Survivin Expression in Sentinel Lymph Nodes From Melanoma Patients TO THE EDITOR: We have read with interest the report by Gradilone et al 1 in the January 15, 2003 issue of the Journal of Clinical Oncology. It provides evidence about the correlation of survivin expression in melanoma sen- tinel lymph nodes and patients’ clinical outcome. Gra- dilone et al included 36 patients with different melanoma American Joint Committee on Cancer (AJCC) stages, from stage IA to III. Patients with sentinel nodes, in which survivin expression was detected, had a worst outcome: 72% of the samples were positive for survivin, and among these patients, 61.5% died or progressed versus 0% among survivin-negative patients. Although we congratulate the original finding, we believe that some considerations must be taken into account. First, although a role of apoptosis for melanoma genes (including the survivin gene) has been poorly stud- ied in primary and metastatic melanoma, the authors started investigating their role as a prognostic marker in melanoma sentinel nodes, furthermore including hema- toxylin and eosin–negative lymph nodes in their analy- ses. Although tyrosinase was positive in all cases tested, this does not mean that melanoma cells were present in all nodes. Tyrosinase containing nevocytes that have migrated from cutaneous nevi are found in the capsule of the lymph nodes of 24% of patients undergoing lymphadenectomy. Schwann cells of nerves that traverse lymph nodes could also express tyrosinase. Additionally, macrophages that have scavenged partially melanized melanosomes from melanoma cells disrupted in tissues peripheral to the lymph nodes or within the node itself may contain detectable m-tyrosinase. For these reasons, the finding of an enhanced signal for m-tyrosinase after reverse transcriptase polymer- ase chain reaction (RT-PCR) of a lymph node does not guarantee that the source of this signal is a metastatic mel- anoma cell. Immunohistochemical staining was carried out to identify the cell histotypes within the metastatic lymph nodes that were positive for survivin, bax, and bcl-2. Au- thors found that all three proteins were positive, corre- sponding with the infiltrating cells, whereas lymphocytes were totally negative. It is not clear, in our opinion, how this could be so, since Table 1 of the Gradlione et al 1 article shows that only four AJCC stage III patients were included, and immunohistochemical staining was per- formed on only one stage III case, the result being nega- tive for the three proteins. Second, Gradilone et al explain in their methods section that tyrosinase and melanoma-inhibitory activity expression had been investigated without any further explanation of re- sults of the correlation with clinical outcome of these genes, or if it could be a possible preselection test before survivin RT-PCR. Multiple studies have analyzed the upstaging value of multiple-marker RT-PCR. Ten percent to 47% of patients with histologically-negative sentinel nodes were upstaged using multimarker techniques. 2-4 The use of multiple mel- anoma markers, such as tyrosinase and MAGE, in sentinel nodes has been shown by other investigators to have a significant correlation with outcome 5-6 Third, although 52.9 months of median follow-up seems enough, patients with a short follow-up were also included (19, 19, 20, and 20 months, respectively). Further- more, at the time of performing the statistical survival analysis, Gradilone et al used Fisher’s test instead of the Kaplan-Meier method. Fisher’s test does not consider follow-up time. So if you do not include time in statistical analysis, it could be possible that patients with longer follow-up are distributed in the survivin-positive group, while patients with a very short follow-up are in the survivin-negative group. In this study, for the 10 patients with survivin-negative results, seven had a follow-up time shorter than the median follow-up time (42, 44, 46, 46, 47, 49, and 51 months, respectively), so it could be possible that with a similar follow-up time in both groups, no difference in progression-free survival would be found between posi- tive and negative survivin cases. Despite these theoretical considerations, a new per- spective in melanoma, as in other tumors, is presented here; not only could the presence of melanoma cells in sentinel lymph nodes be the reason for a worse outcome, but the malignant biologic potential of the tumor cells could be an important factor. Further investigation in this setting must be performed to know the real role of these genes as prognostic markers. A higher number of patients with a long follow-up time could be included, making a segregated analysis for hematoxylin- eosin–positive and –negative nodes. Maria Gonza ´lez Cao, Susana Puig, and Begon ˜a Mellado Oncology Department, IDIBAPS, Hospital Clinic, Barcelona, Spain ■■■ JOURNAL OF CLINICAL ONCOLOGY C O R R E S P O N D E N C E VOLUME 22  NUMBER 13  JULY 1 2004 2751 Journal of Clinical Oncology, Vol 22, No 13 (July 1), 2004: pp 2751-2756