Long-term outcomes of nodal surveillance practices in sentinel node positive melanoma: An early post MSLT-II cohort

Abstract

Background:

The Second Multicenter Selective Lymphadenectomy Trial (MSLT-II), published in 2017, demonstrated equivalent melanoma-specific survival between nodal surveillance and completion lymph node dissection (CLND) for sentinel lymph node (SLN) positive melanoma. This study evaluated outcomes of nodal surveillance in an early post-MSLT-II institutional cohort.

Methods:

Included patients received nodal surveillance from 2017 to 2023. Primary outcomes were nodal basin and any site recurrence. Kaplan Meier curves and Cox proportional hazard models were used to evaluate recurrence-free survial and associated factors.

Results:

This 212 patient cohort (median age 61 years, 57.7 % male, 96.7 % white) had median primary tumor depth of 1.9 mm and one positive SLN. Sixty-three patients (29.7 %) recurred at 23 month median follow-up. Sixty percent received adjuvant therapy. Any site and nodal basin recurrence-free survival were 58.3 % and 80.9 % at 3 years. On adjusted analysis, older age and head/neck primary site were associated with worse recurrence-free survival.

Conclusion:

Long-term outcomes at a single institution were comparable to clinical trial findings. Nodal surveillance remains a feasible management strategy for SLN + melanoma.

1. Introduction

The post-operative management of sentinel lymph node positive (SLN+) melanoma has shifted toward active nodal surveillance with nodal ultrasound and physical examination in recent years. Few patients now undergoing completion lymph node dissection (CLND) for micrometastatic disease identified on SLN biopsy.¹ This shift was prompted by landmark multicenter randomized controlled trials including the Second Multicenter Selective Lymphadenectomy Trial (MSLT-II) and German Dermatologic Cooperative Oncology Group Study (DeCOG-SLT) demonstrated no difference in melanoma-specific survival for patients randomized to nodal surveillance versus CLND, with reduced rates of postoperative lymphedema in the surveillance group.^2,3

Implementation of nodal surveillance in real-world practice, however, has presented challenges not reported in trial settings, with early work demonstrating reduced adherence to nodal surveillance, which may be multifactorial involving patient coordination, provider preferences among both surgical and medical oncology, as well as surveillance imaging methods themselves.⁴ It is unknown whether clinical trial outcomes will be recapitulated in pragmatic settings where nodal surveillance adherence may be worse, or how broad uptake of adjuvant systemic therapy will affect oncologic outcomes. To characterize longer-term outcomes in a contemporary, real-world cohort, we evaluated an early post-MSLT-2 cohort treated at a large tertiary care center in southeastern United States. The primary objective was to describe recurrence free survival and to compare this with trial outcomes. Secondary objectives were to identify factors that are associated with recurrence, describe patterns of recurrence (i.e. local/in-transit, nodal, distant, or multiple sites), and management of recurrent disease.

2. Methods

2.1. Study population

Study approval was obtained from the University of Alabama at Birmingham (UAB) Institutional Review Board. Patients were identified using cancer registry data and billing codes for melanoma and sentinel lymph node biopsy using the UAB Enterprise Data Warehouse, which stores clinical and billing data. The study cohort included adult patients with cutaneous melanoma and a positive sentinel lymph node biopsy (SLNB) from July 2017 to June 2023 who underwent active nodal surveillance (i.e. omission of completion lymph node dissection (CLND)). All patients underwent perioperative staging imaging to exclude metastatic disease prior to nodal surveillance and inclusion within the cohort. Those with distant metastatic disease identified on perioperative staging studies were excluded. A list of ICD and CPT codes used to identify this cohort can be found in Supplemental Table 1. Additionally, free text queries of pathology reports were used as an additional way to capture node-positive cases. Patient, disease, and treatment data were manually abstracted from clinically available data including patient demographics, operative reports, pathology reports, surveillance imaging, adjuvant therapy, recurrences, management strategies for recurrent disease, and vital statistics. Potential exposures (independent variables), such as tumor location and pathologic features, were collected to evaluate their association with the study endpoint/dependent variable (recurrence). Patient with missing values for selection of nodal surveillance versus completion lymph node dissection were excluded (N = 3). Otherwise, there were no missing data elements.

