ORIGINAL ARTICLE. Cervical Sentinel Lymph Node Biopsy for Melanomas of the Head and Neck and Upper Thorax

Transcription

1 ORIGINAL ARTICLE Cervical Sentinel Lymph Node Biopsy for Melanomas of the Head and Neck and Upper Thorax Jeffrey D. Wagner, MD; Hee-Myung Park, MD; John J. Coleman III, MD; Charlene Love, RN; John T. Hayes, MS Objective: To describe a clinical experience with sentinel lymph node biopsy (SLNB) of head and neck nodal basins for clinical stage I melanomas draining to these areas. Design: Consecutive clinical case series with a mean follow-up of 10.7 months. Setting: University tertiary care referral medical center. Patients: Seventy patients with clinical stage I cutaneous melanoma who underwent SLNB of cervical and/or parotid lymph node basins. Interventions: Patients underwent same-day preoperative technetium Tc 99m lymphoscintigraphy followed by SLNB using gamma probe and blue dye (66 patients) and blue dye alone (4 patients). Patients with histological evidence of tumor (hereinafter positive ) according to SLNB results underwent modified cervical completion lymph node dissection, including parotidectomy as appropriate. Patients without histological evidence of tumor (hereinafter negative ) according to SLNB results were followed up clinically without undergoing completion lymph node dissection. Main Outcome Measures: The rates of SLNB success, SLNB positivity, completion lymph node dissection positivity, complications, and SLNB false-negative results were determined by clinical follow-up. Results: Locations of melanomas in the 70 patients were the face (n = 20), neck (n = 14), ear (n = 9), scalp (n = 9), and upper thorax (n = 18). Locations of basins that underwent biopsy (n = 104) were in the cervical (n = 68), parotid (n = 19), and axillary (n = 17) regions. The mean Breslow thickness was 2.1 mm (range, mm). Sentinel lymph node biopsy was successful in 103 basins (99%). The mean number of sentinel lymph nodes per basin was 2.5 (range, ). Positive sentinel lymph nodes were found in 12 patients (17%) and 15 basins (14%). Sentinel lymph node biopsy results correlated with the American Joint Committee on Cancer tumor stage (P =.05) and a Breslow thickness of 1.23 mm or greater (P =.03). Additional tumor-containing nodes were noted in 5 (42%) of the 12 patients who underwent completion lymph node dissection, and these results correlated with the presence of multiple positive sentinel lymph nodes (P =.01). There were complications in 3 patients (4%) (seromas in 2 patients and temporary spinal accessory nerve paresis in 1 patient). One nodal recurrence in a basin that was negative according to SLNB results (SLNB with blue dye only) was noted (false-negative rate, 2%). The results of SLNB were accurate in 69 patients (99%). Conclusions: Sentinel lymph node biopsy using lymphoscintigraphy and blue dye to manage cutaneous melanomas draining to the head and neck nodal areas is reliable and safe. Sentinel lymph node biopsy results correlated with a Breslow thickness of 1.23 mm or greater and the American Joint Committee on Cancer tumor stage. Completion lymph node dissection is recommended after determining positive SLNB results. Arch Otolaryngol Head Neck Surg. 2000;126: From the Indiana University Cancer Center Interdisciplinary Melanoma Program, Departments of Surgery and Plastic and Reconstructive Surgery (Drs Wagner and Coleman and Ms Love), Radiology (Dr Park), and Biostatistics (Mr Hayes), Indiana University School of Medicine, Indiana University Purdue University at Indianapolis. THE WORLDWIDE incidence of cutaneous melanoma continues to rise about 6% per year. 1 Melanoma of the skin is now the sixth most common cancer in the United States. 2 Melanoma rates are rising faster than any cancer in men and are second only to lung cancer in women. 1 An estimated new cases of invasive melanoma accounting for 7300 deaths were seen in the United States in By virtue of its metastatic potential, melanoma accounts for the vast majority of deaths from cutaneous malignancies. Management algorithms for cutaneous melanoma have undergone a paradigm shift during the past 5 years. Although melanoma remains a disease treated primarily by surgery, trends have been toward less radical excision, fewer elective lymph node dissections (ELNDs), and a more aggressive diagnostic approach. The advent of lymphatic mapping and sentinel lymph node biopsy (SLNB) to stage regional lymph nodes and the use of interferon as the first effective postsurgical adjuvant therapy have had a major impact on the care of melanoma in the United States and elsewhere. 313

2 PATIENTS AND METHODS The study design was a retrospective analysis of a consecutive clinical case series. The Indiana University Cancer Center Interdisciplinary Melanoma Program computerized database was searched to identify patients with clinical stage I cutaneous melanoma at Indiana University Medical Center, Indianapolis, and affiliated hospitals between November 1, 1994, and March 1, All patients who underwent SLNB of at least one cervical and/or parotid nodal basin to stage nonpalpable regional lymph nodes were identified. Hospital and outpatient clinical records were reviewed and pertinent demographic, radiological, surgical, clinical, and histopathological data were recorded for analysis. SLNB TECHNIQUE Early in this series, our SLNB protocol evolved from the use of blue dye alone to the use of both blue dye with intraoperative radiolocalization of SLNs using a handheld gamma probe. Regardless of harvest technique, all patients underwent preoperative lymphoscintigraphy to identify the basin(s) at risk for nodal disease. The