1355 Tumor-Associated Antigen TA-90 Immune Complex Assay Predicts Subclinical Metastasis and Survival for Patients with Early Stage Melanoma Mark C. Kelley, M.D. Ralph C. Jones, M.D., LCDR Rishab K. Gupta, Ph.D. Reynold Yee, B.S. Stacey Stern, M.S. Leslie Wanek, Ph.D. Donald L. Morton, M.D. MC USN Roy E. Coats Research Laboratories, John Wayne Cancer Institute, Saint John’s Health Center, Santa Monica, California. Presented as a poster at the 49th Annual Cancer Symposium of the Society of Surgical Oncology, Atlanta, Georgia, March 21–24, 1996. Supported by Grants CA-12582, CA-29605, and CA-0989 from the National Cancer Institute, and funding from the Wrather Family Foundation, Los Angeles, California. Mark Kelley’s current address: Division of Surgical Oncology and Section of Surgical Sciences, Vanderbilt University Medical Center, 21st Ave. South, Nashville, TN 37232-2729. Ralph C. Jones’s current address: Department of Surgery, National Naval Medical Center, 8901 Wisconsin Ave., Bethesda, MD 20889-5600. Address for reprints: Donald L. Morton, M.D., John Wayne Cancer Institute, 2200 Santa Monica Blvd., Santa Monica, CA 90404. The views expressed herein are solely those of the authors and do not necessarily reflect the views of the U.S. Navy, the U.S. Army, the Uniformed Services University of Health Sciences or the Department of Defense. Received December 19, 1997; revision received April 3, 1998; accepted April 3, 1998. © 1998 American Cancer Society BACKGROUND. TA-90 is a tumor-associated antigen first identified in the urine and sera of patients with metastatic melanoma. In the early stages of disease, TA-90 is present in circulating immune complexes (ICs) that may be detected with an antigen specific enzyme-linked immunosorbent assay (ELISA). In this study, the authors evaluated the efficacy of the TA-90 IC assay in detecting subclinical metastasis of early stage melanoma and predicting the survival of patients with this disease. METHODS. Archival sera were collected preoperatively from 114 patients who underwent wide excision with or without regional lymphadenectomy in the treatment of clinical Stage I melanoma. Sera were analyzed for TA-90 IC in a blinded fashion, and results were correlated with the patient’s clinical course as determined by database and chart review. Subclinical metastases were considered present at the time of surgery if the lymphadenectomy specimen was pathologically positive and/or the patient subsequently developed recurrence. RESULTS. The TA-90 IC assay predicted subclinical metastasis in 43 of 56 patients (P , 0.0001), with 14 false-positive and 13 false-negative results. Sensitivity and specificity for the detection of occult metastasis were 77% and 76%, respectively. Positive and negative predictive values were 75% and 77%, respectively. Fifteen of 18 tumor positive regional lymph node basins (83%) and 34 of 46 recurrences (74%) were accurately predicted when considered independently (P , 0.004). Preoperative TA-90 IC status was also highly correlated with survival: 5-year overall and disease free survival rates were 63% and 46%, respectively, for the TA-90 IC positive group, compared with 88% and 82%, respectively, for the TA-90 IC negative group (P 5 0.0001). A multivariate analysis with standard prognostic variables identified preoperative TA-90 IC status as a strong, independent prognostic factor for both overall and disease free survival. CONCLUSIONS. To the authors’ knowledge, TA-90 is the first tumor marker that accurately predicts subclinical metastatic disease and survival for patients with early stage melanoma. For this reason, the TA-90 IC assay has the potential to improve dramatically the prognostic evaluation of patients with this disease. Its role in postoperative risk stratification and early detection of recurrence is being evaluated in a prospective study. Cancer 1998;83:1355– 61. © 1998 American Cancer Society. KEYWORDS: melanoma, tumor marker, tumor-associated antigen, TA-90 immune complexes, occult metastasis, prognostic factor. T umor markers may be used to detect occult metastatic disease or recurrence, determine prognosis, and monitor response to therapy.1,2 Some of the markers described for melanoma have been correlated with disease stage and recurrence, but none have achieved 