Protocol. This trial protocol has been provided by the authors to give readers additional information about their work.

Transcription

1 Protocol This trial protocol has been provided by the authors to give readers additional information about their work. Protocol for: Faries MB, Thompson JF, Cochran AJ, et al. Completion dissection or observation for sentinel-node metastasis in melanoma. N Engl J Med 2017;376: DOI: /NEJMoa

2 This supplement contains the following items: 1. Original protocol, final protocol, summary of changes 2. Original statistical analysis plan, final statistical analysis plan, summary of changes

3 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 JOHN WAYNE CANCER INSTITUTE CONFIDENTIAL Multicenter Selective Lymphadenectomy Trial II (MSLT-II): A Phase III Multicenter Randomized Trial of Sentinel Lymphadenectomy and Complete Lymph Node Dissection versus Sentinel Lymphadenectomy Alone in Cutaneous Melanoma Patients with Molecular or Histopathological Evidence of Metastases in the Sentinel Node Sponsor: John Wayne Cancer Institute 2200 Santa Monica Boulevard Santa Monica, CA Original Version January 2, 2003 Amendment #1 May 18, 2004 CONFIDENTIALITY STATEMENT The information contained in this document is privileged and contains confidential information. Any other distribution, copying or disclosure, in part or in whole, is strictly prohibited unless such disclosure is required by federal regulations or state law. Persons to whom the information is disclosed must be informed that all of the information is confidential and may not be further disclosed by them in part or in whole. Page 1 of 66

4 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 TABLE OF CONTENTS 1. PROTOCOL SYNOPSIS Background Primary Secondary Trial Design Number of Subjects Safety Measurements Adverse Events Surgery-related Adverse Events Designated Sites BACKGROUND AND SIGNIFICANCE Background and MSLT-I The Sentinel Node Concept and MSLT-I Continued Evolution of Sentinel Lymphadenectomy Molecular Staging of SN Nodal Ultrasound Evaluation Detection of Subclinical Micrometastasis in Melanoma by Evaluation of Blood and Serum Markers Can immunologic and molecular tumor markers identify subclinical melanoma metastases and patients at high risk for postoperative recurrence? Clinical implications of ultrastaging subclinical micrometastatic melanoma by immunologic, molecular and proteomic markers TA90-IC assay S-100 and MIA asays Prognostic assays based on serum DNA microsatellites with loss of heterozygosity (LOH) Serum proteomics Interaction Between Tumor Burden and Endogenous Immune Response as a Determinant of the Clinical Outcome in Melanoma Melanoma Differentiation-Associated Antigens Endogenous TA-90 Immune Response and Survival Ganglioside Antigens Experimental Plan for Assessment of Immune Response TRIAL DESIGN AND RANDOMIZATION Enrollment Enrollment Prior to LM/SL Enrollment Following LM/SL Randomization Sentinel Node Negative Subjects Non-MSLT-2 Centers Adjuvant Therapy Designated Sites INCLUSION AND EXCLUSION CRITERIA Inclusion Criteria Exclusion Criteria STUDY PROCEDURES Enrollment Screening Phase Pre-LM/SL LM/SL H&E and IHC of SN RT-PCR Assay Randomization Phase Eligibility Verification CLND Observation with Nodal US Follow Up Visits Subject Termination/Completion Page 2 of 66

5 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, Study Completion Early Withdrawal Due to AE Early Withdrawal Due to Withdrawal of Consent Screen Failure SN-negative Group Study Procedures Physical Examinations Adverse Events Quality of Life Questionnaires Radiological Evaluations Safety Labs Nodal US Blood Specimens Shipment of Pathology Specimens LM/SL Specimens CLND Specimens Primary Tumor Specimens ASSESSMENT OF EFFICACY Primary Endpoint Secondary Endpoints Recurrence Classification by Location Recurrence Classification by Size TREATMENT OF SUBJECTS Interventional and Operative Procedures Biopsy of Primary Tumors Lymphoscintigraphy Operative Procedures Guidelines Special situations Pathology Procedures Primary melanomas Sentinel node examination method Pathologic examination of completion lymphadenectomy nodes Molecular Biology Procedures Primary Melanoma Molecular Studies SAFETY Adverse Events Surgery-related Adverse Events Required Details Intensity Causality Serious Adverse Events STATISTICAL CONSIDERATION Statistical Design Major Outcomes Primary outcome o Melanoma-specific survival Secondary outcomes Pathology study outcomes Immunology and molecular outcomes Quality of life Safety data Statistical Analysis Subject accrual, withdrawal and baseline characteristics Statistical analysis for the primary outcome death from melanoma Statistical analysis for the other time-to-event secondary outcomes Statistical analysis for the observational study Statistical analysis for assessing the prognostic significance of new markers Page 3 of 66

6 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, Statistical analysis for quality of life (QOL) data Statistical analysis for safety data Sample Size OVERSIGHT COMMITTEES Pathology Oversight Committee (POC) Primary lesion Regional node metastases Surgical Oversight Committee (SOC) Independent Data Monitoring Committee Responsibilities Current IDMC members ETHICAL AND REGULATORY CONSIDERATONS Informed Consent Institutional Review Curriculum Vitae Laboratory Certification and Appropriate Normal Values Patient Privacy Record Retention Publication Rights Conflicts of Interest Financing Regulatory Requirements Study Monitoring REFERENCES...59 APPENDIX APPENDIX APPENDIX APPENDIX Page 4 of 66

