Long-term outcomes after near-infrared sentinel lymph node mapping in non–small cell lung cancer

Abstract

Objective

To report the first analysis of long-term outcomes using near-infrared (NIR) image-guided sentinel lymph node (SLN) mapping in non–small cell lung cancer (NSCLC).

Methods

Retrospective analysis of patients with NSCLC enrolled in 2 prospective phase 1 NIR-guided SLN mapping trials, including an indocyanine green (ICG) dose-escalation trial, was performed. All patients underwent NIR imaging for SLN identification followed by multistation mediastinal lymph node sampling (MLNS) and pathologic assessment. Disease-free (DFS) and overall survival (OS) were compared between patients with NIR⁺ SLN (SLN group) and those without (non-SLN group).

Results

SLN detection, recurrence, DFS, and OS were assessed in 42 patients with NSCLC who underwent intraoperative peritumoral ICG injection, NIR imaging, and MLNS. NIR⁺ SLNs were identified in 23 patients (SLN group), whereas SLNs were not identified in 19 patients enrolled before ICG dose and camera optimization (non-SLN group). Median follow-up was 44.5 months. Pathology from NIR⁺ SLNs was concordant with overall nodal status in all 23 patients. Sixteen patients with SLN were deemed pN0 and no recurrences were, whereas 4 of 15 pN0 non-SLN patients developed nodal or distant recurrent disease. Comparing SLN versus non-SLN pN0 patients, the probability of 5-year OS is 100% versus 70.0% (P = .062) and 5-year DFS is statistically significantly improved at 100% versus 66.1% (P = .036), respectively. Among the 11 pN+ patients, 7 were in the SLN group, with>40% showing metastases in the SLN alone.

Conclusions

Patients with pN0 SLNs showed favorable disease-free and overall survival. This preliminary review of NIR SLN mapping in NSCLC suggests that pN0 SLNs may better represent true N0 status. A larger clinical trial is planned to validate these findings.

Graphical abstract

Lung cancer is the leading cause of cancer-related mortality in the United States, with more than 155,000 deaths expected in 2017¹. Despite early detection and surgical resection, nearly 30% of patients with early-stage lung cancer recur and 5-year survival remains between 60% and 80%.^2-4 Prognosis is strongly linked to the detection of metastatic disease in local and regional nodes, and survival is reduced by even the presence of occult micrometastatic disease.^5,6

Variability in lymphatic pathways and differences among surgeons in surgical lymph node staging remain critical barriers to accurate nodal staging in non-small cell lung cancer (NSCLC), resulting in understaging and the potential for residual occult nodal disease.^7-9 Furthermore, extensive lymphadenectomy (LAD) may result in unnecessary mediastinal dissection in true pathologic node negative (pN0) cases as suggested by findings of the ACOSOG Z0030 randomized trial comparing mediastinal lymph node sampling (MLNS) with LAD in early-stage disease.

The ability to systematically identify the first tumor-draining lymph nodes (ie, sentinel lymph nodes [SLNs]) has the potential to improve the accuracy of staging in NSCLC by scrutinizing those nodes most likely to harbor occult metastatic disease. Focused assessment of the first tumor-draining lymph node(s) could identify a subset of patients harboring occult nodal disease who may benefit from an extensive LAD and/or adjuvant therapy.¹¹ On further optimization of SLN mapping technology, true pN0 patients may be spared a complete LAD because this potentially compromises subsequent nodal staging for a second primary lung cancer, a relatively common occurrence among patients with a previous early-stage lung cancer.¹² Although SLN sampling has the potential to change the current standard of care, long-term outcomes from SLN staging in patients with NSCLC have not been reported.

Our group has previously reported the safety and feasibility of intraoperative minimally invasive near-infrared (NIR) SLN mapping in lung cancer.^13,14 Using an optimized dose of NIR-detectable indocyanine green (ICG) dye, SLNs were identified in 100% of patients after transpleural peritumoral injection of ICG and in 80% of patients after transbronchial ICG injection. All patients also underwent subsequent MLNS, and SLN disease was concordant with the overall nodal status in all NSCLC cases.

