Sentinel lymph node biopsy is associated with increased cost in higher risk thin melanoma

Abstract

Introduction:

National Comprehensive Cancer Network guidelines recommend that sentinel lymph node biopsy (SLNB) be discussed with patients with thin melanoma at higher risk for lymph node metastasis (T1b or T1a with positive deep margins, lymphovascular invasion, or high mitotic index). We examined the association between SLNB and resource utilization in this cohort.

Methods:

We conducted a retrospective cohort study of patients that underwent wide local excision for higher risk thin melanomas from 2009 to 2018 at a tertiary care center. Patients who underwent SLNB were compared to those who did not undergo SLNB with regard to resource utilization, including total hospital cost.

Results:

A total of 70 patients were included in the analysis and 50 patients (71.4%) underwent SLNB. SLNB was associated with increased hospital costs ($6700 vs. $3767; p < .01) and increased operative time (68.5 vs. 36.0 min; p < .01). This cost difference persisted in multivariable regression (p < .01). Of patients who underwent successful SLN mapping, 3 out of 49 patients had a positive SLN (6.1%). The cost to identify a single positive sentinel lymph node (SLN) was $47,906.

Conclusion:

In patients with a higher risk of thin melanoma, SLNB is associated with increased cost despite a low likelihood of SLN positivity. These data better inform patient-provider discussions as the role of SLNB in thin melanoma evolves.

1 |. INTRODUCTION

In clinical stage I/II melanoma, sentinel lymph node (SLN) status is the strongest predictor of melanoma-specific survival.¹ Guidelines from the National Comprehensive Cancer Network (NCCN) suggest that clinicians consider performing sentinel lymph node biopsy (SLNB) in melanoma patients when the probability of positivity is greater than 5%.^2,3 For patients with T1b melanoma or T1a melanoma with other adverse features (mitotic index ≥ 2/mm², lymphovascular invasion, etc.), the probability of positivity is 5%–10% and the NCCN recommends discussing and considering SLNB on an individualized basis.^2,4,5 SLNB may also be considered when there is clinical suspicion of a thicker lesion in the setting of a positive deep margin on initial biopsy. Recent studies have demonstrated that SLNB is currently performed in approximately 30%–35% of patients in this higher risk thin melanoma cohort.⁶

However, SLNB is a relatively high-cost intervention that contributes to the overall cost of cancer care for patients with melanoma. In 2003, the cost of SLNB and wide local excision (WLE) in patients with thin melanoma was estimated to be $10,000–15,000 compared to $1000–2000 for WLE alone.⁷ While the cost has decreased in more recent studies, the cost-effectiveness of SLNB in thin melanoma remains a debated topic.⁸ Multiple factors affect the impact of SLNB in this cohort, including the survival benefit conferred by adjuvant therapy for node-positive patients, the equivalent survival outcomes associated with immediate completion lymph node dissection (LND) compared to observation and delayed LND, and the increasing efficacy of systemic treatment for recurrent disease.^9–16

In addition to the cost of SLNB at the patient level, there are important fiscal considerations at the level of the hospital system. It is estimated that there will be 100,350 cases of cutaneous melanoma in the United States in 2020, with approximately 70% of clinical stage I and II patients having T1 lesions.^17,18 As such, a significant proportion of the $3.2 billion spent annually on melanoma treatment can be attributed to patients with thin melanoma.¹⁹ With these considerations in mind, a detailed understanding of the costs of SLNB in patients with higher risk thin melanoma will help guide patient-provider discussions around the use of SLNB and help physicians quantify the impact of SLNB on resource utilization and cost at the level of hospital systems.

Contemporary data on the cost of SLNB in thin melanomas, especially patients with higher risk thin melanomas, are lacking. In addition, the majority of cost analyses for SLNB have utilized surrogates of hospital costs (charges, cost-to-charge ratios, or reimbursements) rather than actual hospital costs.²⁰ The objective of this study is to assess the cost of SLNB in patients with higher risk thin melanoma using a unique dataset that includes actual hospital cost data as calculated by cost-accounting algorithms that account for the supplies, labor, equipment, and facilities associated with each procedure.

