Validation of the lymph node ratio as a prognostic factor in patients with N1 nonsmall cell lung cancer

Authors

  • Sirisha Jonnalagadda BA,

    1. University of Medicine and Dentistry of New Jersey-Robert Wood Johnson Medical School, Piscataway, New Jersey
    Search for more papers by this author
  • Jacqueline Arcinega MPH,

    1. Division of General Internal Medicine, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Cardinale Smith MD,

    1. Division of Hematology and Oncology and Palliative Care Medicine, Mount Sinai School of Medicine, New York, New York
    Search for more papers by this author
  • Juan P. Wisnivesky MD, DrPH

    Corresponding author
    1. Division of General Internal Medicine, Mount Sinai School of Medicine, New York, New York
    2. Division of Pulmonary and Critical Care Medicine, Mount Sinai School of Medicine, New York, New York
    • Department of Medicine, Mount Sinai School of Medicine, One Gustave L. Levy Place, Box 1087, New York, NY 10029
    Search for more papers by this author
    • Fax: (212) 824-2317


  • The first author was responsible for the study concept and design, analysis and interpretation of the data, and drafting the article; the second author was responsible for the study concept and design, critical revision of the article for important intellectual content, final approval of the article, and administrative, technical, and logistic support; the third author was responsible for critical revision of the article for important intellectual content and final approval of the article; and the fourth author was responsible for the study concept and design, analysis and interpretation of the data, drafting the article, statistical expertise, and final approval of the article.

Abstract

BACKGROUND:

The number of positive lymph nodes (LNs) has been proposed as a prognostic indicator in N1 nonsmall cell lung cancer (NSCLC). However, the number of positive LNs is confounded by the number of LNs resected during surgery. The lymph node ratio (LNR) (the ratio of the number of positive LNs divided by the number of LNs resected) can circumvent this limitation. The prognostic significance of the LNR has been demonstrated in elderly patients with NSCLC. The objective of the current study was to evaluate whether a higher LNR is a marker of worse survival in patients with NSCLC aged ≤65 years who have N1 disease.

METHODS:

The Surveillance, Epidemiology, and End Results database was used to identify 4004 patients who underwent resection for N1 NSCLC. Patients were classified into 3 groups according to LNR (≤0.15, 0.16-0.5, and >0.5). Associations of the LNR with lung cancer-specific and overall mortality were evaluated using the Kaplan-Meier method. Stratified and Cox regression analyses were used to assess correlations between the LNR and survival after adjusting for other prognostic factors.

RESULTS:

Unadjusted analysis indicated that a higher LNR was associated with worse lung cancer-specific survival (P < .0001) and overall survival (P < .0001). Stratified and multivariate analyses also indicated that the LNR was an independent predictor of survival after controlling for potential confounders.

CONCLUSIONS:

The current results confirmed that the LNR is an independent prognostic factor for survival in patients with N1 NSCLC. This information may be used to identify patients who are at greater risk of cancer recurrence. Cancer 2011;. © 2011 American Cancer Society.

The extent of lymph node (LN) involvement is a well established prognostic factor for patients with nonsmall cell lung cancer (NSCLC). Patients with N0 disease (no regional LN involvement) or N1 disease (involvement of ipsilateral intrapulmonary, peribronchial, or hilar LNs) are amenable to surgery with the possibility of achieving good long-term outcomes.1, 2 Conversely, patients with N2 disease (involvement of ipsilateral mediastinal or subcarinal LNs) and N3 disease (involvement of contralateral LNs) usually receive chemotherapy and/or radiation and have 5-year survival rates of 15% to 30%.3

Patients with NSCLC who have N1 disease have variable outcomes. The risk of recurrence for patients with stage II disease can range widely from 7% to 55%.4 This heterogeneity in outcomes has important implications for decisions about the use of adjuvant treatments and postoperative surveillance, and it makes discussions about prognosis with patients and their families more difficult.

Previous studies have suggested that the number of positive LNs may provide independent prognostic data in patients with NSCLC.5-11 However, the maximal number of positive LNs is limited by the number of LNs sampled during surgery. Although the literature recommends resection of ≥10 LNs for accurate NSCLC staging, there is considerable practice variability, limiting the use of the actual number of positive LNs as a prognostic factor.11 This limitation can be addressed with the LN ratio (LNR), which is the ratio of the number of positive LNs to the total number of LNs resected during surgery. It has been demonstrated that the LNR can predict survival in colon, esophageal, gastric, bladder, and breast cancer.12-17 A recent study indicated that the LNR is also an independent predictor of survival in elderly patients (≥65 years) with N1 NSCLC.18 However, those results have not been validated in an independent population or in patients aged ≤65 years with NSCLC. In the current study, we used population-based data to assess the prognostic value of the LNR in an independent group of patients aged ≤65 years who had N1 NSCLC.

