A recursive partitioning algorithm identified patient age as an important determinant of nodal positivity after mitotic rate of 1 mitotic figure/mm2 or more and thickness of 0.76 mm or more. Numbers in parentheses are 95% CIs.
Lymph node metastases decreased with increasing patient age across traditional pathologic risk groups. Patients with T1a tumors 0.50 to 0.75 mm not shown; nodal positivity rate was less than 3% for all ages in this group. Error bars indicate 95% CI.
eFigure. Flow Diagram for Patient Selection for Study
eTable. Characteristics of Patients Not Undergoing Regional Nodal Evaluation
Customize your JAMA Network experience by selecting one or more topics from the list below.
Sinnamon AJ, Neuwirth MG, Yalamanchi P, et al. Association Between Patient Age and Lymph Node Positivity in Thin Melanoma . JAMA Dermatol. 2017;153(9):866–873. doi:10.1001/jamadermatol.2017.2497
Is patient age associated with rate of lymph node positivity in clinically localized thin melanoma?
In this cohort study of a national data set of 8772 patients undergoing curative resection of thin melanoma, older age was significantly associated with reduced risk of lymph node metastases. Patients younger than 40 years with category T1b tumors 0.50 to 0.75 mm, who would generally not be recommended sentinel lymph node biopsy, had a lymph node positivity rate of 5.6%; conversely, patients 65 years or older with T1b tumors 0.76 mm or larger, who would generally be recommended for sentinel lymph node biopsy, had a lymph node positivity rate of only 3.9%.
Current guidelines for sentinel lymph node biopsy may be overly restrictive in younger patients and overly permissive in older patients.
More than half of all new melanoma diagnoses present as clinically localized T1 melanoma, yet sentinel lymph node biopsy (SLNB) is controversial in this population given the overall low yield. Guidelines for SLNB have focused on pathologic factors, but patient factors, such as age, are not routinely considered.
To identify indicators of lymph node (LN) metastasis in thin melanoma in a large, generalizable data set and to evaluate the association between patient age and LN positivity.
Design, Setting, and Participants
A retrospective cohort study using the National Cancer Database, an oncology database representing patients from more than 1500 hospitals throughout the United States, was performed (2010-2013). Data analysis was conducted from October 1, 2016, to January 15, 2017. A total of 8772 patients with clinical stage I 0.50 to 1.0 mm thin melanoma undergoing wide excision and surgical evaluation of regional LNs were included for study.
Main Outcome and Measures
The primary outcome of interest was presence of melanoma in a biopsied regional LN. Clinicopathologic factors associated with LN positivity were characterized, using logistic regression. Age was categorized as younger than 40 years, 40 to 64 years, and 65 years or older for multivariable analysis. Classification tree analysis was performed to identify high-risk groups for LN positivity.
Among the study cohort (n = 8772), 333 patients had nodal metastases, for an overall positivity rate of 3.8% (95% CI, 3.4%-4.2%). A total of 4087 (54.0%) patients were women. Median age was 56 years (interquartile range [IQR], 46-67) in patients with negative LNs and 52 years (IQR, 41-61) in those with positive LNs (P < .001). In multivariable analysis, younger age, female sex, thickness of 0.76 mm or larger, increasing Clark level, mitoses, ulceration, and lymphovascular invasion were independently associated with LN positivity. In decision tree analysis, patient age was identified as an important risk stratifier for LN metastases, after mitoses and thickness. Patients younger than 40 years with category T1b tumors 0.50 to 0.75 mm, who would generally not be recommended for SLNB, had an LN positivity rate of 5.6% (95% CI, 3.3%-8.6%); conversely, patients 65 years or older with T1b tumors 0.76 mm or larger, who would generally be recommended for SLNB, had an LN positivity rate of only 3.9% (95% CI, 2.7%-5.3%).