Given that nodal disease was not identified in any other node within the subsequent MLNS specimen in patients with pathologically negative SLNs, we hypothesized that if negative SLNs represent true pN0 status, these patients would show improved disease-free survival (DFS) and overall survival (OS) without evidence of subsequent nodal recurrence. Thus, the aims of the current study are to evaluate the long-term incidence of recurrence and survival outcomes of patients enrolled in our ongoing phase 1 trials of NIR SLN mapping in patients with NSCLC.

Methods

This study is a retrospective analysis of clinical outcomes from the prospective National Cancer Institute–funded NIR SLN mapping trial in NSCLC (ClinicalTrials.gov, NCT00264602) approved by the Dana-Farber Cancer Institute Internal Review Board (Boston, Mass) as well as a pilot trial in navigational bronchoscopy-guided SLN mapping approved by the Partners Internal Review Board (Boston, Mass). Patient accrual began in February 2009 and is ongoing. Eligible patients included those with known or suspected NSCLC undergoing lung resection. The technique of transpleural or transbronchial peritumoral injection of ICG and NIR imaging was performed as previously described.^13,14

All patients enrolled for NIR lymphatic mapping with known or suspected NSCLC were considered eligible for outcome analysis. One patient died during the immediate postoperative hospitalization of morbidity unrelated to the NIR study and was excluded from long-term outcome analysis. A patient with a suspected T4 lesion who received adjuvant chemotherapy and 2 patients with a Pancoast tumor and subsequently identified positive surgical margins were also excluded from analysis of recurrent disease given the high risk of local recurrence.

Initial patients were enrolled in a dose-escalation trial, and SLNs were not identified in the early patient cohort before ICG dose optimization. These patients underwent MLNS alone and served as the control group (non-SLN). With standardization of ICG dosing, SLNs were reliably identified, with these patients constituting the experimental SLN group.

Preoperative nodal staging was assessed via endobronchial ultrasonography (EBUS) and/or cervical mediastinoscopy (c-med) in patients based on clinical criteria. Importantly, all patients in this study had pathologic nodal staging via MLNS at the time of the lung resection as is standard of care, regardless of whether an SLN was identified. Routine microscopic evaluation of all nodes in the SLN and/or MLNS specimens was performed to evaluate for evidence of metastatic nodal disease in order to (1) assess whether SLN status is predictive of overall nodal status found via MLNS and (2) stratify patients as pN0 or pN⁺ based on the absence or presence of metastatic nodal disease, respectively.

Duration of follow-up was defined as the interval between the date of surgery and the last patient contact. Disease recurrence was characterized as local (ipsilateral parenchymal recurrence), regional (nodal), distant (extrathoracic and/or contralateral metastatic disease), or a combination of these categories and was defined as either a biopsy-proven recurrence or imaging features highly suspicious for recurrence with clear radiologic and clinical documentation that resulted in a change in treatment plan.

A Mann-Whitney U test was used for continuous data. A Fisher exact test and χ² analysis were used for unordered categorical data, and a Jonckheere-Terpstra test was used for ordered categorical data. Kaplan-Meier survival analysis was performed to evaluate DFS and OS with log-rank test for significance. For DFS analysis, patients were censored at the date last seen alive or at the date of death from other causes. GraphPad Prism 5.0 (La Jolla, Calif) or SAS (Cary, NC) were used for all analyses.

Results

Forty-two patients undergoing peritumoral ICG injection, MLNS, and tumor resection, with or without SLN identification, were included for outcome analysis with a median follow-up of 44.5 months (interquartile range [IQR] 19-69 months). Patient groups and outcomes for each group are detailed in Figure 1. Initial patients enrolled were in a dose-escalation trial, and SLNs were not reliably identified before ICG dose optimization (n = 14) or in an additional 5 patients because of technical issues with ICG injection or the imaging system. These patients underwent MLNS alone and served as the non-SLN control group (n = 19). With optimization of ICG dosing, at least 1 SLN was identified in 23 patients with 17 SLNs in the N1 station and 12 SLNs in the N2 station and a subsequent MLNS was performed. The detection of metastatic disease within the SLN in all pN⁺ cases, and the absence of disease in the SLN when all other nodes were negative, yielded a sensitivity, specificity, positive predictive value, and negative predictive value for NIR-identified SLN mapping of 100%.

