Evaluation of Lymph Node Ratio Association With Long-term Patient Survival After Surgery for Node-Positive Merkel Cell Carcinoma

This cohort study of 1274 patients who underwent surgery for node-positive Merkel cell carcinoma evaluates the association between lymph node ratio and long-term survival.

Key Points

Question

Is the ratio of positive lymph nodes associated with long-term patient survival after surgery for node-positive Merkel cell carcinoma?

Findings

This cohort study of patient records in the National Cancer Database and the Surveillance, Epidemiology, and End Results Registry found that the ratio of positive lymph nodes was significantly associated with long-term patient survival. Patients with high lymph node ratios experienced a survival advantage when treated with adjuvant chemoradiation therapy over adjuvant radiation therapy alone.

Meaning

Lymph node ratio may provide a useful indicator for both patient counseling and treatment planning after surgery for node-positive cases of Merkel cell carcinoma.

Abstract

Importance

Merkel cell carcinoma (MCC) carries the highest mortality rate among cutaneous cancers and is rapidly rising in incidence. Identification of prognostic indicators may help guide patient counseling and treatment planning. Lymph node ratio (LNR), the ratio of positive lymph nodes to the total number of examined lymph nodes, is an established prognostic indicator in other cancers.

Objectives

The primary objective was to evaluate the association between LNR and patient survival after surgery for node-positive MCC. The secondary objective was to evaluate whether the survival rates associated with adjuvant therapies vary by patient LNR status.

Design, Setting, and Participants

Retrospective cohort study of patients with node-positive MCC treated with surgery and lymphadenectomy. We queried the National Cancer Database (NCDB) and the Surveillance, Epidemiology, and End Results (SEER) registry for patient records. Data originated from 2004 through 2017 for the NCDB and from 1973 through 2016 for the SEER registry. The SEER registry comprises a population-based US cohort while cases from the NCDB include all reportable cases from Commission on Cancer–accredited facilities and represents approximately 70% of all newly diagnosed cancers in the United States. All data analysis took place between August 1, 2018, and February 11, 2019.

Exposures

The ratio of positive lymph nodes to the total number of examined lymph nodes, LNR, was stratified into quartiles.

Main Outcomes and Measures

Overall survival (NCDB) and disease-specific survival (SEER).

Results

We identified 736 eligible cases in the NCDB and 538 eligible cases in the SEER registry. Among these 1274 patients, the mean (SD) age was 71.1 (11.5) years, and 401 (31.5%) were women. After controlling for clinical and tumor factors including AJCC N staging, patient LNR of 0.07 to 0.31 (hazard ratio [HR], 1.37; 95% CI, 1.03-1.81) and greater than 0.31 (HR, 2.84; 95% CI, 2.10-3.86) was associated with significantly worse survival than an LNR less than 0.07. Univariate supplementary analysis performed in the SEER data set revealed a similar association of LNR with disease-specific survival. For patients with an LNR greater than 0.31, treatment with surgery and adjuvant chemoradiation therapy was associated with improved survival compared with surgery and adjuvant radiation therapy alone (HR, 0.61; 95% CI, 0.38-0.97), while this was not found for patients with an LNR of 0.31 or lower (HR, 0.93; 95% CI, 0.65-1.33).

Conclusions and Relevance

For lymph node–positive MCC, LNR offers a potentially prognostic metric alongside traditional TNM staging that may be useful for both patient counseling and treatment planning after surgery.

Introduction

Merkel cell carcinoma (MCC), first described in 1972, is a highly malignant neuroendocrine skin cancer.¹ The etiological factors of MCC are well defined, with the majority of cases linked to Merkel cell polyomavirus and the remainder to UV-induced DNA damage.^{2,3,4,5,6,7,8} Since its first description, regularly conducted studies of MCC epidemiology in the United States have indicated a rapid rise in incidence.^9,10,11,12 Recent work in this area reported a US incidence of 2488 cases per year (0.7 cases/100 000 person-years) and predicted a rise in this rate to 2835 cases per year by 2020 and 3284 by 2025.¹³ Studies conducted in France and Australia have indicated a similar trend.^14,15 This is particularly concerning given the higher mortality rate for MCC than for other cutaneous cancers, including malignant melanoma.^16,17,18 Such high mortality makes accurate prognostic staging especially valuable. The AJCC Cancer Staging Manual criteria currently categorize MCC cases that have undergone nodal metastasis based on whether positive nodes are notable on both clinical and pathologic examination or pathologic examination alone.¹⁹