Conclusions and Relevance
Patient age is an important factor in estimating lymph node positivity in thin melanoma independent of traditional pathologic factors. Age therefore should be taken into consideration when selecting patients for nodal biopsy.
Approximately 85% of new invasive melanoma diagnoses present with stage I and II disease.1 Among these, more than 70% are category T1 tumors (≤1.0 mm) and, despite the overall excellent prognosis for this subgroup of patients, those with T1 tumors represent the largest public mortality burden due to their overall high prevalence.2 However, the use of sentinel lymph node biopsy (SLNB) remains controversial in this population due to overall low yield of identifying occult nodal disease. Proper patient selection for SLNB is critical for reducing cost and perioperative morbidity while simultaneously recognizing patients in whom SLNB carries value. These points are magnified given the volume of patients with clinically localized T1 melanoma seen in consultation annually.
Current guidelines published by the National Comprehensive Cancer Network (NCCN) help inform clinicians in properly selecting patients with thin melanoma for SLNB.3 These guidelines are based largely on the probability of identifying nodal disease; SLNB is generally not recommended in lesions unlikely to yield nodal metastasis. A yield of 5% is commonly recognized as a threshold to support use of SLNB, since this is the approximate rate of both false negativity and more common regional morbidities, such as infection and seroma, associated with the procedure.4 Current NCCN guidelines recommend performing SLNB for T1b melanoma 0.76 mm or larger and not performing SLNB in any melanoma smaller than 0.76 mm. For T1a melanoma larger than 0.76 mm, the NCCN recommends that clinicians discuss and consider SLNB for patients.3
Several studies have investigated prognostic factors for SLN positivity in patients with thin melanoma with variable results.5-12 Varied results may in part be due to the relative rarity of SLN metastases in T1 tumors and, therefore, the need for large sample size for statistical power. In particular, age has been variably reported to be associated with lymph node positivity in this subgroup of patients. Some investigators have found younger age to be associated with higher SLN positivity rates in thin melanoma,12-16 but others have shown no such association.17,18 A recent meta-analysis by Cordeiro and colleagues19 characterizing risk factors for SLN positivity in thin melanoma was unfortunately unable to evaluate the effect of age due to variable reporting across studies. Balch et al20 identified a pattern of decreasing SLN positivity with increasing age in their cohort. Cavanaugh-Hussey et al21 reported a similar finding in the Surveillance, Epidemiology, and End Results study. However, these studies examined all tumor thicknesses and did not focus on thin primaries. This issue has thus not been fully explored in thin melanoma, a population in whom it is particularly relevant since the use of SLNB is more controversial. A possible association between patient age and nodal positivity is clinically important since advanced age may often correlate with additional comorbidities. Possible overutilization of SLNB in this older cohort may be particularly relevant given anesthesia-associated risks with the SLNB procedure.
In this study, we sought to identify clinicopathologic factors associated with regional lymph node positivity in patients with thin melanoma and, in particular, to evaluate its association with age, using a large cohort of patients from the National Cancer Database (NCDB), an oncology database representing patients from more than 1500 hospitals throughout the United States. We limited our study to tumors 0.50 to 1.0 mm because nodal metastases in lesions smaller than 0.50 mm have been shown to be rare.22 We evaluated our results in the context of current NCCN guidelines to help inform current clinical practice and decision making.
Data for this study were obtained from the NCDB melanoma participant user file. The NCDB is a joint endeavor of the American Cancer Society and the American College of Surgeons’ Commission on Cancer, using hospital registry data from institutions accredited by the Commission on Cancer. It is estimated that data within the NCDB represent approximately 70% of new cancer diagnoses in the United States annually.23 The institutional review board at the University of Pennsylvania reviewed the study and concluded that it did not require ongoing oversight. Data are deidentified.
