Advertisement

Lung Cancer in the Very Young: Treatment and Survival in the National Cancer Data Base

Open ArchivePublished:April 18, 2016DOI:https://doi.org/10.1016/j.jtho.2016.03.023

      Abstract

      Introduction

      Young patients with lung cancer represent a distinct subset of patients with this disease. The National Cancer Data Base includes patients of all ages and contains detailed staging, treatment, and survival information. The objective of this study was to examine treatment patterns and outcomes in young patients with non–small cell lung cancer (NSCLC).

      Methods

      The National Cancer Data Base was queried for NSCLC cases from 2003 to 2009. Younger patients were defined as those aged 20 to 46 years. Older patients were defined as those aged 47 to 89 years. Patient demographics, tumor characteristics, treatment, and survival were analyzed. The primary outcomes were 5-year overall and relative survival.

      Results

      The study included 173,856 patients; 5657 were 20 to 46 years of age. Younger patients were treated differently and received more aggressive therapy at each stage. At stage I, 64% of younger patients received surgery only versus 55% of the older patients (p < 0.0001). Younger patients had improved survival at all stages. This effect was more pronounced at earlier stages (the hazard ratios for the older group were 1.84, 1.62, 1.18, and 1.14 for stages I through IV, respectively [all p < 0.0001]). The absolute differences in 5-year overall survival between the younger and older groups were 25% for stages I and II but only 9% and 2% for stages III and IV, respectively.

      Conclusions

      Overall and relative survival in younger patients with NSCLC is better than in older patients, with greater benefit seen in earlier stages. Despite having fewer comorbidities and undergoing more aggressive treatment, younger patients with advanced-stage NSCLC have only marginally better overall and relative survival than older patients.

      Keywords

      Introduction

      Lung cancer is the leading cause of cancer mortality in the United States.
      • Siegel R.L.
      • Miller K.D.
      • Jemal A.
      Cancer statistics, 2015.
      Although lung cancer classically remains a disease of older patients, there is a distinct and significant subset of patients in whom lung cancer is diagnosed at a young age. Some studies have concluded that outcomes between older and younger patients are similar,
      • Gadgeel S.M.
      • Ramalingam S.
      • Cummings G.
      • et al.
      Lung cancer in patients <50 years of age: the experience of an academic multidisciplinary program.
      • Maruyama R.
      • Yoshino I.
      • Yohena T.
      • et al.
      Lung cancer in patients younger than 40 years of age.
      • Mauri D.
      • Pentheroudakis G.
      • Bafaloukos D.
      • et al.
      Non-small cell lung cancer in the young: a retrospective analysis of diagnosis, management and outcome data.
      • Skarin A.T.
      • Herbst R.S.
      • Leong T.L.
      • et al.
      Lung cancer in patients under age 40.
      whereas other studies have shown improved survival in younger patients.
      • Lara M.S.
      • Brunson A.
      • Wun T.
      • et al.
      Predictors of survival for younger patients less than 50 years of age with non-small cell lung cancer (NSCLC): a California Cancer Registry analysis.
      • Nugent M.D.
      • Edney M.T.
      • Hammerness P.G.
      • et al.
      Non–small cell lung cancer at the extremes of age: impact on diagnosis and treatment.
      • Subramanian J.
      • Morgensztern D.
      • Goodgame B.
      • et al.
      Distinctive characteristics of non-small cell lung cancer (NSCLC) in the young: a Surveillance, Epidemiology, and End Results (SEER) analysis.
      • Thomas A.
      • Chen Y.
      • Yu T.
      • et al.
      Trends and characteristics of young non-small cell lung cancer patients in the United States.
      • Kuo C.-W.
      • Chen Y.-M.
      • Chao J.-Y.
      • et al.
      Non-small cell lung cancer in very young and very old patients.
      These studies have shown that younger patients with lung cancer are more likely to be female, be non-smokers, and present at more advanced stages of disease. Most of the studies are either single-institution studies or use the Surveillance, Epidemiology, and End Results (SEER) database, which does not provide detailed treatment data unless paired with the Medicare-linked database. The Medicare data do not have much information on young patients. The National Cancer Data Base (NCDB) has an inherent benefit over other databases because of the inclusion of patients of all ages with a wealth of cancer-related information, making it well suited for examining the care of younger patients with non–small cell lung cancer (NSCLC). We studied differences in patient, tumor, and treatment characteristics between younger and older patients to better understand the factors that affect survival in this unique set of patients.