2.2. Outcomes

The primary end point was disease recurrence at any site (categorized as local or in-transit, nodal basin only, distant, or multi-site). Multi-site recurrences could involve any combination of sites, such as local/in-transit + nodal or nodal + distant. Recurrences were confirmed by review of pathology reports, clinical documentation, and imaging studies, with preference for confirmation by pathology when available. Time to disease recurrence was defined as number of days from SLN biopsy to first confirmed detection of melanoma recurrence at any site. Only the site of initial recurrence was evaluated. Participants were censored at their last active (clinical) follow-up, and only known follow-up time points were included in the time-to-event analysis. Secondary outcomes included disease and treatment factors associated with time to recurrence and method of recurrence detection.

2.3. Nodal surveillance

Nodal surveillance measures included nodal basin ultrasounds, cross-sectional imaging, and physical exams. Nodal US were performed under a protocol at the reporting institution every 4–6 months for five years, overseen primarily by surgical oncologists and surgical advanced practice provider. Cross-sectional imaging surveillance modalities included whole-body positron emission tomography (PET), computed tomography (CT) of the chest/abdomen/pelvis, and magnetic resonance imaging (MRI) of the brain, when applicable. Care provided at outside facilities was included if documentation was available. As this study was not performed under a prospective surveillance protocol, there was significant variation in use of cross-sectional imaging, which was often performed by local oncologists and was not universally available for review.

2.4. Statistical analysis

Statistical analysis was performed using Stata 17 statistical software (StataCorp LLC, College Station, TX, USA) and R studio (version 4.1.5). Descriptive statistics included categorical measures, reported as frequency (percentage), and continuous variables which were reported as median (interquartile range). Bi-variate analyses included chi-squared independence tests for categorical data and Wilcoxon rank-sum tests for continuous variables, all of which had non-parametric distributions. Time to any-site and nodal recurrence were evaluated using Kaplan-Meier survival curves. Factors associated with any site recurrence free survival were evaluated using Cox proportional hazard models.

3. Results

3.1. Descriptive statistics

There were 212 included patients (Table 1). Participants had a median age of 61 years and were predominantly male (57.5 %), white (96.7 %), and not Hispanic or Latino (91.0 %). Most primary tumors were located on the trunk (40.1 %) or lower extremity (28.8 %), followed by tumors of the upper extremity (17.0 %), and head/neck (14.2 %). Of these tumors, 45.8 % were ulcerated and 10.8 % had satellites or microsatellites. These patients had a median Breslow depth of 1.9 mm and a median of one positive SLN. Sentinel lymph node biopsy sites included axillary (94, 44.3 %), inguinal (66, 31.1 %), cervical (30, 14.2 %), more than one nodal basin (20, 9.4 %), or other/not available (2, 1.0 %). Patients had a median of one positive sentinel lymph node (IQR 1–1). Positive SLN basins included axillary (100, 47.2 %), inguinal (72, 34.0 %), cervical (29, 13.7 %), more than one basin (8, 3.7 %), and other (3, 1.4 %). The median nodal tumor diameter was 1.0 mm (IQR 0.2–2.7). Fifteen patients (7.1 %) had extranodal extension. BRAF mutations were detected in 53 (25.0 %) of tumors. One hundred and twenty-seven patients (59.9 %) received adjuvant systemic therapy (116, 91.3 % received single agent checkpoint inhibitor therapy), 3 (1.4 %) received nodal basin radiation, and 4 (1.9 %) received primary site radiation. Reasons for adjuvant radiation to the primary site included high risk features (microsatellites, close margins with inability to re-excise, anal primary site). Reasons for nodal basin irradiation included anal primary site (2) and presence of extranodal extension in a positive node, though it is unclear why this patient received adjuvant nodal radiation without completion lymph node dissection.

Table 1. Cohort patient and tumor characteristics.

Data presented as median with interquartile range (IQR) for continuous measures and n (%) for categorical measures.

