Sentinel Lymph Node Biopsy for Cutaneous Melanoma: The Stanford Experience, 1997-2004

Objective  To review sentinel lymph node (SLN) data from Stanford University Medical Center from January 1, 1997, to January 1, 2004, including rates of SLN positivity according to 2002 American Joint Committee on Cancer (AJCC) tumor classification, relation to other clinical and pathologic prognostic factors, and rates and sites of melanoma recurrence.

Design  Retrospective case series.

Setting  Stanford University Medical Center and Stanford melanoma clinics.

Patients  A total of 274 consecutive patients with primary melanoma who underwent SLN biopsy (SLNB) between January 1, 1997, and January 1, 2004, or who were referred to the Stanford melanoma clinics after SLNB and were followed up through March 2005.

Interventions  All patients underwent standard wide local excision of their primary tumors and SLNB with intradermal injection of isosulfan blue dye and/or technetium sulfur colloid.

Main Outcome Measure  Rates of SLN positivity per 2002 AJCC tumor classification, relation to other clinical and pathologic prognostic factors, and rates and sites of melanoma recurrence in node-negative and node-positive patients.

Results  Positive SLNs were detected in 39 (15%) of 260 cases, including 0 (0%) of 45 for cutaneous melanomas 1.0 mm thick or less (T1), 21 (18%) of 115 for melanomas 1.01 to 2.0 mm thick (T2), 12 (19%) of 64 for melanomas 2.01 to 4.0 mm thick (T3), and 5 (16%) of 32 for melanomas thicker than 4.0 mm (T4). Median Breslow depths were 1.89 mm for SLN-positive biopsy specimens and 1.50 mm for SLN-negative biopsy specimens (P = .07). The recurrence rate was 46% among SLN-positive patients, with a median time to recurrence of 8 months. Bivariate analysis revealed SLN positivity to be associated with AJCC tumor classification (P = .02), location on the trunk (P = .03), and presence of ulceration (P = .03). By multivariate logistic regression, ulceration (P = .01) was predictive of SLN positivity, whereas SLN status (P< .001), ulceration (P = .02), and location (P = .03) were predictive of recurrent disease.

Conclusion  Data from the past 8 years confirm the accuracy and prognostic value of SLNB in cutaneous melanoma and the low rate of regional nodal recurrence for SLN-negative patients.

The incidence of melanoma is rising more rapidly than that of any other cancer.¹ The lifetime incidence of melanoma in the United States is 1 in 68.² Although all stages of melanoma are increasing in incidence, earlier diagnosis has contributed to an increasing proportion of thin, early-stage melanoma with greater likelihood for cure.³

Sentinel lymph node (SLN) status is the most powerful prognostic indicator for recurrence and survival for individuals diagnosed as having primary melanoma.^4-6 With minimal morbidity, SLN biopsy (SLNB; also known as sentinel lymph node dissection [SLND]) reliably detects subclinical nodal involvement,^7-10 provides accurate and cost-effective¹¹ staging, and identifies patients who may benefit from standard adjuvant or experimental therapy.^3,12 Whether SLNB confers a survival benefit awaits confirmation from several ongoing clinical trials,¹³ including the Multicenter Sentinel Lymphadenectomy Trial,⁷ the Sunbelt Melanoma Trial,¹⁴ and the Florida Melanoma Trial.¹⁵

A number of studies have sought to identify predictors of SLN status, with variable findings.^4,16-29 Factors identified as predictors of SLN positivity include ulceration,^{4,16-20,23-27,29} tumor location,^4,16,20,28 patient age,^{16,17,20,22,27} tumor mitotic rate,^20-23,25 and tumor subtype.^27,28,30-34 Increasing tumor thickness has uniformly been associated with SLN positivity.^16-21 Most series report SLN positivity at less than 5% for melanomas 1.0 mm thick or less (T1), 8% to 12% for melanomas 1.01 to 2.0 mm thick (T2), 20% to 30% for melanomas 2.01 to 4.0 mm thick (T3), and more than 30% for melanomas thicker than 4 mm (T4),^{16,20,30,35-37} with a median thickness of 3.0 to 4.0 mm for primary melanoma with positive SLNs.^4,16,30,35 At Stanford University Medical Center (SUMC), we analyzed our SLNB experience during an 8-year period, including rates of positivity per 2002 American Joint Committee on Cancer (AJCC) tumor classification,³⁸ relation to other clinical and pathologic prognostic factors, and rates and sites of melanoma recurrence in node-negative and node-positive patients.