      Materials and Methods

      Data Source

      The NCDB is a hospital-based tumor registry run by the American College of Surgeons and the American Cancer Society. Currently there are approximately 1500 Commission on Cancer–accredited facilities that are mandated to report all new cancer cases to the NCDB. These hospitals represent approximately 30% of all hospitals in the United States and are estimated to capture approximately 70% of all newly diagnosed lung cancer cases.
      • Lerro C.C.
      • Robbins A.S.
      • Phillips J.
      • et al.
      Comparison of cases captured in the National Cancer Data Base with those in population-based central cancer registries.
      The SEER database captures only approximately 28% of all cancer cases.
      • Lerro C.C.
      • Robbins A.S.
      • Phillips J.
      • et al.
      Comparison of cases captured in the National Cancer Data Base with those in population-based central cancer registries.
      The NCDB captures detailed staging information as well as information on initial treatment and survival. The data used in this study are derived from a de-identified NCDB file, and thus, this study is exempt from Institutional Review Board approval.

      Patient Selection

      The NCDB participant user file for lung cancer was queried for all patients between the ages of 20 and 89 years inclusively in whom NSCLC was diagnosed from 2003 through 2009. Patients were excluded if their lung cancer was not their first primary cancer, if they were missing clinical stage information or tumor size, if they received nonstandard of care therapy (e.g., clinical trial, immunologic, hormonal, or ablative therapy and wedge resections), if their first course of treatment was unknown, and if they had positive margins or received postoperative radiation therapy alone without postoperative chemotherapy (Fig. 1). The NCDB began capturing comorbidity data in the form of a modified Charlson-Deyo score in 2003, and the seventh edition of the American Joint Committee on Cancer (AJCC) cancer staging manual came out in 2010, so patients in whom cancer was diagnosed before 2003 and after 2009 were excluded to ensure complete comorbidity data and to avoid the changes to the staging system. Patients were also excluded if there was no information on survival and if there was a mismatch between the clinical tumor, node, and metastasis indicators and overall clinical stage. This mismatch occurred if the combination of the clinical tumor, node, and metastasis indicators for a particular patient were not consistent with the resultant clinical stage. These mismatches were likely to due to coding errors or other nuances and were eliminated to avoid any confounding data. Patients were defined as young if their age at diagnosis was greater than two standard deviations less than the mean age of diagnosis. Therefore, patients aged 46 years or younger at the time of diagnosis were considered young. The older group comprised patients aged 47 to 89 years.
      Figure thumbnail gr1
      Figure 1Exclusion criteria and study design. Age is represented in years. NSCLC, non–small cell lung cancer; TNM, tumor, node, and metastasis; PORT, postoperative radiation therapy.

      Data Elements

      The primary outcomes were 5-year overall and relative survival. Independent variables included sex, race, ethnicity, facility location, proximity to an urban area, insurance type, income level, education level, modified Charlson-Deyo score, facility type, tumor size, histologic diagnosis, primary tumor site, tumor grade, laterality, AJCC sixth-edition clinical stage, initial treatment, and presence or absence of palliative care. The NCDB uses a modified Charlson-Deyo score, with patients with two or more comorbidities grouped together. Tumor size was treated as a categorical variable. Although there is significant overlap, tumor size was chosen instead of clinical tumor stage to separate the effects of tumor size from other tumor characteristics that are encompassed in the tumor stage. Furthermore, with the transition from sixth-edition AJCC staging to seventh-edition staging, many of the definitions for tumor stage changed and most of these changes were related to tumor size.
      • Greene F.L.
      • Page D.L.
      • Fleming I.D.
      • et al.
      AJCC Cancer Staging Manual.
      • Edge S.
      • Byrd D.
      • Compton C.
      • et al.
      AJCC Cancer Staging Manual.
      Additionally, in the upcoming eighth edition of the lung cancer staging system, it appears that there will be further modifications based on tumor size.
      • Goldstraw P.
      • Chansky K.
      • Crowley J.
      • et al.
      The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition of the TNM classification for lung cancer.
      Therefore, tumor size was thought to be more applicable to the current AJCC seventh-edition staging definitions than to the AJCC sixth-edition tumor stage.
      • Greene F.L.
      • Page D.L.
      • Fleming I.D.
      • et al.
      AJCC Cancer Staging Manual.
      • Edge S.
      • Byrd D.
      • Compton C.
      • et al.
      AJCC Cancer Staging Manual.
      Initial treatment was determined using specific data fields in the NCDB to determine the individual therapies received as well as the time after diagnosis at which each therapy was initiated. Patients who had therapy that began more than 180 days from initial diagnosis were excluded. The use of adjuvant and neoadjuvant therapy was calculated by comparing the therapy start dates for surgery and the additional therapy. Chemotherapy was defined as having received multiagent chemotherapy. Radiation therapy was defined as having received at least 4500 cGy combined between the initial and boost dose, and only patients who received radiation to the lung and chest wall were included. Missing data for each variable were coded as unknown and included in the multivariable models.

