The morass of tumor markers

T I B S - May 1985

182

Opinion to monitor some of the markers. This commercial enterprise encourages the production of more 'tumor markers'. Borek To complete the 'authentication' of a myth, the conclusions must be pubThe other problem with most of the lished. I believe that with the pullulation tumor markers used today is intrinsic in of journals not under the auspices of their nature. Most of them are proteins, professional societies, but of commercial and proteins are the end products of the publishing houses, standards have fallen molecular mechanisms of every ceil. A very low in some cases. An ambitious mammalian cell is endowed with the review writer can then enshrine the capability of producing perhaps 10000 myth for ever as fact. This puts an enoror more different proteins. Unless we mous burden on those who referee chance upon a protein which either papers for journals. In a paper from causes a malignancy or is a universal Nature I was asked to referee, I found concomitant of malignant tissues, look- tremendous lacunae both in the design ing for protein products which qualify as of the experiments and in the conclusion tumor markers in terms of the require- based upon those experiments; I made ments listed above is hopeless. the appropriate recommendations. In 1965, Phil Gold and his associates About four weeks later, I received the in Montreal isolated a protein from same paper from another journal. At colon cancer by differential immunologi- first I thought I would decline because cal elution. Subsequently, this antigen the author would be exposed to double was found in embryonic tissue and the jeopardy, but curiosity made me read it; intestinal lining of infants. Gold's esti- I found that absolutely no changes had mate of the usefulness of this marker, been made on the basis of my rectermed carcinoembryonic antigen (CEA), ommendations. At a conference I was rather modest. However, investiga- attended, the author presented the very tors who came into the field later have same data and the very same conmade exaggerated claims for it, but it dusions. often happens that the acolytes exceed the claims of the prophet 2. A ray of holm Recently, I looked at a review article As the eyes are said to be the mirror by someone active in this field. It had of the soul, so the urine can be conthe ambitious title, 'CEA, the universal sidered to be the mirror of mammalian tumor marker '3. Of the 35 malignancies metabolism. This has been known for a mentioned in this article, only two long time, indeed as recently as 100 (colon carcinoma and choriocarcinoma) years ago the taste of the urine was used were described as producing the tumor to detect diabetes mellitus. Fortunately, marker in 100% of the cases; at the more sophisticated sensors than the other extreme were testicular teritomas physician's tongue have since been with only 9% positive tumor markers developed. Using only the volume of (from the original paper, it seems that urine that would have adhered to the even these were positive only in organ hapless physician's finger, we can now culture). Well, if that is a tumor marker, determine whether or not a patient has a then the keen eye of a pathologist peer- malignancy. There are two highly proming through a microscope is better. At ising tumor markers, neither of which any rate, I would not want to have my are proteins. The first is the group of malignancy managed by a marker which modified nucleosides from the breakis 91% false-negative. The author of the down of tRNA (see TIBS, January 1977 review confirmed Pasteur's aphorism, for details); the second is neopterin. 'Nothing gets as complicated as the The extraordinarily complex structure arguments of a dying theory', because of tRNA may reflect its many functions. he recommended that three different Its primary sequence consists of about protein markers be monitored on the 80 of the four major bases (A,C,G,U) same patient to follow his disease. In found in other RNAs and a variety of America, these measurements would modified bases not found anywhere else. cost a minimum of $150 each time. Now These modified bases increase in numthat commercial firms have realized that ber with the complexity of the organism; this is a profitable area, kits are available Escherichia coli tRNA contains only two