Following human subjects approval from SUMC for participation in the multicenter SLN Working Group Data Acquisition Project, a retrospective cohort review was conducted of patients who underwent SLNB as part of melanoma staging from January 1, 1997, through January 1, 2004, and were followed up in the Stanford Multidisciplinary Melanoma Clinic or Pigmented Lesion and Cutaneous Melanoma Clinic through March 31, 2005. Individuals who underwent SLNB after January 2004 were excluded to maximize length of follow-up. A total of 274 patients were identified, 255 (93%) of whom had their SLNB performed at SUMC. In 19 cases, SLNBs were performed at other institutions, and SUMC pathologists reviewed all primary tumor and SLN histopathologic specimens when care was initiated at SUMC. The SLNs were identified in 241 (95%) of 255 SLNBs performed at SUMC and 19 of 19 outside SLNBs for a total of 260 SLNBs, which comprised the study sample.

Consistent with published guidelines,^6,38-40 SLNB was recommended for pathologic staging of the regional nodal basin(s) in patients with primary melanoma 1.0 mm thick or more or less than 1.0 mm in cases in which adverse histopathologic prognostic factors were present, such as ulceration, regression, or Clark level IV/V. Completion LND (CLND) was recommended for cases in which positive SLNs were identified. Clinicopathologic characteristics were analyzed, including sex, age, tumor location, Breslow depth, ulceration, histologic subtype, and mitotic rate. Cases with recurrence were classified according to anatomic location of first recurrence, including local or cutaneous, in-transit, regional nodal, and visceral. Local recurrence was defined as satellite recurrence within the excision scar, excluding persistent disease type. In-transit recurrence was defined as dermal lymphatic recurrence between the primary scar and regional nodal basin. Regional nodal recurrence was defined as recurrence in the regional lymph nodal basin and visceral recurrence as distant skin, nodal, and/or systemic metastasis. Attempts were made to contact all patients with 2 years or less of documented follow-up to determine disease-free vs recurrent status.

All patients underwent standard wide local excision of their primary tumors and SLNB with intradermal injection of isosulfan blue dye and/or technetium sulfur colloid. Informed consent was obtained for all procedures. Of the 255 SLNBs performed at SUMC, 193 (76%) were performed with both isosulfan blue dye and technetium sulfur colloid, 41 (16%) with radiocolloid alone, and 4 (2%) with blue dye alone. Data regarding the types of tracers used in 17 SUMC cases and the 19 outside cases were not available. When radiocolloid was used, preoperative lymphoscintigraphy was performed and intraoperative mapping was obtained with a handheld gamma probe. Counts were measured in vivo and ex vivo. The SLNs were defined as any blue-stained nodes and/or nodes with radioactive counts more than 2 times background as measured by the gamma counter in counts per second, and/or counts greater than 10% of the most radioactive node ex vivo.^41-43

Primary tumor and SLN specimens were examined in the SUMC Department of Pathology. The SLNs were serially sectioned and fixed in paraffin, and each section was then analyzed on 2 hematoxylin-eosin–stained levels. Immunohistochemical studies, including S100, HMB-45, and Melan-A (Dako, Carpinteria, Calif), were performed for further evaluation in cases where the routine hematoxylin-eosin stain did not reveal SLN metastases (Ventana Medical Systems, Tucson, Ariz). Appropriate positive and negative controls were included. Histopathologic characteristics, including tumor thickness, ulceration, melanoma subtype, and SLN status, were available for most cases. When available, for all SLN-positive melanoma and all thickness-matched SLN-negative cases, slides were retrospectively re-reviewed by a single dermatopathologist (A.U.) to confirm the tumor mitotic rate when previously reported and to obtain tumor mitotic rate when not previously reported. Mitotic figures were counted per millimeter squared, beginning in the fields with the greatest number of mitotic figures.