unfiltered technetium Tc 99m sulfur colloid radiolabel (37-74 MBq [1-2 mci]) was injected intradermally in 2 to 4 divided doses at the tumor site 2 to 4 hours before surgery. Continuous imaging with a large field-of-view gamma camera was performed for 2 to 2.5 hours. After the induction of anesthesia, 0.5 to 1.0 ml of isosulfan blue dye (Lymphazurin; Zenith Parenterals, Rosemont, Ill) was injected intradermally at the site of the primary melanoma. All basins identified by lymphoscintigraphy were explored through limited incisions using the handheld gamma probe (C-Track; Care Wise Medical Products, Morgan Hill, Calif). In the parotid region, SLNB was performed with enucleation of the SLN without formal parotidectomy or full facial nerve dissection. All blue nodes were removed as SLNs. Ex vivo SLNs residual node basin radioactivity ratios were determined using the gamma probe. If necessary, additional radioactive nodes were removed as SLNs until the ratio of the most radioactive ex vivo SLN residual node basin was 10:1 or greater. If frozen-section or permanent-section analysis of the SLN demonstrated metastatic melanoma (hereinafter positive ), CLND was performed on the involved basin(s). For cervical basins, CLNDs were guided by the preoperative lymphoscintigram. Patients with melanomas of the face, neck, or scalp metastatic to the parotid nodes underwent total parotidectomy, sparing the facial nerve, and either supraomohyoid cervical or modified CLND. Patients with primary tumors in the head and neck in whom the SLNB results demonstrated metastatic disease to any level of the neck underwent spinal accessory nerve sparing cervical CLND without parotidectomy, unless the lymphoscintigram demonstrated drainage through the parotid gland area. Patients with melanomas of the trunk draining to the posterior triangle nodes underwent posterolateral cervical node dissection, sparing the submandibular triangle nodes. HISTOLOGICAL ANALYSIS OF SURGICAL SPECIMENS The SLNs that were grossly suspected of metatastic tumor were submitted for intraoperative frozen-section analysis. If the SLN was grossly unsuspected and frozen-section analysis did not show metastatic melanoma (hereinafter negative ), SLNs were fixed in formalin and submitted for 1-mm step sections. These sections were analyzed with hematoxylin-eosin stain. Sentinel lymph node specimens that were negative for metastases by this analysis had additional sections stained with immunohistochemical markers reactive to S100 protein and/or HMB-45 antigen. Non-SLN and CLND specimens were analyzed in a routine fashion after formalin fixation, with 1 to 3 sections from the node reserved for hematoxylin-eosin staining. STATISTICAL ANALYSIS Univariate analyses were performed to determine possible relationships between several independent variables and the dependent variables of the histological results of SLNB and CLND. We analyzed the following variables that were potentially predictive of SLN and CLND positivity: sex, age, tumor location, growth phase, tumor regression, histological lymphocytic tumor infiltrate, histological melanoma subtype, mitotic index, tumor ulceration, Clark level, Breslow thickness, American Joint Committee on Cancer (AJCC) 12 tumor stage, number of basins that underwent biopsy, number of SLNs harvested, and presence of multiple positive SLNs. The Pearson 2 test, Fisher exact test, and exact Mantel-Haenszel test were used for univariate analysis, as appropriate. A multivariate regression model was then devised to analyze the independent variables with the SLNB result (dependent variable). A univariate P.25 was required to be included in the multivariate model. Both backward and forward selection procedures, with a significance level of P<.10, were used. Classification tree analysis was performed to allow categorization of the continuous variable of Breslow thickness to identify an efficient cutoff point for predicting the SLNB positivity rate. Nodal metastases from melanoma have been shown to occur in a somewhat orderly fashion. 3-5 The sentinel lymph node (SLN) in regional lymph node basins can be harvested and analyzed to predict the histological status of the entire basin. The histological absence of metastatic melanoma in the SLN indicates absence of disease in the remaining nodal basin. Therefore, complete nodal staging can be achieved with SLNB results. Because nodal status is one of the most important predictors of survival in melanoma, SLNB has significant clinical implications for prognosis and therapy. Melanomas of the head and neck are considered to be more aggressive than those on the extremities, 6 and the management of patients with clinically unsuspected lymph nodes is more problematic. 7 A lack of convincing evidence for the survival benefit of traditional ELND, greater unpredictability of lymphatic drainage, and special anatomic considerations, such as the facial nerve within the parotid gland, have discouraged the elective 314