Variable	Cohort (n = 212)
Age, Years	61 (49–70)
Gender
Female	90 (42.5 %)
Male	122 (57.5 %)
Race
Asian	1 (0.5 %)
Black	1 (0.5 %)
White	205 (96.7 %)
Unknown	5 (2.4 %)
Ethnicity
Hispanic or Latino	4 (1.9%)
Not Hispanic or Latino	193 (91.0%)
Unknown	15 (7.1 %)
Travel distance, miles	74.7 (38.7–104.5)
Tumor location
Head or Neck	30 (14.2 %)
Upper Extremity	36 (17.0 %)
Trunk	85 (39.6 %)
Lower Extremity	61 (28.8 %)
Breslow Depth, mm	1.9 (1.2–3.3)
Presence of Ulceration	97 (45.8 %)
Presence of Microsatellites	23 (10.8 %)
Presence of In-Transit Disease	5 (2.4 %)

3.2. Disease recurrence and survival outcomes

Mean follow-up time was 726 days (23 months, range 0–2130). Four patients did not follow up after surgery and did not contribute to follow-up time in survival analyses. These patients were included in the characteristics of patients not undergoing completion lymph node dissection but did not contribute any follow-up time to the survival analyses. 41 patients had at least 3 years of follow-up data and 11 patients had at least 5 years of disease-free follow-up. There were 63 patients (29.7 % of overall cohort) who recurred during follow-up, at a median of 378 (224–565) days from SLN biopsy. Sites of initial recurrence were local or in-transit (10, 15.9 %), nodal basin only (19, 30.2 %), or distant/multi-site (34, 53.9 %) (Table 3). Among those with multi-site disease, the recurrence patterns were local/in-transit + nodal (N = 2), local/in-transit + nodal + distant (N = 1), local/in-transit + nodal (N = 2), local/in-transit + nodal + distant (N = 1), local/in-transit + distant (N = 2), and nodal + distant (N = 4). Twenty-five patients first recurred at distant sites only. At 3 years (N = 41), recurrence-free survival was 58.3 %, and nodal basin recurrence-free survival was 80.9 %. At 5 years (N = 11), recurrence-free survival was 54.7 %, and SLN basin recurrence-free survival was 79.1 %. Any-site and nodal basin recurrence-free survival curves are depicted in Fig. 1.

Table 3. Sites of disease recurrence.

Data are presented as n (%) for categorical measures.

Recurrent Disease Characteristics	Cohort (n = 63)
Initial Sites of Recurrence
Limited to the Regional Nodal Basin	19 (30.2 %)
Isolated Local or In-Transit Disease	10 (15.9 %)
Distant/Multisite Disease	34 (53.9 %)

Fig. 1.

Recurrence free survival (A) and sentinel node basin recurrence free survival (B).

On bivariate analysis, factors associated with recurrence were older age (p = 0.019), tumor ulceration (p = 0.03), larger maximum tumor dimension/diameter in a positive node (p = 0.031) and presence of BRAF mutation (p < 0.001) (Table 2). Age was treated as a continuous variable for risk adjustment purposes and was significant, meaning that for each one year increase in age at surgery, there was a 3 % reduction in risk of recurrence during follow-up. In the Cox model of recurrence-free survival which included disease and treatment factors, non-head and neck primary sites were associated with improved recurrence-free survival (upper extremity: HR 0.31, 95 % CI 0.12–0.81; lower extremity: HR 0.38, 95 % CI 0.17–0.83, trunk: HR 0.47, 95 % CI 0.23–0.95)) while older age was associated with worse RFS (HR 1.02, 95 % CI 1.01–1.05). Tumors of the trunk and upper and lower extremities had 53 %, 69 %, and 62 % lower risks of recurrence, respectively, in comparison to head and neck primary sites. There were no significant associations based on sex, Breslow depth, ulceration, satellitosis, in-transit disease, number of positive SLN, or receipt of adjuvant systemic therapy (Fig. 2).

Table 2. Factors associated with any-site disease recurrence.

Data are presented as median IQR for continuous measures and n (%) for categorical measures.