Notably, these staging criteria do not consider the number or ratio of positive nodes in the prognosis of MCC cases. For cases with pathologically confirmed lymph node metastasis, National Comprehensive Cancer Network guidelines²⁰ recommend treatment with a lymph node dissection. In other cancers for which lymph node dissections are performed, several studies have pointed to lymph node ratio (LNR), the ratio of the number positive lymph nodes identified to the total number of lymph nodes examined, as a significant prognostic factor.^{21,22,23,24,25} In some of these cases, the total number of positive lymph nodes was also prognostic of long-term patient survival, although this was typically less predictive than LNR.^{26,27,28,29,30,31} The association of LNR and the total number of positive lymph nodes with survival after lymph node dissection for node-positive MCC has not been well studied.

Additionally, it is not known whether LNR could be used to stratify patients for optimal treatment strategies in MCC. While adjuvant radiation therapy has been demonstrated to improve outcomes,³² and is recommended for treatment of the primary tumor in cases regardless of nodal metastasis,²⁰ the role of systemic chemotherapy is not established in MCC. Interestingly, in other cancers, LNR has been shown to predict the benefit of postoperative adjuvant therapy.^33,34,35

In this study, we evaluate the association of LNR with survival after lymph node dissection for node-positive MCC. As a secondary objective, we evaluate whether the survival rates associated with adjuvant therapies vary by patient LNR status. To accomplish this, we examined a cohort of 736 patients from the Commission on Cancer’s National Cancer Database (NCDB) who were diagnosed between 2004 and 2014, inclusive. We also performed a supplementary analysis of 538 cases from the Surveillance, Epidemiology, and End Results (SEER) registry.

Methods

Data Sources

Data originated primarily from the NCDB with a supplemental analysis performed using the SEER data set. The NCDB data originated from 2004 through 2017, with diagnostic years 2004 through 2014. The NCDB is a nationwide clinical surveillance resource data set that includes approximately 70% of all newly diagnosed cancers in the United States from over 1500 cancer programs, as previously described.³⁶ SEER cases include 28% of cancers in the US population and are selected to be representative of the entire US population.³⁷ SEER data originated from 1973 through 2016, with diagnostic years 1973 through 2015. These registries have been used in a number of studies of cancer prognostic factors.^{38,39,40,41,42,43} This study was determined to be exempt from institutional review by the Yale Human Investigation Committee (No. 2000023704).

Study Population

We identified MCC cases with a primary site in the skin using the International Classification of Diseases for Oncology 3 histology code 8247 and primary site codes C44.0 through C44.9. We excluded patients if they were younger than 18 years at the time of diagnosis, did not have pathologic evidence of nodal disease, had unknown T staging data, did not receive definitive surgery, did not receive a lymph node dissection, had missing data on adjuvant therapy, had any other cancers, had distant metastases, or had missing or incomplete follow-up. With SEER data, we excluded patients who were younger than 18 years at the time of diagnosis, did not have pathologic evidence of nodal disease, did not receive definitive surgery, did not receive a lymph node dissection, or had missing or incomplete follow-up.

Statistical Analysis

Patients were defined as having received chemotherapy if they ever received any chemotherapy, regardless of the type or number of agents. We considered patients to have received radiation therapy if they received external beam irradiation. We defined patients as having had an adequate lymph node dissection if they had 5 or more examined lymph nodes, as described previously.⁴⁴ Lymph node ratio was calculated as the ratio of the number of positive lymph nodes detected to the number of lymph nodes examined. We categorized LNR into 3 groups according to quartiles with patients in the bottom quartile, middle 2 quartiles, and top quartile of LNR grouped together. The dividing points between these groups were quartile 1, 0.07 and quartile 3, 0.31. In our final sample, 99.5% (n = 732) of patients did not undergo immunotherapy, and 0.5% (n = 4) had unknown immunotherapy status. Results were unchanged when patients with unknown immunotherapy status were excluded from the analysis.

