringe by a 10-cm length of 0.38 mm (internal diameter) polyethylene tubing. The position of the needle in the left ventricular cavity was confirmed by the appearance of bright red (oxygenated) blood in the tubing, before and after injection. Two minutes after injection, the popliteal lymph nodes and lungs were removed and placed in 70% ethanol, then 1min gamma counts were obtained. Lymph-node histology After gamma counting, the lymph nodes were weighed, formalin-fixed, wax-embedded, cut into 5-bm sections and stained with hematoxylin and eosin. The extent of nodal involvement was recorded at 1 , 2, 3 and 4 weeks after FWI. The numbers of blood vessels containing MT-100-TC cells were counted. The vessels were small veins (mean diameters 23.3 2 3.2 bm; median 17.8 bm, range 9.5 to 143 km) identified on the basis of thin basement membranes and incomplete or no smooth muscle, endothelium and occasional intravascular erythrocytes and leukocytes. The observed vessels were either within the hilar regions of lymph nodes or in peripheral to large nodal metastases. Individual MT-100-TC cells were identified on the basis of their large nuclei (12 -C 3 bm diameter) and distinct nucleoli. Although small clumps and single cancer cells may be difficult or impossible to identify within solid tissues in routinely stained sections, they were comparatively easily recognized lying within small blood-vessels. Lymph-node vascular densities Sections of popliteal lymph nodes taken 1, 2 , 3 and 4 weeks after FWI were examined at X400 magnification, using a grid eyepiece of 0.0625-mm2 projected area to make counts of blood vessels within lymph nodes. “Primary” foot-pad tumors In these and other experiments (Ward and Weiss, 1989a), 4 weeks after FWI, 14/14 animals had pulmonary metastases. Using the same criteria as described above, counts were made of the proportions of tumor-associated blood vessels containing cancer cells, in sections from 8 of these “primary” foot-pad lesions. Lung histology Sections were prepared from lungs to detect the possible presence of micrometastases, in the absence of overt lesions. RESULTS
Bloodflow to lymph nodes This is expressed as normalized gamma counts on lymph
Received: February 20, 1990 and in revised form April 24, 1990.
453
LYMPHOGENOUS A N D HEMATOGENOUS METASTASIS
nodes, following left ventricular injections of radio-labelled microspheres. The results summarized in Table I show an increased blood flow to the ipsilateral popliteal nodes (i.e., increased microsphere retention) over the 1- to 4-week period after FWI.Over this period, no significant counts above background (105 cpm) were obtained in the contralateral popliteal nodes. Lymph-node histology Popliteal lymph nodes removed from animals 1 to 4 weeks after FWI were examined histologically, and the results are summarized in Table 11. Cancer cells were first observed in the ipsilateral popliteal nodes 2 weeks after FWI.After 3 weeks, although all ipsilateral nodes were largely replaced by cancer cells, the node capsules were generally intact. Infiltration of cancer cells had occurred through the hilar regions, into the perinodal fat and along the fascia1 planes of muscle. At 4 weeks, normal nodal tissues were entirely replaced by tumor, with mean cortical width of 0.5 ? 0.07 cm (n = l l ) , containing morphologically intact cancer cells. There were central, amorphous necrotic cores, measuring 1.53 2 0.01 cm (n = 9) in diameter. In the cortices, the cancer cells were arranged in loose strands, interspersed with extravasated erythrocytes. In most cases, the metastatic tumors had extended into the adjacent fat and muscle, to completely surround the blood vessels in these sites. The contralateral popliteal nodes were seen to contain scattered cancer cells 3 weeks after FWI in 1/12 animals, and by 4 weeks, 319 animals had their contralateral popliteal nodes partially replaced by cancer cells. Intravascular small clumps of cancer cells were detected only in animals 3 weeks after FWI in 3/316 identified vessels (Table 11). In these cases, the blood vessels were observed lying within the fatty tissues. Lymph-node weights As summarized in Table 111, there is a consistent, non-linear increase in the weights of the ipsilateral nodes in contrast to the contralateral nodes, over the 4-week period of observation. Approximately 20% of the volume of the 4-week ipsilateral nodes is due to central, necrotic material. After 3 and 4 weeks, paired t-tests reveal that the ipsilateral nodes weigh significantly more than the contralateral nodes (p = 0.05 and GO.001 respectively).