7 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 Sponsor: Primary Principal Investigator: Medical Monitor: Co-Investigators: Laboratory Director: (RT-PCR & LOH) Laboratory Director: (TA90-IC, anti-ta90 IgG, anti-ta90 IgM, GP-100, TRP-2, S-100, & MIA) Laboratory Director: (anti-gd2 IgM and anti-gt1b IgM) Clinical Sites: John Wayne Cancer Institute 2200 Santa Monica Boulevard Santa Monica, CA Donald L. Morton, M.D. (310) Nora M. Hansen, M.D. (310) (310) FAX (310) hour emergency number Alistair Cochran, M.D. David Elashoff, Ph.D. Robert Elashoff, Ph.D. Richard Essner, M.D. Leland J. Foshag, M.D. Rishab K. Gupta, Ph.D. Dave S. B. Hoon, Ph.D. Mepur Ravindranath, Ph.D. Bret Taback, M.D. Dave S. B. Hoon, Ph.D. Rishab Gupta, Ph.D. Mepur Ravindranath, Ph.D. This list will be maintained in the Operations Manual. Page 5 of 66

8 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 STANDARD PROTOCOL ABBREVIATIONS AE AWD CBC CLND CTO DFS ELISA H&E ICF ICH IHC JWCI LM LN LOH MIA MSLT-I MSLT-II NED NSN OS PE PI RT-PCR SAE SRC SL SN US WLE Adverse Event Alive With Disease Complete Blood Count Complete Lymph Node Dissection Clinical Trials Office Disease-Free Survival Enzyme-linked Immunosorbent Assay Hematoxylin & Eosin staining Informed Consent Form International Conference on Harmonisation Immunohistochemistry John Wayne Cancer Institute Lymphatic Mapping Lymph Node Loss of Heterozygosity Melanoma Inhibiting Activity Multicenter Selective Lymphadenectomy Trial Multicenter Selective Lymphadenectomy Trial II No Evidence of Disease Non-Sentinel Node Overall Survival Physical Examination Principal Investigator Reverse Transciptase Polymerase Chain Reaction Serious Adverse Event Surgical Review Committee Sentinel Lymphadenectomy Sentinel Node Ultrasound Wide Local Excision Page 6 of 66

9 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, PROTOCOL SYNOPSIS 1.1 Background New data indicate that by histopathological and molecular techniques, approximately 70-80% of melanoma patients with micrometastases in the sentinel nodes (SNs) have no other involved nodes. A complete lymph node dissection (CLND) may not be necessary in these patients because sentinel lymphadenectomy (SL) may have already removed all the nodes containing tumor metastases. The Multicenter Selective Lymphadenectomy Trial II (MSLT-II) is a randomized clinical trial designed to determine whether the melanoma-specific survival associated with intraoperative lymphatic mapping and sentinel lymphadenectomy (LM/SL) alone improves survival compared with LM/SL plus CLND in subjects with SN metastases detected by histopathologic or molecular techniques. 1.2 Trial Objectives Primary To determine whether LM/SL and CLND in cutaneous melanoma subjects with evidence of SN metastases by histopathologic or molecular evaluation will lead to longer melanoma-specific survival than that observed after LM/SL alone and postoperative observation with serial nodal ultrasound Secondary 1. To determine whether in cutaneous melanoma subjects with evidence of metastases in the SN by histopathologic or molecular techniques, LM/SL and CLND provide longer disease-free survival (DFS) than LM/SL and postoperative observation with serial nodal ultrasound. 2. To compare the frequency of same-basin recurrence after LM/SL without CLND with the frequency of non-sentinel node (NSN) metastases detectable by histopathology in CLND specimens. 3. To assess prospectively the prognostic accuracy of molecular staging (RT-PCR) of the SN. 4. To determine if new histopathologic techniques applied to SN predict the presence of metastases in the NSN, the likelihood of recurrent melanoma and of death from melanoma. 5. To determine if indices of the endogenous immune response to melanoma-associated antigens can predict regional or distant subclinical metastasis of melanoma before and after LM/SL. 6. To assess subjects blood before and after surgeries, for molecular markers (DNA, RNA, and proteomic markers [serum protein]) of melanoma. 7. To assess subjects blood before and after surgeries, for melanoma tumor markers (MIA, S- 100, TA90-IC) and evaluate their prognostic significance. 8. To assess primary tumors of subjects for tumor DNA markers and evaluate their capacity to predict disease outcome. 9. To retrospectively assess LNs from CLND by molecular analysis for histopathologically inapparent occult metastases. 10. To assess and compare the quality of life of subjects undergoing CLND or observation after LM/SL. Page 7 of 66

