Relationship of total serum sialic acid to sialylglycoprotein acute-phase reactants in malignant melanoma.

Reported elevations of total serum sialic acid may be a result of shed tumour-related membrane sialyglycoprotein and/or concurrent elevation of non-specific, acute-phase reactant sialoglycoprotein. To clarify further the specificity and sensitivity of serum sialic acid monitoring, analyses of sialic acid by the thiobarbituric acid method and acute-phase reactants by radial immunodiffusion were made using the same malignant melanoma patients' sera. Preliminary studies of IgG, IgA, IgM, ceruloplasmin and C-reactive protein suggested that these would not be valuable monitors of tumour burden. Single serum samples from 59 melanoma patients and age- and sex-matched controls were further examined for sialic acid, alpha 1-acid glycoprotein, alpha 1-antitrypsin, haptoglobin, and alpha 2-macroglobulin. Patients were grouped according to tumour burden. In pairwise statistical tests, differences between groups tended to be greater for sialic acid than for acute-phase reactants. On discriminant analysis , sialic acid was clearly the most significant single discriminator between groups, with an F statistic of P < 0.00005. Although alpha 1-acid glycoprotein was quite strongly correlated with sialic acid, it was not such a good discriminator and did not add significantly to the predictive power of sialic acid alone.

ACUTE-PHASE REACTANTS (APR) can serve as effective monitors of tumour burden. Features of special interest are the relative ease of measurement and freedom from the restrictions of histological specificity of most immunodiagnostic tests. Broad clinical applicability is suggested by the variety of neoplasms where APR concentration is related to tumour burden, including leukaemia (Child et al., 1977) gliomas (Weiss et al., 1979) and carcinomas of breast (Coombes et al., 1977) bowel (Ward et al., 1977b), lung (Hollinshead et al., 1977) ovary (Mueller et al., 1971) cervix (te Velde et al., 1979) and prostate (Ward et al., 1977a). In each case serum concentration of APR has been directly related to tumour burden.
We have been examining the role of tumour-related sialylglycoproteins as relatively nonspecific tumour markers. Increased membrane density of sialic acid (N-acetyl neuraminic acid) and associated elevation of sialyltransferase activity has been reported in a variety of malignant and transformed cells (Mabry & Carubelli, 1972;Van Beek et al., 1973). We and others have found significant in vitro tumour cell production of sialylglycoproteins and relatively rapid appearance in tissue-culture media (Bhavanandan et al., 1977;Grim et al., 1976). This suggests that elevated serum sialylglycoproteins or related sialyltransferase activity might be a common feature in cancer patients. We have shown that there is a strong correlation between serum sialic acid and tumour burden in malignant melanoma, and that this relatively simple measurement appears to be a better correlate than related sialyltransferase activity (Silver et al.,1 979a). In addition to the sialic acid and sialyltransferase of tumour-cell origin there is a nonspecific component. Both serum sialic acid and sialyltransferase activity can act as nonspecific acute-phase reactants (Silver et al.,1 979a). In the case of sialic acid this can be explained by the demonstrated sialic acid content, of some acute-phase reactant (APR) glycoproteins (Koj, 1974).
Since APRs have already been suggested as valuable monitors of tumour burden, it would be of interest to investigate the relationship of total serum sialic acid and APR in cancer patients. The purpose of this study was to evaluate both serum sialic acid and known APR sialylglycoproteins as monitors of tumour burden in malignant melanoma. The APRs selected for study included the following: al-acid glycoprotein (AGP), al-antitrypsin (AAT), haptoglobin (HPT), cA2-macroglobulin (AMCG), ceruloplasmin (CPL), and immunoglobulins G, A, and M. C-reactive protein (CRP) was also considered, although its sialic acid content is not significant (Fischer & Gill, 1975).