TABLE 11- LYMPH-NODE [NVOLVEMENTAND INCIDENCE OF INTRAVASCULAR CANCER CELLS AT SPECIFIC TIMES AlTER FOOT-WEB INJECTIONS OF MT-100-TC CARCINOMA CELLS Weeks after foot-web injections
1 2 3 4
Intravascula? cancer cells
Nodal involvement’ Ipsilateral
Contralateral
(incidence)‘
- (0112) (17117) (16/16) (12/12)
++ +++
++++
bsilateral Contralateral -
- (0/7) - (0/10) (1/12) (3/9)
+
++
0/196 0/171 3/316 01161
0/134 0/85 0/122 0/215
+
+
++ +
TABLE 111 - POPLITEAL SODE-WEIGHTS AT SPECIFIC TlMhS AFTER FOOT-WEB INJECTIONS OF MT-100-TC CARCINOMA CELLS Popliteal node weights Weeks after gm ? SE (n) foot-web injection bsilateral Contralateral
Controls 1 2 3 4
0.006 0.013 0.019 0.09 4.66
k
0.002 (4)
+. 0.001 (7)
*k 0.005 (6) 0.04 (9)
k
0.5
(4)
0.004 0.01 0.01 0.006 0.01
f 0.001 (4) f 0.002 (7)
*f 0.003 (6) 0.001 (9) f 0.002 (4)
TABLE IV - VASCULAR DENSITIES IN IPSILATERAL AND CONTRALATERAL POPLITEAL LYMPH NODES/TUMORS AT SPECIFIC TIMES AFTER FOOT-WEIB INJECTIONS OF MT-100-TC CARCINOMA CELLS Weeks after foot-web injection
Controls 1 2 3
4
Vascular densities’ -C SE (n)’
t-tests Ipsilateral
Ipsilateral
*
0.89 0.08 0.45 0.05 (144;5) 0.64 k 0.07 ( 183;6) 1.04 2 0.09 (187;16) 1.34 0.18 (205:12)
* *
vs.
Contralateral
contralateral p -
(223;15) 0.60 2 0.09 (97;4) 0.61 2 0.07 (142;6) 1.15 k 0.10 (164; 12) 1.02 2 0.11 (168: 10)
-
0.117 0.765 0.423 0.142 ~
Lymph-node vascular densities The vessel counts per unit areas of the lymph nodes at different times after FWI are summarized in Table IV. One week after FWI the mean vascular densities in the ipsilateral nodes showed a statistically significant decrease over those in the untreated controls @ < 0.0001). The mean densities showed statistically significant increases between 1 and 2 weeks (p = 0.036) and 2 and 3 weeks @ = O.OOOS), but not between 3 and 4 weeks (p = 0.15). The contralateral nodes TABLE I - BLOOD-FLOW TO LYMPH NODES’ Weeks after foot-web iniection
Ipsilateral popliteal node gamma counts corn 2 SE (n)*
Controls3 1 2
approx. bkgd4 (4) 216 67 (7) 233 f 94 (6) 1112 2 304 (9)
3 4
*
3832 2 1971 (4)
‘Gamma counts in the ipsilateral popliteal nodes, made 2 min after left ventricular injection of radiolabelled microspheres (approx. 250,000 cpm), in tumor-bearing animals at different times after foot-web injections.-’Minus background (105 cpm). -31.e., non-tumor-bearing animal~.-~cpmnot significantly above background.
-
‘Extent of nodal involvement: - no cancer cells present; scattered cancer cells present in node; + + normal nodal tissues partially replaced by cancer cells; + + normal nodal tissues largely replaced by cancer cells; + normal nodal tissues completely replaced by cancer cells.- Incidence of involved n~des.-~Incidenceof intravascular clumps of cancer cells in peripheral and hilar blood vessels.