2 |. METHODS

Patients that underwent WLE for higher risk thin melanoma at a single institution from January 2009 to December 2018 were identified within the electronic medical record by current procedural terminology (CPT) code corresponding WLE and International Classification of Disease (ICD) 9/10 code corresponding to cutaneous melanoma (Epic Clarity; Epic Systems Corporation). CPT and ICD 9/ 10 codes utilized for patient identification are listed in Table S1. Inclusion criteria were T1b or T1a melanoma (American Joint Committee on Cancer, 8th edition) with high-risk features on preoperative biopsy (positive deep margins, lymphovascular invasion, or mitotic index ≥ 2/mm²).²¹ Exclusion criteria were patients less than 18 years, patients with known regional or distant disease before surgery, and patients who were upstaged to T2 or greater lesions on final pathology. We also did not capture data on patients who underwent WLE in the clinic.

Cost data representing actual hospital costs were extracted from the institutional legacy data warehouse. Data from these sources were merged using two patient identifiers to create the final dataset for analysis. Individual components of hospital cost (direct variable cost, direct fixed cost, indirect variable cost, and indirect fixed cost) were also obtained to provide more granular cost data. Direct variable cost includes items directly related to patient care (medications, intravenous fluids, staplers, etc.).²² Direct fixed costs are also associated with each patient encounter and include capital equipment and labor (nursing, pharmacy, etc.). Indirect variable cost includes facility costs that change with hospital volume (electricity, laundry, etc.). Indirect fixed costs include static facility costs (building maintenance, construction, etc.). Costs were adjusted for medical price inflation using the producer price index for hospital inpatient care provided by the Bureau of Labor Statistics and are reported in 2018 dollars.^23,24 Cost data were linked to cancer outcomes data and hospital encounter data from an institutional cancer database and the electronic medical record.

All patients who underwent SLNB had preoperative lymphoscintigraphy with Tc-99m sulfur colloid performed the day before, or the day of, surgery. Blue dye (isosulfan blue or methylene blue, at surgeon discretion) was routinely injected at the primary site as an adjunct to the radiotracer. All sentinel lymph node specimens were reviewed in a standard fashion using both hematoxylin and eosin and immunohistochemistry using a pan-melanoma cocktail stain by a dedicated surgical pathologist.

Patients undergoing SLNB were compared to those that did not undergo SLNB with pathologic data and cost as outcome metrics. Mann-Whitney U test and χ² test were used where appropriate for baseline characteristics and bivariate outcomes. All statistical tests were conducted using two-tailed tests and a p-value of less than .05 was considered significant. Multivariable linear regression was performed for total hospital cost controlling for predictors of cost. All potentially significant variables were included in this model without selection because the primary objective was to obtain unbiased estimates of the association. All statistical analyses were performed in SAS 9.4 (SAS Institute Inc) and R 3.6.1 (R Foundation for Statistical Computing).

3 |. RESULTS

A total of 70 patients met eligibility criteria for this study, 20 who underwent WLE alone and 50 who underwent WLE with an SLNB. In the entire cohort, reasons for inclusion were T1b disease (71.4%), T1a lesions with positive deep margins (27.1%), and T1a lesions with high mitotic index (1.4%). There were no T1a patients with lymphovascular invasion. Patients who underwent SLNB were younger than patients who did not undergo SLNB (49.5 vs. 61.5 years; p = .03; Table 1), and more likely to have T1b lesions (82.0% vs. 45.0%, p < .01). There was no difference in patient gender, body mass index, American Society of Anesthesiologists class, insurance type, or admission status between groups.

TABLE 1.

Bivariate analysis of patient and case factors in SLNB and non-SLNB cases

	No SLNB (N = 20)		SLNB (N = 50)
	n	Summary	n	Summary	p
Age	20	61.5 (52.0-74.5)	50	49.5 (39.0-63.0)	.03m
Sex	20		50		.45c
Male		8 (40.0)		25 (50.0)
Female		12 (60.0)		25 (50.0)
Race	20		50		.23c
White		19 (95.0)		49 (98.0
Unknown		1 (4.8)		1 (2.0)
Body mass index	20	27.5 (23.8-31.8)	47	26.6 (24.2-31.5)	.59c
ASA class	20		50		.36c
1		0 (0.0)		6 (12.0)
2		16 (80.0)		33 (66.0)
3		4 (20.0)		10 (20.0)
4		0 (0.0)		1 (2.0)
Insurance type	20		50		.26c
Capitated		2 (10.0)		8 (16.0)
Contracted/FFS		9 (45.0)		30 (60.0)
Government		9 (45.0)		11 (22.0)
Other		0 (0.0)		1 (2.0)
Patient class	20		50		.31c
Outpatient		16 (80.0)		45 (90.0)
Short stay		4 (20.0)		4 (8.0)
Inpatient		0 (0.0)		1 (2.0)
Risk Factor	20		50		<.01c
T1a + pos. margin		11 (55.0)		8 (16.0)
T1a + mitosis ≥ 2		0 (0.0)		1 (2.0)
T1b		9 (45.0)		41 (82.0)

Note: Values presented as median (interquartile range) or N (column %).