Statistical analyses were determined a priori. Kaplan-Meier analyses stratified by LNR were performed. Survival analysis for the NCDB was performed using overall survival, whereas, owing to its availability in the SEER data set, survival analysis for SEER patients was performed using disease-specific survival. We also conducted multivariable survival analyses using Cox proportional hazards models. All patient, tumor, and treatment characteristics outlined in Table 1 were tested for appropriateness for inclusion in these models using Akaike information criterion minimization.⁴⁵ This methodology ensures parsimony of the multivariable model. We additionally conducted a univariate Kaplan-Meier analysis stratified by total positive lymph node number. For this analysis, we analyzed survival trends for each number of positive lymph nodes and found that 1 to 2, 3 to 4, and 5 or more positive lymph nodes could be used as distinct categories of measurement because patients in each category had similar outcomes. Among patients with known outcomes, data were missing for 33.5% of patients (n = 370). Patients with missing data were slightly younger: mean (standard error [SE]) age, 69.7 (0.6) years vs 71.5 (0.4) years for those without missing data. Patients with missing data also had a slightly higher proportion of T3/T4-stage disease (24.8%; n = 29) and N2/N3 disease (8.2%; n = 17) than those without missing data. Patient sex, race, comorbidity score, insurance status, LNR, and long-term survival did not vary significantly between patients who were vs were not missing data. Using the SEER data, we conducted a univariate Kaplan-Meier analysis stratified by LNR, according to the cut points used in the NCDB data. Statistical significance was determined at the P < .05 level. Data analysis was performed using STATA, version 13 (StataCorp LP).

Table 1. Characteristics of the NCDB Analytic Sample.

Characteristic	Patients, No. (%) (n = 736)
Age, median (SD), y	73 (11.3)
Sex
Male	496 (67.4)
Female	240 (32.6)
Race
White	697 (94.7)
Black	9 (1.2)
Hispanic	12 (1.6)
Asian/Pacific Islander	5 (0.7)
Other/unknown	13 (1.8)
Insurance
Private	205 (27.8)
Medicare	494 (67.1)
Other government	22 (3.0)
None	4 (0.5)
Unknown	11 (1.5)
Charlson/Deyo score
0	547 (74.3)
1	145 (19.7)
2	35 (4.8)
≥3	9 (1.2)
Facility type
Academic	399 (54.2)
Other	337 (45.8)
Pathologic T stage
1	360 (48.9)
2	259 (35.2)
3	69 (9.4)
4	48 (6.5)
Pathologic N stage
1A	163 (22.2)
1B	164 (22.3)
1, NOS	377 (51.2)
2/3	32 (4.4)
Lymph node ratio, median (SD)	0.15 (0.24)
Treatment combination
Surgery only	211 (28.7)
Surgery + RT	390 (53.0)
Surgery + CRT	135 (18.3)

Abbreviations: CRT, chemoradiation therapy; NCDB, National Cancer Database; NOS, not otherwise specified; RT, radiation therapy.

Results

The characteristics of the NCDB study sample are detailed in Table 1. All results used NCDB data unless otherwise specified. Most tumors were limited in size, with the majority receiving an AJCC T classification of 1 or 2 (84.1%; n = 619). There was nearly an identical number of patients with N1A (22.2%; n = 163) and N1B (22.3%; n = 164) disease, with 4.4% of patients classified as having N2/N3 disease (n = 32). The median LNR was 0.15, with approximately a quarter of patients having an LNR less than 0.07 and a quarter greater than 0.31. While surgery with adjuvant radiation therapy was the most common treatment modality (53.0%; n = 390), a substantial proportion of patients received either surgery alone (28.7%; n = 211) or surgery with adjuvant chemoradiation therapy (18.3%; n = 135).