1356 CANCER October 1, 1998 / Volume 83 / Number 7 clinically useful accuracy.3–5 The need for a sensitive and accurate marker of occult metastatic melanoma is probably greatest when selecting patients most likely to benefit from postoperative adjuvant therapies,6 such as melanoma vaccines7 and high dose interferon a-2b,8 after surgical resection of clinical Stage I (CS-I) melanoma. Urine tumor-associated antigen (U-TAA) is a high molecular weight glycoprotein identified in the urine of patients with metastatic melanoma.9 It comprises multiple subunits, including an immunogenic 90-kD subunit designated TA-90.10 TA-90 is expressed by 71% of melanoma cell lines and tumor biopsies and up to 70% of breast, colon, and lung carcinomas and soft tissue sarcomas.10,11 Immunochemical analysis, including antibody blocking studies showing lack of cross-reactivity, indicate that TA-90 is distinct from known tumor markers, such as carcinoembryonic antigen (CEA), prostate specific antigen (PSA), CA-15-3, a-fetoprotein, and other oncofetal antigens (R. K. Gupta, unpublished data). TA-90 is also a major target for host antitumor immune responses. Nearly all melanoma patients whose tumors express this antigen develop anti-TA-90 immunoglobulin (Ig)M and/or IgG antibodies spontaneously.12,13 Patients who receive an allogeneic polyvalent melanoma cell vaccine (PMCV) also develop increased humoral and delayed-type hypersensitivity (DTH) responses to TA-90.12,13 In an analysis of patients receiving PMCV after surgical treatment for regional metastatic melanoma, a high IgM and DTH response to TA-90 correlated directly with survival, whereas a strong IgG response correlated inversely with survival. These observations were the basis for a highly predictive immunologic survival model for patients treated with PMCV.13 TA-90 is found in the serum of melanoma patients as a free antigen and as an immune complex (IC) with IgG antibody.14 Although TA-90 also forms an IC with IgM antibody in melanoma patients, assays to detect this IgM-IC are less reliable. Patients with extensive tumor burden express high levels of free TA-90 and low levels of TA-90 IgG-IC; the opposite is true of patients with a low tumor burden.14 An antigen specific enzyme-linked immunosorbent assay (ELISA) was developed to detect TA-90 IgG-IC in the sera of cancer patients.11 In the current study, this TA-90 IC assay was used to detect subclinical metastasis in patients with CS-I melanoma and predict their survival. METHODS Patient Population The John Wayne Cancer Institute (JWCI) prospective melanoma database was used to identify individuals who underwent definitive surgical treatment of World Health Organization (WHO) CS-I melanoma between January 1, 1980, and December 31, 1990. This period was chosen to guarantee that all patients had been treated at least 6 years previously and therefore had been followed long enough to allow for recurrence.15,16 The study population comprised 114 randomly selected patients who had biopsy-proven primary melanoma with no clinical or radiographic evidence of metastasis, and from whom serum samples had been obtained and cryopreserved no more than 30 days before surgery. All patients underwent wide excision of the primary site with a surgical margin $2 cm; 71% of primary lesions were intermediate in thickness (0.75– 4.0 mm) (Table 1). One hundred one patients also underwent elective regional lymphadenectomy, preceded in 25 cases by intraoperative lymphatic mapping and sentinel lymph node dissection.17 Eighteen patients who underwent lymphadenectomy had tumor in a single lymph node (n 5 9), in 2– 4 lymph nodes (n 5 6), or in $5 lymph nodes (n 5 3). Of the 9 patients with a single tumor-involved lymph node, 5 had only a single micrometastasis (,2 mm in greatest dimension) in the sentinel lymph node. Twenty-eight patients underwent postoperative adjuvant therapy: 11 received bacillus CalmetteGuérin (BCG) and 17 received PMCV plus BCG.7 Recurrence was recorded and correlated with preoperative TA-90 IC status. The minimum postoperative follow-up was 72 months for surviving patients, and the median follow-up was 91 months (range, 4 –181 months) for the entire group. All patients gave informed consent to participate in a prospective study of TA-90 antigen expression, which