10 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, Trial Design MSLT-II is a phase III multicenter, prospective, randomized clinical trial to determine if LM/SL and CLND for tumor-positive SN improves melanoma-specific survival compared to LM/SL alone in subjects with histopathologic or molecular evidence of metastases in the SN. 4,200 subjects with cutaneous melanoma will be initially accrued; 3,500 from MSLT-II centers (see Operations Manual) prior to LM/SL; and 700 from other centers following a tumor-positive SN biopsy. Only subjects with a tumor-positive SN by either H&E, IHC or RT-PCR will be randomized. Subjects are randomized to either CLND or observation with serial ultrasound of the basin that underwent SL. We anticipate that approximately 1,925 will have a tumor-positive SN, and will accept randomization. Trial accrual is expected to continue for 7 years, with an additional 3-year follow up period. Figure 1 shows the stratification and the overall treatment plan following randomization. 1.4 Number of Subjects 4,200 subjects will be enrolled. 1,925 subjects will be randomized. No more than 50% of subjects randomized (962 subjects) may come from an individual MSLT-II center. Enrollment will terminate when 1,925 subjects have been randomized. 1.5 Safety Measurements Safety will be evaluated by review of AEs, clinical laboratory tests, and physical examinations Adverse Events Specific adverse events will be recorded for one year after Day 0 of the study. These include persistent blue staining at the injection site, lymphedema, weakness, dysesthesia, and inability to work or permanent disability Surgery-related Adverse Events Other specific adverse events, related to surgery, will be recorded from enrollment through 120 days following Day 0 of the study. These include allergic reaction to blue dye, wound separation, seroma/hematoma, hemorrhage, infection, skin graft failure, thrombophlebitis, urinary tract infection, pneumonia, and cardiac complications. 1.6 Designated Sites Designated sites will be obtaining blood specimens that will be shipped to JWCI (as well as additional pathological specimens) for serum archiving and the following assays: o Serum DNA/LOH (as well as fixed tissue LOH) o Serum Proteomics o Antibody Titer (TA90 IgG, TA90 IgM, IgM, GT1b IgM, GP-100, and TRP-2) o TA90-IC o S-100/MIA Page 8 of 66

11 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 Diagnosis of Localized Melanoma LM/SL With Tumor-Positive SN by H&E, IHC or PCR Stratification Factors SN tumor positivity by H&E/IHC or PCR Breslow Thickness (<1.20 mm, mm, >3.5mm) Site of LM/SL (MSLT Center or non-mslt Center) Randomize CLND Observation Serial Nodal US LM/SL Basin Figure 1: General principles for the randomized multicenter trial evaluating completion lymph node dissection in melanoma subjects after sentinel lymphadenectomy. 2. BACKGROUND AND SIGNIFICANCE 2.1 Background and MSLT-I The treatment of regional lymph node basins in patients with clinically localized melanoma has been controversial since 1892, when Snow first advocated routine elective lymph node dissection in patients with clinically negative (non-palpable) lymph nodes. In the intervening period, only two treatment options were available: immediate elective lymph node dissection (ELND) or observation with complete dissection if lymph node metastases became clinically evident. The proponents of complete elective dissection favored early removal of regional lymph nodes based on the hypothesis of an orderly progression of tumor from the primary site to regional lymph nodes and then systemically. By the time patients present with gross nodal disease, their tumors have generally already spread systemically, and can no longer be cured by lymph node dissection. However, if regional metastatic disease could be removed prior to systemic spread, the patient would Page 9 of 66

12 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 be cured. There was considerable evidence to suggest that this hypothesis might be correct. For example, retrospective studies have reported that patients undergoing immediate lymphadenectomy survive better than patients who after observation were treated with therapeutic lymph node dissection (TLND) when they developed node metastases Elective lymph node dissection reportedly improves survival over at least 5-10 years in certain subgroups of melanoma patients. Balch et al. 1,2 reported that improved survival was not evident until 5-8 years of follow-up and then only in patients with primary melanomas of 1.5 to 3.99 mm thickness. Four prospective randomized trials of elective lymphadenectomy in the management of stage I melanoma have been reported. Veronesi and the World Health Organization 11,12 reported a randomized prospective study involving 17 cancer centers in 12 countries. They concluded that the addition of elective lymph node dissection to wide excision did not statistically improve patient prognosis as compared to other patients treated by wide excision alone followed at 3-month intervals. This study has been criticized for non-uniform quality of the surgical procedures, failure to randomize patients according to lesion thickness and level of invasion, and a large number of women in the cohort that may have reduced the therapeutic benefits of ELND. A Mayo Clinic study 13 also failed to show a significant difference between ELND and TLND. The most recent report by Balch and colleagues from the Melanoma Intergroup Trial 14 found no overall benefit of ELND, but several subgroups of patients (1-2 mm thick melanoma; age under 60 years and non-ulcerated primaries) appeared to derive significant benefit. Although these were prospectively stratified groups, since the subgroupings of interest were not part of the original hypothesis, the significance of the results is not clear. The more recent WHO trial, while not demonstrating a survival benefit of ELND, found the subgroup of ELND patients with tumor-positive NODES had significantly better survival than those patients who underwent wide excision and delayed TLND. Thus, the role of surgery in the management of the regional nodes in early-stage melanoma remains hotly debated between advocates of two alternative hypotheses as shown in Figure 2. 14A Incubator Hypothesis Marker Hypothesis >90% Primary Melanoma Rarely (<10% ) in T4 primary Simultaneous lympatic and blood borne spread Primary Melanoma Latent growth in SN Subsequent distant Metastasis Direct blood borne spread to distant sites Sentinel Node Metastasis Distant Metastasis Figure 2. According to the incubator hypothesis, primary melanoma usually metastasizes first via the lymphatics to the SN, where it may grow but remain latent ( incubate ) before spreading to distant sites. Thus early removal of the SN at this stage, before there is distant spread, should prevent distant metastases. Alternatively, according to the marker hypothesis, a primary melanoma metastasizes simultaneously via lymphatic and hematogenous routes. Thus finding tumor cells in the SN is merely a marker of a primary melanoma that has the capacity to metastasize. Removal of the SN in this case is unlikely to influence the growth of distant metastases and would have no therapeutic effect. However, the absence of melanoma cells from the SN would indicate a primary melanoma that is unlikely to spread to distant sites. 2.2 The Sentinel Node Concept and MSLT-I Page 10 of 66