Figure 1.

Study design and characteristics of sentinel lymph node (SLN) and non–sentinel lymph node groups. NIR, Near-infrared; MLNS, mediastinal lymph node sampling.

The demographic and pathologic characteristics of the SLN and non-SLN groups are listed in Table 1. By definition, patients in the SLN group have a slightly shorter follow-up because they represent the subsequent cohort of patients after dose escalation of ICG. Patients were predominantly female (76.2%) and current or former smokers (92.9%), with primary lung adenocarcinoma (85.7%) being the most common histologic diagnosis. The high incidence of sublobar resections in this retrospective analysis is secondary to the presence of pulmonary limitations, other clinical comorbidities, and/or radiographic characteristics that warranted a limited resection. There was preferential enrollment of patients with these characteristics for the localization trial detailed in Hachey et al. to show that this technology could be used to improve wedge resection and staging of lung lesions that are difficult to identify.¹⁴ There were no statistically significant differences in patient characteristics between the SLN and non-SLN groups, including extent of resection (sublobar vs lobar) or histologic subtype or grade, with the exception that the SLN group received a higher dose of ICG as expected during dose optimization.

Table 1. Demographic and pathologic characteristics of SLN and non-SLN cohorts.

Individual characteristics	SLN group (n = 23)	Non-SLN group (n = 19)	P value
Mean age, y (SD)	63 (8.2)	66 (11.1)	.37
Female sex, n (%)	18 (78.3)	14 (73.7)	1.00
Smoking status, n (%)
Never smoker	2	1
Current/former	21 (91.3)	18 (94.7)	1.00
>30 pack-y	13 (61.9)	10 (55.5)	.75
<30 pack-y	8 (38.1)	8 (44.4)
Mean ICG dose, mg (SD)	1.73 (1.0)	0.85 (1.06)	.01
Method of ICG injection			.17
Transpleural	14	16
Transbronchial	9	3
Extent of resection			.64
Wedge	10 (43.5)	6 (31.6)
Segmentectomy	3 (13.0)	3 (15.8)
Lobectomy	10 (43.5)	9 (62.5)
Pneumonectomy	0 (0)	1 (5.3)
Histologic diagnosis, n (%)			.53
Adenocarcinoma	20 (87.0)	16 (84.2)
Squamous cell carcinoma	2 (8.7)	3 (15.8)
Large cell with small cell foci	1 (4.3)	—
Histologic grade*			.38
Well differentiated	8 (34.8)	4 (21.1)
Moderately differentiated	9 (39.1)	8 (42.1)
Poorly differentiated	5 (21.7)	6 (31.6)
Mean tumor size, cm (SD)†	2.11 (1.5)	2.06 (1.5)	.76
Presence of LVI, n (%)	5 (21.7)	3 (15.8)	.71
Among pN0 cases‡	0 (0)	2 (13.3)	.23
Tumor mutation status, n (%)
EGFR	1 (4.3)	1 (5.3)	1.00
KRAS	5 (21.7)	2 (10.5)	.43

P values < .05 are shown in bold. SLN, Sentinel lymph node; SD, standard deviation; ICG, indocyanine green; LVI, lymphovascular invasion; pN0, pathologic node negative; EGFR, epidermal growth factor receptor; KRAS, Kirsten rat sarcoma viral oncogene.

^*One patient from SLN group without documentation of tumor differentiation.

^†Based on final pathology.

^‡For pN0 cases, n = 16 for SLN group and n = 15 for non-SLN group.