Categorical variables were analyzed by χ² tests. Continuous variables were analyzed by t tests. For proportional data, χ² tests for proportions were conducted. Kruskal-Wallis tests were run on rank scores. Multivariate logistic regression for clinicopathologic features predictive of SLN status was performed using ulceration (present vs absent, excluding cases with missing ulceration status), Breslow depth (excluding cases with unknown Breslow depth), tumor location (trunk vs nontrunk), age, and sex. Multivariate logistic regression for clinicopathologic features predictive of recurrent disease was performed, using SLN status, ulceration (present vs absent, excluding cases with missing ulceration status), Breslow depth (excluding cases with unknown Breslow depth), tumor location (trunk vs nontrunk), age, and sex. Breslow depth and age were treated as continuous variables in multivariate analyses. Analyses were performed using SAS statistical software, release 9.1 (SAS Institute Inc, Cary, NC). P<.05 was considered statistically significant.

Table 1 gives the characteristics of all 260 patients with SLNs identified. There were slightly more male patients than female patients (60% vs 40%). The median age of patients at SLNB was 55 years (mean age, 56 years; age range, 7-85 years). Median follow-up for survivors was 29 months (30 for SLN-negative patients and 25 for SLN-positive patients). Melanomas were nearly equally distributed on the trunk (38%) and the extremities (40%) but were less common on the head and neck (22%). Median Breslow depth was 1.50 mm (mean, 2.30 mm). Overall, one fourth of tumors were ulcerated. Half of the tumors were of the superficial spreading subtype (50%), whereas 11% were nodular and 10% desmoplastic. Cases with SLNB performed at outside institutions had similar sex (32% females), age (median, 52 years), follow-up (median, 36 months), tumor location (26% head and neck, 32% trunk, and 42% extremity), tumor depth (median, 1.90 mm), and ulceration (37% known ulcerated and 37% not specified) compared with SLNBs performed at SUMC.

Table 2 provides the incidence of SLN positivity according to various clinicopathologic features. Overall, positive SLNs were detected in 39 (15%) of 260 cases. None of the 45 melanomas 1.0 mm thick or less had positive SLNs. However, 21 (18%) of 115 melanomas between 1.01 and 2.0 mm thick were associated with positive SLNs (two thirds of which were ≤1.5 mm). Similar rates of SLN positivity were noted in melanomas 2.01 to 4.0 mm thick (19%, 12/64) and more than 4.0 mm thick (16%, 5/32). Median depths of primary tumors were 1.89 mm (mean, 2.55 mm) for SLN-positive cases and 1.50 mm (mean, 2.25 mm) for SLN-negative cases (P = .07 for medians). Rates of SLN positivity per 2002 AJCC tumor classification for cases with SLNB performed at outside institutions were as follows: 0% (0/2) for T1, 38% (3/8) for T2, 40% (2/5) for T3, 33% (1/3) for T4, and (100%) 1/1 for melanoma of unknown depth; the overall rate of SLN positivity was 37% (7/19).

Recurrence rates were evaluated in both SLN-positive and SLN-negative patients. Overall, 48 patients developed recurrences at a median time of 13 months. Although 30 (14%) of 221 SLN-negative patients experienced tumor recurrence (median, 17 months), 18 (46%) of 39 SLN-positive patients experienced tumor recurrence (median, 8 months) (P<.001 for recurrence rate, P = .06 for time).

Among melanomas 1.01 to 2.0 mm thick, the recurrence rate was 10% (9/94) for SLN-negative cases compared with 38% (8/21) for SLN-positive cases. Among melanomas more than 2.0 mm thick, the recurrence rate was 24% (19/79) for SLN-negative cases compared with 53% (9/17) for SLN-positive cases.