13 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 The SN concept is based on the hypothesis that lymphatic metastasis is an orderly process which can be predicted by mapping the lymphatic drainage from a primary tumor to the first or sentinel node (SN) in the regional lymphatic basin. In other words, a primary malignancy will metastasize to the SN before involving other nodes in the basin. This hypothesis was studied in a feline model by Wong, Cagle, and Morton, 15 who found that the lymphatic drainage of a particular part of the skin could be mapped to a single SN using vital blue dye. The feasibility of SL in humans and its technical details were first studied in patients undergoing SL followed immediately by CLND, regardless of the histopathologic status of the SN. 16 In this trial, 237 lymphatic basins were studied in 223 patients with clinical stage I melanoma. A SN was identified in 194 of 237 (82%) nodal basins, with 38 of 194 (22%) SN containing metastases. Only 2 of 194 (1%) basins had immunohistochemical evidence of metastases in non-sentinel nodes (NSN) when examined by immunohistochemical methods when the SN was tumor free. This clearly indicates that when the SN was identified, it predicted the status of the nodal basin in 99% of cases. To determine if the SN could have predicted the status of the nodal basin by chance alone, the relative incidence of SN and NSN metastasis was compared. 16 Forty-seven of 259 (18%) SN had metastases, compared with only 2 of 3079 (<0.1%) NSN. This confirms that the predictive ability of the SN is not due to chance alone and that lymphatic metastasis in melanoma is an orderly process that can be identified through lymphatic mapping. These findings have been readily reproduced at many other melanoma centers throughout the world. Since these preliminary studies of SL for melanoma, we have made considerable progress in improving and standardizing the technique of LM/SL. First, different gamma probes were evaluated. It was found that the gamma probes varied in their characteristics for localizing the three commonly used radiocolloids 17. Each probe had particular detection characteristics that may change with different radiopharmaceuticals. Differences in radioisotope detection may explain intraoperative variation in localization during LM/SL and requires consistent use of the same gamma probe and radiocolloid for optimum results. Next, three different lymphoscintigraphic agents were examined. This study by Glass et al. 18 found that accurate identification of the SN was possible with all three agents but required early imaging. Delayed imaging is unreliable and may lead to incorrect identification of the SN. Another study by Essner et al. 19 found that LM/SL performed on the same day with a single injection of filtered technetium-99 sulfur colloid is the most useful method for probe-directed LM. This technique demonstrated the highest in vivo and ex vivo count ratios and the radioactivity fall-off between the excised nodes and post-excision basins, and concordance between the blue dye and hot nodes. In 1999, Kelemen et al. 20 reported that LM/SL could be cautiously performed in patients who have undergone previous wide local excision (WLE) if the primary resection margin was no greater than 2 cm, and if the wound was not closed with rotational flaps. In summary, the above studies which evaluate the technical aspects of LM/SL indicate that the current technology with respect to lymphoscintigraphy, gamma probes and blue dye detection is satisfactory to consistently identify the SN from center to center with a high degree of accuracy. In fact, an international consensus commentary on methodology has been published. 20A Nonetheless, many questions still remain concerning LM/SL for melanoma. The impact of LM/SL on survival and its therapeutic utility remain open questions. In 1994, we began the Multicenter Selective Lymphadenectomy Trial (MSLT-I) to answer some of these questions. This trial was designed to determine whether there is a survival difference between patients treated by WLE + LM/SL and CLND when metastases were found in the SN versus wide excision alone. Additionally, MSLT-I was designed to determine whether LM/SL provides more accurate staging of melanoma and can better identify patients who require a therapeutic lymph node Page 11 of 66