SLN Identify Metastatic Nodal Disease

Among the 23 patients in whom at least 1 SLN was identified, 7 were found on final pathology to have nodal metastases. Metastases were detected in the NIR-identified SLN(s) in all 7 patients, even although analysis of other nodes within the MLNS specimen detected metastases in only 4 of these patients. Therefore, the SLN was the only node containing metastases in 3 patients despite having nodes sampled previously via c-med or from the same station via MLNS during surgery, as shown in Figure 2. Overall, occult metastatic disease was identified in 30% of the SLN specimens (7 of 23 patients), which was higher than the incidence of metastatic disease detected within the other MLNS-harvested nodes in both the SLN cohort (4 of 23 patients, 17.4%) and the non-SLN cohort (4 of 19 patients, 21%) (P = 1.00). The incidence of nodal disease detected within the MLNS of either group is similar to the 10% to 18% incidence of occult nodal disease commonly reported in the literature for clinical early-stage lung cancer but less than the 30% detected by SLNs in the current study.^15,16

Figure 2.

Level 7 lymph node following near-infrared sentinel lymph node mapping. The blue arrow indicates the sentinel lymph node that was the only lymph node positive for metastatic disease within this nodal station. The white arrow indicates another lymph node in the same station that was near-infrared-negative and did not show evidence of metastatic disease.

Recurrence and Survival Outcomes in NSCLC pN⁺ (Node Positive) Patients

Most node positive cases had N1 disease (73%), with the SLN group containing 5 patients with N1 disease alone, 1 with both N1 and N2 disease, and 1 with N2 disease alone after examination of the SLN and MLNS specimens. Of the non-SLN patients, 3 had N1 disease and 1 patient had N2 disease within the MLNS specimen. All patients with nodal metastasis received appropriate adjuvant chemotherapy or chemoradiotherapy. Details regarding recurrence and survival outcomes for patients with positive nodes are outlined in Table 2.

Table 2. Recurrence and survival in pathologic node positive patients staged with sentinel lymph node (SLN) + mediastinal lymph node sampling (SLN group) versus mediastinal lymph node sampling alone (non-SLN group).

pN+ cohort characteristics	SLN group (n = 7)	Non-SLN group (n = 4)	P value
Number of LNs sampled, mean (SD)*	10.67 (6.6)	15 (11.05)	.45
Number of LN stations, mean (SD)	3.67 (1.75)	5.75 (2.22)	.14
Pathologic N stage, n (%)			1.0
1	6 (85.7)	3 (75.0)
2	1 (14.3)	1 (25.0)
Adjuvant chemo/chemoradiation, n (%)	7 (100.0)	4 (100.0)	1.0
Duration of follow-up from surgery, median, months (range)	44 (10–58)	76.5 (21–93)	.11
Lobectomy, n (%)†	4 (57.1)	4 (100.0)	.24
Adenocarcinoma, n (%)	7 (100.0)	2 (50.0)	.11
Poorly differentiated, n (%)	3 (42.9)	0 (0)	.24
Recurrence, n (%)‡	4 (57.1)	3 (75.0)	1.00
Time to recurrence, median (range)	10 (4-46)	17 (9-40)	.63
Lung cancer-specific mortality, n (%)	3 (42.9)	2 (50.0)	1.00
3-y outcomes§
Overall survival, (%)	71.4	75.0	.31
Disease-free survival, (%)	57.1	50.0	.53

SLN, Sentinel lymph node; LN, lymph node; SD, standard deviation.

^*Values include nodes sampled via cervical mediastinoscopy, endobronchial ultrasonography, sentinel lymph node, and other lymph nodes removed during surgery. Pathology reports that did not include the total number of nodes per station were defaulted to a single node, likely leading to underreporting of the nodes sampled.

^†Lobectomy (including pneumonectomy) versus sublobar resection.

^‡Time in months to initial diagnosis of nodal or distant recurrence.

^§Based on Kaplan-Meier analysis. Groups were compared in terms of time to event using log-rank test. For disease-free survival analysis, events included nodal or distant recurrences and disease-specific death.