Patients included for study were at least aged 18 years and were undergoing wide excision of cutaneous melanoma with surgical evaluation of lymph nodes (International Classification of Diseases for Oncology, 3rd Edition histology codes 8720-8723, 8730, 8740, 8742-8745, 8761, and 8770-8772, with corresponding site codes C440-C449). Included patients were limited to diagnosis in 2010-2013 for consistency in reporting of pathologic variables, subsequent to the American Joint Commission on Cancer staging system update in 2009.24 Breslow thickness for study inclusion was 0.5 to 1.0 mm, as described above. Patients were excluded if they had clinical nodal disease, metastatic disease, palliative surgery, or death within 90 days (eFigure in the Supplement). To address misclassification bias, patients were excluded if data for pathologic variables were internally inconsistent, for example, if listed as pN0 but regional lymph nodes were listed as positive elsewhere.
Clinical and pathologic variables included for analysis were age, sex, primary site, thickness (dichotomized at 0.76 mm), Clark level, mitoses (present or absent, with present defined as ≥1 mitosis/mm2), ulceration, lymphovascular invasion (LVI), and regression. Race was classified as entered into the NCDB participant user file, but was not used in analysis since the cohort was predominantly white. Age was treated as both a continuous and categorical variable in univariate analysis but classified into 3 age groups (<40, 40-65, and ≥65 years) for multivariable analysis. The primary outcome evaluated was presence of metastatic melanoma in regional lymph nodes.
Descriptive statistics are presented as frequencies for categorical variables and medians with interquartile ranges (IQRs) for continuous variables. Univariate and multivariable logistic regression modeling was performed. A multivariable model was created using all variables, and those that were nonsignificant were sequentially removed to create a final reduced multivariable model. Classification tree analysis using a recursive partitioning algorithm was performed to identify groups at high risk for nodal positivity (R, version 3.3.1 statistical package).25,26 Variables included in the classification tree were those significant in the multivariable model. All statistical analyses, except the classification tree, were performed using Stata, version 14.27 Data analysis was conducted from October 1, 2016, to January 15, 2017.
From 2010 to 2013, 8772 patients with cutaneous melanoma 0.5 to 1.0 mm undergoing curative resection with regional nodal evaluation were identified in the NCDB and met study criteria. Descriptive statistics are presented in Table 1. The median age was 56 years (interquartile range [IQR], 46-67) with 1379 (15.7%) and 2655 (30.3%) being younger than 40 years and 65 years or older, respectively. A total of 4685 (53.4%) patients were men, and primary sites of disease included the trunk (3025 [34.5%]), upper extremity (2520 [28.7%]), lower extremity (1822 [20.8%]), and head and neck (1374 [15.7%]). Median Breslow thickness was 0.80 mm (IQR, 0.21), and 43.2% of tumors were Clark level IV/V. Mitoses were present in 4735 (54.0%) of the tumors; ulceration and LVI were much less common (679 [7.7%] and 122 [1.4%], respectively).
The median age among 9587 patients who did not undergo regional nodal evaluation but otherwise met inclusion and exclusion criteria was 62 years (IQR, 50-74), with 1085 (11.3%) and 4282 (44.7%) being younger than 40 years and 65 years or older, respectively. Characteristics of this group are reported in the eTable in the Supplement. Patients who did not undergo nodal evaluation generally had lower risk tumors, with significantly thinner lesions, and less commonly had presence of mitoses, ulceration, higher Clark level, or LVI. They did, however, have a higher frequency of head and neck lesions (21.5% vs 15.7%).