14 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 dissection. Patients with Clark level III and Breslow thickness 1.00 mm or greater, or with Clark level IV or V melanoma (regardless of Breslow thickness) were randomly assigned to either WLE alone or WLE, SL with CLND if the SN contained tumor. Seventeen centers completed the learning phase of the trial and randomized patients. The MSLT-I trial, after 8 years of accrual, randomized 2001 patients between WLE alone versus WLE and SL. The overall successful identification of the SN was 96%, thus confirming the successful technology transfer from JWCI to the other participating centers. In 1999, Morton et al. 21 reported a study validating the accuracy of LM/SL for early-stage melanoma by comparing the experience with LM/SL at the JWCI with other centers from MSLT-I. While the success ratios of identifying SN were equivalent at JWCI and other sites, Morton found that the combination of radiocolloid with blue dye was more useful for finding a SN (99.1%) than blue dye alone (95.2%). After a center had completed a 30-case learning phase, the success of SN identification in the MSLT group was independent of the center s case volume or experience in the MSLT. In addition, the incidence of SN metastases was not significantly different between MSLT (18.7%) and the JWCI (22.6%) (p=0.104) experience. Among the 182 patients with positive nodes, 68% had a single positive node and 32% had two or more positive nodes (20% of patients had 2 positive SNs and an additional 12% had positive NSN found at CLND), again confirming the JWCI experience. The slightly higher nodal positivity in the JWCI group can be explained by the greater proportion of patients with adverse risk factors for nodal metastasis (i.e., younger age, truncal primary and higher Clark level). Based upon the small number of events (deaths or recurrences) at this time, among the 2001 MSLT patients we estimate that an additional 5 years of follow-up will be necessary to fully evaluate the therapeutic utility and survival data of LM/SL. 2.3 Continued Evolution of Sentinel Lymphadenectomy We have made several important observations of SN biology from the MSLT-I. The SN contains tumor by IHC and H&E staining in approximately 20% of melanoma patients. If tumor is present in the SN, CLND is viewed by many as the standard of care. However, CLND may overtreat a significant number of patients. New data indicate that by both histopathological and molecular techniques, approximately 70-80% of patients with metastases in the SNs have tumor cells limited to the SN with no involvement of NSN. Therefore, CLND may not be necessary in these patients since the SL may have already removed all the nodes containing melanoma metastases. Also, it is possible that LM/SL alone may be more effective than routine elective CLND because it maintains regional immunity by leaving more of the regional lymph nodes intact. Since the vast majority of such patients have no disease beyond the SN, it is unclear which patients require a completion lymphadenectomy to improve survival. In fact, only 10-12% of patients in the MSLT-I trial and from the JWCI experience have additional tumor-positive nodes when the CLND specimen is evaluated by standard H&E techniques (10.5% of 323 patients with positive SN seen at the JWCI, and 12% of 182 patients with tumor-positive SN from the MSLT-I trial). However, the incidence of NSN involvement increases to 20% when CLND specimens are evaluated by IHC (Cochran, unpublished observations). There may be identifiable clinicopathologic factors determining the patients who are at greatest risk to have disease beyond the SN. Such data would refine the indications for completion lymphadenectomy, potentially sparing an additional morbid procedure in the vast majority of patients with a tumor-positive SN, and better identifying those at risk for recurrence. If it were possible to accurately identify the minority of patients with tumor in NSNs, it would be possible to avoid CLND in those whose NSNs are free of tumor. We have looked at the relationship of Breslow thickness to NSN tumor positivity (Table 1) and found that with increasing tumor thickness above 3.5 mm, there is an increase in NSN tumor positivity. However, the JWCI Page 12 of 66

15 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 database and Balch et al. 14 show no survival benefit of SL or ELND over wide excision alone in melanoma thicker than 3.5 mm. Thus, even though thicker lesions are more likely to have NSN involvement, early removal of these nodes by CLND appears to provide no survival benefit presumably because the melanoma has already spread to distant sites. We will use Breslow thickness to stratify subjects for MSLT-II because it is a pre-lm/sl correlate of NSN involvement. Breslow (mm) < >3.5 # NSN Metastasis 4/32 4/37 14/64 21/52 % NSN Metastasis Table 1. Relationship of Breslow thickness to NSN metastasis in patients with positive SN The most important measure of the accuracy of the SN procedure is the long-term recurrence rate in the lymphatic basin of patients with pathologically negative SN. Four studies have reported samebasin recurrence rates of zero to 5% after excision of histologically tumor-free SN, but median followup was short in these reports. Since regional recurrence in melanoma may require more than 5 years to become clinically apparent, longer follow-up is required to determine the true nodal failure rate. Regional nodal recurrence after excision of a tumor-free SN suggests incorrect identification of the SN and/or occult metastases in additional sentinel or NSNs. However, another explanation is failure to identify micrometastases (<2 mm) in the SN due to histopathologic sampling limitations. Theoretically, this can be minimized by serial sectioning and immunohistochemical staining of the SN. However, this is very labor intensive. Theoretically, a 6-mm node can be sliced into 1,500 fourmicron sections, but typically only ten or twenty of these sections are evaluated by pathological examination. This inspects less than 0.5% of the entire node. Serial sectioning to extinction and examination of every section would avoid the risk of missing metastases but is prohibitively expensive. Although sensitive tests such as immunohistochemistry and RT-PCR technology can enhance tumor detection, they are only effective if the region of the node most likely to contain tumor is examined. 2.4 Molecular Staging of SN In an effort to improve the detection of occult metastases in the SN and to minimize nodal recurrence we have developed molecular techniques to identify tumor cells in the SN. We have found that histopathology appears to understage SN micrometastasis in about 30% of patients and that these molecular metastases are clinically significant. We previously demonstrated that we can use RT-PCR of fresh frozen sections of the SN to detect occult metastases. We studied fresh frozen SN in 72 patients 28 and found that multi-marker expression in histopathologically negative SN was highly correlated with recurrence (p=0.005). Those patients who had no positive markers had virtually 100% survival. We concluded that standard histopathology underestimates the true number of metastases. Expression of multiple molecular tumor markers in the SN more accurately reflects micrometastases and may be a more powerful predictor of disease relapse than histopathologic staining. We have demonstrated that a single molecular tumor marker is unsatisfactory. Because handling frozen tissues and sections is logistically problematic and would not be possible in a multicenter, worldwide trial, we have developed molecular techniques to reproducibly extract and quantify RNA from paraffin embedded tissue. 14A Our techniques for multi-marker staging from paraffin-embedded sections can identify clinically relevant metastases. It is estimated that up to 30% of patients undergoing LM/SL may have clinically relevant SN micrometastases in the SN which are detectable only by more sensitive molecular staging. Our findings dramatically show that if SN are Page 13 of 66