Lymph node ratio was found to be significantly associated with patient survival. Kaplan-Meier univariate survival curves stratified by LNR are presented in Figure 1A and demonstrate significant differences in survival (P < .001). Three- and 5-year survival rates (SE) for patients with an LNR less than 0.07 were 68.7% (3.7%) and 58.9% (4.2%), respectively; for patients with an LNR between 0.07 and 0.31, they were 55.7% (2.7%) and 45.3% (3.0%); and for patients with an LNR greater than 0.31, they were 27.1% (3.4%) and 20.3% (3.2%).

Figure 1. Kaplan-Meier Survival Curves Stratified by Patient LNR.

A, Univariate overall survival from the National Cancer Database analytic sample. B, Disease-specific survival from the Surveillance, Epidemiology, and End Results registry analytic sample. LNR indicates lymph node ratio.

Multivariate analysis, controlling for factors including AJCC pathologic N stage, also indicated that LNR was significantly associated with survival (Table 2). Patients with LNRs between 0.07 and 0.31 (hazard ratio [HR], 1.37; 95% CI, 1.03-1.81) or greater than 0.31 (HR, 2.84; 95% CI, 2.10-3.86) had significantly poorer outcomes than those with LNRs less than 0.07. Subgroup analysis performed for patients with confirmed N1B (LNR, <0.07: Reference; LNR, 0.07-0.31: HR, 2.29; 95% CI, 0.97-5.40; LNR >0.31: HR, 5.23; 95% CI, 2.31-11.85) or N1, NOS disease (LNR, <0.07: Reference; LNR, 0.07-0.31: HR, 1.46; 95% CI, 1.02-2.09; LNR, >0.31: HR, 2.82, 95% CI, 1.90-4.19) also demonstrated that LNR was significantly associated with patient survival. For patients with N1B disease, mean (SE) 3- and 5-year survival rates were 64.2% (11.2%) and 56.2% (12.4%), respectively, for patients with LNRs less than 0.07; 61.5% (5.7%) and 52.5% (6.9%), respectively, for patients with LNRs between 0.07 and 0.31; and 20.3% (5.4%) and 13.0% (5.0%), respectively, for patients with LNRs greater than 0.31. We were unable to perform subgroup analysis for patients with stage N1A disease owing to insufficient event numbers nor for patients with stage N2/N3 disease owing to an insufficient subgroup sample size.

Table 2. Multivariate Analysis of Characteristics Associated With Survival^a.

Characteristic	Hazard Ratio (95% CI)	P Value
Age	1.03 (1.02-1.04)	<.001
Sex
Female	1 [Reference]
Male	1.32 (1.06-1.64)	.01
Race
White	1 [Reference]
Black	1.60 (0.67-3.84)	.29
Hispanic	0.33 (0.13-0.82)	.02
Asian/Pacific Islander	1.61 (0.39-6.62)	.51
Other/unknown	1.11 (0.47-2.65)	.81
Insurance
Private	1 [Reference]
Medicare	1.51 (1.13-2.02)	.01
Other government	2.24 (1.15-4.36)	.02
None	2.54 (0.78-8.29)	.12
Unknown	0.62 (0.19-2.09)	.44
Charlson/Deyo score
0	1 [Reference]
1	1.27 (1.00-1.61)	.05
2	1.95 (1.26-3.02)	.003
≥3	1.21 (0.52-2.78)	.66
Pathologic T stage
1	1 [Reference]
2	1.48 (1.18-1.85)	.001
3	1.75 (1.24-2.46)	.001
4	1.66 (1.15-2.38)	.01
Pathologic N stage
1A	1 [Reference]
1B	1.79 (1.24-2.58)	.002
1, NOS	2.07 (1.50-2.86)	<.001
2/3	1.64 (0.94-2.84)	.08
Lymph node ratio
<0.07	1 [Reference]
0.07-0.31	1.365 (1.03-1.81)	.03
>0.31	2.84 (2.10-3.86)	<.001
Treatment combination
Surgery + RT	1 [Reference]
Surgery + CRT	0.81 (0.61-1.08)	.15
Surgery only	1.24 (0.99-1.56)	.07

Abbreviations: CRT, chemoradiation therapy; NOS, not otherwise specified; RT, radiation therapy.