Seven of 8 patients with lymphovascular invasion positive (LVI⁺) tumors in this study developed recurrent disease. Although there was no difference in incidence of LVI⁺ tumors between the groups, metastatic nodal disease was detected within the SLN in all LVI⁺ patients in the SLN group. This finding is in contrast to the non-SLN group, in which only 60% of patients with LVI⁺ tumors were initially staged as pN⁺. All LVI⁺ patients in the non-SLN group, including the 40% deemed pN0, recurred, suggesting that occult nodal disease was present but missed on routine histologic analysis of the MLNS specimen. The only LVI⁺ patient without evidence of recurrent disease is a patient with SLN in which metastatic disease was detected within the SLN alone and who subsequently received adjuvant therapy early in her postoperative course.

There is no significant difference in recurrence rates or survival among pN⁺ patients in the SLN or non-SLN groups, confirming that outcomes in patients with nodal disease are related to the extent of nodal disease itself rather than whether disease is detected within the SLN or in other nodes within the MLNS specimen. Four of the 7 SLN pN⁺ patients developed recurrent disease after adjuvant therapy (57%), including 2 with distant only disease, 1 with regional and distant disease, and 1 with regional disease alone. In the non-SLN group, 3 of the 4 patients (75%) recurred, with 1 patient having recurrence in the regional lymphatics and 2 patients developing distant metastases.

Recurrence and Survival Outcomes for pN0 Patients

Although 4 patients with metastatic disease in the SLN also had metastatic disease in additional nodes, no patient with a pathologically negative SLN had evidence of nodal metastases in any other node within the MLNS specimen. Therefore, a pN0 SLN correlated with the absence of metastatic disease (ie, pN0 status) in the MLNS specimen. This finding leads to the hypothesis that patients with pN0 SLN have more favorable long-term outcomes.

The recurrence rates and survival outcomes for 31 pN0 patients (16 in the SLN group and 15 in the non-SLN control) are detailed in Table 3. The overall median follow-up is 40 months (IQR, 14.5-69 months) with no significant difference in median follow-up between the SLN and non-SLN groups. No evidence of disease recurrence has been found in any patient in the pN0 SLN group. In contrast, despite the nodes in the MLNS specimen being deemed pN0, 4 patients in the non-SLN group have developed nodal and/or distant recurrent disease. An additional patient who developed a local recurrence in the non-SLN group after a segmentectomy was not included because this recurrence was not related to nodal staging. This situation results in a 26.7% recurrence rate within the non-SLN patients versus 0% in the SLN group (P = .04). The median time to recurrence was 11.5 months. Details of the regional and distant recurrences among non-SLN pN0 patients are summarized in Table 4.

Table 3. Recurrence and survival in pathologic node negative patients staged with sentinel lymph node (SLN) + mediastinal lymph node sampling (SLN group) versus mediastinal lymph node sampling alone (non-SLN group).

pN0 cohort characteristics	SLN, pN0 (n = 16)	Non-SLN, pN0 (n = 15)	P value
Number of LNs sampled, mean (SD)*	5 (3.0)	6 (3.5)	.40
Number of LN stations sampled, mean (SD)	3 (1.7)	4 (1.5)	.21
Duration of follow-up from surgery, median, in months (range)	25.5 (9-83)	47 (12-91)	.29
Lobectomy, n (%)†	6 (37.5)	6 (40.0)	1.00
Adenocarcinoma, n (%)	13 (81.3)	14 (93.3)	.60
Poorly differentiated, n (%)	2 (12.5)	6 (40)	.22
Recurrence, n (%)‡	0 (0)	4 (26.7)	.04
Time to recurrence, median (range)	NA	11.5 (9–27)	NA
Lung cancer-specific mortality, n (%)	0 (0)	2 (13.3)	.23
5-y outcomes§
Overall survival, (%)	100.0	70.0	.062
Disease-free survival, (%)	100.0	66.1	.036

P values<.05 are in bold. SLN, Sentinel lymph node; pN0, pathologic node negative; LN, lymph node; SD, standard deviation; NA, not applicable.

^*Values include nodes sampled via cervical mediastinoscopy, endobronchial ultrasonography, sentinel lymph node, and other lymph nodes removed during surgery. Pathology reports that did not include the total number of nodes per station were defaulted to a single node, likely leading to underreporting of the nodes sampled.

^†Lobectomy versus sublobar resection.