Of the 8772 patients included in analysis, 333 were found to have regional nodal metastases on surgical lymph node evaluation, for a rate of 3.8% (95% CI, 3.4%-4.2%). Clinicopathologic factors associated with nodal positivity are reported in Table 2. In univariate analysis, younger age, female sex, deeper Breslow thickness, higher Clark level, presence of mitoses, presence of ulceration, presence of LVI, and regression were associated with nodal metastases. In the reduced multivariable model, factors that remained significantly associated with nodal positivity included younger age (<40 years: odds ratio [OR], 2.04; 95% CI, 1.44-2.90; P < .001; 40-64 years: OR, 1.59; 95% CI, 1.19-2.11; P = .001), female sex (OR, 1.26; 95% CI, 1.00-1.58; P = .04), Breslow thickness ≥0.76 mm (OR, 1.74; 95% CI, 1.36-2.23; P < .001), higher Clark level (level III: OR, 2.07; 95% CI, 1.17-3.63; P = .01; level IV/V: OR, 2.27; 1.30-3.96; P = .003), presence of mitoses (OR, 1.46; 95% CI, 1.13-1.89; P = .003), presence of ulceration (OR, 1.58; 95% CI, 1.11-2.24; P = .01), and presence of LVI (OR, 2.07; 95% CI, 1.06-4.04; P = .03).
A classification tree analysis was performed using a recursive partitioning algorithm (Figure 1). Presence of mitoses was identified as the first split factor in the tree analysis for predicting nodal disease, followed by Breslow thickness 0.76 mm or more. These subgroups were next split by patient age. Among patients with mitogenic tumors 0.76 mm or larger, those younger than 65 years were at higher risk than were those 65 years or older (nodal positivity rate, 6.5% vs 4.1%). Similarly, patients with mitogenic tumors 0.50 to 0.75 mm were at higher risk if they were younger than 40 years (5.8% vs 2.6%). Among patients with absent or unknown mitoses and thickness of 0.76 mm, patients younger than 65 years were at higher risk than were those 65 years or older (4.2% vs 1.3%).
The association between patient age and nodal positivity was next evaluated among risk groups identified by current NCCN guidelines: T1a tumors smaller than 0.76 mm, T1b smaller than 0.76 mm, T1a 0.76 mm or larger, and T1b 0.76 mm or larger (Table 3). Among T1b tumors smaller than 0.76 mm, T1a 0.76 mm or larger, and T1b 0.76 mm or larger, nodal positivity consistently decreased by increasing age group. Among patients with T1b tumors 0.50 to 0.75 mm for whom current NCCN guidelines do not recommend SLNB, nodal positivity was 5.6% in patients younger than 40 years and decreased to 3.0% in patients 40 to 65 years and 2.0% in those 65 years or older. This trend was statistically significant (P = .006).28 In patients with T1a tumors 0.76 mm or larger in whom NCCN guidelines recommend consideration of SLNB, positivity was 6.9% when age was younger than 40 years, 4.2% when age was 40 to 65 years, and 1.1% when age was 65 years or older (P < .001). In patients with T1b tumors 0.76 mm or larger, in whom NCCN guidelines recommend performing SLNB, positivity was 7.1% when age was younger than 40 years, 6.4% when age was 40 to 65 years, and 3.9% when age was 65 years or older (P = .004). A similar trend was not observed among patients with T1a tumors smaller than 0.76 mm (P = .38); however, rates of nodal positivity were no higher than 2.7% in this group. These results are shown in Figure 2.
In the present study, we identify indicators of lymph node positivity in thin (0.50-1.00 mm) melanoma using a large national data set and, in particular, characterize the effect of patient age on nodal positivity. To our knowledge, this is the largest single study cohort addressing this topic to date. Since more than half of new invasive melanoma diagnoses present as clinically localized thin melanoma, identification of subgroups in whom SLNB has a higher yield has important implications for proper use of health care resources. Conversely, recognition of patients in whom the procedure has low yield may help to reduce morbidity. The inverse association of patient age with node positivity is particularly relevant since older patients may be at higher perioperative risk due to the need for anesthesia.
Factors that were found to be independently associated with lymph node metastases in the present study were younger age, female sex, Breslow thickness larger than 0.76 mm, higher Clark level, presence of mitoses, presence of ulceration, and lymphovascular invasion. In a classification tree analysis, patient age was recognized as an important indicator of nodal disease, after mitoses and Breslow thickness. Age was shown to consistently stratify traditional risk groups for risk of nodal spread; among patients with T1b tumors 0.76 mm or larger, T1a tumors 0.76 mm or larger, and T1b tumors 0.50 to 0.75 mm, the rate of nodal positivity decreased consistently with increasing age groups.