We investigated the proportion of the recurrences among all SLN-negative patients that were regional nodal recurrences (vs local or visceral) and whether performance of SLNB at outside institutions affected recurrence rates. Table 3 gives the sites of first recurrence. The recurrence rate in the regional lymphatic basin(s) of all patients with negative SLNs was low (4.5%, 10/221). In 2 of the 10 cases of regional nodal recurrence among SLN-negative melanomas, the SLNBs were performed at outside institutions. For SLNBs performed at SUMC, the rate of regional nodal recurrence for patients with negative SLNs was 3.8% (8/209). In one third (2/6) of the SLN-positive cases in which regional nodal recurrences occurred, patients had declined CLNDs. Overall, 31 (79%) 39 patients found to have positive SLN(s) had subsequent CLNDs. Additional lymph nodes involved by tumor were identified in 7 (23%) of 31 of these cases on subsequent CLND.

Of SLN-positive cases with SLNB performed at SUMC, 28 (88%) of 32 had CLND. Reasons that patients declined CLND included interim worsening of comorbid conditions and concerns regarding postsurgical complications such as lymphedema. The recurrence rate for SLNBs performed at outside institutions was 53% (10/19) (overall 6 of 7 SLN-positive cases recurred compared with 4 of 12 SLN-negative cases; specifically, 0 of 2 T1 tumors recurred compared with 3 of 8 T2 tumors, 6 of 8 T3/T4 tumors, and 1 of 1 with unknown depth), with recurrences at a median time of 8 months. Sites of first recurrence were local (n = 8), in-transit (n = 3), regional nodal (n = 12), and visceral (n = 15) for SLNBs performed at SUMC and local (n = 2), in-transit (n = 0), regional nodal (n = 4), and visceral (n = 4) for SLNBs performed at outside institutions. The overall rate of recurrence in cases performed at outside institutions was significantly higher than for SUMC cases (53% vs 16%, P<.001).

Head and neck melanoma was particularly likely to recur (18/57, 32%), including 13 of 52 SLN-negative cases and 5 of 5 SLN-positive cases. This comprised 13 of the 31 recurrences among all SLN-negative cases and 5 of the 18 recurrences among all SLN-positive cases.

Using bivariate analyses, we evaluated to what degree clinicopathologic features other than tumor thickness correlated with SLN positivity (Table 2). Ulceration (P = .03) was significantly correlated with SLN positivity. Head and neck, trunk, or extremity location reached borderline significance (P = .08). Post hoc analysis demonstrated that trunk vs nontrunk location (P = .03) was significantly correlated with SLN positivity. Although the rate of SLN positivity among head and neck melanoma was only 9% (5/57), 11 (19%) of 57 head and neck cases were desmoplastic melanoma. Desmoplastic subtype (P = .11) was associated with SLN negativity but did not reach statistical significance. Male sex (P = .04) was also associated with SLN positivity. Regarding age, overall, 6 (23%) of 26 patients younger than 35 years had a positive SLNB result. Older patients (>60 years) had lower rates of SLN positivity, despite being more likely to have thick tumors (more than half had T3 or T4 melanoma compared with 28% of patients 60 years or younger). However, there were no statistically significant differences in SLN status when comparing patients younger than 35, 35 to 60, and older than 60 years.

The mitotic rate was inconsistently or incompletely reported in pathology reports for primary tumors. After reanalysis of available archived slides, only 27 (69%) of 39 SLN-positive cases had available information on tumor mitotic rate. Twenty-two (81%) of 27 of these had 1 or more mitoses per square millimeter. In the absence of complete data on tumor mitotic rate, we compared the tumor mitotic rate of SLN-positive melanoma to a sample of SLN-negative cases matched for Breslow depth. The SLN-positive cases had a higher median tumor mitotic rate (2.0 vs 1.0 mitoses per square millimeter, P = .07) than thickness-matched SLN-negative cases.