16 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 histopathologically negative and negative against all markers, patients have 90-95% long-term disease-free survival (DFS) and overall survival (OS) (Figure 4). Conversely, patients with increasing numbers of positive markers have decreasing DFS and OS. This data was developed using a multimarker RT-PCR assay that includes melanoma-associated markers. Melanogenesis markers such as tyrosinase, which may be positive due to the presence of nevus cells, Schwann cells, and nerves in the SN, are not used. This strongly suggests that these reactions are not false positives. In fact, the dramatic inverse correlation between DFS and number of positive markers clearly suggests that these molecular metastases are due to occult melanoma cells that cannot be detected by standard immunohistopathologic techniques. Therefore, it is our plan to utilize molecular staging of the SN as standard technique, which will be evaluated for its prognostic accuracy. Figures 4. Left: Kaplan Meier estimates of disease-free survival (DFS) in patients with H&E/IHC-positive SNs (N= 53) and IHC-negative SNs (N=162) according to SN expression of no mrna markers or at least one mrna marker (tyrosinase, MART-1, MITF, and/or TRP-2). Right: Kaplan Meier analysis of overall survival (OS) according to multimarker qrt and histopathology status of SNs in 215 patients. 14A 2.5 Nodal Ultrasound Evaluation In the past, lymph node basins of patients who did not undergo elective lymph node dissection were followed by clinical examination. More recently, ultrasound has been examined as an adjunct to palpation in an effort to improve detection of nodal metastasis at an earlier stage. Since the size and number of metastases have been correlated with patient survival, it has been hypothesized that earlier detection will translate to earlier treatment with improved outcomes. Several preliminary studies of ultrasound have been conducted to date. Blum et al. 29 used ultrasound to evaluate 1288 patients for evidence of metastases and found ultrasound to have a 89% sensitivity and 99% specificity. A subsequent study by Voit et al. 30 demonstrated a 99% sensitivity and 98% specificity for ultrasound in a prospective evaluation of 829 patients followed over a 4-year period. Characteristics of nodal appearance on ultrasound that have been found to be highly predictive of metastases were described by Moehrle et al. 31 as length to depth ratio of less than 2, hypoechoic center, and absence of hilar blood vessel. The presence of any 2 of these characteristics was found to be 100% sensitive and 96% specific. We plan to use nodal ultrasound to follow SN tumor-positive subjects who are randomized to SL alone and observation only. Those subjects with suspicious nodes by ultrasound will undergo fine Page 14 of 66

17 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 needle aspiration biopsy and if it is positive or if palpable tumor does develop in the lymph node basin, the subject will then undergo CLND. 2.6 Detection of Subclinical Micrometastasis in Melanoma by Evaluation of Blood and Serum Markers Can immu nologic an d mole cular tumor ma rkers id entify subc linical melanoma metastases and patients at high risk for postoperative recurrence? Malignant melanoma metastasizes via the lymphatics to regional lymph nodes and by the vascular system to distant sites, either directly from the primary site or indirectly from the nodal metastases. CLND is undertaken to remove nodal metastases before they spread to distant sites. The patients who are likely to benefit from CLND are those with regional lymph node metastases but no progressively growing metastasis in other sites (see Figure 2, Incubator Hypothesis). The goal of this study is to determine whether immunologic or molecular markers can be used as stratification factors in predicting the outcome of subjects rendered clinically free of disease (NED: no evidence of disease) by surgical resection. Accurate prediction of the presence or absence of subclinical tumor burden in a postoperative patient who is clinically NED will identify patients who are curable by surgery alone and who do not need adjuvant therapy. The successful development of methodology to quantitate the presence of subclinical tumor burden and follow its response to therapy opens up new horizons in melanoma therapy. MSLT-II subjects will be involved in the investigation of these questions. Detection of signs of minimal melanoma in the blood is distinct from, but complementary to, detection of minimal disease in the SN Clinical implications of ultrastaging subclinical micrometastatic melanoma by immunologic, molecular and proteomic markers MSLT-II will identify a substantial population of patients who are without evidence of metastatic melanoma in the SN or distant sites (clinical stage I). The natural history of melanoma suggests that some of these patients will suffer disease recurrence due to subclinical disease that was not clinically apparent. Clinical stage I group subjects will be tested preoperatively and at specified postoperative intervals by assays for immunologic markers such as TA90-IC and molecular markers such as DNA microsatellites for loss of heterozygosity (LOH) and BRAF mutation. These techniques have the potential to identify subjects with minimal residual disease, who need postoperative adjuvant therapy and close clinical surveillance. This would permit major advances in cost-effective design for stratification in clinical trials, potential reduction of the toxicity and overall costs of care. Benefits will accrue not only to the patients with residual disease, who may be more aggressively treated, but also to those without disease, who can be spared costly follow up and potentially morbid treatment that is unlikely to benefit them TA90-IC assay TA-90 is a glycoprotein tumor-associated antigen that is immunogenic and induces formation of endogenous antigen-antibody complexes The TA90-IC assay detects a unique tumor marker that can predict the presence of subclinical micrometastases in early-stage melanoma preoperatively and during postoperative follow up. The TA90-IC assay may detect positive results months before recurrence can be detected by routine clinical radiologic evaluation. Page 15 of 66