was approved by the Institutional Review Boards of the University of California, Los Angeles, and Saint John’s Health Center, Santa Monica, California. TA-90 Immune Complex Assay The preoperative serum sample from each patient was assayed for TA-90 IC using an antigen specific IC ELISA as previously described (Fig. 1).11 Briefly, sera were diluted 1:60 in 0.025 M phosphate-buffered saline and 1% Triton X-100, pH 7.2. A 1:60 dilution was chosen because it gave the optimal discrimination of TA-90 IC values between normal volunteers and cancer patients when serial dilutions of sera were studied. Diluted sera were placed in ELISA plates containing wells coated with murine anti-TA-90 monoclonal antibody (AD1-40F4), incubated for 45 minutes, and washed to remove unbound proteins. Goat antihuman IgG Fab conjugated to alkaline phosphatase was add- Tumor Marker for Early Stage Melanoma/Kelley et al. TABLE 1 Demographic and Pathologic Data for 114 Patients Characteristics Age (yrs) Median Range Primary site Head/neck Trunk Extremity Clark’s level II III IV V Unknown Ulceration Satellitosis Gender Male Female Histologic type SSM Nodular LMM/other Breslow’s depth ,0.75 mm 0.75–1.50 mm 1.51–4.0 mm .4.0 mm Unknown Tumor positive wide excision specimen Pathologic stage IA IB IIA IIB IIIB 1st recurrence: None Local In transit Regional lymph nodes Distant sites No. (%) of patients 48 20–79 17 (15%) 47 (41%) 50 (44%) 1357 was $0.410 optical density (OD) at 405 nm. This value was generated by testing 59 normal volunteers11: their mean OD was 0.169 6 0.080, and none had an OD $0.410, which was .3 standard deviations above the group mean. A cutoff of 0.410 was validated in a separate group of 250 normal controls and 419 patients with a variety of malignancies: only 3% of controls were positive, compared with over 50% of cancer patients.11 Clinical Correlation and Statistical Analysis 9 (8%) 48 (42%) 47 (41%) 6 (5%) 4 (4%) 9 (8%) 4 (4%) 67 (59%) 47 (41%) 55 (48%) 28 (25%) 31 (27%) 15 (13%) 33 (29%) 48 (42%) 14 (12%) 4 (4%) 16 (14%) 8 (7%) 32 (28%) 45 (39%) 11 (10%) 18 (16%) 68 (60%) 6 (5%) 7 (6%) 9 (8%) 24 (21%) SSM: superficial spreading melanoma; LMM: lentigo maligna melanoma. ed; the plates were incubated for 45 minutes and then washed to remove unbound IgG Fab. The addition of 1 mM p-nitrophenyl phosphate in 10% diethanolamine produced a yellow color in the presence of the bound alkaline phosphatase conjugate. After incubation for 1 hour, absorbance was measured at 405 nm. Samples were assayed in duplicate, corrected for nonspecific binding, and compared with standard positive and negative controls. All assays were performed by laboratory personnel with no knowledge of the identity or clinical status of the patient. Sera were considered positive if the absorbance TA-90 IC status was correlated with lymph node status and recurrence. Patients who had pathologically positive lymph nodes and/or later developed recurrence were considered to have had occult metastasis at the time of surgery. Sensitivity, specificity, and predictive values of the TA-90 IC assay for lymph node status, recurrence, and subclinical metastasis were calculated. The significance of the association between TA-90 IC status and these endpoints was determined with Fisher’s exact test. Survival curves for TA-90 IC positive and negative patients were constructed with the SAS statistical software program (SAS Inc., Cary, NC), and statistical significance was evaluated using the log rank test. Univariate and multivariate analyses were performed with the Cox proportional hazards model, using TA-90 IC status (positive vs. negative) and the following standard prognostic variables (Table 1): age (,50 vs. $50 years), gender, Breslow’s depth (,1.5 mm vs. $1.5 mm), Clark’s level (II or III vs. IV or V), regional lymph node status (positive vs. negative), histologic type (superficial spreading vs. other), primary site (extremity vs. other), ulceration, satellitosis, and tumor status of the wide excision specimen (positive vs. negative). Results were considered significant if the P value was ,0.05. RESULTS Correlations Among TA-90 IC Status, Lymph Node Status, and Recurrence The preoperative TA-90 IC assay was positive for 57 patients and negative for 57. When the TA-90 IC status was related to