^‡Time in months to initial diagnosis of nodal or distant recurrence.

^§Based on Kaplan-Meier analysis. Groups were compared in terms of time to event using log-rank test. For disease-free survival analysis, events included nodal or distant recurrences and disease-specific death.

Table 4. Incidence of recurrence among pathologic node negative cases by presence or absence of sentinel lymph node identification.

	Recurrence description
Recurrence type	SLN group, pN0 (n = 16)	Non-SLN group, pN0 (n = 15)
Regional	n = 0	n = 2
	There were no cases of regional recurrence	R1: T1aN0, moderately to poorly differentiated adenocarcinoma, LVI+ pN0 on c-med and MLNS specimens RML lobectomy Perihilar recurrence at 53 mo and subsequent vertebral metastases Undergoing chemotherapy
		R2: T1aN0, poorly differentiated squamous cell carcinoma pN0 on MLNS specimens LUL wedge resection Treated intralobar recurrence but subsequently developed metastasis to locoregional lymph nodes and chest wall Undergoing chemotherapy
Distant	n = 0	n = 2
	There were no cases of distant metastasis	R3: T2bN0, poorly differentiated adenocarcinoma, LVI+ pN0 on c-med and MLNS specimens RUL lobectomy Distant parenchymal recurrence with multiple new PET avid nodules and brain metastases at 9 mo Died of metastatic disease at 47 mo postoperatively
		R4: T2aN0, moderately differentiated adenocarcinoma, KRAS+, VPI+ pN0 on MLNS specimen RUL wedge resection Developed sternal and rib metastases at 14 mo Undergoing chemotherapy

SLN, Sentinel lymph node; pN0, pathologic node negative; R#, individual patient # with a recurrence; LVI, lymphovascular invasion; MLNS, mediastinal lymph node sampling; RML, right middle lobe; LUL, left upper lobe; c-med, cervical mediastinoscopy; RUL, right upper lobe; KRAS, Kirsten rat sarcoma viral oncogene; PET, positron emission tomography; VPI, visceral pleural invasion.

This difference in recurrence does not appear to be secondary to differences in the cohort because there was no statistically significant difference between the SLN and non-SLN pN0 cohorts in terms of the numbers of patients with lobar versus sublobar resections, adenocarcinoma, or tumor differentiation.

Kaplan-Meier curves of OS and DFS for pN0 patients in both the SLN and non-SLN groups are shown in Figure 3. For pN0 patients, the probability of OS at 5 years postoperatively is 100% versus 70.0% for SLN and non-SLN groups, respectively (P = .062), and reaching statistical significance the probability of 5-year DFS is 100% and 66.1%, respectively (P = .036).

Figure 3.

A, Overall survival (P = .062). B, Disease-free survival (P = .036). Shaded area represents the 95% confidence interval. SLN, Sentinel lymph node; pN0, pathologic node negative.

Discussion

NIR imaging is increasingly used for intraoperative tumor localization and SLN mapping.^13,14,17,18 Although a variety of techniques for SLN mapping have been tested,^19,20 ICG injection remains one of the most attractive techniques because it avoids radiation exposure, has a high signal-to-background ratio, and allows realtime intraoperative images to be merged with NIR video during minimally invasive surgery.^11,13,14 Previous reports have focused on specific technical details, performance, and SLN detection yield, rather than the goal of determining whether identified SLNs are predictive of OS and DFS in early NSCLC. Therefore, we conducted the first retrospective analysis of long-term outcomes after NIR-guided SLN staging in patients with NSCLC and showed that patients with pathologically negative SLNs have a statistically significant lower recurrence rate and improved OS and DFS compared with patients deemed pN0 after MLNS alone.