MATERIALS AND METHOD)S
Patientls. Sera from 59 melanoma patients were used to study sialic acid, AGP, AAT, HPT and AMG. In order to maintain continuity with earlier data, the population is the same as used in a previous study (Silver et al., 1979a). Pre-study evaluation and grouping according to objective assessment of tumour burden was as described in detail by Silver et al (1979a). Group I patients had no evidence of disease at the time of serum sampling. Group II patients had a relatively small tumour burden consisting of either primary melanoma, local recurrence, or intransit metastases estimated at less than 5 g. Group III patients all had relativelv advanced regional or distant metastatic disease. clearly greater than a tumour burden of 5 g. Sexand age-matched normal control sera were selected from our serum collection for comparison with each patient group. Addi-tional sera from 10 Group Ifl melanoma patients wNere selected as a preliminary evaluation of CPL, CRP, IgG, IgA, and TgM.
Serum. collectio;i. Whole blood w,as collected in 10 ml glass vacutainer tubes and the samples allowsed to clot at room temperature for 1 h. After centrifugation at 500 g for 10 min., the sera wNere removed in 0-8ml aliquots, placed in polypropylene tubes and stored at -70°C until used. Before assay the serum samples were allowed to thawr at room temperature.
Serum assay. Determination of bound sialic acid was by a modification of the thiobarbituric acid technique (Warren. 1959) as described by us (Silver et al., 1978(Silver et al., , 1979a. In addition to the usual inter-assay and intra-assay controls, analysis included calculation of extinction coefficients for each experiment from a series of standard sialic acid and deoxvribose solutions. The calculation of extinction coefficients and sialic acid concentration was aided by a compuiter programme allowing for minor contamination with interfering substances (Silver et al., 1978).

RESULTS
The preliminary evaluation of immunoglobulins, CPL and CRP is represented in Table I. Neither the immunoglobulins nor CPL were remarkably raised in the face of extensive malignant melanoma. They were not further investigated. However, 9/10 of the same patients' sera did show clear CRP elevations. This was further studied in sera selected from patients with known tuimour burden, evaluated as described above. Again CRP was elevated in association with advanced disease, this time in 7/10 Group III patients. Of the other sera, there were CRP elevations in 0/10 normals, 1/16 Group I patients, and 0/8 Group II patients. Since CRP does not contain significant sialic acid, and elevations were not seen in any limited-disease (Group II) patients, this APR was not further studied. that could contribute significantly to total serum sialic acid content, and might also act as sensitive monitors of tumour burden: AGP, AAT, HPT and AMG. These APRs and sialic acid were measured in each of the 59 melanoma sera from patients with known tumour burden and age-and sex-matched normal controls. The mean serum values are shown in the Figure  Ceruloplasmin 0/8 1/10 3/10 C-reactive protein 0/8 9/10 10/10 of some markers is suggested for Group II patients over normal controls or Group I. Pairwise statistical evaluation using the Mann-Whitney test is shown in Table II. As suggested by the Figure, AMG is generally a less reliable indicator of tumour burden than other APRs. The remaining APRs and sialic acid showed significant increases in concentration for Group III patients. Similar comparisons between other groups were less remarkable. For most comparisons confidence limits (P values) were less for sialic acid than for the APRs. Since sialic acid was not uniformly superior for all pairwise comparisons, it was felt desirable to test whether sialic acid performed uniformly better overall. In order to assess this, a stepwise discriminant analysis was undertaken in which consecutive entry of the variables was controlled by computer package (Kiecka, 1975). Two analyses were performed. The first included data from the 3 groups of melanoma patients and normal controls; the second examined the melanoma patients alone. In the initial analysis the first variable chosen was sialic acid. Not only was this variable deemed to be the most significant single predictor of discrimination between the groups, but the discrimination was also   Abbreviations: as in Table II. extremely significant (F = 73 6 with 3 and 117 degrees of freedom, P < 0.00005). Very similar results were obtained on the second discriminant analysis, in which attention was restricted to the 3 melanoma groups. Sialic acid was again picked as the best single discriminating variable (F= 63-7 with 2 and 59 degrees of freedom, P < 0 00005). Observations on the within-group correlation matrix showed that sialic acid measurements were quite strongly correlated with AGP, but AGP was not such a good discriminator as sialic acid. This was reinforced elsewhere in the analysis, since AGP was not picked by the program as another variable in the discriminant function.