MATERIALS AND METHODS

Methods

Study patients were identified from the Surveillance, Epidemiology, and End Results (SEER) registry. SEER is a national registry that collects information on incidence, prevalence, and survival of patients with cancer from geographic regions representing approximately 28% of the US population.19 All patients who were included in the study had undergone resection (lobectomy or pneumonectomy) and had N1 NSCLC confirmed on pathologic staging between 1988 and 2007. The study was limited to patients with primary disease who were not diagnosed from death certificate data or autopsy.1 Patients who received preoperative radiation therapy (which can lead to down staging of LN involvement) and those who had incomplete information on tumor size, tumor extension, and LN involvement also were excluded from the study. The final cohort included 4004 patients.

The SEER registry provides detailed information regarding the extent of LN involvement, the number of positive LNs, and the number of LNs sampled during surgery. By using this information, the LNR was calculated for each patient. Consistent with previous studies, patients were classified into 1 of 3 LNR groups: ≤0.15, 0.16 to 0.5, and >0.5.18

Sociodemographic characteristics (age, sex, race/ethnicity, and marital status) were obtained from the SEER registry. Patients were staged according to the seventh edition of the American Joint Committee on Cancer Cancer Staging Manual (the tumor, lymph node, metastasis [TNM] classification) using SEER data on tumor extension and LN metastasis.20 Histology subtypes were classified as adenocarcinoma, squamous cell carcinoma, large cell carcinoma, bronchioalveolar carcinoma, or other. Information regarding surgical treatment was ascertained from the site-specific surgery codes available in SEER. By using these data, patients were classified as having undergone lobectomy (codes 20-45) or pneumonectomy (codes 50-70). The receipt of preoperative and postoperative radiation therapy (external-beam radiation) also was ascertained from SEER data.

The primary outcome was lung cancer-specific survival, because it allowed us to control for unrelated causes of death. We performed secondary analyses assessing the association of the LNR with all-cause mortality. To estimate lung cancer-specific survival, patients who died from causes other than lung cancer were classified as censored at the date of death. Cause of death information in SEER is provided by the National Center for Health Statistics and is obtained from state death certificates. Survival was calculated as the interval from the date of diagnosis to the date of death or the last follow-up date available in the SEER registry (December 31, 2007).

Statistical Analysis

Differences in the baseline characteristics across the 3 LNR groups were evaluated using the chi-square test. Survival curves were estimated for patients in the 3 LNR groups using the Kaplan-Meier method and were compared with the log-rank test. Survival also was estimated for the 3 LNR groups after stratifying by age, sex, race, histology, tumor site and status, type of surgery, and receipt of radiation therapy to assess whether prognostic differences across the 3 LNR groups remained significant after controlling for these confounders. Survival rates were calculated up to 15 years after diagnosis to avoid reporting survival estimates based on a small number of observations. Cox regression analysis was used to assess the association between the LNR and survival after adjusting for other prognostic factors. The assumption of proportionality of hazards was evaluated using log-log plots of survival curves. All analyses were performed using the SPSS statistical software package (IBM, Chicago, Ill) with 2-sided P values.

RESULTS

We identified 4004 patients with N1 NSCLC from the SEER database. Overall, 34%, 43%, and 23% of the study patients had an LNR of ≤0.15, 0.16 to 0.5, and >0.5, respectively. There were differences in the distribution of sociodemographic characteristics (sex, race/ethnicity, marital status) and tumor characteristics (histology, location, status, size) between the LNR groups (P < .05 for all comparisons) (Table 1). In addition, patients in the highest LNR group were more likely to have adenocarcinoma (P < .0001) or an upper lobe tumor (P < .0001) and more often underwent lobectomy (P < .0001) or received postoperative radiation therapy (P < .0001).

Table 1. Baseline Characteristics of Patients Aged ≤65 Years With Nonsmall Cell Lung Cancer and N1 Lymph Node Involvement
 No. of Patients (%) 
CharacteristicLNR ≤0.15, N=1346LNR 0.16-0.50, N=1737LNR ≥0.5, N=921P
  1. LNR indicates lymph node ratio.