3 surgical staging of regional nodal basins in the head and neck. Sentinel lymph node biopsy is more challenging in the head and neck region than elsewhere. Several reports 7-11 of SLNB for head and neck melanomas have recently appeared in the literature. The purposes of this study are to describe our clinical experience with patients who underwent SLNB of cervical and parotid nodal basins for melanomas of the head and neck and upper thorax that drain to these areas; to confirm the success rate, accuracy, false-negative rate, and morbidity of SLNB in head and neck nodal basins; and to identify possible predictors of the results of SLNB and completion lymph node dissection (CLND) in this population. RESULTS Two hundred seventy-five consecutive SLNB procedures were performed in 275 patients with clinical stage I cutaneous melanoma to stage regional lymph node basins. The study population comprised 70 patients who underwent SLNB of at least one cervical or parotid lymph node basin, as directed by the results of preoperative lymphoscintigraphy. The mean age of the study population was 54 years (range, 4-92 years). Additional characteristics of the study population are detailed in Table 1. Seventy SLNB procedures were performed to stage 104 regional head and neck nodal basins in the study population (66 with blue dye plus gamma probe and 4 with blue dye only). The mean Breslow thickness was 2.1 mm (range, mm). At least 1 SLN was successfully localized and harvested in 103 basins (99%). The SLN could not be identified intraoperatively in 1 cervical basin (0.9%). A total of 320 SLNs were removed (mean SLNs per basin, 2.5; range, ). Twelve melanomas (17.1%) and 15 basins (14.4%) had positive SLNB results (3 patients each had 2 positive basins). The locations of the 15 positive basins were in the cervical (n = 9), parotid (n = 4), and axillary (n = 2) regions. Sentinel lymph node biopsy positivity rates for AJCC tumor stages T1 to T4 were 0%, 13%, 21%, and 38%, respectively. The mean Breslow thickness of 2.40 mm for SLNB positive melanomas was not significantly greater than 2.10 mm for SLNB-negative melanomas (t test, P =.31). Comparison of the study population with the 205 patients who underwent SLNB of melanomas draining only to the non head and neck basins is shown in Table 2. The comparison of the 2 groups showed that the Breslow thickness and proportion of ulcerated melanomas were similar. The head and neck SLNB group had a significantly greater proportion of male patients than the non head and neck group. In addition, the head and neck group had significantly more nodal areas that underwent biopsies, more total SLNs removed, more SLNs per basin removed, and more nodes removed at CLND than the non head and neck group. Patients undergoing head and neck SLNB were more likely to have a positive non-sln found in the CLND specimen than patients undergoing non head and neck SLNB (42% vs 24%), but this difference was not significant (P =.28). Univariate analyses of several patient and tumor categorical variables by SLNB result for the 70 melanomas are shown in Table 1. Sentinel lymph node biopsy positivity was positively correlated with increasing AJCC tumor stage. To further investigate the relationship of tumor thickness to SLNB result, we performed a classification tree analysis of tumor thickness. A classification tree analysis demonstrated that a Breslow thickness of 1.23 mm was the optimal cutoff point for predicting SLNB result. This cutoff point yielded a sensitivity of 0.92, a specificity of 0.43, a positive predictive value of 0.25, and a negative predictive value of 0.96 and was significantly associated with the SLNB result (P =.03). Table 1 also shows univariate analysis of a variety of other patient and tumor factors (independent variables) potentially associated with the SLNB result (dependent variable). Age was analyzed as a categorical variable using the mean age of 54 years as a cutoff point. The Clark level was analyzed as an ordinal variable, and the growth phase was analyzed as a categorical variable (vertical component present vs absent). The mitotic index was classified as high ( 6 mitoses per high-power field ), intermediate (3-5 mitoses per high-power field), or low ( 2 mitoses per high-power field) and analyzed as an ordinal variable. Because a trend toward significance was noted in the mitotic index as an ordinal variable (P =.09), we also analyzed mitotic index as a categorical variable (high vs intermediate and low; P =.06). Lymphocytic tumor infiltrate was graded as brisk, moderate, minimal or none, or absent and analyzed as an ordinal variable. Tumor ulceration and regression were graded as present or absent. For some variables, some of the values were unknown. Unless indicated, patients with unknown values were excluded from analysis. No variable was significantly associated with SLNB result, although a high tumor mitotic index showed a trend toward statistical significance (P =.06). Multivariate logistic regression was performed to demonstrate the possible relationship(s) between 8 variables with a univariate P.25 and the SLNB result. Tumor thickness was modeled in 3 different ways: as a continuous variable, as a categorical variable using AJCC tumor stage cutoff points, and with the classification tree cutoff point of 1.23 mm. By these multivariate analyses, only AJCC tumor stage and categorical Breslow thickness remained in their respective multivariate models (Table 1). Twelve patients with a positive SLNB specimen underwent CLND in 15 positive basins. Five patients (42%) had at least one additional positive non-sln in one CLND specimen each. The locations of the positive CLND basins were in the axillary (n = 2), parotid (n = 2), and neck (n = 1) regions. The relationship between patient and tumor variables and positive CLND result after positive SLNB result was investigated by univariate analysis (Table 3). Some patients had unknown values and were excluded from the analysis. Only the presence of multiple positive SLNs achieved statistical significance for predicting CLND result (P =.01). The morbidity rate for all SLNB procedures was 4% (3 of 70 patients). Two patients developed seromas at the SLNB site that required needle aspiration. One patient had a temporary spinal accessory nerve paresis that resolved in 4 months. No temporary or permanent facial nerve weakness occurred in relation to SLNB of the parotid region. Two of the 3 patients who underwent facial nerve sparing total parotidectomy after a positive result of SLNB of the parotid node had mild partial 315