~~
‘Mean number blood vessels per 0.0625 mm2.-*n = (total number of fields examined; total number of nodes examined).
also showed a decreased vascular density compared with controls @ = 0.03), after 1 week; no change was observed between 1 and 2 weeks @ = 0.93); there was a significant increase between 2 and 3 weeks (p < O.OOOl), and no significant change between 3 and 4 weeks (p = 0.38). As shown in Table IV, no statistically significant differences were observed between the mean vessel densities in the tiniepaired ipsi- and contra-lateral nodes. “Primary” foot-pad tumors In sections of foot-pad tumors from 8 animals bearing pulmonary metastases, 82 of 3 14 identifiable blood-vessels contained cancer cells. Lung histology Pulmonary micrometastases were detected in 4/4 rats only after 4 weeks following FWI. DISCUSSION
Initially, cancer cells enter the bloodstream and lymphatic
454
WEISS AND WARD
system by independent processes. Subsequently, cells in the lymphatic system may enter the bloodstream mainly via the thoracic, subclavian and right lymphatic ducts, which terminate in the venous system at the junctions of the internal jugular and subclavian veins (Haagensen, 1972). Conversely, cells in the bloodstream may seed lymph nodes via the nodal arteries. Although various attempts have been made to demonstrate lymphatico-venous communications, the demonstration of lymph-to-blood passage of cancer cells has been much less convincing than blood-to-lymph passage (Weiss, 1980). Metastases may themselves generate further metastases (Weiss, 1985a), which disseminate via the blood-stream, the lymphatic system and other routes (Weiss, 19856). In transplanted tumors, which in some ways are analogous to metastases with respect to neovascularization, cancer cells enter blood-vessels in the tumor periphery (Warren et al., 1978). In S.C. T241 tumors in mice, a tumor neovasculature appeared after 4 days, and cancer cells were detected in vessel perfusates 1 day later; a linear relationship was observed between the densities of perfused vessels and the concentrations of effluent cells (Liotta et al., 1974). The extent of cancer-cell release into the bloodstream may be considerable. Thus, MTW9 mammary carcinomas growing in subcutaneous pouches in the rat released more than lo6 cancer cells per gram of tumor per 24 hr (Butler and Gullino, 1975); similar results were obtained with 3LL and B16 melanoma in mice (Glaves, 1983). The specific question addressed here is whether or not vascularized metastases in lymph nodes also release sufficient numbers of cancer cells into the bloodstream to affect the progression of hematogenous metastasis. Jirtle (1981) estimated the blood flow to lymphatic metastases from the MT mammary carcinoma by arrest of 25-pm diameter, radioactive microspheres injected into the left ventricle of rats. In this and the MTW-9B rat mammary carcinoma, relative blood flow was reported as inversely proportional to tumor weight. In the present studies, which were made with a different tumor and smaller (15 pm) microspheres, the results are different in some respects from those reported by Jirtle. Thus, we also observed a progressive increase in lymph-node blood flow (Table I) associated with increasing total lymph-node weight (Table HI). However, the relationship between blood flow and tumor weight was not linear; when the logarithm of tumor weight was regressed against blood flow (cpm), a highly significant correlation coefficient was obtained (r = 0.99; p = 0.003), indicating a complex relationship between tumor volume and vascularity. In spite of the complexity of the relationship, increases in blood flow to the lymph-node metastases over time were associated with increased nodal weight which, to a major extent, reflected increased tumor burden. The increased blood flow did not appear to depend to a major extent on increased vascular density because, on the one hand, no statistically significant increases in vascular density were observed between 3 and 4 weeks (Table IV) although, during this time, there was an approximately 3-fold increase in blood flow (Table I). On the other hand, the approximate 2-fold increase in vascular density occumng between 2 and 4 weeks corresponded to an approximately 16-fold increase in blood flow. As the lymph-node metastases increased in size, many more blood vessels were present. Based on the modest increases in blood-vessel density, the probability of an individual cancer cell coming into contact with a vessel closely associated with a tumor is expected to be increased by only a small amount; however, in terms of the whole cancer, more cancer cells are expected to come into contact with the tumor-associated vasculature as a whole, because more vessels and cancer cells are present.