Before ICG optimization in our dose-escalation NIR SLN mapping trial, SLNs were not identified, and therefore the non-SLN cohort was used as a control group for the subsequent cohort in which SLNs were identified. The groups were comparable given that all patients met the same enrollment criteria, received a peritumoral injection of ICG followed by NIR imaging for an SLN, and underwent standard MLNS and surgical resection of the tumor. Pathologic nodal status of both SLN and MLNS specimens was determined by a lung pathologist in a blinded fashion. There have been no nodal or distant recurrences or NSCLC-related deaths in any patient with an identified SLN found negative for malignancy (pN0 SLN). In contrast, 4 of 15 pN0 patients in the non-SLN group developed nodal or distant recurrence despite the pN0 status of the MLNS specimen. This difference suggests that the SLN status is a better representation of the true overall nodal status. Furthermore, with a median follow-up of more than 3 years, the pN0 SLN outcomes are better than those reported for early-stage NSCLC in previous studies.^2,3

Although this is a small initial study, the difference in recurrence and DFS between the SLN and non-SLN groups is statistically significant and may be to the result of several factors. SLN mapping guides the surgeon to the first tumor-draining lymph node(s) at greatest risk for metastatic disease. The ability to focus histologic analysis on tumor-specific SLNs, and thus identify true pN0 patients, may also at least partially explain the lower recurrence rates and better DFS noted in the pN0 SLN group, particularly because standard assessment of MLNS nodes was performed in a blinded fashion for both the SLN and non-SLN patients. The better outcomes among the pN0 SLN group do not seem to be secondary to bias within the cohorts because there was no statistically significant difference between the SLN and non-SLN pN0 cohorts in terms of number of patients undergoing lobar versus sublobar resections, pathologic diagnosis, tumor differentiation, or LVI. However, the higher rate of identification of occult pN⁺ disease among patients with tumors of worse prognosis in the SLN versus non-SLN groups (60% vs 0% in poorly differentiated tumors and 100% vs 33% in LVI⁺ tumors) does suggest that the improved outcomes seen in the SLN pN0 cohort may result from shifting of previously deemed pN0 patients with undetected nodal disease into the correct pN⁺ cohort.

SLN studies performed in other solid organs have shown the not-infrequent presence of more than 1 SLN (usually only 2).²¹ This situation has been shown to occur in the case of distally bifurcated lymphatic channels but also is likely to occur in the lung given the dual drainage patterns of the peribronchial and subpleural lymphatic pathways. We believe that the ability of NIR SLN mapping to identify SLNs within both N1 and N2 nodal stations as a function of individual patient differences in lymphatic drainage may reduce the potential for inadequate surgical nodal staging and missed occult disease as a result of surgeon variability in the aggressiveness and accuracy of nodal staging.^9,22

SLN technology may identify occult nodal disease that might otherwise be missed, altering the treatment course by necessitating adjuvant chemotherapy. LVI⁺ tumors have been associated with the development of early occult metastatic disease within lymph nodes and this correlates with our finding that of the 11 pN⁺ patients 6 had evidence of LVI on pathology.²³ In the SLN group, metastatic disease was identified in the SLN in all 5 patients with LVI⁺ tumors. This finding is in marked contrast to the non-SLN group in which 2 of the 3 LVI⁺ patients were deemed pN0 after standard histologic analysis of the MLNS specimen. Both of these patients subsequently developed metastatic nodal disease, suggesting that missed occult nodal disease in the MLNS specimen was responsible for the poorer outcomes in the non-SLN pN0 cohort. Although multiple previous studies have indicated worse outcomes associated with LVI, administering adjuvant chemotherapy based on LVI tumor status alone has not entered standard clinical practice.²⁴ Therefore, the ability of SLN mapping to detect early isolated metastases in the SLN may offer a method of earlier detection and, consequently, early initiation of adjuvant treatment. The only patient alive and disease free with an LVI⁺ tumor received early chemoradiation therapy for metastatic disease found only in the SLN, despite multiple nodes from the same level 7 station being negative in the MLNS specimen (Figure 2). Without targeted assessment of the SLN, adjuvant chemoradiation therapy would not have been initiated in the early postoperative period.