^aFacility type was excluded from the multivariate model after Akaike information criterion minimization.

We also found that the outcomes following the various treatment combinations varied by patient LNR (Table 3). In a multivariate survival analysis of patients with high LNRs (ie, above 0.31), we found that treatment with surgery and adjuvant chemoradiation therapy was associated with significantly improved survival compared with treatment with surgery and adjuvant radiation therapy alone (HR, 0.61; 95% CI, 0.38-0.97). We also found that for those with an LNR of 0.31 or lower, increased patient comorbidities were associated with poorer survival.

Table 3. Multivariate Analysis of Characteristics Associated With Survival by Patient Lymph Node Ratio.

Characteristic	Lymph Node Ratio ≤0.31		Lymph Node Ratio >0.31
	Hazard Ratio (95% CI)	P Value	Hazard Ratio (95% CI)	P Value
Age	1.04 (1.02-1.05)	<.001	1.02 (0.10-1.04)	.10
Sex
Male	1.27 (0.96-1.67)	.10	1.36 (0.93-1.98)	.12
Female	1 [Reference]		1 [Reference]
Race
White	1 [Reference]		1 [Reference]
Black	1.61 (0.62-4.20)	.33	15.75 (1.87-132.69)	.01
Hispanic	Insufficient numbersa	NA	0.33 (0.12-0.91)	.03
Asian/Pacific Islander	1.09 (0.15-7.97)	.94	2.42 (0.29-19.88)	.41
Other/unknown	0.73 (0.18-3.01)	.66	2.02 (0.69-5.90)	.20
Insurance	Excludedb	NA
Private			1 [Reference]
Medicare			1.61 (0.92-2.84)	.10
Other government			3.44 (1.27-9.30)	.02
None			1.79 (0.40-8.02)	.45
Unknown			Insufficient numbersa	NA
Charlson/Deyo score			Excludedb	NA
0	1 [Reference]
1	1.46 (1.09-1.95)	.01
2	2.22 (1.31-3.79)	.003
≥3	0.90 (0.32-2.52)	.84
Pathologic T stage
1	1 [Reference]		1 [Reference]
2	1.65 (1.25-2.18)	<.001	1.47 (0.99-2.19)	.054
3	1.96 (1.24-3.11)	.004	1.63 (0.96-2.79)	.07
4	2.19 (1.43-3.36)	<.001	1.08 (0.50-2.34)	.85
Pathologic N stage
1A	1 [Reference]		1 [Reference]
1B	1.33 (0.86-2.07)	.21	5.04 (1.93-13.15)	.001
1, NOS	1.823 (1.283-2.591)	.001	4.68 (1.82-12.04)	.001
2/3	1.37 (0.70-2.66)	.36	4.04 (1.21-13.49)	.02
Lymph node ratio
<0.07	1 [Reference]
0.07-0.31	1.32 (0.10-1.75)	.054	NA	NA
Treatment combination
Surgery + RT	1 [Reference]		1 [Reference]
Surgery + CRT	0.93 (0.65-1.33)	.69	0.61 (0.38-0.97)	.04
Surgery only	1.32 (0.10-1.74)	.053	1.34 (0.86-2.09)	.20

Abbreviations: CRT, chemoradiation therapy; NA, not applicable; NOS, not otherwise specified; RT, radiation therapy.

^aInsufficient subgroup event numbers for multivariate analysis.

^bVariable was excluded from multivariate model after Akaike information criterion minimization.