the lymph node status of the 101 patients who underwent elective lymphadenectomy, 38 of 83 with tumor free regional lymph nodes were TA-90 IC positive (Table 2). It is noteworthy that 27 of 38 TA-90 IC positive, lymph node negative patients subsequently developed recurrence, compared with 9 of 45 TA-90 IC negative, lymph node negative patients. This suggests that the TA-90 IC assay may have detected systemic occult disease. All 3 patients with tumor positive lymph nodes but a negative TA-90 IC assay had only a single lymph node micrometastasis 1358 CANCER October 1, 1998 / Volume 83 / Number 7 FIGURE 1. This diagram represents the TA-90 antigen specific immune complex assay. Diluted serum samples are added to plates coated with murine anti-TA-90 monoclonal antibody. After incubating for 45 minutes and washing to remove unbound protein, goat antihuman immunoglobulin Fab conjugated with alkaline phosphatase is added, and the plates are incubated for 45 minutes and washed. Then p-nitrophenyl phosphate is added and absorbance measured at 405 mm after 1 hour. (in the sentinel lymph node). Thus, the TA-90 IC assay predicted lymph node status with a sensitivity of 83% and a specificity of 54%; its positive and negative predictive values were 28% and 94%, respectively. Forty-six patients developed local (n 5 6), intransit (n 5 7), regional lymph node (n 5 9), or distant (n 5 24) recurrence. The sensitivity and specificity of the TA-90 IC assay for predicting recurrence were 74% and 66%, respectively (Table 2). Positive and negative predictive values were 60% and 79%, respectively. The high false-positive rate of the TA-90 IC assay was explained in part by the absence of recurrence in several patients with lymph node metastases. Occult metastasis, defined as a tumor positive lymphadenectomy specimen and/or postoperative recurrence of disease, was identified in 43 TA-90 IC positive patients but only 13 TA-90 IC negative patients (Table 2). The sensitivity and specificity of the TA-90 IC assay for occult metastasis were 77% and 76%, respectively; corresponding positive and negative predictive values were 75% and 77%. Correlation Between TA-90 IC Status and Survival Five-year disease free survival was 82% for TA-90 IC negative patients but only 46% for the TA-90 IC positive group (P 5 0.0001; Fig. 2). The median disease free survival was 53 months and .120 months for the TA-90 IC positive and negative groups, respectively. The 5-year rate of overall survival was 63% for TA-90 IC positive patients versus 88% for TA-90 IC negative individuals (P 5 0.0001; Fig. 3). The median overall survival was 91 months and .170 months in the TA-90 IC positive and negative cohorts, respectively. Analysis of TA-90 IC Status as a Prognostic Factor for Survival Univariate analysis identified TA-90 IC status as a significant prognostic factor for overall and disease free survival (P 5 0.0001). As expected, Breslow’s depth was also a significant predictor of both overall and disease free survival (P 5 0.0001). The histologic type of melanoma was significant for overall and disease free survival, whereas a tumor-positive, wide ex- Tumor Marker for Early Stage Melanoma/Kelley et al. 1359 TABLE 2 Preoperative TA-90 Immune Complex Status as a Predictor of Lymph Node Status and/or Recurrence TA-90 IC Lymph Nodes (n 5 101) Positive Negative Recurrence (n 5 114) Positive Negative Subclinical metastasisa (n 5 114) Positive Negative Positive Negative 15 38 3 45 34 23 12 45 43 14 13 44 Sensitivity/ specificity P value 83%/54% 0.004 74%/66% ,0.0001 77%/76% ,0.0001 a Subclinical metastasis is defined as the presence of lymph node metastases and/or subsequent recurrence. FIGURE 3. The correlation between preoperative TA-90 immune complex (IC) status and overall survival is shown. Individuals with a positive TA-90 IC assay had a significantly lower median overall survival (91 vs. .170 months) and 5-year overall survival (63% vs. 88%) than those with a negative TA-90 assay. men remained a significant prognostic factor for recurrence in multivariate analysis. No other factors (including lymph node status) predicted recurrence or survival in this cohort. Similar results were obtained using the TA-90 IC OD and number of involved lymph nodes as continuous variables (data not shown). DISCUSSION