We used these findings to evaluate current recommendations for SLNB in the context of patient age. Many clinicians consider a node positivity threshold of 5% for offering SLNB.29,30 Current NCCN guidelines recommend that clinicians discuss and offer SLNB for patients with T1b tumors 0.76 to 1.0 mm. However, in the present study, patients in this group 65 years or older were found to have nodal positivity rates of only 3.9% (95% CI, 2.7%-5.3%). In fact, patients 65 years or older did not surpass the 5% threshold in any of the risk groups. Considering that wide excision of primary melanoma without SLNB can often be performed under local anesthesia, these findings are particularly clinically relevant since older patients are more likely to have comorbidities that can increase their risk for higher levels of anesthesia.
Conversely, SLNB is generally not recommended in patients with melanoma smaller than 0.76 mm due to low yield. We found that, within this group, patients younger than 40 years with T1b tumors 0.50 to 0.75 mm had nodal positivity of 5.6% (95% CI, 3.3%-8.6%), suggesting that SLNB should perhaps be considered in this group. Lastly, in patients with T1a tumors 0.76 to 1.0 mm, the NCCN guidelines recommend consideration of SLNB; patient age may play an important role in decision making for SLNB to better inform the potential benefits of the procedure, particularly in older patients in whom procedural risk may already be increased.
Previous studies identifying factors associated with SLN positivity in thin melanoma have largely been single-institutional studies. Results have thus been inconsistent and varied across studies, but have reported thickness, Clark level, mitotic rate, ulceration, LVI, sex, and age as risk factors for SLN positivity.5-12 Among the larger studies, Han and colleagues29 found Breslow thickness of 0.75 mm or more, Clark level IV/V, and ulceration to be associated with SLN metastases in a multi-institutional study. The association between sex and SLN positivity has been variably reported, with most associations noted in univariate but not in multivariable analysis. In a recent meta-analysis pooling these reports, Cordeiro and colleagues19 found thickness 0.75 mm, Clark level IV/V, mitoses, and microsatellites to be associated with SLN positivity. Pertinent to the present study, this meta-analysis was unable to evaluate age, given variable reporting across individual studies. Moreover, the meta-analysis was performed at the study level, without aggregating individual patient data. We were unable to assess the effect of microsatellitosis, as these data are not collected by the NCDB; this factor would, in any case, be expected to be rarely identified in thin lesions and would classify these patients as having stage III disease even before nodal staging.
The use of the NCDB affords us at least 2 important strengths, the first being a large sample size with the accompanying statistical power. Equally important is generalizability, given that the database includes hospitals across the United States with, no doubt, variation in practice patterns, surgical techniques, and pathologic analysis. In the present study, identifying factors indicative of nodal metastases in patients with thin melanoma is generally consistent with what has been previously reported from other single institution experiences.5-12
We acknowledge limitations to this study, including those inherent to any retrospective analysis. The possibility of misclassification and selection bias highlight the need for further verification of our study findings with prospective, institutional data. The NCDB, as is any database, is subject to misclassification bias during coding that cannot be independently verified. To address this issue, internal consistency between variables was evaluated and patients in whom pathologic data were incongruent were excluded. However, there remain certain data elements that we are unable to cross-verify, including tumor thickness; specifically, thin tumors transected on biopsy may lead to misclassification of their thickness. The NCDB recommends coding thickness as the greatest measured thickness from any procedure performed on the lesion, without adding measurements from different procedures. Although this recommendation may understage the disease in some patients, it may also prevent overstaging in patients for whom pathology reports do not specify whether residual tumor on re-excision should be additive or not to prior biopsy. Arguing against any significant misclassification of thickness is the fact that the nodal positivity rates and rate of biopsy performance among tumors smaller than 0.76 mm seem to be concordant with large, published institutional series.19
There is a certain element of selection bias regarding who should undergo SLNB; clinicians are more likely to recommend the procedure if they have a higher level of suspicion of nodal positivity. Consistent with this, patients who did not undergo lymph node evaluation but otherwise met study criteria had thinner lesions with lower Clark level and, less commonly, presence of mitoses, ulceration, or LVI. Therefore, nodal positivity rates are likely enriched in the study group. Fewer patients 65 years or older underwent nodal evaluation (38.3%) compared with younger age groups (<40 years, 55.9%; 40-64 years, 52.8%), although it is unclear whether this reflects a lower suspicion of nodal disease, worse comorbidities, or another reason. Although less frequent use of SLNB in older patients may be appropriate given our findings, the 38.3% rate of procedure performance may be considered high when the yield of LN metastases did not exceed 5% in any risk group 65 years or older.