Multivariate logistic regression for clinicopathologic predictors of SLN status (Table 4) demonstrated statistical significance only for ulceration (odds ratio, 2.82; 95% confidence interval, 1.28-6.23; P = .01). Multivariate logistic regression could not be performed with Breslow depth modeled as 1 mm or less vs more than 1 mm, because there were 0 cases of SLN-positive melanoma less than 1 mm in our data set. Multivariate logistic regression for clinicopathologic predictors of recurrent disease (Table 5) demonstrated statistical significance for SLN status (P< .001), ulceration (P = .02), and location (P = .04).

We retrospectively analyzed a series of 274 consecutive patients with primary cutaneous melanoma who underwent SLNB at SUMC (or post-SLNB referral to the Stanford melanoma clinics) from January 1, 1997, to January 1, 2004. Subsequent follow-up was extended through March 31, 2005, to allow for meaningful analysis of disease recurrence. Our analysis focused on rates of SLN positivity in relation to clinical and pathologic prognostic factors, as well as rates and sites of melanoma recurrence. Comparable to similar series in the literature, our data confirm the accuracy and prognostic value of SLNB in primary cutaneous melanoma, as well as the low rate of regional nodal recurrence for SLN-negative cases. Our institutional experience is consistent with the accepted practice of SLNB as a staging technique for patients with melanoma 1.0 mm thick or more.

Consistent with other published data, SLN status was the strongest predictor of recurrence.^4-6 In turn, primary tumor ulceration was most predictive of SLN positivity and was also independently predictive of disease recurrence. The rate of regional nodal recurrence among SLN-negative patients was 3.8% for SLNBs performed at SUMC, similar to other series in the literature when considering follow-up time.^35,36,44-47 Compared with other populations, the SUMC population is unusual in several ways, including having greater numbers of head and neck melanoma (22% vs 0%-16%),^{4,23,24,26,28-31,35,36,44-46,48-51} more cases of desmoplastic melanoma (10% vs 0%-2%),^{19,26,28-31,44,45,50,52} fewer cases of nodular melanoma (11% vs 19%-52%),^{19,26,29-31,44,45,50,52,53} and older patients (median age, 55 vs 47-52 years).^{16,17,23,24,26,28,30,36,46,48,50,53} Moreover, among melanomas more than 2.0 mm thick, the rate of ulceration (36%) was unusually low compared with other series,^5,54,55 which demonstrate 47% ulceration for melanomas 2.01 to 4.0 mm deep and 63% to 73% ulceration for tumors larger than 4.0 mm.

Analysis of our institutional experience has several limitations. As a retrospective single institution experience, the statistical power of our study is limited by sample size and follow-up. Although our cohort extends across 8 years, the median follow-up for survivors was 29 months. This is comparable to a number of series in the literature that comment on recurrence.^35,44,46,47 Nevertheless, longer follow-up is needed to confirm our findings. The limitations of our data underscore the need for larger cohorts, multicenter collaboration for data collection, and completeness of data sets for clinicopathologic characteristics and prognostic factors such as mitotic rate. For example, for patients 35 years or younger, increasing mitotic rate has recently been shown to better predict SLN positivity than increasing Breslow depth.²⁰

With minimal morbidity, SLNB provides early detection of occult regional nodal metastatic melanoma. Although the prognostic role of SLNB as a staging technique is established, controversy remains over its therapeutic benefit. By providing regional nodal staging, SLNB identifies patients who may benefit from standard or experimental systemic adjuvant therapy. The best evidence that suggests that SLNB in conjunction with surgical therapy may offer a survival benefit comes from the Intergroup Melanoma Trial,^56-58 a multicenter, randomized trial of more than 700 subjects with intermediate-thickness melanoma, comparing elective LND (ELND) (lymphoscintigraphy-directed immediate LND at the time of wide local excision of the primary melanoma) in subjects with clinically nonpalpable lymph nodes to observation, with therapeutic LND performed at the time of gross nodal involvement. The trial demonstrated improved 10-year survival for ELND with preoperative lymphoscintigraphy in prospectively stratified subsets of patients with nonulcerated melanomas (n = 543; 84% vs 77% survival; P = .03; 30% reduction in mortality), melanomas 1 to 2 mm thick (n = 446; 86% vs 80% survival; P = .03; 30% reduction in mortality), or melanomas on the trunk (n = 385; 84% vs 78%; P = .05; 27% reduction in mortality).