18 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 TA90-IC status is an important prognostic factor independent of other characteristics such as Breslow thickness and SN status As shown in Figure 5, the sensitivity and specificity for detecting subclinical disease which will lead to recurrence within 5 years is 75-80% in recent larger series of postsurgical stage I/II patients. 100 AJCC STAGE I & II 80 TA-90 IC Negative N = 114 Percent Disease-free p< TA-90 IC Positive N = Disease-free Survival (months) Figure 5. TA90-IC status is an independent risk factor for recurrence after surgical resection of AJCC stage I-II melanoma. Kelley et al. 32 evaluated preoperative serum TA90-IC levels for predicting the survival of patients with early-stage melanoma. Archival sera that had been collected preoperatively from 114 patients who underwent wide excision with or without CLND were selected. Sera were analyzed for TA90-IC in a blinded fashion, and results were correlated with clinical course. Subclinical metastases were considered present at the time of surgery if the CLND specimen was pathologically positive and/or the patient subsequently developed recurrence. Of 56 patients with subclinical metastasis, the TA90-IC assay was positive in 43 (p<0.0001). Sensitivity and specificity for the detection of occult metastasis were 77% and 76%, respectively. Preoperative TA90-IC status was also highly correlated with survival: 5-year OS and DFS rates were 63% and 46%, respectively, for the TA90-IC positive group, compared with 88% and 82%, respectively, for the TA90-IC negative group (p=0.0001). A multivariate analysis with standard prognostic variables identified preoperative TA90-IC status as a strong independent factor for both OS and DFS. In a separate study, Kelley et al. 33 used the TA90-IC assay in a blinded fashion to examine multiple archival serum samples prospectively collected during postoperative surveillance of 166 patients with AJCC stage I, II or III melanoma. Results were correlated with disease recurrence and survival. Respective 5-year DFS rates were 74% and 24% (p=0.0001) for TA90-IC negative and positive patients. The TA90-IC assay was a significant predictor of survival for both stage II and III patients. Multivariate analysis identified TA90-IC status as the strongest independent prognostic factor for both OS and DFS compared to standard prognostic factors such as Breslow thickness and site of primary. The TA90-IC assay was elevated in 54 (77%) of 78 patients who developed recurrent disease, becoming positive 19 7 months before clinical evidence of recurrence. Overall, the assay detected recurrence with a sensitivity of 78% and specificity of Page 16 of 66

19 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, %. This study s findings indicate that the TA90-IC assay can accurately predict survival and suggest the presence of subclinical disease after surgery for melanoma. This approach has potential in selecting patients for adjuvant therapy S-100 and MIA asays Subsequent to the reports by our group (Cochran et al. 39,40 ) S-100 became a widely accepted histopathological marker for malignant melanoma. Recently, it has been shown that S-100B, a subunit of the S-100 protein, is detectable in serum of melanoma patients and correlates with stage and progression of the disease. Serum S-100 levels appear to be relevant for prediction of survival and might be useful to guide adjuvant therapy. Melanoma inhibiting activity (MIA), a small (11kD) soluble protein that was initially identified and isolated from spent culture medium of melanoma cells lines, 41 is believed to be restricted to melanoma. An assay has been developed to quantitate the MIA protein in serum and determine the clinical significance of MIA levels. MIA is detectable in serum of advanced-stage melanoma patients and its level has been correlated to response to therapy and to disease recurrence. 41A Recently, a number of reports 42 have compared the clinical utility of S-100 and MIA in melanoma patients. All of these reports conclude that both serum markers are valuable for monitoring therapy and detecting tumor progression in advanced-stage patients, but their usefulness in predicting recurrence and death during the early postoperative period (when adjuvant therapy would be most useful) has not been well studied prospectively. Subjects entered in MSLT-II at the major multicenter sites will have their blood specimens cryopreserved for later testing to determine whether the various immunologic marker assays give independent or similar results in assessment of postoperative risk of recurrence. Blood will be collected and cryopreserved preoperatively and on postoperative months 4, 8, and 12. Each serum sample will be tested for TA90-IC, S-100 and MIA. The MIA will be measured by one-step ELISA with two monoclonal antibodies (MAB 1A12 and MAB 2F7; Boehringer Mannheim). S-100 in serum samples will be determined by an immunoluminescent assay kit from Byk-Sangtec Diagnostics. TA90-IC will be tested as previously described Prognostic assays based on serum DNA micro satellites with lo ss of heteroz ygosity (LOH) Dr. Hoon s laboratory initially discovered that free tumor-specific circulating DNA can be detected within the plasma/serum of melanoma patients. 42A These DNA markers increase in number with disease progression. Tumor DNA markers are not detected in normal, healthy donors. Chromosome instability (allelic imbalance) is frequent and has specific patterns in melanoma tumors. We have established that these allelic imbalances (DNA LOH) and BRAF mutations can be detected in serum/plasma. Preliminary data suggest that serum LOH and BRAF mutation markers may be highly specific prognostic markers for disease recurrence. These DNA markers in AJCC stage III melanoma patients have been correlated with poor prognosis. The studies are designed to use these unique surrogate molecular markers to predict disease recurrence. The serum Page 17 of 66