The m o r a s s of tumor markers Ernest 'Current tumor markers are neither highly sensitive nor tumor specific.' A. L. PohP The hope of finding some unique metabolic products or unique components of malignant cells circulating in body fluids which can be measured is as old as modem biochemistry. The term 'tumor marker' for such an entity was coined by Morton K. Schwartz of the Sloan Kettering Institute. An effective tumor marker should be specific for malignancy, provide a minimum of false-positives and false-negatives, indicate the extensiveness of the malignancy, and preferably diminish or disappear after effective therapy. At a recent International Conference held in Vienna under the auspices of the Society For Early Detection of Cancer, someone calculated that there were reports on almost 90 different tumor markers. Unfortunately, none of these putative tumor markers meet the above qualifications even partly. There are two major reasons for this. The most obvious is the pressing need for a scientist to get his work published in order to obtain funds. This makes it tempting for eager individuals to publish results prematurely, without establishing whether the results are statistically meaningful. For example, I recently read the following about a new tumor marker: '99% of normals had values ranging from 1 to 35'; numbers of individuals, means and standard deviations were not reported (this was in an important journal of gynecology from a famous institution). I was appalled by the total lack of assessments of the significance of the scant data on which conclusions were drawn. About fifty years ago, my mentor for my graduate studies was the distinguished biochemist, H. T. Clarke. Although he was a courteous English gentleman, he would have flayed me alive in front of my colleagues had I given a report with data like that. E. Borek is at the Department of Molecular Biology, A M C Cancer Research Center and Hospital, 6401 West Colfax Avenue, Lakewood, Colorado 80214, USA. 1985. Elsevier Science Publishers B . V. . Amsterdam

0376 - 5067/85/$02A~1

183

T I B S - May 1985 or three, yeast tRNA five or six and as much as 20% of mammalian tRNA may be modified bases. The origin of these modified bases was elucidated about 20 years ago, as a result of a fortuitous observation. I had been working for some time on the effect of starving lysogenic microorganisms of nutrients. Fortunately, from the collection of the late Jaques Monod, we had two mutants of E. coli which were both lysogenic and auxotrophic - one for leucine, the other for methionine. We examined the DNA, RNA and protein content of the starved microorganisms. Nothing unusual was observed with the leucine auxotroph, but the methionine-requiring mutant was found to have 50% more RNA than it had before starvation4. This genetic aberration became known as Relaxed Control and has proved to be widely seminal. Since the majority of the modified bases in tRNA are methylated, and since methionine is a universal methyl donor, it occurred to me that the RNA accumulated in the methionine-starved organism must lack methylated bases. This was later confirmed. This methyldeficient RNA could be used to search for enzymes which introduce methyl groups. The discovery of these 'methyl transferases' for tRNA was followed by that of similar enzymes for DNA and rRNA. As a corollary, if a modified amino acid in a protein does not have a specific tRNA, then it must be modified (acetylated, phosphorylated, etc.) after translation. Of course, many such modifying enzymes are now known. The compelling impetus to study the tRNA methyltransferases in tumor tissue was provided by the studies of the British pathologist, Peter Magee 5. Magee found that dimethylnitrosamine methylates to a much greater extent the tRNA of the target tissue than its DNA. In turn, the tRNA methyltransferases are abnormally hyperactive in every malignant tissue. Transfer RNAs of tumor tissue The finding of aberrant tRNAmethylating enzymes in tumor tissue prompted the study of the tRNAs themselves. Surprisingly, only a few of the tRNAs in the malignant tumor were found to differ in structure from those in the normal tissue. Recently, Guy Dirheimer of Strasbourg isolated 18 different tumor-specific tRNAs with modifications which differed from those in normal counterparts. On the other hand, we found with the aid of Japanese

colleagues, that the primary sequence (i.e. before modification) is the same~. Ex~n of IR1NAbreakdown products by eano~r patients The excretion of abnormally high amounts of modified purines and pyrimidines by cancer patients was called to my attention by the late Alexander Guttman, an outstanding clinical investigator. He had been studying excretion of 7-methylguanosine by gout patients and observed very high concentrations of modified bases and of more species, in addition to 7-methylguanosine, in a patient with gout and leukemia. Since we had just recently discovered the method of their synthesis in nucleic acids and since one of the excretion products was N2N2-dimethylguanosine which is present only in tRNA, the origin of the excretion products became obvious. During the formation of tRNA, these modified bases must be positioned with extraordinary accuracy by highly specific enzymes. To prevent their incorrect insertion into new macromolecules after their release by tRNA turnover, mammalian cells lack the kinases needed to convert the modified base nucleoside or deoxynucleoside into the appropriate triphosphate. Since these bases are very stable and cannot be metabolized or recycled, they must be excreted. It occurred to me that, if the method of determination of these modified bases and nucleosides were improved, it could be used to diagnose cancer. To this end, the National Cancer Institute awarded a contract for improving methodology to Charles Gehrke of the University of Missouri. T. Phillip Waalkes, who at that time was on the staff of the NCI, was responsible for the project. In the section to follow, I describe work done in collaboration with them. Nucleoside markers The excretion of nucleosides was followed before, during and after therapy in a malignancy that responds well to chemotherapy. Within five days of starting therapy in six patients with Burkitt's lymphoma, nucleoside excretion returned to normal and remained normal as long as the subjects were in remission. How normal nucleoside excretion is restored after effective chemotherapy is obscure. An intrusion by the chemotherapeutic agents into the aberrant metabolism of the tumor tissue is implied since not all tumor cells are eradicated so quickly. Several other malignancies exhibit the same diminu-