Abbreviations: ASA, American Society of Anesthesiologists; FFS, fee for service; SLNB, sentinel lymph node biopsy.

^mMann-Whitney U test (p values).

^cPearson’s χ² test (p values).

Among patients who underwent SLNB, mapping was successful in 98% (49/50 patients; Table 2). The nondiagnostic study was due to the nonvisualization of an SLN on lymphoscintigraphy. In patients with successfully mapping, 46 patients had a negative SLN (93.9%) and 3 patients had a positive SLN (6.1%). The outcomes for patients with a positive SLN are shown in Table 3; two patients underwent subsequent LND that identified no additional positive nodes and remain disease free, and one patient underwent ultrasound surveillance according to Multicenter Selective Lymphadenectomy Trial II protocol and developed regional recurrence. With a median follow-up of 17 months for the entire cohort, there was only one distant recurrence and melanoma-specific death from either group, and this occurred in the SLNB cohort (data not shown).

TABLE 2.

Resource utilization and clinical outcomes in SLNB and non-SLNB cases

	No SLNB (N = 20)		SLNB (N = 50)
	n	Summary	n	Summary	p
Operative time (min)	20	36.0 (27.0–46.5.0)	50	68.5 (54.0–97.0)	<.01m
Cost components ($)	20		50
Direct fixed cost		230 (200–320)		502 (401–604)	<.01m
Indirect fixed cost		1696 (1142–1927)		2444 (1878–2750)	<.01m
Direct variable cost		1412 (1266–1895)		3196 (2665–3699)	<.01m
Indirect variable cost		463 (272–505)		597 (437–698)	<.01m
Total cost		3767 (3412–4300)		6700 (5409–7728)	<.01m
SLNB result	n/a		50		n/a
Positive		n/a		3 (6.0)
Negative		n/a		46 (92.0)
Nondiagnostic		n/a		1 (2.0)

Note: Values presented as median (interquartile range) or N (column %).

Abbreviations: n/a, not applicable; SLNB, sentinel lymph node biopsy.

^mMann-Whitney U test (p values).

TABLE 3.

Outcomes of patients with positive SLN

	Case year	Risk factor	Postoperative follow-up	Local recurrence	Distant recurrence	Systemic therapy	Follow-up duration
Patient 1	2018	T1b	Observation, delayed LND	Yes	No	No	19 months
Patient 2	2016	T1b	Immediate LND	No	No	No	41 months
Patient 3	2015	T1b	Immediate LND	No	No	No	54 months

Abbreviations: LND, lymph node dissection; SLN, sentinel lymph node.

Patients who underwent SLNB had a higher median total hospital cost ($6700 vs. $3767; p < .01) and increased operative time (68.5 vs. 36.0 min; p < .01). Each subcomponent of the total cost was also higher in patients that underwent SLNB, with direct variable cost ($3196 vs. $1412; p < .01) and indirect fixed cost having the largest differences ($2444 vs. $1696; p < .01). Using these costs, the additional cost to identify a single positive SLN in our 49 patients undergoing SLNB was $47,906. The cost difference between groups persisted on multivariable linear regression for cost, adjusting for clinically relevant variables (+$3524; p <.01; Table 4). No other factor included in the model was found to be significantly associated with hospital costs.

TABLE 4.

Multiple linear regression for total hospital cost

	Parameter estimatea	p
SLNB vs. no SLNB	+$3524	<.01i
Age	+$24	.20i
Body mass index (BMI)	+$1	.99i
ASA Class I vs. ASA Class III	−$158	.89i
ASA Class II vs. ASA Class III	−$406	.56i
Inpatient vs. short stay	+$1219	.54i
Outpatient vs. short stay	−$670	.35i

Abbreviations: ASA, American Society of Anesthesiologists; SLNB, sentinel lymph node biopsy.

^aParameter estimate for linear regression for total hospital cost controlling for sentinel lymph node biopsy, age, BMI, ASA rating, and patient admission class.

ⁱMultiple linear regression.