Age, y    
 ≤50255 (19)305 (18)160 (18).48
 51-60609 (45)838 (48)427 (46) 
 ≥60482 (36)594 (34)334 (36) 
Women526 (39)707 (41)427 (46).002
Race/ethnicity    
 White1110 (83)1363 (79)661 (72)<.0001
 African American111 (8)200 (11)131 (14) 
 Hispanic52 (4)60 (3)36 (4) 
 Other73 (5)114 (7)93 (10) 
Married908 (68)1112 (64)571 (62).02
Histology    
 Adenocarcinoma646 (48)890 (51)546 (59)<.0001
 Squamous cell carcinoma498 (37)568 (33)214 (23) 
 Large cell carcinoma71 (5)99 (6)60 (7) 
 Bronchioalveolar cell carcinoma58 (4)89 (5)53 (6) 
 Other73 (6)91 (5)48 (5) 
Tumor location    
 Upper lobe772 (58)1001 (58)507 (55)<.0001
 Middle lobe43 (3)76 (4)52 (6) 
 Lower lobe404 (30)525 (30)315 (34) 
 Other location127 (9)135 (8)47 (5) 
Tumor classification    
 T1A147 (11)187 (11)121 (13).01
 T1B207 (15)265 (15)174 (19) 
 T2154 (11)192 (11)112 (12) 
 T2A415 (31)598 (35)284 (31) 
 T2B173 (13)215 (12)103 (11) 
 T3250 (19)280 (16)127 (14) 
Type of surgery    
 Lobectomy937 (70)1286 (74)770 (84)<.0001
 Pneumonectomy409 (30)451 (26)151 (16) 
Postoperative radiation therapy    
 Yes384 (28)670 (39)438 (48)<.0001
 No962 (72)1067 (61)483 (52) 

Unadjusted survival analysis indicated that a higher LNR group was associated with worse lung cancer-specific survival (P < .0001) (Fig. 1A) and overall survival (P < .0001) (Fig. 1B). The median lung cancer-specific survival was 7.2 years, 4.3 years, and 2.6 years for patients in the 0.15 LNR, 0.16 to 0.5 LNR, and >0.5 LNR groups, respectively. Stratification by tumor status, an established prognostic factor, also indicated a significant association between a higher LNR and worse lung cancer-specific and overall survival (Fig. 2A-F). Similar results were obtained in stratified analyses by age, sex, marital status, tumor size, type of surgery, and receipt of radiation therapy (Table 2).

Figure 1.

Analyses of (A) lung cancer-specific survival and (B) overall survival are illustrated according to the lymph node ratio (LNR).

Figure 2.

(A-F) Lung cancer-specific survival and overall survival are illustrated according to the lymph node ratio (LNR) for patients with (A,B) T1 disease, (C,D) T2 disease, and (E,F) and T3 disease.

Table 2. Stratified Comparisons of Median Lung Cancer-Specific Survival With Lymph Node Ratio Group in Patients With Nonsmall Cell Lung Cancer
 Median Lung Cancer-Specific Survival, y 
CharacteristicLNR ≤0.15LNR 0.16-0.50LNR ≥0.5P
  1. LNR indicates lymph node ratio.

Age, y    
 ≤5010.084.083.00<.0001
 51-607.254.332.75<.0001
 ≥605.504.332.25<.0001
Women7.504.752.83<.0001
Race/ethnicity    
 White7.084.332.50<.0001
 African American5.584.172.92.01
 Hispanic 3.583.42.15
 Other7.584.922.92.005
Married7.254.502.67<.0001
Histology    
 Adenocarcinoma5.754.172.58<.0001
 Squamous cell carcinoma11.004.672.42<.0001
 Large cell carcinoma3.172.422.50.73
 Bronchioalveolar cell carcinoma7.335.002.67<.0001
 Other11.173.923.58.07
Tumor location    
 Upper lobe9.255.252.67<.0001
 Middle lobe10.084.924.25.13
 Lower lobe5.753.582.17<.0001
 Other location5.003.332.92.47
Tumor classification    
 T1A 6.754.42<.0001
 T1B5.585.173.33.001
 T28.005.752.42<.0001
 T2A13.334.332.58<.0001
 T2B5.922.502.25.001
 T34.582.921.75<.0001
Type of surgery    
 Lobectomy8.004.502.83<.0001
 Pneumonectomy5.173.581.75<.0001
Postoperative radiation therapy    
 Yes7.004.082.50<.0001
 No7.174.582.75<.0001

