Metastasis: Incidence and Mechanism
As noted earlier, a metastatic tumor is one that originates in and is physically separated from the primary neoplastic growth. The capacity of malignant tumors to metastasize is one of the major features of their lethality to the host.
Routes of Metastasis
Anatomically, a number of pathways for metastatic cells are possible. The most obvious is the blood circulation, in which neoplastic cells gain entrance to the vascular system by invasive and related processes and then are carried by the bloodstream to new sites, where they initiate growth. Usually this new growth begins in small capillaries, where neoplastic cells are caught and begin to invade through the capillary endothelium to initiate a new growth. It is obvious from a variety of studies that the number of cells that enter the bloodstream is far greater than the number that ever give rise to metastatic lesions. The phenomenon of “canceremia” or neoplastic cells in the blood has been studied in the past through the use of blood filtrates in which the number of neoplastic cells per milliliter of blood may be estimated. In advanced cases of neopla- sia this number is quite high, whereas in early tumor growth there may be virtually no tumor cells within the blood vascular system. Recently, sensitive systems have been able to detect as few as one epithelial tumor cell per milliliter of whole blood (Denis et al., 1997; Komeda et al.,1995). In this study, the presence of neoplastic cells in the blood was generally correlated with the presence of metastatic lesions, but not entirely. With the polymerase chain reaction (PCR), patients exhibiting the presence of metastases showed a significantly higher level of the hepato- cyte-specific alpha-fetoprotein mRNA in blood. Furthermore, from experimental studies, it would appear that only when a neoplasm reaches a certain critical size does it begin to shed cells into the bloodstream. Of the cells entering the bloodstream, considerably fewer than 0.01% ever give rise to any metastatic lesions (cf. Sellwood et al., 1969). Classical clinical observations have indicated that carcinomas metastasize more frequently via the lymphatic system, whereas sarco- mas spread more commonly through the blood vascular system. However, the presence of nu- merous anastomoses between venous and lymphatic vessels tends to invalidate this concept (Fisher and Fisher, 1966).
Another common route of metastasis is the lymphatic system, in which the flow of neo- plastic cells, although considerably slower than in the bloodstream, is probably of a similar mag- nitude. Some neoplasms may implant in other sites by mere physical movement from one site to another. This is commonly seen in certain ovarian or gastrointestinal neoplasms, with resultant implantation of neoplastic cells from one side of the peritoneal cavity to the other. A similar sort of implantation is probably the mechanism responsible for the appearance of accessory spleens or splenosis—i.e., intraabdominal splenic tissue separate from the spleen itself (Carr and Turk,1992).
With the advent of numerous surgical techniques, a new pathway of metastasis has be- come apparent. If the surgeon is not careful, the operative knife may enter the neoplasm and become covered with living cells, which may be carried to and implanted at another site in the surgical field. This phenomenon has accounted for the reappearance of tumors in the operative site some months or years after the initial surgery. In addition, manipulation of the neoplasm during surgery may initiate both vascular and lymphatic metastatic lesions. However, now that these problems are apparent, most surgeons take extreme care to prevent the occurrence of such metastasis during curative resections by isolating the neoplastic lesion prior to extirpation and by thorough washing of the wound site.
In experimental oncology, the analog to surgical metastasis is the tumor transplant. Clearly, in the original behavioristic classification of neoplasms, the distinguishing feature of the metastatic capability of the malignant neoplasm is reflected in its transplantability. In some in- stances, however, it has been possible to transplant neoplasms that have not been shown to be behavioristically malignant in the original primary host. This further points out the artificiality of the behavioristic classification to the experimental oncologist.