4 Table 1. Distribution of Categorical or Ordinal Variables by Sentinel Lymph Node Biopsy (SLNB) Result in 70 Patients Who Underwent Head and Neck SLNB* No. (%) of SLNB Specimens P Variable Negative Positive Univariate Analysis Multivariate Analysis Age, y (categorical) (85) 5 (15) (81) 7 (19) Sex Female 18 (90) 2 (10) Male 40 (80) 10 (20) AJCC tumor stage T1 10 (100) 0 (0) T2 20 (87) 3 (13) T3 23 (79) 6 (21) T4 5 (62) 3 (38) Breslow thickness, mm (categorical) (96) 1 (4) (75) 11 (25) Histological subtype Desmoplastic 1 (100) 0 (0) Lentigo maligna 4 (100) 0 (0) Nodule 10 (83) 2 (17) Superficial spreading 34 (85) 6 (15) Unclassified 9 (69) 4 (31) Vertical growth phase Present 41 (79) 11 (21) Absent 17 (94) 1 (6) Mitotic index (ordinal) High 3 (50) 3 (50) Intermediate 12 (92) 1 (8) Low 39 (83) 8 (17) Unclassified 4 (100) 0 (0) Mitotic index (categorical) High 3 (50) 3 (50) Others 86 (55) 9 (14) Lymphocytic tumor infiltrate Brisk 8 (89) 1 (11) Moderate 11 (79) 3 (21) Minimal or none 24 (77) 7 (23) Indeterminate 15 (100) 0 (0) Clark level II 4 (100) 0 (0) III 16 (89) 2 (11) IV 35 (80) 9 (20) V 2 (67) 1 (33) Tumor ulceration Present 17 (74) 6 (26) Absent 24 (86) 4 (14) Tumor regression Present 2 (50) 2 (50) Absent 54 (86) 9 (14) Tumor location Face 17 (85) 3 (15) Neck 11 (79) 3 (21) Scalp 6 (67) 3 (33) Ear 9 (100) 0 (0) Upper thorax 15 (83) 3 (17) No. of SLNs removed 1 5 (71) 2 (29) Multiple 53 (84) 10 (16) No. of basins that underwent biopsy 1 34 (87) 5 (13) 2 23 (83) 5 (17) (33) 2 (67) *Unknown values were excluded from analysis unless indicated. Negative indicates results showed no evidence of tumor; positive, results showed evidence of tumor; ellipses; not applicable; AJCC, American Joint Committee on Cancer 12 ; and SLN, sentinel lymph node. Analysis performed using Fisher exact test. Analysis performed using Mantel-Haenszel test. Analysis performed using exact logistic regression test. Analysis performed using Pearson 2 test. 316

5 temporary facial nerve weakness that recovered spontaneously in each patient within 3 weeks. At the 10.7-month median clinical follow-up, 5 patients (7%) had developed recurrent melanoma and 3 had died of disseminated melanoma. Three patients had distant metastatic failure; one of these patients also had a recurrence in the axillary region in the basin that had a negative result of SLNB using blue dye only. Intransit metastasis occurred in one patient, and one nodal basin recurrence was noted in a patient after a positive cervical SLNB result followed by complete modified neck dissection. Therefore, the number of false-negative SLNB results in patients with drainage to head and neck nodal basins compared with clinical follow-up was 1 (1%). The SLNB procedure was accurate for staging in 69 (99%) of 70 patients and 103 (99%) of 104 regional nodal basins. COMMENT Morton et al 3 introduced the concept of SLNB in This concept has its origin from early work with cutaneous lymphoscintigraphy, which was initially proposed as a means of localizing lymph node basins for ELND. Research 3 in animals and humans has demonstrated well-defined pathways leading from each cutaneous territory (cutaneous lymphosomes) to specific regional nodes, the SLNs. Preoperative lymphoscintigraphy is critical for the success of SLNB in patients with truncal and head and neck melanomas. The lymphatic drainage patterns of the head and neck and upper thorax are complex and not easily predicted. 7-11,13 The value of traditional lymphoscintigraphy for definition of potentially equivocal or multiple draining basins has been clearly demonstrated. Without this test, nodal surgery may be misdirected in more than 50% of axial melanomas. 13 Sentinel lymph nodes were first demonstrated using vital blue dye injected intradermally at the site of the melanoma. Initially, SLNB using this technique was successful in only about 80% of patients undergoing the procedure. 3 Subsequently, several groups 4,5,7,14-17 validated the use of the gamma probe for radiolocalization of the SLN. In experienced hands, success rates now approach 100% because of improvements in preoperative lymphoscintigraphy and intraoperative mapping techniques, including the use of the gamma probe. Although there is some debate about the relative importance of dye vs the gamma probe, the independent evolution of SLNB at various centers supports the complementary use of both. 14,18,19 Analysis of the SLN was shown to be a sensitive indicator of metastatic melanoma in regional nodes. 3-5,20 The absence of metastatic melanoma in the SLN indicates a high likelihood of absence of disease in the remaining nodal basin. Therefore, complete nodal staging can be achieved with SLNB results. Sentinel lymph node biopsy provides staging information essentially equivalent to CLND with less morbidity. With a sensitivity of about 95% in experienced hands, 3 SLNB can virtually eliminate unnecessary lymph node dissections. For these reasons, SLNB has essentially replaced ELND as the technique of choice for staging regional lymph node basins at most major melanoma centers. This report confirms the diagnostic accuracy and low morbidity of SLNB for melanoma staging in the head and Table 2. Comparison of Patients in the Head and Neck Group With the Patients in the Non Head and Neck Group Who Underwent Sentinel Lymph Node Biopsy (SLNB)* Characteristic Head and Neck SLNB Group Non Head and Neck No. of patients NA Age, mean, y Sex, male-female 2.5:1 1.2:1.02 Breslow thickness, mean, mm Ulcerated tumor, % No. of SLNB basins, mean No. of SLNs removed, mean No. of SLNs per basin, mean Positive SLNB, % Positive CLND, % No. of CLND nodes *The P value was calculated using the t test. SLN indicates sentinel lymph node; positive, results showed evidence of tumor; CLND, completion lymph node dissection; and NA, not applicable. neck nodal basins. The low false-negative rate and high overall accuracy rate are similar to those in previous reports Our experience confirms findings of others that SLNB of intraparotid nodes can be safely performed without a formal parotidectomy. 