FIGURE 2. The correlation between preoperative TA-90 immune complex (IC) status and disease free survival is shown. TA-90 IC positive (optical density $0.410) individuals had a significantly lower median survival (53 vs. .120 months) and 5-year disease free survival (46% vs. 82%) than TA-90 IC negative patients (P 5 0.0001). cision specimen was a significant predictor of disease free survival. Multivariate analysis also identified TA-90 IC as an independent prognostic factor for overall (P 5 0.0167) and disease free (P 5 0.0084) survival. Breslow’s depth remained an independent predictor of overall (P 5 0.0034) and disease free (P 5 0.0080) survival, but the histologic type was no longer significant. It is noteworthy that the tumor status of the wide excision speci- A significant proportion of patients thought to have localized, early stage melanoma actually have subclinical metastasis.15,16 Although surgical lymph node staging can detect regional lymph node metastases, there is no way to detect occult disease accurately at other sites. This distinction had only prognostic significance until a recent study demonstrated that high dose interferon a-2b improved the survival of patients with high risk melanoma.8 Because the survival benefit conferred by high dose interferon is only modest and the regimen is quite toxic and expensive,18 identifying the patients with occult metastatic disease who would be most likely to benefit from such therapy would be of great potential utility. In this study, the TA-90 IC assay had a sensitivity of 77% and a specificity of 76% for occult melanoma, which were far greater than percentages reported with tumor markers for other malignancies.1,2 The sensitivity of the assay may be limited by the fact that not all melanomas express TA-90, and some patients may develop immune complexes containing IgG antibod- 1360 CANCER October 1, 1998 / Volume 83 / Number 7 ies that recognize an epitope which is spatially close to the epitope recognized by the mouse monoclonal antibody used to capture the TA-90-IgM IC. The second limitation could be addressed by an ELISA utilizing a capture antibody that recognizes a spatially distinct epitope. The specificity for detecting lymph node metastasis was somewhat lower than that for occult metastasis because several TA-90 IC positive patients without regional lymph node metastases ultimately developed recurrence, suggesting that the assay was actually detecting distant metastatic disease. In addition, all three TA-90 IC negative patients with tumorinvolved regional lymph nodes had only a solitary lymph node micrometastasis (in the sentinel lymph node), which may have been too small a tumor burden for the assay to detect. Although the result of the TA-90 IC assay is positive in most patients with lymph node metastases, it does not approach the sensitivity and specificity of lymph node dissection and therefore cannot be suggested as a substitute for surgical staging, particularly staging based on a focused immunohistochemical analysis of the sentinel lymph node.17,19 In the current study, the TA-90 IC assay might have yielded false-negative results for patients whose lymphadenectomy specimens were examined by standard histologic techniques prior to routine use of sentinel lymphadenectomy and immunohistochemistry. Use of the TA-90 IC assay to analyze a single preoperative blood sample distinguished 2 subgroups with a 25% difference in 5-year survival. It may be possible to increase the assay’s diagnostic accuracy by eliminating confounding variables associated with preoperative evaluation. For example, after biopsy but prior to wide excision, the assay may detect residual melanoma at the primary site rather than occult metastatic disease. In this study, 14 patients had microscopic tumor in the wide excision specimen; although this was not strongly correlated with TA-90 IC status (data not shown), it was a potential source of falsepositive results. In addition, TA-90 IC formed in response to a primary melanoma may persist after excisional biopsy and thus be a source of false-positive results. Multivariate analysis demonstrated that TA-90 IC was surpassed only by Breslow’s depth as an independent predictor of recurrence and survival. Regional lymph node status was not