Experimental oncologists have known for years that some neoplasms may regress in vivo. This regression may be seen readily in tumors transplanted into new hosts. The host recognizes the tumor as foreign tissue and thus tends to reject it immunologically. In the human, examples of tumor regression are quite rare and many such examples may be explained on the basis of host immunity to the neoplasm, although in some instances such a mechanism may not explain the peculiar behavior of the neoplasm (O’Regan and Hirshberg, 1993). This is particularly true when certain types of neoplasms appear to differentiate into adult benign tissues after initially having demonstrated their malignant potential both morphologically and biologically (cf. Bolande, 1985). The mechanism for this phenomenon has not yet been explained, but it is un- likely that it is mutational. One of the earliest examples of a “controlled” reversion of differenti- ation of a neoplasm to normal tissue is seen in the plant. The crown gall tumors briefly discussed in Chapter 4 and several other plant neoplasms (including a teratoma) can, under controlled con- ditions, differentiate and produce a normal plant (Braun, 1972; Sacristán and Melchers, 1977). However, studies by Braun and Wood (1976) demonstrated that the differentiation of neoplastic plant cells to a normal phenotype may occur without the loss of the Ti-plasmid (Chapter 4), so that, under appropriate conditions such as those seen in cell culture, these cells will again ex- press their neoplastic phenotype. Yang et al. (1980) have also shown that the generation of nor- mal-appearing plants from a number of crown gall neoplasms is accompanied by the loss of Ti- plasmid sequences in the plant DNA; this indicates the necessity of the plasmid for the mainte- nance of the transformed state in vivo. Thus, the mechanism of the reversion of plant neoplasms to a normal phenotype may reflect changes in the genetic apparatus of the cell, including loss of oncogenic sequences (Yang et al., 1980) or changes in methylation of DNA, both mechanisms resulting in changes in the cellular production of auxins or plant hormones. Neoplastic plant cells are capable of producing their own supply of auxins, while nonneoplastic cells require ex- ogenous sources of these regulatory substances (cf. Syôno and Fujita, 1994).
There have been a number of examples of regression resulting from differentiation of neo- plastic cells of both benign and malignant neoplasms in experimental animals. Seilern-Aspang and Kratochwil (1962) demonstrated that carcinogen-induced epidermoid carcinoma in the newt in many instances will differentiate to normal-appearing cells, which ultimately disappear from the organism. Cooper and Pinkus (1977) have demonstrated that the intrauterine transplantation of basal cell carcinomas in the rat causes such cells to differentiate into keratinizing foci and cysts, with eventual complete loss of the malignant phenotype. Pierce (1974) was among the first to emphasize that both benign and malignant neoplasms commonly exhibited the capability for differentiation even to the point of terminal cell types incapable of further replication. As described in Chapter 5, Stewart and Mintz (1981), Papaioannou et al. (1975), Rossant and McBurney (1983), and others have demonstrated the differentiative capabilities of teratocarci- noma cells by their inoculation into developing blastocysts of genetically distinct animals, with subsequent development of chimeras composed of cells of both teratocarcinoma and host lin- eage, all of which exhibit a normal phenotype. Even in the adult, neoplastic cells may colonize their tissue of origin and differentiate morphologically into their normal adult counterpart, e.g., rat hepatocytes (Coleman et al., 1993). The induction of differentiation of neoplasms by chemi- cals is now a well-known phenomenon, usually occurring most readily in vitro (Freshney, 1985; Chapter 14). Such chemical induction of differentiation has been studied most extensively in hematopoietic and lymphoid tissues both in humans and experimental animals (Sachs, 1993). The potential for therapeutic intervention in neoplasia has even been realized in a few diseases, such as acute promyelocytic leukemia (Chapter 6; Breitman et al., 1994).
In the human, neuroblastoma of childhood, which in most cases is highly malignant and derived from neuroblasts of the adrenal medulla, may in a small percentage (about 7%) of all cases spontaneously regress either by differentiation to a benign ganglioneuroma or by disap- pearance of the neuroblastoma cell (Pahlman and Grotte, 1982). Evans (1982) has pointed out that it is mainly infants with an early stage of the disease who survive. In particular, almost 90% of patients exhibiting the clinical picture of a stage IV-S (De Bernardi et al., 1992) neuro- blastoma survive free of their neoplasm with little or no therapeutic intervention. Spontaneous differentiation of neuroblastomas both in vivo and in vitro has been described both with and without treatment (Evans et al., 1976). A histologically related neoplasm, retinoblastoma, may also undergo spontaneous regression even in patients with the genotype for this disease (Gal- lie et al., 1982). Bolande (1985) has pointed out the potential significance of the regression, especially by the process of differentiation, of several types of malignant neoplasms in early life in the human, suggesting that specific repressive influences or factors of the internal envi- ronment may exert effects on neoplasms similar to those seen in the teratocarcinoma chimeras (see above).