      Statistical Analysis

      Bivariate analysis of the independent variables was done using the chi-square test to compare characteristics between the two age groups. Differences in treatment between the groups were analyzed at different stages using the chi-square test. Survival analysis was performed using the Kaplan-Meier product-limit technique and compared by the log-rank test. Patients were stratified by clinical stage. Relative survival analysis was performed using life table data from the Centers for Disease Control and Prevention from 2003 through 2011.
      • Arias E.
      United States life tables, 2011.
      Relative survival was defined as the ratio of the observed to expected survival rates for patients with a particular age, sex, race, and year of death and is a surrogate marker for cancer-specific survival.
      • Howlader N.
      • Ries L.A.G.
      • Mariotto A.B.
      • et al.
      Improved estimates of cancer-specific survival rates from population-based data.
      Race was not included in the relative survival analysis because race as defined by the life table data did not match the race data as defined in the NCDB. Thus, our relative survival analysis was based on age, sex, and year of death. This analysis was also stratified by clinical stage. A Cox proportional hazards model was used to analyze predictors of survival at each stage. All independent variables, including age group, were entered into the model. A backward elimination technique was used with a threshold of p less than 0.10 for inclusion in the final model. No variables were forced into the model. All statistical analysis except that of relative survival was performed using the standard package SAS version 9.4 (SAS Institute, Cary, NC). Relative survival analysis was performed using the “relsurv” package in R version 3.2.2 (R Foundation for Statistical Computing, Vienna, Austria.). A p value less than 0.05 was considered significant. All statistical tests were two sided.

      Results

      Patient Population

      The final study population consisted of 173,856 patients identified using the demographic, tumor, and treatment characteristics as already described. Patient characteristics are presented in Table 1. There were 5657 patients between the ages of 20 and 46 years (mean age 42.4, standard deviation 4.0) and 168,199 patients between the ages of 47 and 89 years (mean age 68.7, SD = 9.9). The younger patients were more likely to be female (50% versus 46%, p < 0.0001), less likely to be white (77% versus 86%, p < 0.0001), and healthier as measured by a Charlson-Deyo score of 0 (76% versus 58%, p < 0.0001). Their tumors were more likely to be adenocarcinoma (49% versus 39%, p < 0.0001), were less likely to be squamous cell carcinoma (16% versus 30%, p < 0.0001), tended to present with more tumors larger than 7 cm (17% versus 12%, p < 0.0001) and more node-positive disease (60% versus 51%, p < 0.0001), and consequently were more likely to present at a higher stage (35% versus 31% stage III, p < 0.0001, and 39% versus 31% stage IV, p < 0.0001). Younger patients began their treatment sooner than older patients (mean 30.2 days versus 38.1 days, p < 0.001).
      Table 1Patient Characteristics
      CharacteristicAge, y
      20–4647–89
      n%n%
      SexMale27764990,96154
      Female28815177,23846
      RaceNonwhite12772324,19614
      White43807714,00386
      EthnicityNon-Hispanic491787147,65188
      Hispanic203442423
      Unknown537916,30610
      RegionNortheast10831931,40119
      Midwest15582847,11428
      South23504265,43939
      West6661224,24514
      Facility typeCommunity cancer program6911221,61713
      Comprehensive cancer program28635197,23958
      Academic/research program20933749,05129
      Other1002920
      Median income<$38,00014212535,92821
      $38,000–47,99914982642,90826
      $48,000–62,99913562443,10626
      ≥$63,00011662140,61924
      Unknown216456383
      High school education≥21%12662231,43119
      13–20.9%16853047,52928
      7–12.9%16542952,82431
      <7%8411530,86718
      Unknown211455483
      Charlson-Deyo score043007697,87558
      110851948,83329
      2+272521,49113
      Histologic diagnosisAdenocarcinoma27834965,92239
      Squamous cell carcinoma8891650,32630
      Large cell carcinoma325669244
      Other14392537,85723
      Bronchioloalveolar carcinoma221471704
      Grade1281586885
      29111634,85321
      320823753,27732
      4164334762
      Unknown22193967,90540
      Tumor locationMain bronchus/carina/hilum295558543
      Upper lobe/lingula34756197,70358
      Middle lobe227470754
      Lower lobe11542045,89327
      Overlapping103223761
      NOS403792986
      LateralityOther382778855
      Right33005894,19256
      Left19753566,12239
      Tumor size0–2 cm10461831,75619
      2–3 cm10441834,39320
      3–5 cm16132953,02832
      5–7 cm10061828,14817
      >7 cm9481720,87412
      Clinical N017633169,72641
      1499915,8259
      219963551,23030
      39091617,58610
      X490913,8328
      Clinical stageI11272051,6131
      II347611,5277
      III19623551,74331
      IV22213952,96831
      SurgeryNo surgery408372123,84074
      Segmentectomy28014261
      Lobectomy12282237,56322
      Extended Lobectomy125223911
      Pneumonectomy193329792
      TreatmentS8511533,67620
      S-C342665094
      C-S108216591
      S-CR7218681
      CR-S201416471
      R149314,9999
      CR14332529,54018
      C14582630,52518
      No Treatment10431848,77629
      Mean time from diagnosis to treatment (days)30.238.1
      All5657100168,199100
      Note: Insurance status, urban/rural setting, and palliative care were analyzed and removed from reporting as they were not predictors of mortality.
      NOS, not otherwise specified; S, Surgery; C, chemotherapy; R, radiation; CR, chemoradiation.