significant, possibly because 9 of 18 lymph node positive patients (50%) had only a single lymph node metastasis, and 5 of these metastases were micrometastases in the sentinel lymph node. Patients with micrometastatic deposits in a single lymph node have a much better prognosis than patients with multiple or macroscopically positive lymph nodes.15,20 In summary, the TA-90 IC assay can accurately detect subclinical metastasis and predict the survival of patients with early stage melanoma. Further investigation is needed to define its role in postoperative risk stratification and early detection of recurrent disease, and the authors have undertaken a large prospective study of serial TA-90 IC values obtained 1–3 months after surgical resection of early stage melanoma. Prospective validation of the results reported in this article is underway as part of an international multicenter trial of sentinel lymphadenectomy performed on patients with CS-I melanoma. REFERENCES 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. Norton JA, Fraker DL. Tumor markers. In: Sabiston DS. Sabiston textbook of surgery. New York: W. B. Saunders, 1996. Pandha HS, Waxman J, Sikora K. Tumor markers. Br J Hosp Med 1994;51:297–303. Schadendorf D, Zuberbier T, Diehl S, Schadendorf C, Czarnetzki BM. Serum manganese superoxide dismutase is a new tumor marker for malignant melanoma. Melanoma Res 1995;5:351–3. Horikoshi T, Ito S, Wakamatsu K, Onodera H, Eguchi H. Evaluation of melanin-related metabolites as markers of melanoma progression. Cancer 1994;73:629 –36. Miliotes G, Lyman GH, Cruse CW, Puelo C, Albertini J, Rappaport D, et al. Evaluation of new putative tumor markers for melanoma. Ann Surg Oncol 1996;3:558 – 63. Hansson J. Systemic therapy of malignant melanoma. Med Oncol 1997;14:73– 81. Morton DL, Barth A. Vaccine therapy for malignant melanoma. CA Cancer J Clin 1996;46:225– 49. Kirkwood JM, Strawderman MH, Ernstoff MS, Smith TJ, Borden EC, Blum RH. Interferon alfa-2b adjuvant therapy of high-risk resected cutaneous melanoma: the Eastern Cooperative Oncology Group Trial EST-1684. J Clin Oncol 1996; 14:7–17. Rote NS, Gupta RK, Morton DL. Tumor-associated antigens detected by autologous sera in the urine of patients with solid neoplasms. J Surg Res 1980;29:18 –22. Euhus DM, Gupta RK, Morton DL. Characterization of a 90 –100 kDa tumor-associated antigen in the sera of melanoma patients. Int J Cancer 1990;45:1065–70. Gupta RK, Morton DL. Monoclonal antibody based ELISA to detect glycoprotein tumor-associated antigen-specific immune complexes in the sera of cancer patients. J Clin Lab Anal 1992;6:329 –36. Euhus DM, Gupta RK, Morton DL. Induction of antibodies to a tumor-associated antigen by immunization with a whole melanoma cell vaccine. Cancer Immunol Immunother 1989;29:247–54. Jones RC, Kelley MC, Gupta RK, Nizze JA, Yee R, Leopoldo Z, et al. Immune response to polyvalent melanoma cell vaccine in AJCC stage III melanoma: an immunologic survival model. Ann Surg Oncol 1996;3:437– 45. Gupta RK. Circulating immune complexes in malignant melanoma. Dis Markers 1988;6:81–96. Tumor Marker for Early Stage Melanoma/Kelley et al. 15. Balch CM, Soong SJ, Milton GW, Shaw HM, McGovern VJ, Murad TM, et al. A comparison of prognostic factors and surgical results in 1,786 patients with localized (stage I) melanoma treated in Alabama, USA and New South Wales, Australia. Ann Surg 1982;196:677– 84. 16. Morton DL, Davtyan DG, Wanek LA, Foshag LJ, Cochran AJ. Multivariate analysis of the relationship between survival and the microstage of primary melanoma by Clark level and Breslow thickness. Cancer 1993;71:3737– 43. 17. Morton DL, Wen D-R, Wong JH, Economou JS, Cagle LA, Storm FK, et al. Technical details of intraoperative lym- 1361 phatic mapping for early-stage melanoma. Arch Surg 1992; 127:392–9. 18. Balch CM, Buzaid AC. Finally, a successful adjuvant therapy for high-risk melanoma [editorial]. J Clin Oncol 1996;14:1–3. 19. Cochran AJ, Wen D-R, Morton DL. Occult tumor cells in the lymph nodes of patients with pathological stage I malignant melanoma. Am J Surg Pathol 1988;12:612– 8. 20. Day CL, Mihm MC, Lew RA, Harris MN, Kopf AW, Fitzpatrick TB, et al. Prognostic factors for patients with clinical stage I melanoma of intermediate thickness (1.51–3.99 mm). Ann Surg 1982;195:35– 43.