Analysis of data from SEER using disease-specific survival, owing to its availability in this data set, also revealed LNR to be significantly associated with patient survival (P < .001). Survival curves for patients in the SEER registry stratified by LNR are presented in Figure 1B. Mean (SE) 3- and 5-year disease-specific survival rates were 74.2% (5.2%) and 68.2% (5.8%), respectively, for patients with LNRs less than 0.07; 61.4% (3.3%) and 53.6% (3.6%), respectively, for patients with LNRs between 0.07 and 0.31; and 33.8% (4.2%) and 27.2% (4.2%), respectively, for patients with LNRs greater than 0.31.

We additionally repeated our analyses according to the gross number of positive lymph nodes. Our univariate and multivariate survival analysis results were similar to those obtained with the LNR analysis. Survival curves according to the gross number of positive nodes are presented in Figure 2. Mean (SE) 3- and 5-year survival rates were 64.5% (2.5%) and 55.3% (2.8%), respectively, for patients with 1 to 2 positive lymph nodes; 43.2% (4.6%) and 33.0% (4.6%), respectively, for patients with 3 to 4 positive lymph nodes; and 29.8% (3.4%) and 21.3% (3.3%), respectively, for patients with 5 or more positive lymph nodes. Multivariate analysis of patients with 5 or more positive lymph nodes found improved survival for patients receiving adjuvant chemoradiation therapy instead of radiation therapy alone (HR, 0.76; 95% CI, 0.46-1.23; P = .26), but the difference was not significant.

Figure 2. Kaplan-Meier Univariate Overall Survival Stratified by the Total Number of Positive Lymph Nodes.

Discussion

The presence of lymph node metastases is an established adverse feature in cases of MCC.^{46,47,48,49,50} Current AJCC pathologic nodal staging classifies node-positive patients as N1A if the positive nodes are detectable only pathologically, N1B if they are detectable both clinically and pathologically, N2 if there are in-transit metastases without nodal metastases, and N3 if there are in-transit metastases with nodal metastases.¹⁹ The total number of positive lymph nodes and the LNR are not a component of the current staging system.

In the present study, we demonstrate that LNR after lymph node dissection is significantly associated with survival rates in MCC. Notably, LNR was found to be significantly associated with survival after controlling for AJCC pathologic N classification, suggesting that LNR may be used as a feature for risk stratification of patients in addition to traditional TNM staging. While this is the first analysis, to our knowledge, to demonstrate this association in cases of MCC, work in other cancers has shown a similar association for LNR. In cases of epithelial ovarian carcinoma,⁵¹ breast cancer,⁵² pancreatic cancer,⁵³ colorectal cancer,⁵⁴ and gastric cancer,⁵⁵ LNR was demonstrated to be more strongly associated with survival rates than traditional AJCC pN staging.

We also found that the total number of positive lymph nodes was similarly associated with survival rates, in addition to traditional AJCC pathologic N classification. This has been previously demonstrated in other cancers including cervical cancer,⁵⁶ prostate cancer,⁵⁷ and both medullary and papillary thyroid cancer.^58,59 However, in several studies where both the total number of positive lymph nodes and the LNR were included in the analysis, it was found that LNR was more strongly associated with patient survival than the number of positive nodes.^{26,27,28,29,30,31}

In cancers for which LNR is an established prognostic feature, it has also been shown to be beneficial in treatment decision making. In oral cavity cancer, for which the standard therapy is definitive surgery followed by adjuvant radiation therapy or chemoradiation therapy based on the presence of adverse features such as extranodal extension, high LNR could be used to determine which patients would most benefit from postoperative radiation therapy (PORT).³⁴ Similarly, cervical cancer cases with high LNR have been shown to exhibit a more marked benefit with postoperative adjuvant therapy than cases with low LNRs.³⁵ Notably, in non–small cell lung cancer, where the value of PORT is not established, it has been shown that PORT is associated with improved survival in patients with node-positive disease and high LNR (≥0.50) but not in patients with LNRs less than 0.50.³³ In the case of MCC, the role of PORT is well established and has been shown to significantly improve both long-term survival and locoregional control.^{60,61,62,63,64,65,66} The role of adjuvant systemic therapy is less well defined. Previous studies have shown no benefit to treatment with postoperative chemoradiation therapy over radiation therapy alone.^67,68,69 However, the findings of the present study demonstrate that in patients with an LNR greater than 0.31, treatment with adjuvant chemoradiation therapy was associated with significantly improved survival compared with radiation therapy alone. Additionally, no improved survival was found for patients with LNRs of 0.31 or less treated with surgery and adjuvant chemoradiation therapy vs adjuvant radiation therapy alone, suggesting less benefit with triple-modality therapy for this lower-risk group. It should be noted that in both the present study and those previously published, the agents used in conjunction with radiation therapy were cytotoxic; it is not yet known how newer biologic agents, such as avelumab, currently approved for metastatic MCC, may be used in the treatment of node-positive MCC.⁷⁰