A number of other rather common neoplasms in the adult human have been shown to re- gress spontaneously in rare instances (cf. Lewison, 1976). In the human, many primary lesions of malignant melanoma regress, reportedly as high as almost 60% (McGovern et al., 1983). However, the prognosis in patients with regressing primaries is usually less favorable than in those where no regression of the primary lesion is seen. Prehn (1996) has suggested that the reason for this is that the regression is the result of a relatively strong immune reaction, but that surviving neoplastic cells are more aggressive. However, Kelly et al. (1985) found no difference in prognosis between melanomas exhibiting regression and those that did not show such an ef- fect. Interestingly, in miniature swine, regression of spontaneous melanoma occurs with nearly a 100% incidence (Oxenhandler et al., 1982). An interesting phenomenon has been described for renal cell carcinomas, in that removal of the primary tumor may lead to disappearance of pulmo- nary metastases (Markewitz et al., 1967; Fujita et al., 1988; Vogelzang et al., 1992). In addition, several isolated cases of spontaneous regression of hepatocellular carcinomas have been re- ported (cf. Kaczynski et al., 1998). In the special case of the chorionepithelioma, a malignant neoplasm of the placenta, spontaneous regression may be accounted for by host immunity, as the tumor tissue is actually foreign to the maternal immune system. In the male, a morphologically similar neoplasm that may arise from host tissues does not regress spontaneously.
The reader should keep in mind that spontaneous regressions of neoplasms in the human are exceedingly rare. However, the spontaneous regression of neoplasia in the human is now a well-documented albeit very sporadic phenomenon (Challis and Stam, 1990). A variety of theo- retical mechanisms have been proposed to explain such occurrences (Stoll, 1992; Noda, 1993), including alterations in signal transduction, cell adhesion, and immunological, hormonal, and even “psychoneuroimmunological” mechanisms. With a few exceptions of demonstrable en- hancement of immunological responses to specific neoplasms and the direct effect of chemicals on specific genes to induce differentiation, no satisfactory explanation has been forthcoming in cases where the diagnosis is clear. In the human, there have been at least two examples in which regression of neoplasms is a demonstrable component of their natural history. These include basal cell carcinoma (Curson and Weedon, 1979; Hunt et al., 1994) and intraepithelial neoplasia (carcinoma in situ) of the uterine cervix (Hanselaar et al., 1998). Differentiation of neoplastic cells, both in the human and in lower animals, has been described repeatedly, although in the vast majority of cases only a portion of the total neoplastic cell population differentiates to the terminal state. In some cases, differentiation, especially in lymphoid neoplasms, may result in a more malignant neoplasm (cf. Lynch, 1995). As discussed earlier (Chapter 7), preneoplastic le- sions in the absence of promoting agents naturally regress, primarily through the mechanism of apoptosis. Regression of cells in the stage of progression, however, appears to use more varied mechanisms. The natural history of neoplastic development must take into account the potential for regression or differentiation of established neoplasms. Such considerations may alter our ear- lier considerations (Chapter 2) of the arbitrary nature of distinguishing benign from malignant neoplasms when viewed as dynamic processes traversing the stages of promotion and progres- sion in neoplastic development. At least theoretically, some benign neoplasms may, in essence, be differentiated cell populations (such as the ganglioneuroma resulting from differentiation of the neuroblastoma) and thus will never develop beyond that stage. On the other hand, some le- sions considered benign neoplasms may be cells in the stage of promotion, reflecting preneo- plastic lesions (e.g., hepatic adenomas in patients receiving synthetic estrogens; Chapter 7). Unfortunately, it is not possible by means of histopathology to distinguish readily between be- nign neoplasms with the potential for progression to malignancy and those resulting from the differentiation of neoplastic cells. Further studies at the molecular level may allow a more defin- itive distinction.