If cancer cells in fact enter the bloodstream directly and frequently from lymph-node metastases, it might reasonably be expected that they would have been detected in the hilar and other lymph-node-associated vessels. However, it is extremely difficult to detect invasion of small vessels in tumor sections stained with hematoxylin and eosin (Martin et al., 1987). We have attempted to avoid this problem by confining our search for intravasated cancer cells to the easily defined hilar vessels of the lymph nodes and the peripheral vessels of advanced nodal metastases, since cancer cells entering less well defined vessels within the tumor itself must ultimately disseminate via these vessels. However, between 2 and 4 weeks, when large numbers of cancer cells were present in the ipsilateral nodes, intravascular cancer cells were observed in only 3 of a total of 648 vessels examined (Table 11); in these cases, observed 3 weeks after FWI, small clumps of cancer cells were observed lying within 3 vessels associated with fat, to which the “fixed” nodes were adherent and which cancer cells were invading. Using the same criteria for blood-vessel and cancer-cell identification, significantly more (82 of 314; p = 0.001) blood vessels associated with metastasizing, “primary” footpad lesions contained cancer cells than did similar vessels associated with involved lymph nodes. However, the presence of intravascular cancer cells is not necessarily predictive of hematogenous metastasis (Weiss, 1990). Bioassays previously made on right ventricular blood from rats bearing MT- 100-TC tumors in their foot-pads revealed tumorigenic cancer cells in 4/8 host animals after 3 and 4 weeks, and lung metastases were observed in 10/10 animals after 4 weeks. However, the lung metastases were of the diffuse type traditionally associated with lymphogenous metastasis, as distinct from the discrete nodular lesions associated with hematogenous metastasis (Ward and Weiss, 1989a). In the present study, pulmonary micrometastases were only detected after 4 weeks (4/4 animals) by microscopy. In contrast to this limited blood-borne dissemination, extensive lymphatic metastasis, extending up to the axillary nodes, was observed after 4 weeks. These observations are in accord with those previously described in detail elsewhere (Ward and Weiss, 19896). On the basis of the present experiments, the sequence of metastasis-related events occurring in the popliteal lymph nodes may be reconstructed. Tumors seeded in the nodes by lymphogenous metastasis from foot-pad tumors grow and become progressively more vascularized. However, in spite of dramatic increases in blood-flow to the involved lymph nodes, cancer cells are not detected within associated blood vessels, in “mobile” nodes which are not adherent to adjacent structures. Therefore, although in other experiments (Ward and Weiss, 1 9 8 9 ~ viable ) cancer cells were recovered from the venous blood draining foot-pad tumors, these cells must have entered the bloodstream via veins draining the “primary” tumors, as distinct from those draining the lymph-node metastases. The vast majority of cancer cells entering the bloodstream from many types of cancers are killed in the microvasculature by a variety of fast and slow mechanisms (Weiss et al., 1989), which are major contributors to “metastatic inefficiency” (Weiss, 1990). However, passage through the lymphatic system is expected to be much less traumatic (Weiss, 1980). This differential trauma may well account for minimal (micrometastatic) involvement of the lungs in the presence of massive lymph-node involvement. The results are also in accord with the view that, until the degree of lymph-node involvement progresses to lymph-node “fixation”, when cancer cells invade extranodal tissues and intravasate at these sites, hematogenous and lymphogenous metastasis remain independent processes, proceeding at different rates. Even when distant lymph-node metastasis occurs, and can-
LYMPHOGENOUS AND HEMATOGENOUS METASTASIS
cer cells are released into the major ducts terminating in the subclavian veins, most of the intravenous cells will be killed in the pulmonary microvasculature (Weiss, 1990). Although some might consider that a transplantable mammary carcinoma growing in the hind foot-pad of a rat is a less than ideal model for human breast cancer, our results nonetheless provide a possible explanation for some limited aspects of human disease. Thus, in a large series of breast cancer patients with regional node involvement, following surgery, 24% survived for 20 years (Axtell et al., 1976). These patients survived presumably because, at the time of operation, metastases were confined to the regional lymph nodes and, although cancer cells were probably entering the bloodstream, they did not
455
form hematogenous metastases. Analogous results were obtained in the rat model system following surgery (Ward and Weiss, 1989b). Fixation of involved lymph nodes adversely affects prognosis in humans and, as in the rats used in the present experiments, probably indicates an increased risk of hematogenous metastasis. ACKNOWLEDGEMENTS
We thank Dr. J.P. Harlos for help in statistical analyses, and Ms. G. Elkin and Mr. D. Graham for their technical assistance This work was partially supported by a grant to L. Weiss from the NCI, DHHS.