A number of theories have been proposed to explain the prognostic utility of LNR. Urban et al³³ suggest that LNR may provide a proxy of the body’s immune response to cancer, pointing to the improved prognosis associated with tumors with infiltrating lymphocytes and the much lower rate of nodal metastasis noted in cervical cancers with tumor-infiltrating CD8⁺ lymphocytes.^71,72 More likely, LNR provides a measure of the likelihood of persistent disease after surgery, whether in the form of microscopic residual disease or micrometastases. This may be the driving factor behind the association of increased LNR with increased likelihoods of disease recurrence.^{73,74,75,76,77} This may also explain the added benefit of adjuvant chemoradiation therapy over adjuvant radiation therapy alone that we demonstrate in patients with high LNR, with the added chemotherapy potentially helping to treat micrometastases that were more likely to be present in these high LNR cases. Notably, there is no evidence that cytotoxic chemotherapy supports a durable response in stage IV (metastatic) cases of MCC, and it is therefore more regularly used as a palliative therapy for these patients.⁷⁸ In fact, previous work has reported poorer outcomes for patients undergoing chemotherapy, although these analyses notably did not control for presenting stage.⁴⁹ We have considered that cytotoxic agents may be more effective in treating subclinical micrometastases than the (likely larger) metastases detected on imaging or examination in stage IV disease. However, further studies would be necessary to clarify this possibility. The association of high LNR with persistent disease may also underlie the greater predictive value of LNR over total positive node count alone, since patients with the same number of positive nodes but more extensive dissections are less likely to have further undetected disease.

Limitations

There are several limitations to this study. First, because of the institutional and intersurgeon variability in techniques used to perform lymph node dissections, strict LNR prognostic cutoffs may not be broadly applicable. Instead, individual institutions and surgeons using the same techniques may find value in comparing the LNR of new patients with their historical patient cohorts. As more established guidelines are adopted regarding lymph node dissections for MCC, discrete prognostic LNR cutoffs may become more feasible.

We were also unable to control for several variables, such as sun-exposure history and Merkel cell polyomavirus positivity, owing either to their absence from the data sets or to a large amount of missing data. While cases with missing data did not have significant differences in LNR or survival, the effect of their exclusion is unclear. Additionally, we were unable to assess the effect of immunomodulatory therapeutics such as avelumab because they were not used in our analytic sample.⁷⁰ Furthermore, lack of data on local recurrence precluded our ability to assess the impact of LNR on disease recurrence. We were additionally unable to determine disease-specific outcomes for our NCDB cohort because the survival outcomes reported in the database represent all-cause survival, while SEER data presented an opportunity to assess disease-specific survival. Finally, we are unable to determine how many of the cases captured in our NCDB cohort were also present in our SEER cohort because the data between the 2 cannot be linked. Based on the case collection methodology of these 2 data sources, the overlap in cases is limited to individuals treated at Commission on Cancer–accredited cancer centers within the SEER 18 registry regions.

Conclusions

In conclusion, this is the first study to our knowledge to demonstrate the association of LNR with survival for node-positive cases of MCC. Our data suggest that LNR may be used to risk-stratify patients after surgery. Our results also suggest that more intensive treatment with surgery and adjuvant chemoradiation therapy rather than surgery and adjuvant radiation therapy alone may lead to a more pronounced survival benefit for cases with a high LNR (>0.31).