The LNR also was an independent predictor of lung cancer-specific and overall survival in Cox regression analyses that were adjusted for potential confounders (Table 3). Compared with patients in the lowest LNR group (≤0.15), there was a 1.28 (95% confidence interval [CI], 1.15-1.43) and 1.96 (95% CI, 1.74-2.22) increased hazard of lung cancer mortality in the 0.16 to 0.5 LNR group and the ≥0.5 LNR group, respectively. A similar pattern was observed in analyses using all-cause mortality as an outcome; compared with the ≤0.15 LNR category, an increased hazard of overall mortality of 1.25 (95% CI, 1.14-1.38) and 1.82 (95% CI, 1.63-2.03) was observed among patients in the 0.16 to 0.5 LNR group and the ≥0.5 LNR group, respectively.

Table 3. Adjusted Association Between the Lymph Node Ratio and Survival in Patients With N1 Nonsmall Cell Lung Cancer
 HR (95% CI)
VariableLung Cancer-Specific SurvivalOverall Survival
  1. HR indicates hazard ratio; CI, confidence interval.

Lymph node ratio  
 ≤0.15ReferenceReference
 0.16-0.51.28 (1.15-1.43)1.25 (1.14-1.38)
 >0.51.96 (1.74-2.22)1.82 (1.63-2.03)
Age1.02 (1.00-1.04)1.02 (1.01-1.04)
Women0.87 (0.80-0.96)0.86 (0.79-0.93)
Race/ethnicity  
 WhiteReferenceReference
 Black1.02 (0.88-1.18)1.11 (0.98-1.26)
 Hispanic0.90 (0.71-1.16)0.99 (0.80-1.22)
 Other0.87 (0.73-1.04)0.88 (0.74-1.03)
Married0.89 (0.81-0.97)0.84 (0.78-0.92)
Histology  
 AdenocarcinomaReferenceReference
 Squamous cell carcinoma0.73 (0.65-0.82)0.80 (0.73-0.89)
 Large cell carcinoma1.09 (0.91-1.31)1.12 (0.96-1.32)
 Bronchioalveolar carcinoma0.85 (0.68-1.05)0.92 (0.76-1.11)
 Other0.94 (0.78-1.15)0.90 (0.75-1.08)
Tumor classification  
 T1AReferenceReference
 T1B1.31 (1.09-1.57)1.29 (1.10-1.52)
 T21.34 (1.10-1.63)1.30 (1.09-1.55)
 T2A1.38 (1.17-1.62)1.35 (1.17-1.56)
 T2B1.67 (1.38-2.02)1.59 (1.34-1.88)
 T31.91 (1.59-2.29)1.72 (1.46-2.02)
Tumor location  
 Upper lobeReferenceReference
 Middle lobe1.00 (0.80-1.25)0.98 (0.80-1.20)
 Lower lobe1.20 (1.09-1.33)1.24 (1.13-1.35)
 Other location1.06 (0.89-1.26)1.12 (0.96-1.31)
Type of surgery  
 LobectomyReferenceReference
 Pneumonectomy1.22 (1.09-1.37)1.22 (1.10-1.35)
Postoperative radiation therapy1.07 (0.97-1.17)1.05 (0.97-1.14)

DISCUSSION

Patients with N1 NSCLC have heterogeneous outcomes. A recent study demonstrated that the LNR is a prognostic factor in elderly patients with N1 disease.18 In the current study, we validated those findings in a large, population-based cohort of patients aged ≤65 years with N1 NSCLC. We observed a consistent increased risk of mortality with higher LNRs after controlling for other prognostic factors from current lung cancer staging classifications. Moreover, the increased hazard of death among patients with an LNR ≥0.5 was equivalent to that among patients with T3 disease, which is an established prognostic factor. These results suggest that the LNR can be used to stratify patients with N1 disease into subgroups with different risks of lung cancer recurrence after resection.

It has been demonstrated that the LNR is an important prognostic factor in several malignancies, including breast, bladder, colon, esophageal, and gastric cancers.12-17 Similarly, Bria et al observed an association between the LNR and lung cancer outcomes in a cohort of 415 patients with N1 and N2 disease.11 More recently, a study using the SEER-Medicare registry indicated that the LNR was associated independently with overall and lung cancer-specific survival in a large cohort of elderly patients.18 Although those results support the usefulness of the LNR, there is a need to validate the findings in an independent population before they can be adopted in clinical practice. Moreover, elderly patients with lung cancer may be less likely to experience cancer recurrence, because they have a greater chance of dying from competing risks. Thus, it is also important to replicate these findings among younger patients with N1 NSCLC.