tion of markers after effective therapy. Thus, the oncologist can treat patients on an individual basis7. A beautiful experiment was designed by colleagues at the University of Umea to evaluate the effectiveness of carcinoembryonic antigen (CEA), tumorderived polypeptide antigen (TPA) and platelet antigen (placental alkaline phosphatase, PLAP) as biological markers in malignant lymphomas. Their conclusions in toto are as follows: 'We have evaluated CEA, TPA, PLAP in urine from patients with three different kinds of malignant lymphomas. Six modified nucleosides, ~, MJA, M~G, mq, m2G, and m~G were analysed in the urine, from the same group of patients. The histological diagnoses were histiocytic lymphoma, lymphocytic lymphoma and Hodgkin's disease. The patients were classified into four different clinical stages. Consecutive samples were analyzed before and during ongoing radiotherapy and chemotherapy and during the post-treatment period. Our results showed that TPA and PLAP had limited value as biological markers for patients with malignant lymphomas. For CEA, a possible correlation with clinical stage was observed only in patients with Hodgkin's disease. The modified nucleosides, especially 0, showed a correlation with clinical stage for patients with all three diagnoses. Elevated levels of ~ in urine were: in healthy adults 4%, in patients in clinical stage one, 14% and in patients with advanced disease, 62%. Six cases showed a good correlation between the change in clinical stage upon treatment and the parallel change in the level of ~ in the urine. Our results suggest that modified nucleosides, especially ~ are valuable as biological markers for patients with malignant lymphomas TM. Gerhardt Schrch, the Director of the Children's Oncology Hospital in Dortmund used multivariant analysis. In other words, he used more than one nucleoside for making decisions on the status of the child. He found that: 'if we include more than one modified RNA catabolite in the evaluation of a patient, the method can be used with better than 95% confidence for a number of different malignancies '9. Another very important contribution comes from the laboratory of Gisella Nass who studied the appearance of tumor markers in mice with an implant of carcinogen. The mice excrete the markers long before a true tumor is evident or can be palpated ~°. Walter Kersten found that asbestos

184

T I B S - May 1985

workers who had no malignancy did have high concentrations of markers, but in some instances the markers disappeared. We found essentially the same thing in our laboratory. This implies the possibility of reversal of the path to a devastating malignancy. The animal system studied by Nass would be ideal for examining the reversibility of the aberrant metabolism which leads to malignancy. In collaboration with John Brewer, a distinguished gynecologist, we followed six women with choriocarcinoma by determining the nucleoside markers before therapy and then 7-8 days after therapy. We determined the nucleoside markers whereas Brewer determined the standard marker, human chorionic gonadotrophin (HCG). The nucleoside markers returned to normal in seven days; yet HCG concentrations remained high. Normally, gynecologists continue to feed methotrexate to patients until the HCG values return to normal or near normal - this may take weeks, if not months. Brewer decided that, according to the nucleoside markers, the women's malignancies had been eliminated and, therefore, stopped feeding methotrexate but followed the clinical progress of the women for two years. They remained

normal; two of them had normal deliveries, one had an elective abortion and the others were normal. Therefore, if our colleagues in gynecology are willing to try this new method of following effectiveness of therapy, we may be able to spare women the debilitating exposure to the hepatoxic agent, methotrexate n.

more workers are needed in this field, because a great many questions remain to be answered. If one-tenth of the effort which has been expended on the protein markers is devoted to these two nonprotein markers in the coming years, we will reap far greater benefits for our patients.

