In addition to explaining the basis for the majority of cases of hypercalcemia of malignancy, the presence of high levels of PTHRP in the plasma indicates the presence of a neoplasm in the host, although not specifically defining the histogenetic type of neoplasm (Figure 17.12). Thus, PTHRP is one of a large number of substances that are secreted, are present on or within the neoplastic cell, or may be produced by normal cells in response to the presence of or factors in or released by the neoplasm. This large group of such substances are termed tumor markers, since they indicate the presence or specific characteristics of one or more neoplasms. Generally speaking, tumor markers may be characterized at virtually all levels of genetic expression. This is seen diagrammatically in Figure 17.13. Markers at the genomic level include cytogenetic markers, loss of heterozygosity (LOH), and specific mutations. Messenger RNAs produced spe- cifically in neoplastic cells may be used as markers in tissue sections, while proteins within and secreted by neoplastic cells can be used as markers in tissue sections as well as in body fluids. Products of metabolism and/or specific enzymic function, such as polyamines, lactate, etc., may also be considered as tumor markers. As seen in Chapters 15 and 16, intracellular tumor markers are very useful in our understanding of the critical alterations that give rise to the neoplastic cell. However, a number of these changes have found usefulness in cytological and histological diag- noses, and materials secreted or released from neoplastic cells have been measured in blood, urine, cerebrospinal fluid, etc. It is the diagnostic, prognostic, and screening usefulness of tumor markers that are considered here.
Tumor Markers in Clinical Studies
Tumor markers useful in the diagnosis of specific neoplasms date back to the middle of the nine- teenth century with the description of Bence-Jones urinary proteins as markers for multiple my- eloma (cf. Virji et al., 1988). Since that time, a large number of markers have been developed
Figure 17.13 Tumor markers occur in or are produced from neoplastic cells at several levels of genetic expression. DNA-based markers may be the result of specific nucleotide alterations, sequence alterations, as well as cytogenetic changes. When transcribed, many such genomic alterations will be reflected both at the mRNA and protein level (transcription and translation). Markers at the protein and enzyme level may be detected within the cell on sample sections or within body fluids, having been released from normal or dying neoplastic cells. Low-molecular-weight products include abnormal nucleic acid bases, polyamines, amine hormones, and others. (Modified from Lehto and Pontén, 1989, with permission of the authors and publisher.)
both from experimental studies and studies directly in the human. However, only a relative few tumor markers in body fluids are generally useful for screening, diagnostic, therapeutic, and prognostic studies. Tissue markers may be more successful in the future, particularly with the advances of methodologies allowing for the identification of mutational, cytogenetic, mRNA, and protein markers within tissue sections.
The important criteria for the usefulness of a tumor marker are its sensitivity and speci- ficity with respect to identifying the neoplastic cell in which or from which the marker arose. A sensitive marker detects a high percentage of patients with the disease, while a specific marker is present only in cancer patients and not in those without cancer (Magdelénat, 1992). Table 17.9 shows an outline and definition of sensitivity and specificity as predictive values in neo- plasia. The values in Table 17.9 are linked to both sensitivity and specificity, but also to the prevalence (P) or incidence of the disease in the population studied. Knowing this latter param- eter and calculating sensitivity and specificity, one may obtain a “positive predictive value” as shown in the following equation:
Key: TP, true positive or patients having cancer; FP, false positive or patients expressing the marker but not having cancer; FN, false negative or patients not expressing the marker but having cancer; TN, true negative, patients not expressing the marker and not having cancer.
Adapted from Magdelénat, 1992, with permission of the author and publisher.
where S equals sensitivity, Sp equals specificity and P equals prevalence. The positive predictive value is extremely useful in screening large groups of patients for the purpose of monitoring the presence of neoplasia at an early time point when the disease is most treatable (cf. Nielsen and Lang, 1999). An example of using this information to determine the predictive value in screen- ing for ovarian cancer with a specific tumor marker, CA 125, a serum protein, may be seen in Table 17.10. As noted in the table, screening in the general population gives a very high predic- tive value far in excess of the actual prevalence. However, by screening only women with pelvic masses, the number of false positives can be dramatically reduced.