The mechanisms of lymphatic metastasis are not yet well understood, but the interplay between primary tumors and the SLN is increasingly recognized and merits further study. Early genomic and molecular changes in cancer cells promote evasion of apoptosis, downregulation of cell adhesion molecules, and crosstalk with immune cells in the tumor microenvironment, all of which serve to prime cells for motility and metastatic potential.^25,26 Important for lymphatic metastasis, cancer cells and tumor-associated macrophages produce various growth factors that promote the development of new lymphatic vessels and the remodeling of existing channels, thereby increasing lymphatic flow to tumor-draining lymph nodes and creating a lymphatic niche for metastatic tumor cells.²⁷ Once established within SLNs, cancer cells may continue to secrete growth factors that recruit lymphatic vessels beyond the SLN, resulting in distant organ dissemination, pointing to the significance of the sentinel lymph node.

The current study provides evidence that SLNs can be reliably identified in patients with NSCLC and presents an opportunity for focused study of the first population of tumor cells that metastasize from the primary tumor given the differential gene expression profiles and mutation status for key genes associated with carcinogenesis and metastasis.^28-33 Genomic profiling of metastatic disease within SLNs has significant potential to identify specific mutations and biomarkers that can increase our understanding of early metastatic and primary tumor behavior in patients with lung cancer.

Although we achieved our goal of evaluating the long-term recurrence and survival of patients after the NIR-mediated identification of SLNs, the current study is limited in its evaluation of overall outcomes and generalization to the NSCLC population as a whole given the small sample size, relatively short follow-up, high incidence of sublobar resections, and predominance of women and smokers within the cohorts. The current study is not a randomized study of SLN versus MLNS alone because all patients underwent MLNS or a formal LAD as per standard of care. Thus, we do not intend, nor do we have sufficient data, to propose that SLN mapping alone is equivalent to complete MLNS, particularly if nodal disease is present. However, our preliminary SLN data seem to support that definitive targeted SLN mapping may improve our ability over standard MLNS to identify patients who are truly pN0. If a pathologic negative SLN can be established to reflect a true overall pN0 status in subsequent prospective multicenter trials, additional aggressive LAD may not be necessary in future patients with small tumors, and selective SLN mapping would still allow targeted nodal staging if a patient were to subsequently develop a second primary tumor. In contrast, evidence of metastatic disease within the SLN is indicative of more advanced disease and advocates for a change to a neoadjuvant approach or a possible change in surgical resection with a subsequent radical LAD for both nodal staging and nodal clearance, depending on the nodes involved.

This is the first study to report long-term recurrence and survival outcomes in patients with NSCLC after NIR-guided SLN mapping. We have identified favorable outcomes without evidence of nodal recurrence or distant metastasis in patients with pathologic negative SLNs, thus supporting our hypothesis that negative tumor-associated SLNs may be representative of the true overall nodal stage in patients with resectable NSCLC. Similarly, if metastatic nodal disease was detected in any nodes, it was always detected in the SLN, thus serving as an accurate marker for more advanced disease and necessitating an LAD to characterize the extent of metastatic disease. Larger, multicenter trials using NIR-guided SLN mapping in patients with early lung cancer are planned both to establish the reproducibility and accuracy of SLN mapping and to begin the necessary molecular and biochemical analysis on tumor-specific SLN to improve our understanding of the tumor–nodal relationship that sanctions early nodal metastasis.

Central Message.

In non–small cell lung cancer, pathologically negative sentinel lymph nodes found by near-infrared sentinel node mapping show favorable outcomes including overall and disease-free survival.

Perspective.

Non–small cell lung cancer (NSCLC) prognosis is strongly linked to detection of metastatic regional lymph nodes (LN). Recurrence, despite negative nodes, suggests occult disease. Near-infrared image-guided sentinel lymph node (SLN) mapping offers an opportunity for targeted LN sampling and focused pathologic assessment. To date, long-term outcomes with SLN staging in NSCLC have not been reported.

Abbreviations and Acronyms

C-med

cervical mediastinoscopy

DFS

disease-free survival

EBUS

endobronchial ultrasonography

ICG

indocyanine green

IQR

interquartile range

LAD

lymphadenectomy

LVI

lymphovascular invasion

MLNS

mediastinal lymph node sampling

NIR

near-infrared

NSCLC

non-small cell lung cancer

OS

overall survival

pN0

pathologic node negative

SLN

sentinel lymph node