Our results support future adoption in clinical practice of the LNR as a prognostic factor for N1 NSCLC. The LNR provides physicians independent information that should lead to more accurate staging and can assist physicians with targeting adjuvant treatment for their patients who have NSCLC. Patients with a higher LNR appear to be at an increased risk of recurrence and should be considered for aggressive postoperative treatment to improve their long-term outcomes. Current treatment guidelines for N1 NSCLC recommend surgical resection followed by adjuvant, platinum-based chemotherapy.21 However, there are risks of acute toxicity and long-term adverse effects from adjuvant chemotherapy.22 Patients with advanced age, multiple comorbidities, or poor performance status may not be able to tolerate or experience the long-term benefits of adjuvant chemotherapy. Thus, physicians may consider withholding adjuvant therapy for these patients if they have an LNR <0.15. Although validated biomarkers may be available in the future, the prognostic value of the LNR can help treating physicians provide patients with more accurate information regarding their chances of recurrence, regardless of the number of LNs dissected. This information can assist patients and their families make important treatment decisions and long-term plans.

The use of postoperative radiation therapy (PORT) for patients who have lung cancer with LN involvement remains controversial.23, 24 The PORT meta-analysis produced no evidence of a clear benefit from PORT.23 Similarly, in a retrospective study using the SEER database, no significant improvement in survival was observed with PORT among patients with N1 NSCLC.24 In our study, patients with a higher LNR were more likely to receive PORT. However, a higher LNR was associated with worse survival in analyses that were stratified and adjusted for PORT receipt. Future studies evaluating the role of PORT should control for the LNR, which is a potential confounder of the association between the receipt of radiation therapy and survival.

The number of N1 LNs also has been proposed as a prognostic factor for patients with resected NSCLC. Marra et al observed an association between the number of positive LNs and survival in a cohort of patients with resected N1 NSCLC.25 Similarly, a study in Japan demonstrated that a higher number of positive LNs resulted in worse survival among patients with N1 and N2 disease.26 However, the maximal number of positive LNs is determined by the number of LNs sampled during surgery. Because many patients with lung cancer (>50%) undergo sampling of <10 LNs, this criteria may not be applicable to all patients in routine clinical practice. Although future efforts should focus on adequate sampling of LNs for patients who undergo lung cancer resection, the LNR can be used for predicting the risk of recurrence regardless of the extent of LN sampling.

There are some limitations to the current study. There was a significant difference in baseline characteristics between patients in the 3 LNR groups. Because several of these factors are established prognostic indicators for resected NSCLC, there is a possibility of confounding. However, stratified and multivariate analyses revealed a consistent association between the LNR and survival, suggesting that the LNR is an independent prognostic factor.

The assessment of lung cancer-specific survival requires accurate information about the cause of death. The SEER Program determines the cause of death from death certificates, which are a potential source of misclassification. However, the underlying cause of patients dying from lung cancer progression was identified as relatively accurate in a study using a larger registry.27 In addition, we confirmed our results in secondary analyses using all-cause mortality, which should not be subject to misclassification. SEER does not include information on chemotherapy; however, our study excluded patients who had received neoadjuvant radiation therapy, which usually is given in conjunction with neoadjuvant chemotherapy, decreasing the possibility that these patients were included in our cohort. Although less detailed than analyses from single institutions, an advantage of using registry data includes the large sample size and the improved generalization of the study findings.

The LNR may provide useful prognostic information to make important treatment decisions for patients with N1 NSCLC. Our findings indicated that the LNR is an independent prognostic factor in patients aged ≤65 years with resected N1 NSCLC, validating findings from previous studies. Patients with a higher LNR have worse outcomes and should be considered for more aggressive treatment and increased surveillance for recurrence. Our results provide support for incorporation of the LNR into future lung cancer staging systems.

FUNDING SOURCES

This study was funded by the Doris Duke Foundation for Clinical Research and in part, by the National Cancer Institute (5R01CA131348-03).

CONFLICT OF INTEREST DISCLOSURES

Juan Wisnivesky is a member of the EHE International research board, has received lecture honorarium from Novartis Pharmaceuticals, and has received a Chronic Obstructive Pulmonary Disorder research grant from Glaxo-Smith-Kline.

Ancillary