In order to screen large populations effectively for a neoplasm, a very high degree of spec- ificity (greater than 95%) and good sensitivity would be required considering the relatively low
Table 17.10 Results of Screening for Ovarian Cancer with the Tumor Marker CA 125 in Several Different Populations of Women
prevalence of the neoplasm in the general population, as noted above. In fact, a simple calcula- tion shows that in ideal conditions (1% incidence of the disease, 99% sensitivity and specificity), the frequency of false positives would be in the neighborhood of 50% (Magdelénat, 1992). Per- haps the best example where screening for a specific tumor marker has had a major impact in mortality reduction is the assay of human chorionic gonadotropin (HCG) in choriocarcinoma. The relatively high incidence of the neoplasm (5% to 10%) in women who have had an hydati- form mole, together with the excellent sensitivity and specificity of the assay and the high chemosensitivity of the neoplasm, has resulted in the combination of these factors leading to excellent detection and care of patients with this neoplasm.
One may also look at this problem in another way, as seen in Figure 17.14. In this in- stance, if one assumes a bell-shaped curve of the distribution of marker values for diseased and nondiseased individuals with a slight overlap as shown in the figure, one may calculate a “cutoff” value beyond which one considers marker values under the curve (3,4) as positive. However, in selecting such a value, one must take into account a number of other factors such as (1) the risk of complications associated with the screening test, (2) the dollar cost of the screen- ing test and subsequent cost for further testing, (3) the subject’s anxiety as well as anticipated medical and economic burdens, and (4) the social and public health impact of either missing positive cases or identifying truly nondiseased patients as being diseased (Makuch and Muenz,1987). These are questions involving the “risk/benefit” ratio, discussed in Chapter 13 but in a different context.
A major use of several tumor markers is in following the course of the disease after ther- apy. As long as the neoplasm is present in substantial amount in the organism, the marker will also, of course, be present in the tissue but also in secreted markers or those released by the neoplasm into body fluids. On removal of the neoplasm, if almost all or all of the neoplastic tissue has been surgically excised, the level of the marker will fall to near normal or normal levels. If there is regrowth of the primary or metastatic lesions of the neoplasm, in almost all instances the level of the marker itself will also increase and indicate to the clinician that signifi- cant neoplastic growth is occurring. A diagram of what might be expected in the circumstance of primary therapy with remission following recurrence in relation both to tumor burden and ex- pression of the tumor marker is seen in Figure 17.15. As noted from the figure, the appearance of the marker in most instances usually occurs after significant growth of the neoplasm has oc-
Figure 17.14 Potential outcomes in classifying patients into one group or another based on the true dis- tribution of tumor marker values in two distinct populations: (1) true negatives; (2) false negatives; (3) false positives; (4) true positives. (Adapted from Makuch and Muenz, 1987, with permission of the authors and publisher.)
Figure 17.15 Graphical relationships of marker expression in the serum and tumor growth or burden. Note that there is a direct parallel between the level of expression of the marker and the tumor burden, although in primary treatment, which many times is surgical, there is a very abrupt drop in tumor burden because of removal of the neoplasm, whereas secondary treatment may be chemotherapy and/or radiation leading to a somewhat slower loss of tumor burden. Regrowth of the neoplasm a second time usually causes demise of the patient. (Adapted from Magdelénat, 1992, with permission of the author and publisher.)
curred. Following primary therapy with excision and/or destruction of the vast majority of neo- plastic cells, the level of the marker decreases in accord with the loss of the tumor burden. Serum markers decrease dramatically following the removal of the neoplasm, usually with half-lives of less than 1 week (cf. Duffy, 1996; Takashi et al., 1989). If the neoplasm is not completely re- moved, some low level of the marker will remain in body fluids and then increase again as the residual neoplasm grows, as noted in the figure. In this way, the clinician may monitor the effec- tiveness of therapy as well as recurrence of the neoplasm.
As Virji et al. (1988) have pointed out, to date no tumor marker has been shown to be specific or sufficiently sensitive to be used in the detection and screening of neoplasms in the general population. However, some years ago, Weber (1982) proposed a list of characteristics for detecting and utilizing biological markers in monitoring human neoplastic disease (Table17.11). The table includes not only screening and diagnosis but also the following therapy by using a tumor marker, as discussed below. With the ready availability of biopsy material from a variety of neoplasms, which in the past could not be had, an additional requirement would be its detection in tissue sections and samples of the neoplasm itself. The rather arbitrary upper limit of 200 million cells for detection does not indicate a lower limit. Utilizing tissue sections and modern methodologies (see below), it is now possible to detect markers in only a few cells or in extremely low concentrations in body fluids. Recently, Hayes et al. (1996) have proposed a tumor marker utility grading system (TMUGS) for the evaluation of the clinical utility of tumor markers and the establishment of an investigational program for evaluation of new tumor markers.