Incidences and Sites of Metastases of Neoplasms
As stated previously, the number of metastatic cells that give rise to metastatic lesions is ex- tremely small. Metastases usually appear in the adjacent lymph nodes or in the lung or liver, the site depending on where the tumor originates and its histogenetic origin. Certain neoplasms may have a predisposition, because of their anatomical location, to metastasize to certain organs. Cancer of the lung not uncommonly metastasizes to the brain, possibly because of neoplastic invasion into the pulmonary veins, from which metastatic cells may enter the carotid arteries and pass to the brain. Carcinoma of the breast regularly metastasizes to the adrenal glands, and carcinoma of the prostate frequently metastasizes to bones, particularly the vertebrae. Relatively few metastatic lesions occur in some organs—for example, the spleen and thymus. In the case of the thymus, this is probably because of its peculiar vascular system. Neoplastic cells may have difficulty in establishing metastatic lesions in the spleen because of its unique anatomical struc- ture and its function in eliminating senescing or otherwise damaged cells (Bishop and Lansing,1982). Figure 10.9 is a diagram of the organ distribution pattern of metastases from a variety of cancers.
Although the anatomical location of the organ in which metastases colonize can in many instances be explained by vascular connections of the primary tumor and adjacent lymphoid structures, in many circumstances the relative high incidence of metastasis of specific neoplasms to specific organs cannot be explained on this basis. Furthermore, it is possible, by using trans-planted highly malignant murine neoplasms, to select variant metastatic cells that have a high predilection for growth in specific organs (Nicolson, 1978; Fidler et al., 1978). For example, clones of cells may be obtained from the parent neoplasm by repeated selection in vivo. These clones have a propensity for growth in the liver, the lung, or the brain. In recent years the mech- anism for such organ selectivity has been studied rather intensively with respect to metastasis occurring through the vascular system. The critical tissue in organ-specific metastasis is the en- dothelium that lines the interior of blood/lymph vessels and the heart. The vascular endothelium provides a protective and antithrombogenic surface as well as acting as a vehicle for transport of various substances from or into the bloodstream. In this process the endothelium performs dis- tinctive biological functions at different vascular sites and in individual organs (Fishman, 1982). Such functions appear to involve specific structural and chemical microdomains of the endothe- lial cell surface (cf. Pauli et al., 1990). Some of these domains may be expressed in an organo- typic and vessel caliber–dependent fashion, whereas others are transient and regulated in a positive or negative manner depending on the functional needs of the tissue (cf. Pauli et al.,1990). The biochemical diversity of normal endothelial cells has been verified at the molecular level, demonstrating clear differences in carbohydrate and other moieties exposed on the surface of the cell. Included in these molecular arrays are the familiar cell adhesion molecules, integrins, cadherins, immunoglobulin superfamily members, selectins, and CD44. However, since the met- astatic cell is floating freely within the bloodstream, lymph, or body cavities, the metastatic cell must find a suitable molecular interaction with an endothelial cell and subsequently utilize its invasive properties to extend through the vessel wall, through the basement membrane, and into the ECM in order to colonize the tissue. This process, which is quite analogous to that seen in Figure 10.8, is depicted in Figure 10.10, where the interaction is termed a “docking and locking” of the metastatic cell (Honn and Tang, 1992). Initial interaction of the neoplastic cell with the endothelium is likely to occur through members of the immunoglobulin superfamily, selectins and CD44, as well as interactions of these and other molecules with integrins. A variety of sur- face carbohydrate determinants on neoplastic cells as well as the endothelium appear to deter- mine the initial aspects of successful metastasis (cf. Dennis and Laferte, 1987; Kannagi, 1997; McEver, 1997). The initial recognition or “docking” step depicted in Figure 10.10 occurs be- tween the moving neoplastic cell and the underlying endothelial cell. This interaction is medi- ated by relatively weak and transient adhesive forces of carbohydrate-carbohydrate and/or carbohydrate-protein interactions involving primarily selectin recognition of carbohydrates and members of the immunoglobulin superfamily recognizing specific integrin receptors (Honn and Tang, 1992; Miyasaka, 1995). This process is very similar to “homing” of lymphocytes to spe- cific areas of the organism (Roos, 1993). However, as with the altered genetic expression so characteristic of the stage of progression, neoplastic cells often employ ectopic expression of adhesion molecules to facilitate their interaction with endothelium and ultimately during the process of invasion (Tang and Honn, 1995). An excellent example of this is noted in the altered expression of CD44, the hyaluoranate receptor. This surface molecule contains 20 exons, and different tissues express different splice forms of the molecule. At least 20 different CD44 iso- forms have been described based on alterations in the content of 10 variable exons occurring in the center of the molecule (Tuszynski et al., 1997). Malignant neoplasms express a wide variety of the splice variants, a number of which have been shown by transfection and other similar experiments to increase neoplastic cell growth and metastasis dramatically in vivo (Sleeman et al., 1995; Cooper and Dougherty, 1995).
