When discussing cancer, the most frequent misunderstanding is that cancer is a rare disease occurring in a few hundred persons per 100,000 of the population. The number per 100,000 is the annual incidence rate, but the incidence rate during a lifetime may be approximately 60-70 times higher.29 Cancer will develop in one in four or/five persons in Japan during their lifetime.29 Thus, the incidence rate of cancer from the viewpoint of a cumulative rate up to age 74 years is approximately one in four in Japan .3° Furthermore, this rate has most likely increased over the last 10 years. These findings indirectly suggest the accuracy of our proposed method to detect cancer at a high rate in the combined stages IV and V, assuming that most subjects in Stage IV pass into Stage V with the passage of time and clinical cancer will develop eventually in subjects in Stage V.

According to a report by Pluygers et al.," simultaneous measurements with 10 tumor markers were made on 2000 supposedly healthy subjects, with follow-up studies repeated over a period of 1-6 years. As a result, they found abnormal levels in one or more markers in 481 subjects (24.1 %) among 2000. This rate agrees with the 29.3% rate (623 of 2126) obtained for tumor stages IV and V in our study. In addition, the cancer detection rate in the abnormal group was 4.6% (22 in 481) and approximately 14 times the rate of the normal group (5 in 1519,0.33%), which was lower than that in the current study of 0.67% (10 of 1503). Their detection rate of 1.4% (27 of 2000) in their study population for cancer was the result of 1-6 years of continuous study.

In the current study, 28 (1 case was determined according to a report received from the subject's family) of 95 tumor stage V subjects were histopathologically or morphologically confirmed as having cancer during a period of 5-7 years after the screening, resulting in a much more efficient detection rate than that of Pluygers et al.31 In one postoperative patient, the tumor stage changed yearly from tumor stage 11 to tumor stage III, then to tumor stage IV, and finally to tumor stage V. In addition, the results of cancer incidence at each stage (Table 4) and the average time from the tumor marker combination assay to the time when clinical cancer appeared support the validity of the proposed model and the cutoff values adopted. As shown in Figure 2, tumor stage III shows an outstandingly high percentage among all age groups. Two reasons can be considered. First, some biologic and immunologic activity will stop microcancer growing at the 106 cell number level.²⁸

Second, in our assay system, the reversibility of each tumor stage was different, that is, 5 of 6 subjects in tumor stage III, 1 of 2 in tumor stage IV, and 1 of 10 in tumor stage V. The high reversibility of tumor stage III may explain the high percentage of this tumor stage in all age groups.

A survey of pathologic reports of latent cancer at autopsy revealed that approximately 30% of the cases of latent cancer were observed at autopsy in subjects in their 50s, and 50% of latent cancer was observed in subjects in their 70s.28 If we can consider that tumor stage IV coincides with latent cancer and tumor stage V with clinical and preclinical cancer, the combined percentage of tumor stage IV and tumor stage V was 36% in subjects in their 50s and 43% in their 70s in our study. These rates correspond to those of latent and preclinical cancer observed at autopsy, respectively. Our growing model of the natural history of cancer is schematic, so it will require modification according to the results obtained in the current study, for example, to include a long stationary phase in each stage. However, the tumor marker combination assay used in this way can be extremely efficient in cancer screening and risk assessment.

We credit three reasons for the successful results of our tumor marker diagnosis. First, cancer grows with a cancer-supporting stroma and cancer vessels because cancer is a neoplasm. Thus, we considered not only oncofetal antigen, but also oncoplacental antigen and cancer vessel-related substances to be important in cancer detection. Thus, not only tumor-specific tumor markers, but also tumor-associated tumor markers and growth-related tumor markers are important in tumor marker diagnosis. The results of our tumor marker risk assessment support this hypothesis. Concerning the order of appearance of the tumor markers, growth-related and associated tumor markers appear early (tumor stage III) in microcancer development, followed by specific tumor markers (tumor stage IV). These results provide further support for our hypothesis.

Second, the choice of tumor markers in the combination assay was appropriate, because three different types of tumor marker were included: tumor-specific tumor markers (CEA, CA 19-9, TPA, and HSAP), tumor-associated tumor markers (FT, FT/Fe, sialic acid, and IAP), and growth-related tumor markers (RNase and ALP isoenzyme)." From the standpoint of cancer screening, it is not necessary to fixate on cancerous organs. Tumor markers that are not organ specific are also important for determining whether cancer exists or not. Furthermore, the sensitivity and specificity of our tumor marker combination assay was ascertained in a cooperative study between our clinic, the National Cancer Institute (Bethesda, Maryland), and the Mayo Clinic (Rochester, New York) in 1986 using sera from subjects with or without cancer.² The results were summarized as follows; sensitivity, 80-90%; specificity, 84.085.0%; and accuracy, 83.3-88.0%. The cancer detection rate in tumor stage V was 29.5% in the current study. The
difference between this value and the National Cancer Institute result, that is, 80-90%, may be due to the finding that the