Recurrence Variable	No Recurrence (n = 149)	Recurrence (n = 63)	P-value
Tumor Location
Head or Neck	17 (11.4 %)	13 (20.6 %)	0.21
Lower Extremity	43 (28.9 %)	18 (28.6 %)
Trunk	60 (40.3 %)	25 (39.7 %)
Upper Extremity	29 (19.5 %)	7 (11.1 %)
Melanoma Histologic Subtype
Acral Lentiginous	12 (8.1 %)	6 (9.5 %)	0.41
Desmoplastic	1 (0.7 %)	0 (0 %)
Lentigo Malinga	2 (1.3 %)	2 (3.2 %)
Nodular	28 (18.8 %)	17 (27.0 %)
Superficial Spreading	63 (42.3 %)	25 (39.7 %)
Unknown	43 (28.8 %)	13 (20.7 %)
Depth of Tumor (Breslow) (mm) (IQR)	1.88 (1.2–3.0)	2.2 (1.2–4.0)	0.13
Presence of Ulceration	61 (40.9 %)	36 (57.1 %)	0.030
Presence of Microsatellites	13 (8.7 %)	10 (15.9 %)	0.13
Presence of In-transit disease	4 (2.7 %)	1 (1.6 %)	0.63
Positive Surgical Margin	13 (8.7 %)	7 (11.1 %)	0.59
Maximum Dimension of Tumor in a Positive Node (IQR)	0.9 (0.2–2.1)	1.7 (0.65–3.0)	0.031
Presence of Extranodal Extension	10 (7.9 %)	5 (7.9 %)	0.77
BRAF Mutation Identified	30 (20.1 %)	23 (36.5 %)	0.001
Adjuvant Systemic Therapy Received	87 (58.4 %)	40 (63.5 %)	0.95
Median Follow Up Time (days)	527	729	0.01

Fig. 2.

Factors associated with recurrence-free survival.

3.2.1. Management of recurrent disease

Recurrent disease was surgically excised in 28 cases (44.4 %) with complete excision in 26 cases (92.9 %). This consisted of therapeutic lymphadenectomy of a regional nodal basin for eighteen patients (28.6 % in the recurrence group). Systemic therapy was utilized for recurrent disease for 48 patients (76.2 %), and radiation therapy was performed for 22 patients (34.9 %). Intralesional therapy was administered to 10 patients with recurrent disease (15.8 %). Distant metastases were detected in 40 (63.5 %) patients who experienced recurrences (18.9 % of the full cohort). At the time of data collection, 27 (12.7 %) patients in the overall cohort were deceased, with 10 deaths directly attributable to melanoma.

4. Discussion

Results from MSLT-II and DeCOG-SLT have led to rapid de-implementation of CLND and adoption of nodal surveillance for SLN + melanoma.^2,3,5 Considerations of the real-world efficacy and practicality of this management strategy are essential to its continued use.⁶ As we are now approaching six years of follow-up for this early post-MSLT-II cohort, the need to examine the alignment of real-world outcomes with those achieved in trials is apparent, and the identification of trends in recurrence could influence management choices moving forward.

Subsequent observational studies have sought to validate the findings from these landmark trials. A 2021 international post-trial cohort found that active surveillance had been adopted for most SLN-positive patients.⁷ Disease-free survival and distant metastasis-free survival did not differ by nodal management or adjuvant treatment. It was concluded that initial real-world outcomes aligned with trial results, including among those who received adjuvant therapy.⁷ Our results are concordant in a cohort with a longer follow-up duration. Recurrence-free survival was 58 % at 3 years compared to 63 % in MSLT-II. 3-year nodal basin recurrence-free survival was 81 % in our cohort versus 77 % in MSLT-II cohort.

The status of sentinel node positivity may not be the only prognostic marker for the development of recurrent or invasive disease. For example, some SLN-positive high-risk groups have been excluded from trials, such as those consisting of patients with microsatellites, extranodal extension (ENE) in the SLN, and those with >3 positive SLNs.⁸ These patients possessing high-risk features were evaluated for similar trial outcomes in a multi-institutional cohort and found to have greater recurrence risks, but results overall supported the use of nodal surveillance in these groups.⁸ Among our nodal surveillance cohort, no significant differences in rates of recurrence were noted based on the presence of microsatellites and ENE, though this is likely attributable to small sample size. Though individuals with MSLT-II exclusion criteria are not well studied, a large multicenter study did address this unique population, finding overall higher rates of recurrence, but no significant differences for node dissection versus surveillance groups.⁸ Our inclusion of these patients reflects the findings from this study and may contribute to the generalizability of our study findings.