9 Caution is advised in this endeavor, however. Our technique uses only blunt dissection with a fine-tipped hemostat, avoids cautery within the parotid gland, and is aided by a nerve stimulator and loupe (magnification 3.5). We recommend superficial parotidectomy if the SLN cannot be confidently localized and harvested with these maneuvers. The follow-up duration in this report was not sufficient to detect all possible melanoma recurrences in SLNB-negative nodal basins. Nonetheless, these data show that there was more morbidity, and extensive elective CLND and parotidectomy procedures can be replaced by SLNB without compromising the detection of occult metastatic disease. The SLNB positivity rate of 17% in this series is similar to that noted in other head and neck SLNB series. 7,9-11 The SLN positivity rate for all tumor sites in various reports 3-5,7-11,14-17,20-22 is between 12% and 36%. Generally, the SLNB positivity rate for head and neck melanomas seems to be slightly lower than that reported in series that included all tumor sites. Although the reason for this observation is unclear, it may be explained by differences in known prognostic factors between groups, which are possibly related to the generally lower threshold for performing SLNB for high-risk areas, such as the head and neck. A positive correlation between SLN positivity and tumor thickness has generally been observed; SLNB positivity rates are less than 5% for AJCC T1 melanomas and approach 50% for AJCC T4 lesions. 22 Thus, SLNB is likely to be a low-yield procedure in most thin melanomas. Conversely, a recent study 23 showed that nodal status is an independent significant prognostic factor in patients with thick melanomas. A rational argument can be made for staging thick melanomas, which ordinarily would not be treated with ELND, with SLNB. P 317

6 Table 3. Distribution of Categorical or Ordinal Variables by Completion Lymph Node Dissection (CLND) Result in 70 Patients Who Underwent Head and Neck Sentinel Lymph Node Biopsy (SLNB)* No. (%) of CLND Specimens Variable Negative Positive P (Univariate Analysis) Age, y (categorical) 54 3 (60) 2 (40) 54 4 (57) 3 (43) Sex Female 0 (0) 2 (100) Male 7 (70) 3 (30).15 AJCC tumor stage T1 0 (0) 0 (0) T2 2 (67) 1 (33) T3 5 (83) 1 (17).24 T4 0 (0) 3 (100) Breslow thickness, mm (categorical) (100) 0 (0) (54) 5 (46) Histological subtype Desmoplasic 0 (0) 0 (0) Lentigo maligna 0 (0) 0 (0) Nodule 1 (50) 1 (50).47 Superficial spreading 5 (83) 1 (17) Unclassified 1 (25) 3 (75) Vertical growth phase Present 7 (64) 4 (36) Absent 0 (0) 1 (100).42 Mitotic index (ordinal) High 1 (33) 2 (67) Intermediate 0 (0) 1 (100) Low 6 (75) 2 (25).31 Unclassified 0 (0) 0 (0) Mitotic index (categorical) High 1 (33) 2 (67) Others 6 (67) 3 (33).52 Lymphocytic tumor infiltration Brisk 1 (100) 0 (0) Moderate 2 (67) 1 (33) Minimal or none 4 (57) 3 (43).99 Indeterminate 0 (0) 0 (0) Clark level II 0 (0) 0 (0) III 1 (50) 1 (50) IV 6 (67) 3 (33).99 V 0 (0) 1 (100) Tumor ulceration Present 4 (67) 2 (33) Absent 2 (50) 2 (50) Tumor regression Present 1 (50) 1 (50) Absent 6 (67) 3 (33) Tumor location Face 2 (67) 1 (33) Neck 2 (67) 1 (33) Scalp 1 (33) 2 (67).99 Ear 0 (0) 0 (0) Upper thorax 2 (67) 1 (33) No. of SLNs removed 1 1 (50) 1 (50) Multiple 6 (60) 4 (40) No. of positive SLNs 1 7 (86) 1 (12) Mulitiple 0 (0) 4 (100).01 No. of basins that underwent biopsy 1 4 (80) 1 (20) 2 1 (20) 4 (80) (100) 0 (0) *Unknown values were excluded from analysis unless indicated. Negative indicates results showed no evidence of tumor; positive, results showed evidence of tumor; AJCC, American Joint Committee on Cancer 12 ; and SLN, sentinel lymph node. Analysis performed using Fisher exact test. Analysis performed using exact Mantel-Haenszel test. Analysis performed using Pearson 2 test. 318

7 A recent report 22 of our overall experience with SLNB for all tumors showed a significant correlation between SLNB positivity and increasing tumor thickness, when analyzed as both continuous and categorical variables, and between SLNB positivity and AJCC tumor stage. Similar associations were noted in the head and neck subset by categorical assignment only. The less convincing relationship in the head and neck subset may be related to smaller numbers of patients or possibly to other confounding factors. In this report, 11 (92%) of all positive SLNB results were noted in patients with tumor thickness of 1.23 mm or greater. Classification tree analysis of the overall group 22 and the head and neck subset showed the Breslow thickness cutoff point of 1.23 mm to be a significant cutoff point for prediction of SLNB result. This cutoff point is obviously influenced by our prior selection biases for the procedure. The ideal criteria for selection for SLNB are not known, and therefore any cutoff points are somewhat arbitrary. However, this cutoff point may serve as a useful initial guideline to counsel patients about their likelihood of having a positive SLNB result. NO OTHER TRADITIONAL melanoma prognostic indicators achieved statistical significance for prediction of SLNB result in the head and neck subset, although a high mitotic index showed a trend toward significance. This association has been previously reported 24 and is not surprising because the mitotic index has been shown to be predictive of survival in patients with melanoma. 