Moreover, the World Health Organization Melanoma Trial No. 14,⁵⁹ which randomized 240 patients to immediate ELND vs delayed LND in the setting of clinically palpable metastasis, demonstrated improved 5-year survival for cases in which microscopic regional nodal disease was identified by ELND compared with patients who underwent delayed therapeutic lymphadenectomy for macroscopic or palpable regional nodal metastasis (48% vs 27%, P = .04). However, interpretation of this study is limited by the lack of preoperative lymphoscintigraphy ELND study arm.

Although the independent prognostic information gathered from SLNB is not refuted, some argue against routinely using SLNB for patients with primary cutaneous melanoma deeper than 1.0 mm.^60-63 Those against the procedure emphasize the lack of randomized controlled trials that demonstrate therapeutic value alone or in combination with adjuvant medical or surgical therapy, including interferon or CLND for SLN-positive disease.^60-62 Some also question the theoretical basis of the procedure, suggesting that metastatic cells are not delayed for a significant duration in the sentinel node to allow for complete capture and removal via SLNB.^60,61,63 It has been debated whether the peace of mind offered by SLNB as a prognostic procedure is worth its morbidity and cost.⁶³ Forthcoming data from the national, multicenter trials should assist in determining the therapeutic impact of SLNB on patient outcome; in any case, careful discussion between practitioner and patient regarding the pros and cons of the SLNB procedure is of utmost importance.

At SUMC, our recommendations regarding SLNB always involve careful consideration of the extent of surgery involved; potential complications, including increased scar formation, extremity edema, and seroma formation (although relatively uncommon); and significant medical comorbidities or advanced age, which may preclude patients from standard or experimental systemic adjuvant therapy. We believe that all of these factors need to be taken into account before recommending the procedure. Sentinel lymph node biopsy should be conducted according to strict protocols and involve experienced institutions and physicians.

It is difficult to interpret differences between cases performed at SUMC and those at outside institutions given selection bias and small sample size. Our analysis of cases in which SLNB was performed at outside institutions demonstrated significantly higher overall rates of recurrence. However, detailed analysis of these cases demonstrated a similar prognostic importance of ulceration and Breslow depth on SLN status and of ulceration, Breslow depth, and SLN status on disease-free survival. Furthermore, review of cases performed at outside institutions demonstrated that many of the recurrences occurred before transfer of care to SUMC. It is our experience that expertise in SLNB is associated with a significant learning curve and requires a multidisciplinary approach and team of specialized physicians in the fields of dermatology, surgery, nuclear medicine, dermatopathology, and medical oncology. Therefore, it may best be suited for academic and/or large multispecialty centers. Because weekend courses are increasingly available for training in the SLNB technique, complication and recurrence rates may differ between established melanoma centers and community-based hospitals less experienced with the technique or with fewer cases of melanoma.

Our SLNB experience confirms the prognostic value of SLNB in primary cutaneous melanoma and its use as a valid staging technique for patients with melanoma 1.0 mm thick or more. Our data document a low rate of regional nodal recurrence for SLN-negative cases, comparable to similar series in the literature.^35,36,44-47 In the absence of completed clinical trials, single-institution experiences add value by demonstrating the reliability and reproducibility of this procedure. The potential survival benefit of SLNB will be clarified by the results of several ongoing clinical trials,¹³ including the Multicenter Selective Lymphadenectomy Trial,⁷ the Sunbelt Melanoma Trial,¹⁴ and the Florida Melanoma Trial.¹⁵

Correspondence: Susan M. Swetter, MD, Department of Dermatology, Stanford University Medical Center, 900 Black Wilbur Dr, Room W0069, Stanford, CA 94305 (sswetter@stanford.edu).

Accepted for Publication: April 23, 2004.

Acknowledgment: We acknowledge the assistance of Christle Layton, BS, in patient follow-up.

Financial Disclosure: None.