Primary tumor location may also have prognostic value, as evidenced by our results and the work of others. The finding that head and neck primary tumors are associated with heightened recurrence risk is supported by prior work.^9–13 Proposed reasons for higher rates of nodal recurrence include more complex lymphatic mapping and drainage pattern variability.¹⁰

A notable difference between our cohort and the trial is that a larger proportion of patients received adjuvant therapy (60 %) compared with 7 % in MSLT-II.² In the adjusted model, there was a trend toward improved recurrence-free survival among patients who received adjuvant systemic therapy, although results were not significant in this non-randomized, single-institution study. Importantly, all major immunotherapy trial cohorts underwent CLND so the strategy of nodal surveillance with adjuvant systemic therapy has not been studied in a randomized fashion. Adjuvant systemic therapy trials reported an approximately 50 % reduction in recurrence. Given this, we might expect better, rather than comparable, recurrence-free survival in this cohort, which had greater utilization of adjuvant systemic treatment. A prior single institution post-MSLT-II cohort study found that adjuvant therapy after a positive SLN biopsy without CLND was an effective strategy to decrease recurrences in those with stage III and IV disease compared with active surveillance alone.¹⁴ This study’s sample size is insufficient to address the effect of adjuvant systemic therapy, and it should be evaluated in future trials and/or larger, multi-institutional cohorts. Further, additional studies are warranted to characterize the SLN + populations who are most at risk for recurrence in order to better select those who would most benefit from adjuvant therapy.

This study has several limitations, including its single-institution, retrospective observational design. Due to the nature of this study design, receipt of adjuvant therapies were not randomized, limiting our ability to draw conclusions on how they affected disease recurrence. With the small size of our cohort, including a significant rural population, these results may not be generalizable to SLN + melanoma cohorts at other institutions with different practice patterns. Many patients in this cohort chose to receive medical oncology care closer to their area of residence, and a large proportion were subsequently lost to follow-up.

These trends limited our full knowledge of recurrences, systemic treatment administration, and whether deaths were attributable directly to melanoma. Our institution is the only NCI-designated Comprehensive Cancer Center in the region and is a referral center for complex cases, including disease recurrence. This may result in overestimation of recurrence risk if patients with recurrence were more likely to re-present at our institution than those without recurrence. At the same time, there is a risk of underestimation of death due to melanoma if such patients were managed locally. With the nature of the study design, it is recognized that patients who were treated more recently have decreased follow-up time, and those who were lost to follow-up do not have data on vital and recurrence status. The authors appreciate that this cohort’s catchment area is even more rural than other centers, likely reducing follow-up duration due to more patients receiving follow-up care outside the study institution.

Finally, the feasibility of nodal surveillance adherence is essential to interpreting ‘real-world outcomes. While these outcomes were not the focus of this study, a similar cohort from this institution demonstrated a 34 % rate of adherent surveillance.⁴ Further, a subsequent multi-site study demonstrated that this rate is on par with that of comparable US institutions, who reported median adherence of 42 %.¹⁵ Establishing realistic expectations for adherence among these institutions where malignant SLN + melanoma is most managed is essential to discern true real-world outcomes outside of the trial setting. Prior work has demonstrated that melanoma is one of the most referred/highly centralized cancer types in the US.¹⁶ The use of this cohort from an NCI-designated comprehensive cancer center is generalizable to the places in the US where most melanoma management occurs.

5. Conclusion

Long-term follow-up outcomes for melanoma nodal surveillance at a single institution demonstrated similar any-site and nodal basin recurrence-free survival to MSLT-II results in this early post-trial cohort, affirming that nodal surveillance remains a feasible management strategy for SLN + melanoma in real-world settings. Future multi-institutional studies may be warranted to evaluate barriers to nodal surveillance and describe long-term outcomes for nodal surveillance in a larger and more diverse cohort.

Data availability

The research data used to generate this study includes sensitive or confidential information regarding patient health.