20 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 from 300 MSLT-II subjects who have histopathology-positive SNs (n=150) or histopathologynegative but RT/PCR-positive SNs (n=150) will be collected and cryopreserved pre-lm/sl and at months 0, 4, 8, and 12. Serum DNA markers will be evaluated; if the results are highly significant, the study will be extended to include additional MSLT-II subjects. This model will be used to determine whether LOH is useful as an independent prognostic factor in a multicenter analysis Serum proteomics Serum proteomics of tumor-bearing patients and follow up of high-risk benign tumors has been shown to be predictive of disease outcome. We will assess serum proteomics to identify potential new markers using specific affinity Protein Chips. These studies will be further followed up to identify specific proteins of predictive value. The initial phase of the pilot trial will be on the blood specimens used for LOH analysis of subjects histopathology or RT-PCR positive SN. 2.7 Interaction Between Tumor Burden and Endogenous Immune Response as a Determinant of the Clinical Outcome in Melanoma Increasing tumor burden, whether indicated by increasing Breslow thickness of the primary melanoma or by increasing volume of tumor in the SN, has been consistently correlated with increased risk of metastasis to SN and NSN, increased levels of immune suppression in the SN as compared to NSN, increased recurrence and decreased survival. Yet patients with identical clinical characteristics may have dramatically different clinical courses, with one patient surviving 10 years while another expires within 1 or 2 years. We hypothesize that the endogenous host immune response to tumor antigens present in the autologous melanoma may be in part responsible for these differences in clinical course. Because the quantification of tumor burden by immunologic and molecular tumor marker assays measures only one aspect of the tumor host relationship, we hypothesized that the endogenous immune response induced by immunogenic tumor antigens in the patient s neoplasm might retard the tumor growth and influence the clinical outcome. We have evidence that the growth of autologous tumor in the patient stimulates a wide range of immune responses to endogenous antigens present on the neoplasm. We have summarized below evidence for the endogenous immune response to melanoma differentiation antigens, TA-90, and ganglioside antigens. Our hypothesis is that there is a natural, ongoing endogenous immune response to cancer antigens of the growing neoplasm, and the effectiveness of the host in mounting this immune response determines the long-term clinical result of surgical therapy in patients whose melanomas have the capacity to metastasize. Antibodies to tumor antigens also may be indicators of subclinical tumor burden but not modulators of the tumor host relationship. In this case, the tumor-associated antigens are not capable of rendering the cell vulnerable to cytotoxic immune attack by host defenses Melanoma Differentiation-Associated Antigens We assessed antibody responses to potential melanoma differentiation antigens (Huang et al. 43 ). The open reading frame sequences of multiple melanoma-associated antigens have been cloned and expressed as recombinant proteins in E. coli. Purified recombinant antigens were employed to detect antibodies in sera of melanoma patients and normal healthy donors. In melanoma patients, sera were tested 1-4 months following surgery for stage II/III disease. By affinity ELISA and Western blotting, all recombinant antigens were shown to be immunogenic. The main subclass of Page 18 of 66

21 MORD-LM/SL-CLND-1102 Original January 2, 2003 Amendment #1 May 18, 2004 antibody response to these antigens was IgG. Most importantly this study demonstrated endogenous anti-trp-2 and anti-gp100/pmel17 IgG responses in melanoma patients. Differences between normal donors and melanoma patients were significant, with p of < The results suggest that melanoma differentiation antigens are autoimmunogenic and can be used as potential surrogates for evaluating host specific immunity in melanoma patients Endogenous TA-90 Immune Response and Survival It has been determined that the endogenous immune response to tumor-associated antigen TA-90 correlates with survival independent of standard prognostic factors in early-stage melanoma patients. Despite complete surgical resection, certain patients with early-stage melanoma will develop metastatic disease. Currently there are no means for predicting which patients with earlystage lesions will develop metastases. TA-90 is expressed by melanoma cells, and the immune response to TA-90 can be detected in the sera of melanoma patients. The detection of elevated IgM anti-ta-90 antibody titers has been shown to correlate with survival in response to vaccine therapy. However, the ability of these assays to predict survival in early-stage melanoma patients independent of vaccine therapy has not been previously investigated in a multicenter trial. Litvak et al. 44 determined whether the endogenous immune response to TA-90 correlates with survival in early-stage melanoma patients independent of other prognostic factors, such as tumor thickness, level of invasion, ulceration, site of primary lesion, gender and age. Two groups of patients with mm thick melanoma were selected randomly from the JWCI database of over 11,000 melanoma patients. The first group of patients (Group 1; n=50) died of metastases within 7 years after complete surgical treatment. The second group of patients (Group 2; n=50) did not develop recurrence and lived at least 10 years. These groups were matched for Breslow thickness, Clark s level, ulceration, site of primary, gender, and age. Archival subject sera collected postoperatively were assessed in a blinded manner for anti-ta-90 IgM and IgG responses by colorimetric ELISA assays. The median anti-ta-90 IgM titer was significantly lower in Group 1 (1:150) than Group 2 (1:800) (p=.0001). Patients in Group 1 also were less likely to have a positive (>1/800) IgM response (12%) versus patients from Group 2 (62%) (p=.001). Patients with a high IgM exhibited enhanced survival (Figure 6). The detection of the absence of endogenous immune response in melanoma patients may help predict outcome and identify patients who are at high risk of recurrence and need adjuvant therapy. Page 19 of 66