Using a large national database, we identify factors associated with regional lymph node positivity in patients undergoing wide excision of clinically localized melanomas 0.5 to 1.0 mm and describe the effect of age on nodal positivity. To our knowledge, this is the largest single study cohort addressing these questions to date and is clinically relevant, since most new melanoma diagnoses are clinically localized T1 lesions. Factors associated with nodal positivity included younger age, Breslow thickness 0.76 mm or larger, higher Clark level, presence of mitoses, presence of ulceration, and lymphovascular invasion. With use of the classification tree analysis, patient age was shown to be an important discriminant of nodal positivity among patients with mitogenic tumors 0.50 to 0.75 mm and 0.76 to 1.0 mm and nonmitogenic tumors 0.50 to 0.75 mm. Risk of node positivity among traditional risk groups, on which guidelines for SLNB are based, were consistently stratified by age. Patients 65 years or older were shown to have a lower risk of nodal disease across risk groups, and patients younger than 40 years had a higher risk. Consideration of SLNB should be given for patients younger than 40 years with T1b tumors 0.50 to 0.75 mm. Conversely, guidelines may be overly permissive in older patients with tumors 0.76 mm or larger. Consideration of patient age may allow more appropriate allocation of SLNB, thus reducing both health care costs and possible unnecessary patient morbidity.
Accepted for Publication: April 27, 2017.
Corresponding Author: Giorgos C. Karakousis, MD, Department of Surgery, Hospital of the University of Pennsylvania, 3400 Spruce St, 4 Silverstein Bldg, Philadelphia, PA 19104 (firstname.lastname@example.org).
Published Online: July 19, 2017. doi:10.1001/jamadermatol.2017.2497
Author Contributions: Drs Sinnamon and Karakousis had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: Sinnamon, Neuwirth, Xu, Fraker, Karakousis.
Acquisition, analysis, or interpretation of data: Sinnamon, Neuwirth, Yalamanchi, Gimotty, Elder, Xu, Kelz, Roses, Chu, Ming, Karakousis.
Drafting of the manuscript: Sinnamon, Gimotty, Kelz, Karakousis.
Critical revision of the manuscript for important intellectual content: All authors.
Statistical analysis: Sinnamon, Gimotty.
Administrative, technical, or material support: Neuwirth, Xu, Kelz, Karakousis.
Supervision: Roses, Fraker, Karakousis.
Conflict of Interest Disclosures: None reported.
Funding/Support: Dr Gimotty was supported by The Abramson Cancer Center of the University of Pennsylvania Core Support Grant P30-CA016520.
Role of the Funder/Sponsor: The Abramson Cancer Center of the University of Pennsylvania had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.
Disclaimer: The National Cancer Data Base (NCDB) is a joint project of the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society. The CoC’s NCDB and the hospitals participating in the CoC NCDB are the source of the deidentified data used herein; they have not verified and are not responsible for the statistical validity of the data analysis or the conclusions derived by the authors.