25,26 Tumor ulceration has been shown to be an important prognostic factor for melanoma recurrence 6 and also is an independent predictor of a positive SLNB result in several series Our overall experience with SLNB shows that tumor ulceration and a high mitotic index are independent predictors of SLNB positivity. 22 This finding suggests that patients with ulcerated tumors and those with lesions with a high mitotic index are at higher risk for occult nodal metastases and should be considered for SLNB, regardless of tumor thickness. Completion lymph node dissection after a positive SLNB result in the head and neck yielded additional positive non-slns in 41% of the patients in this series. The positivity rate for CLND following a positive SLNB result varies from 7% to 35% in various reports. 3-5,7,8,10,11,20-22 These numbers probably underestimate the true CLND positivity rate because analysis of CLND specimens has not been as rigorous as the analysis of SLNB specimens. Therefore, the true CLND positivity rate is unknown. The relatively low incidence of positive CLND results in some series has led to speculation that CLND may be unnecessary for some patients with positive SLNB results, such as those with thin melanomas. 21,27 No significant correlation between patient or tumor factors and CLND result was observed. However, the presence of multiple positive SLNs in the head and neck was a significant predictor of the CLND result; in all the patients with more than one positive SLN, at least one additional positive non-sln was discovered in the CLND specimens. In a previous report, 28 our group was not able to demonstrate a convincing correlation between tumor thickness and tumor burden in the regional lymph node basin or between CLND positivity and the number of positive SLNs or the presence of multiple SLN micrometastases. The head and neck region may be special in this regard, with more SLNs per basin, more basins, and a higher overall CLND positivity rate than non head and neck sites. These data illustrate the difficulty in accurately predicting the status of the residual nodal basin after positive SLNB results. Because recent prospective clinical trials 29,30 suggested that occult melanoma nodal metastases are clinically important and that early therapeutic lymphadenectomy may favorably affect survival, CLND seems to be the most appropriate therapy after a positive SLNB result. Considerable variability exists between different institutions SLNB protocols for use of blue dye and/or radiolocalization. Various radiopharmaceuticals, doses, and timing approaches have been successful, but spillover into non-slns occurs over time and may lead to unnecessary harvest of non-slns In high lymphatic flow areas with numerous nodes, such as the head and neck nodal areas, spillover occurs rapidly, even when large particle tracers are used. Early dynamic lymphoscintigrams obtained immediately after dermal tracer injection are key to identify the pattern and order of nodal drainage and to identify the SLN before numerous second echelon nodes appear. Technical variations resulted in numerous definitions of what constitutes an SLN by radiolabeling criteria, each of which is somewhat arbitrary and subject to our limited knowledge of the intralymphatic kinetics of radiopharmaceuticals. For this reason, the most reliable indicator (and, in fact, the unequivocal clinical definition) of SLN status is a blue-stained afferent channel leading to the blue-stained node(s). 31 Cutaneous lymphoscintigraphy is useful for general localization of lymphatic basins and for intraoperative confirmation of SLN identification and removal. Lymphoscintigraphy is not completely reliable in predicting the number of SLNs that will be found. 34 Intraoperative use of a handheld gamma probe in combination with blue dye is extremely useful because it speeds up the procedure considerably and allows a smaller incision with less dissection in pursuit of the blue node (s), which are also usually hot. If the blue-stained SLN cannot be found, hot nodes can be removed and radioactivity count ratios determined. Because of the reason cited above, the possible existence of multiple aberrant patterns of lymphatic drainage, 34 the possibility of multiple SLNs, basins, and intransit nodes, and the general facilitation that radiolabeling affords, we recommend both lymphoscintigraphy and blue dye for all SLNBs. Equally important to the success of SLNB is a thorough pathological analysis of the nodal tissue. The pathologist handling SLN tissues must make a diligent search for nodal micrometastases. The median tumor volume in SLNB-positive lymph nodes is approximately 4.7 mm 3, approximately equivalent to the tumor volume of a solitary 2-mm nodule. 28 In one large series 35 of recurrences after negative SLNB results, the most common reason for 319