4 |. DISCUSSION

The findings in this study quantify the hospital cost associated with SLNB among patients with a higher risk of thin melanoma. SLNB was associated with an unadjusted 78% increase ($2933) in-hospital cost compared to WLE alone in this cohort, despite a low rate of SLN positivity (6.1%). We observed that it cost approximately $50,000 to identify a positive SLN in our cohort of 49 patients that underwent SLNB for higher risk thin melanoma. These differences persisted on both a multivariable regression for cost. Overall, these cost data are intended to better inform patients and providers in this clinical scenario, which is also influenced by multiple therapeutic considerations. These considerations include the benefit of adjuvant therapy, the patient’s ability to tolerate systemic therapy, and the increasingly effective options available for recurrent disease.^10–12

A cost analysis of SLNB in thin melanoma was most recently conducted by Agnese et al.⁷ This cohort included patients with Breslow thickness less than 1.2 mm and adverse pathologic characteristics (Clark Level III/IV, ulceration, lymphovascular invasion, and regression). The authors found that WLE and SLNB incurred a mean charge of $12,193 compared to a mean charge of $1466 for WLE alone.⁷ In their cohort, 2 out of 138 (1.4%) patients that underwent SLNB had a positive node, indicating that 69 patients would need to be evaluated for identifying one patient with a positive sentinel node.⁷ Using these data, the authors estimated that the cost to identify a single node would be $696,600-$1,051,000.⁷ There are important differences in methodology that must be noted when comparing our results to this study. Our study utilized actual hospital cost data rather than charges, which can be affected by patient regional, institutional, and insurance factors and are less representative of the overall cost to the healthcare system. In addition, Agnese et al.⁷ noted that their sentinel node specimens were not reviewed by a dedicated pathologist, which might have contributed their lower than expected rate of SLN positivity. Our observed rate of 6.1% is more in line with previous studies of patients with thin melanoma.^5,25,26 Overall, our study provides an updated cost analysis of SLNB in higher risk thin melanoma that is more consistent with known literature regarding the rate of SLN positivity in this cohort.

The increased cost conferred by SLNB might be attributed to a number of factors, including the cost of radioisotope and nuclear imaging, pathologic processing, and review, and increased operative time. The cost of consumables (isosulfan blue, radioisotope, etc.) and medications are reflected in direct variable cost, which accounted for the largest component of the cost difference associated with SLNB. Increased operative time contributes to higher facility costs, a component of indirect fixed cost—and this was the second highest contributor to the cost difference associated with SLNB. Admission status (outpatient vs. short stay) did not differ significantly between groups, suggesting that this did not account for the observed cost difference. Overall, our results suggest that no single factor explained to the increased hospital cost associated with SLNB in patients with higher risk thin melanoma.

There are important limitations to consider in our present work. The small sample size and single-institution cohort were necessary to utilize our unique hospital accounting dataset, but this also limits the generalizability of our findings. Hospital costs can vary significantly on a regional basis and these differences would not be captured by our analysis. In addition, its short median follow-up time (17 months) precluded meaningful recurrence analysis between groups, though the 5-year recurrence rate in this patient population is generally low.⁵ We also did not include patients that underwent WLE in an office setting; however, including these patients would only increase the cost difference between groups as patients undergoing WLE in the office setting will have even lower costs than what is estimated in this study. Our cost data also do not account for the cost of subsequent follow-up in patients that do not undergo SLNB, including surveillance imaging, or the cost of complications related to SLNB that develop the following discharge. There are currently no NCCN recommendations for surveillance for thin melanoma patients that do not undergo SLNB or those with negative SLNB, suggesting that ultrasound surveillance would not be indicated in 95% of patients in our study. Finally, our cost estimates do not account for the additional imaging and adjuvant therapy that would occur in patients with a positive SLNB. Adjuvant therapy was not routine for most patients with stage III melanoma during the study period and none of the three patients with a positive SLNB in our cohort received adjuvant therapy. As adjuvant therapy is increasingly utilized for stage III melanoma, our hospital cost findings will need to be considered in the context of evolving systemic therapy recommendations.

In summary, this study represents the first cost analysis of SLNB in higher risk thin melanoma to utilize actual hospital cost data. We found that SLNB incurs significant cost despite an overall low likelihood of SLN metastasis in this cohort. This information should be used to inform patient-provider discussions, and also highlights the potential fiscal benefit to the healthcare system by improving the selection of those patients most likely to benefit from this resource-intensive procedure.