REFEREINCES
AXTELL,L.M., ASIRE,A.J. and MYERS,M.H., Cancer patient survival. Report 5 , p. 16, US Dept. Health Education and Welfare (DHEW h b l . No. (NIH) 77-992), Bethesda, MD (1976). BUTLER,T.P. and GULLINO,P.M., Quantitation of cell shedding into efferent blood of mammary adenocarcinoma. Cancer Res., 35, 512-516 (1975). GHOSH,S., ROHOLT,O.A. and KIM, U., Establishment of two nonmetastasizing and one metastasizing rat mammary carcinoma cell lines. In Vitro, 19, 919-928 (1983). GLAVES, D., Correlation between circulating cancer cells and incidence of metastases. Brit. J . Cancer, 48, 665-673 (1983). HAAGENSEN, C.D., General anatomy of the lymphatic system. In: C.D. Haagensen, C.R. Feind, F.P. Herter, C.C. Slanetz, and J.A. Weinberg, (eds.), The lymphatics in cancer, pp. 2 2 4 1 , Saunders, Philadelphia ( 1972). JIRTLE,R.L., Blood flow to lymphatic metastases in conscious rats. Europ. J. Cancer. 17, 53-60 (1981). LIOTTA,L.A., KLEINERMAN, J. and SAIDEL,G.M., Quantitative relationships of intravascular tumor cells, tumor vessels, and pulmonary metastases following tumor implantation. Cancer Res., 34,997-1004 (1974). MARTIN,S.A., PEREZ-REYES, N. and MENDELSOHN, G., Angioinvasion in breast carcinoma. Cancer, 59, 1918-1922 (1987).
WARD,P.M. and WEISS,L., The relationship between lymphogenous and hematogenous metastasis in rats bearing the MT-100-TC mammary carcinoma. Clin. exp. Metasi., 7, 253-264 (19890). WARD,P.M. and WEISS,L., Metachronous seeding of lymph node metastases in rats bearing the MT-100-TC mammary carcinoma: the effect of elective lymph-node dissection. Breast Cancer Res. Treat., 14, 315-3210 (19896). WARREN, B.A., SHUBIK,P. and FELDMAN, R., Metastasis via the blood stream: the method of intravasation of tumor cells in a transplantable melanoma of the hamster. Cancer Lett., 4, 245-251 (1978). WEISS,L., The pathophysiology of metastasis within the lymphatic system. In: L. Weiss, H.A. Gilbert and S.C. Ballon (eds.), Lymphatic system metastasis, pp. 2-40, G.K. Hall, Boston, MA (1980). WEISS,L., The metastasis of metastases. In: Principles ofmetastasis, pp. T (198%). 201-207, Academic Press, Orlando, l WEISS,L., Metastatic pathways. In: Principles of Metastasis, pp. 181194, Academic Press, Orlando, FL (19856). WEISS,L., Metastatic inefficiency. Advanc. Cancer Res., 54, 159-211 (1990). WEISS,L., ORR,F.W. and HONN,K.V., Interactions between cancercells and the microvasculature: a rate-regulator for metastasis. Clin. exp. Metast., 7, 127-167 (1989).