Wte further investigated those APRs
Each marker was also examined to determine how many melanoma sera were outside the normal range established by the normal control sera (Table III). This type of analysis more closely mimics clinical practice, where the significance of a test is usually determined by comparison with a normal range. The upper limit of normal determined by us was similar to that established by others (Koj, 1974;Fischer & Gill, 1975). As detailed in Table III, serum elevations were more frequent for sialic acid than for any of the APRs. The number of patients with elevated sialic acid levels and/or any APR elevation was not significantly greater than those with sialic acid elevations alone (Table IV). DISCUSSION There has been an evolving re-evaluation of the clinical role of tumour markers. The initial prospect of highly sensitive and specific immunochemical assays used as diagnostic screening aids was justifiably met with great interest. Further experience has tempered this enthusiasm. This is best illustrated by accumulated experience with the most thoroughly evaluated immunodiagnostic tests, carcinoembryonic antigen (CEA) and xyl-fetoprotein (AFP).
Most clinical laboratories have found that CEA testing lacks the histological specificity or sensitivity for small tumour burden necessary for diagnostic screening (Dhar et al., 1972). While AFP is more sensitive and specific, its use as a screening test will probably be restricted to well defined high-risk populations (People's Republic of China, 1974;Silver et al., 1973). On the other hand, tumour markers appear well suited to provide important non-diagnostic information on staging, evaluation of prognosis, detection of early recurrence or quantitation of tumour response to treatment (Parks et al., 1974; Herrera et al., 1977). Such information is crucial in the rational application of the proliferating complex treatment programmes available for an increasing number of human neoplasms. Accurate staging and prognosis are especially important in the selection of appropriate high-risk patients for aggressive surgery or potentially toxic adjuvant therapy programmes. Detection of early recurrence and objective assessment of tumour burden in response to treatment are essential for the prompt selection of effective systemic therapy among alternatives. For these examples of post-diagnostic patient management, a tumour marker need not be specific for a given histological type. On the contrary, a relatively nonspecific marker would have a broad range of clinical applicability not presently enjoyed by many "specific" tumour antigens. Potential tumour markers in this category include selected APRs (Ward et al., 1977a), and sialic acid (Silver et al., 1978(Silver et al., , 1979a. Our preliminary investigation eliminated CRP, CPL and the immunoglobulins as unlikely correlates of tumour burden in malignant melanoma patients. In more detailed investigations, serum concentrations of AGP, AAT, HPT and AMG, both AGP and AAT correlated well with tumour burden (Table II). However, increased serum sialic acid was more frequently seen than APR elevations (Table III) and sialic acid results generally showed greater statistical significance (Table II). This was confirmed on discriminant analysis. Sialic acid was picked as the best discriminating variable, and the APRs did not add to the discriminating power of sialic acid alone.
The basis for the apparent superiority of sialic acid as a tumour marker in this study is yet to be determined. It may be that total serum sialic acid is acting as a correlate of combined total APR. There was nothing in the statistical analysis supporting this view. The observed results are perhaps better explained by tumour sialylglycoprotein production. Certainly, increased sialic acid has been repeatedly observed at the tumour-cell surface for a variety of neoplasms, and sialylglycoproteins appear to be rapidly shed by these cells (Bhavanandan et al., 1977;Grim et al., 1976). We ourselves and others have also developed further evidence that the sialylglycoprotein of tumour origin can be quantitated independently of APR (Silver et al., 1979b;Lipton et al., 1978). If total serum sialic acid is a composite of both APR activity and specific tumour-cell production, it might be expected that sialic acid would reflect tumour burden more closely than any of the APRs.