In addition to the molecular species found in tissue-specific endothelial cells, certain tis- sues produce factors that stimulate the growth of metastatic neoplastic cells (cf. Nicolson, 1992). At times, neoplastic cells themselves may produce growth factors that can stimulate their own growth (autocrine growth). In other instances, experiments have demonstrated that both sponta-
Figure 10.9 The distribution pattern of some neoplasms in relation to organ colonized. As noted in the text, direct vascular connections of the primary neoplasm with the target organ cannot fully explain the high incidence of metastasis in some organs. (Adapted from Rusciano and Burger,1992, with permission of the authors and publisher.)
Figure 10.10 Artist’s conception of the docking and locking hypothesis of initial metastatic colonization. In the “docking” phase, neoplastic cells within the vascular system interact initially with the surface of endothelial cells through carbohydrate-carbohydrate and/or carbohydrate-protein recognition interactions. Molecules involved at this stage are cell surface glycoproteins, selectins, or im- munoglobulin cell adhesion molecules present on the neoplastic cells or the endothelial cell or both. Subsequently, platelet/coagulation interactions occur with the neoplastic cell leading to platelet aggregation, activation, and release reactions. At this time the neoplastic cell establishes stable bonds with the endothelial cell during the activation-dependent “locking” phase mediated largely by integrins and modulated by a host of mediators resulting from both activated platelets as well as the neoplastic cells. Thereafter, the metastatic cell induces retraction of the endothelial cell with subsequent release of proteases and extension into the extracellular matrix. EC, endothelial cell; CAM, cell adhesion molecules; SEM, subendothelial matrix. (Adapted from Honn and Tang, 1992, with permission of the authors and publisher.)
neous and transplanted metastatic neoplasms that are seen to metastasize frequently to specific tissues are also found to have their growth stimulated in vitro by extracts from those tissues to which they readily metastasize but not to other tissues (cf. Rusciano and Burger, 1992). An ex- cellent example of tissue-specific growth enhancement of metastases is seen when prostatic ade- nocarcinoma metastasizes to the bone marrow, producing characteristic bone-forming (osteoblastic) lesions. The growth rate of these metastases is significantly more rapid than that of primary prostatic neoplasms. This effect appears to be due to the production in the bone marrow of a growth factor that is able to enhance specifically the growth of prostatic cancer cells (Chackal-Roy et al., 1989).
In most experimentally produced neoplasms, the animal is killed prior to the occurrence of metastatic lesions. Therefore it is not uncommon to find an animal harboring a histologically malignant neoplasm with no evidence of metastatic lesions. Were the animal to be allowed to die of the cancer, the incidence of metastatic lesions would be much higher in such experimental situations. Similarly, in the human, malignant neoplasms of the central nervous system rarely metastasize, since the lethal changes occurring in the confined space of the skull kill the host before the neoplasm has had an opportunity to express its metastatic capabilities (Alvord, 1976).
In the past it was felt that lymph nodes presented a barrier to the spread of metastatic cells (Cady, 1984). Several experimental findings have seriously questioned this concept and indicate that many neoplastic cells initially retained in lymph nodes maintain only a temporary residence there, and the nature of the neoplastic cells may be at least as great a determinant of their resi- dence in lymph nodes as are the biological and mechanical properties of such structures (Fisher and Fisher, 1967; Kohno et al., 1979). Substantial evidence has accumulated that trauma may increase the number of metastatic cells (if such trauma is directed at the primary neoplasm) and may increase the chance of successful metastatic growth at a site distant from the primary neo- plasm (Fisher et al., 1967).