      Overall and Relative Survival

      Stage-specific unadjusted overall 5-year survival was significantly better for younger patients at each stage (log-rank test, p < 0.001 for all stages), with a larger magnitude of effect observed at the earlier stages (Fig. 2). The median survival times for younger patients at stages I through IV were more than 60, 59.0 (95% confidence interval [CI]: 44.7–82.8), 18.4 (95% CI: 17.1–19.4), and 8.8 (95% CI: 8.3–9.3) months. For the older patients, the median survival times for stages I through IV were 49.3 (95% CI: 48.4–50.1), 21.7 (95% CI: 21.0–22.3), 12.9 (95% CI: 12.8–13.1), and 5.7 (95% CI: 5.6–5.8) months. The absolute difference in 5-year overall survival at stages I and II between the younger and older groups was 25%. At stage III this difference was only 9%, and at stage IV it was only 2%. Relative survival showed a similar trend in the absolute difference in 5-year survival (Fig. 3). The largest differences between overall and relative survival were in stage I and II. Older patients with stage I NSCLC had an overall 5-year survival of 45% but a relative survival of 52%. In comparison, younger patients with stage I NSCLC had an overall 5-year survival of 71% with a relative survival of 70%.
      Figure thumbnail gr2
      Figure 2Kaplan-Meier estimate of 5-year overall survival at each clinical stage, stratified by age group.
      Figure thumbnail gr3
      Figure 3Estimate of 5-year relative survival at each clinical stage, stratified by age group. Relative survival is defined as the ratio of overall survival to expected survival based on life table data (see the text for details).

      Differences in Treatment

      Younger patients were treated differently than older patients at each stage (Fig. 4). At stage I, 64% of younger patients received surgery only versus 55% of the older patients (p < 0.0001). Conversely, at stage I, 14% of the older patients received radiation therapy only compared with 1% of younger patients. At stage II, 29% of younger patients underwent adjuvant therapy whereas only 16% of the older patients received that treatment (p < 0.0001). Conversely, at stage II, 12% of older patients received radiation therapy only compared with 3% of younger patients (p < 0.0001). Younger patients were more likely to receive any treatment across all stages (82% versus 71%, p < 0.0001) and at each individual stage than older patients.
      Figure thumbnail gr4
      Figure 4Treatment by clinical stage for different age groups. S, surgery; S-C, adjuvant chemotherapy; S-CR, adjuvant chemoradiation therapy; C-S, neoadjuvant chemotherapy; CR-S, neoadjuvant chemoradiation therapy; R, radiation therapy; CR, chemoradiation therapy; C, chemotherapy; and Rx, treatment.