Neopterin The other nonprotein marker, which has been introduced by colleagues from the University of Innsbruck, is neopterin. Neopterin is derived from guanosine triphosphate and appears in the urine of the cancer patient as a result of some abnormal metabolism of malig: nant tissue. Unfortunately, data on the various malignancies in which this marker is useful are sparse, but it is very interesting that the concentrations of both nucleoside markers and neopterin are high in subjects with AIDS who, as it is well known, are on the way to various malignancies. I hope that our colleagues and perhaps even more importantly, the funding agencies, will slowly become convinced that more effort and resources should be directed to work on the nonprotein tumor markers. Many

References 1 Pohl, A. L. et al. (1983) Cancer Detection and Prevention 6, 7-20 2 Fuks, A., Shuster, J. and Gold, P. (1980) Cancer Markers (Sell. S., ed.), p. 315, Humana Press 3 Klavins, J. V. (1983) Ann. Clin. Lab. Sci. 13 4 Borek, E. and Kerr, S. J. (1972) Adv. Cancer Res. 15, 163-192 5 Magee, P. N. and Father, E. (1962) Biochem. J. 83, 114 6 Kuchino, Y., Borek, E., Grunberger, D. et al. (1982) Nucleic Acids Res. 10, 6421-6432 7 Borek, E. (1984) Tumour Biol. 5, 1-14 8 Rasmuson, T., Bjork, G. R., Damber, L. etal. (1983) Recent Results Cancer Res. 84, 331-343 9 Muller, J., Erb, N., HeUer-SchOchet al. Recent Results Cancer Res. 84, 317-330 10 Thomale, J. and Nass, G. (1982) Cancer Left. 15, 149-159 11 Borek, E., Sharma, O. K. and Brewer, J. I. (1983) Am. J. Obst. Gynecol. 146, 906-910 12 Hausen, A. et al. (1981) Clin. Chim. Acta 117, 297-305

EmergingTechniques Stereology: a working tool for cell biologists

relevant data relating the structure of cellular components to biochemical functions, and to describe the molecular structure of these components.

K. Schwerzmann and H. Hoppeler What is stereology? Stereology estimates three-dimensional parameters of biological structures from twodimensional pictures obtained in the electron microscope. The structural data can be combined with biochemical measurements to relate quantitatively cellular structures with their biochemical functions in vivo, or to describe the molecular architecture of cellular compartments. At first glance, biochemistry and stereology seem to have little in common. Biochemistry is the science dealing with chemical compounds and products of biological processes, stereology deals with the quantitation of three-dimensional objects from information contained in two-dimensional pictures. Stereology is of no use to the biochemist who studies reactions in a test-tube. If he wants to know what

K. Schwerzmann and H. Hoppeler are at the Anatomisches lnstitut, University o f Berne, CH-3000 Berne 9, Switzerland.

happens in a cell, however, he has to take structural features into account. The cell is a highly ordered structure divided into numerous compartments by intracellular membranes which limit reaction spaces and availability of substrates, and probably affect the sequential order of metabolic processes. Hence, quantitative information concerning, for instance, volume, surface area, and distribution of a given structure in a cell may assist in the elucidation of biochemical processes at the cellular level. Stereology applied to images of cell ultrastructure can be combined with biochemical measurements to obtain

~) 1985.ElsevierSciencePublishersB.V.. Amsterdam 0376- 5067/85/$ff2.f10

Stereology has been described as 'a body of mathematical methods relating three-dimensional parameters of a structure to two-dimensional measurements obtainable on sections of the structure'1. A life scientist uses micrographs, obtained from sections through biological material usually with an electronmicroscope, for measurements which yield the real dimensions of the structure of interest. The principle of stereology is demonstrated in Fig. 1. On a section, the volume V of an object is represented by an area a. Intuitively, the size of the area a is related to the volume V of the object. Likewise, the surface A of an object is represented by the boundary lines l on the section. Stereology defines the relationship between object and profile parameters using mathematical and statistical methods. However, stereology only