8 false-negative results was missed pathologic interpretations. Pathologists are usually experienced with preparing step or serial sections and immunohistochemical staining (S100 protein and/or HMB-45 antigen), but they must be aware of the importance of routinely doing these preparations on SLNs that are negative by standard analysis. 18,19,35,36 The results of enhanced molecular diagnostic assays, such as reverse transcriptase polymerase chain reaction, can upstage patients undergoing SLNB and appear to correlate somewhat with clinical outcome. 37 Reverse transcriptase polymerase chain reaction analysis of SLNB specimens and peripheral blood samples is being tested in ongoing clinical trials, but clinical utility and widespread applicability are uncertain at this time. Until recently, SLNB technology has been available primarily in large melanoma centers. With more surgeons attempting SLNB, it is appropriate to consider the limitations and widespread applicability of the procedure. Lymphoscintigraphy is less reliable after prior wide excision, particularly if extensive reconstructive procedures or prior lymph node surgical procedures have been performed. These procedures may disrupt the lymphatic drainage from the melanoma site, making localization of the SLN more difficult and, in some cases, impossible. Even if the SLN can be localized and harvested, the likelihood that it will accurately reflect the status of the remainder of the basin is diminished. Sentinel lymph node biopsy is a deceptively difficult procedure to master because success is dependent on the use of several disciplines and a gradual learning curve exists. Despite the excellent results in the literature, SLNB has not yet been shown to be reproducible in the hands of the occasional melanoma surgeon. Technical and therapeutic controversies notwithstanding, the histological status of the SLN is clinically relevant for at least 3 major reasons. First, the SLN status has been shown to be an important independent prognostic factor, with a positive result that is predictive of high risk for treatment failure. A recent multicenter study 20 showed the histological status of the SLN to be a more important prognostic factor than tumor thickness. The predictive value of a negative SLNB result is less certain at this time but probably identifies patients with significantly lower risk of recurrence than would be predicted by tumor characteristics. Second, SLNB results identify, with minimal morbidity, those high-risk patients who may benefit from additional therapy, such as early lymphadenectomy or adjuvant interferon therapy, 38 or who may be eligible to participate in clinical trials. Sentinel lymph node biopsy allows the surgeon to operate after histological confirmation, essentially eliminating the problem of negative results of ELND. Last, the psychological benefit for the patient whose SLNB result does not reveal metastases seems significant. A strong argument can be made that SLNB in experienced hands is the ideal clinical approach for staging and management of patients with significant-risk melanoma today. The question of whether SLNB is the standard of care or is still investigational in the management of melanoma is debatable. Despite the confirmed diagnostic accuracy and increasing use of SLNB for melanoma staging, it is important to emphasize that the effectiveness of SLNB as a therapeutic procedure for patients with melanoma is unproven. Although SLNB provides valuable staging information, it cannot directly identify patients who will have melanoma recurrence due to undetected distant disease. The possible therapeutic benefits of the SLNB procedure itself and SLNB-based therapeutic decisions are currently unknown. Because of the lack of procedural standardization, questions about general reproducibility, and absence of prospective clinical trials showing effectiveness of therapy based on SLNB, some experts have recommended against the routine use of SLNBbased therapy outside the investigational setting. 19 CONCLUSIONS Sentinel lymph node biopsy of cervical and parotid nodal areas using the combination of blue dye and intraoperative gamma probe radiolocalization for cutaneous melanoma sites with lymphatic drainage to these areas is a safe and accurate diagnostic procedure. The results of SLNB of the head and neck nodal basins show more positive basins, more SLNs per basin, and a higher percentage of positive CLND specimens than SLNB of other regions. Because SLNB results are most closely associated with tumor thickness and tumor stage, these should be the primary criteria for selection of patients for this procedure. A Breslow thickness of 1.23 mm or greater may serve as a useful guideline for counseling patients about SLNB. After a positive SLNB specimen in head and neck nodal basins, additional tumor-containing nodes may remain in the residual basin in 40% of patients. Because it is difficult to identify these patients, formal CLND is recommended. Accepted for publication July 29, Presented at the annual meeting of the American Head and Neck Society, Palm Desert, Calif, April 25, Corresponding author: Jeffrey D. Wagner, MD, RT 471, Cancer Pavilion Building, 535 Barnhill Dr, Indianapolis, IN ( jdwagner@iupui.edu). REFERENCES 1. Glocker-Reis LA, Miller BA, Hankey BF, et al, eds. SEER Cancer Statistics Review, : Tables and Graphs. Bethesda, Md: National Cancer Institute; Publication Landis SH, Murray T, Bolden S, Wingo PA. Cancer statistics, CA Cancer J Clin. 1999;49: Morton DL, Wen DR, Wong JH, et al. Technical details of intraoperative lymphatic mapping for early stage melanoma. Arch Surg. 1992;127: Reintgen D, Cruse CW, Wells K, et al. The orderly progression of melanoma nodal metastases. Ann Surg. 1994;220: Krag DN, Meijer SJ, Weaver DL, et al. Minimal-access surgery for staging of malignant melanoma. Arch Surg. 1995;130: Soong SJ, Weiss HL. Predicting outcome in patients with localized melanoma. In: Balch CM, Houghton AN, Sober AJ, Soong SJ, eds. Cutaneous Melanoma. 3rd ed. St Louis, Mo: Quality Medical Publishing Inc; 1998: Morton DL, Wen DR, Foshag LJ, Essner R, Cochran A. Intraoperative lymphatic mapping and selective cervical lymphadenectomy for early-stage melanomas of the head and neck. J Clin Oncol. 1993;11: O Brien CJ, Uren RF, Thompson JF, et al. Prediction of potential metastatic sites in cutaneous head and neck melanoma using lymphoscintigraphy. Am J Surg. 1995;170: Ollia DW, Foshag LJ, Essner R, Stern SL, Morton DL. Parotid region lymphatic mapping and sentinel lymphadenectomy for cutaneous melanoma. Ann Surg Oncol. 1999;6:

9 10. Wells KE, Rapaport DP, Cruse CW, et al. Sentinel lymph node biopsy in melanoma of the head and neck. Plast Reconstr Surg. 1997;100: Bostick P, Essner R, Sarantou T, et al. Intraoperative lymphatic mapping for earlystage melanoma of the head and neck. Am J Surg. 1997;174: Fleming ID, Cooper JS, Henson DE, et al (American Joint Committee on Cancer), eds. Cancer Staging Manual. 5th ed. Philadelphia, Pa: Lippincott-Raven; Wells KE, Cruse CW, Daniels S, Berman C, Norman J, Reintgen DS. The use of lymphoscintigraphy in melanoma of the head and neck. Plast Reconstr Surg. 1994; 93: Albertini JJ, Cruse CW, Rapaport D, et al. Intraoperative radiolymphoscintigraphy improves sentinel lymph node identification for patients with melanoma. Ann Surg. 1996;223: Pijpers R, Collet GJ, Meijer S, Hoekstra OS. The impact of dynamic lymphoscintigraphy and gamma probe guidance on sentinel node biopsy in melanoma. Eur J Nucl Med. 1995;22: van der Veen H, Hoekstra OS, Paul MA, Cuesta MA, Meijer S. Gamma probeguided sentinel node biopsy to select patients with melanoma for lymphadenectomy. Br J Surg. 1994;81: Mudun A, Murray DR, Herda SC, et al. Early stage melanoma: lymphoscintigraphy, reproducibility of sentinel node detection, and effectiveness of the intraoperative gamma probe. Radiology. 1996;199: Morton DL, Giuliano AE, Reintgen DS, Roses DF, Ross MI, Thompson JF. Symposium: lymphatic mapping and sentinel node biopsy in patients with breast cancer and melanoma: part 1. Contemp Surg. 1998;53: Morton DL, Giuliano AE, Reintgen DS, Roses DF, Ross MI, Thompson JF. Symposium: lymphatic mapping and sentinel node biopsy in patients with breast cancer and melanoma: part 2. Contemp Surg. 1998;53: Gershenwald JE, Thompson W, Mansfield PF, et al. Multi-institutional melanoma lymphatic mapping experience: the prognostic value of sentinel lymph node status in 612 stage I or II melanoma patients. J Clin Oncol. 1999;17: Joseph E, Brobeil A, Glass F, et al. Results of complete lymph node dissection in 83 patients with positive sentinel nodes. Ann Surg Oncol. 1998;5: Wagner JD, Corbett L, Park HM, et al. Sentinel lymph node biopsy and selective lymphadenectomy for melanoma: recommendations based on experience with 234 consecutive procedures. Plast Reconstr Surg. In press. 23. Heaton KM, Sussman JJ, Gershenwald JE, et al. Surgical margins and prognostic indicators in patients with thick ( 4 mm) primary melanoma. Ann Surg Oncol. 1998;5: Mraz-Gernhard S, Sagebiel RW, Kashani-Sabet M, Miller JR III, Leong SP. Prediction of sentinel lymph node micrometastasis by histological features in primary cutaneous malignant melanoma. Arch Dermatol. 1998;134: Hacene K, LeDoussal V, Brunet M, Lemoine F, Guerin P, Hebert H. Prognostic index for clinical stage I cutaneous malignant melanoma. Cancer. 1983;43: Kopf AW, Gross DF, Rogers GS, et al. Prognostic index for malignant melanoma. Cancer. 1987;59: Haddad FF, Stall A, Messina J, et al. The progression of melanoma nodal metastasis is dependent on tumor thickness of the primary lesion. Ann Surg Oncol. 1999;6: Wagner JD, Davidson D, Coleman JJ III, et al. Lymph node tumor volumes in patients undergoing sentinel lymph node biopsy for cutaneous melanoma. Ann Surg Oncol. 1999;6: Cascinelli N, Morabito A, Santinami M, MacKie RM, Belli F. Immediate or delayed dissection of regional nodes in patients with melanoma of the trunk: a randomised trial: WHO melanoma programme. Lancet. 1998;351: Balch CM, Soong SJ, Bartolucci AA, et al. Efficacy of an elective regional lymph node dissection of 1 to 4 mm thick melanomas for patients 60 years of age and younger. Ann Surg. 1996;224: Morton DL, Bostick PJ. Will the true sentinel node please stand? Ann Surg Oncol. 1999;6: Nathanson DS, Nelson L, Karvelis KC. Rates of flow of technetium 99m labeled human serum albumin from peripheral injection sites to sentinel lymph nodes. Ann Surg Oncol. 1996;3: Glass EC, Essner R, Morton DL. Kinetics of three lymphoscintigraphic agents in patients with cutaneous melanoma. J Nucl Med. 1998;39: Uren RF, Howman-Giles RB, Shaw HM, Thompson JF, McCarthy WH. Lymphoscintigraphy in high-risk melanoma of the trunk: predicting draining node groups, defining lymphatic channels, and locating the sentinel node. J Nucl Med. 1993; 34: Gershenwald JE, Colome MI, Lee JE, et al. Patterns of recurrence following a negative sentinel lymph node biopsy in 243 patients with stage I or II melanoma. J Clin Oncol. 1998;16: Cochran AJ, Wen DR, Morton DL. Occult tumor cells in the lymph nodes of patients with pathological stage I malignant melanoma: an immunohistological study. Am J Surg Pathol. 1988;12: Shivers SC, Wang X, Li W, et al. Molecular staging of malignant melanoma: correlation with clinical outcome. JAMA. 1998;280: 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 J Clin Oncol. 1996;14:7-17. Correction Typographical Error. In the Editorial by Johns titled Millennial Milestones That Matter, published in the January issue of the ARCHIVES (2000;126:13), a typographical error appears in the right column, second full paragraph, first word of the fifth line. It should have read modern anesthetics....thejournal regrets the error. 321