      Multivariable Cox Model

      The multivariable Cox proportional hazards model for mortality was performed for each clinical stage and is presented in Table 2. Being in the older group was a significant predictor of mortality at all stages, with a larger magnitude of effect in the earlier stages (hazard ratios [HR] of 1.84, 1.62, 1.18, and 1.14 for stages I through IV, respectively [all p < 0.0001]). Comorbidity was also a significant predictor of mortality at all stages. The HRs for mortality for a Charlson-Deyo score higher than 1 were 1.49, 1.41, 1.42, and 1.42 for stages I through IV, respectively (all p < 0.0001). Other clinically significant predictors of mortality across all stages included male sex, white race, unknown and non-Hispanic ethnicity, receiving care at a nonacademic hospital, increasing tumor size, increasing nodal status, and initial treatment. Other variables were significant predictors of mortality only at specific stages. Tumor grade higher than 1 was predictive of mortality at all stages except stage II.
      Table 2Cox Proportional Hazards Model for Mortality, Stratified by Clinical Stage
      ParameterStage IStage IIStage IIIStage IV
      HRp ValueHRp ValueHRp ValueHRp Value
      Age, y20–46 (ref)1.001.001.001.00
      47–891.84<0.00011.62<0.00011.18<0.00011.14<0.0001
      SexFemale (ref)1.001.001.001.00
      Male1.21<0.00011.15<0.00011.14<0.00011.17<0.0001
      RaceNonwhite (ref)1.001.001.001.00
      White1.14<0.00011.110.00111.16<0.00011.14<0.0001
      EthnicityHispanic (ref)1.001.001.001.00
      Non-Hispanic1.20<0.00011.190.02531.19<0.00011.36<0.0001
      Unknown1.21<0.00011.170.06351.20<0.00011.380.4834
      RegionWest (ref)1.001.001.001.00
      Midwest1.060.00181.130.00141.11<0.00011.050.0011
      Northeast1.010.78551.030.4841.030.10020.950.0011
      South1.050.00631.050.16291.040.01291.000.8114
      Median income≥$63,000 (ref)1.001.001.001.00
      $48,000–62,9991.09<0.0001NSNS1.050.0021.000.7758
      $38,000–47,9991.17<0.0001NSNS1.08<0.00011.040.0021
      <$38,0001.22<0.0001NSNS1.10<0.00011.050.0002
      Unknown1.070.8012NSNS1.100.69921.13<0.0001
      High school education>21% (ref)1.001.001.001.00
      13–20.9%1.020.42330.960.16061.050.0008NSNS
      7–12.9%1.040.09460.970.37051.040.0188NSNS
      <7%1.070.01440.940.07811.040.0446NSNS
      Unknown1.600.06141.36<0.00011.240.3779NSNS
      Facility typeAcademic/research program (ref)1.001.001.001.00
      Community cancer program1.20<0.00011.100.00611.17<0.00011.13<0.0001
      Comprehensive cancer program1.12<0.00011.070.00921.11<0.00011.09<0.0001
      Other1.080.60121.340.35861.250.05881.130.2562
      Charlson-Deyo score0 (ref)1.001.001.001.00
      11.16<0.00011.19<0.00011.21<0.00011.24<0.0001
      2+1.49<0.00011.41<0.00011.42<0.00011.42<0.0001
      Histologic diagnosisAdenocarcinoma (ref)1.001.001.001.00
      Bronchioloalveolar carcinoma0.73<0.00010.770.00070.76<0.00010.63<0.0001
      Large cell carcinoma1.140.00011.010.93481.010.64541.090.0001
      Other1.11<0.00011.050.10961.030.04211.08<0.0001
      Squamous cell carcinoma1.11<0.00011.050.10051.000.7431.010.4256
      Grade1 (ref)1.001.001.001.00
      21.29<0.00011.080.19811.130.00051.22<0.0001
      31.37<0.00011.160.0131.16<0.00011.44<0.0001
      41.32<0.00011.220.03361.23<0.00011.45<0.0001
      Unknown1.32<0.00011.120.07141.17<0.00011.38<0.0001
      Tumor locationUpper lobe/lingula (ref)1.001.001.001.00
      Middle lobe1.080.01191.090.14631.050.0671.15<0.0001
      Lower lobe1.11<0.00011.15<0.00011.16<0.00011.020.4987
      Main bronchus/carina/hilum1.040.760.480.00040.990.83951.19<0.0001
      Overlapping1.26<0.00011.040.61841.140.0011.100.0055
      NOS1.21<0.00011.180.00391.14<0.00011.060.0004
      LateralityLeft (ref)1.001.001.001.00
      Right0.980.14661.020.40740.990.57091.010.1288
      Other1.220.07182.120.00021.150.0290.970.2034
      Tumor size0–2 cm (ref)1.001.001.001.00
      2–3 cm1.23<0.00011.130.00371.11<0.00011.050.0054
      3–5 cm1.57<0.00011.24<0.00011.25<0.00011.18<0.0001
      5–7 cm1.93<0.00011.47<0.00011.43<0.00011.29<0.0001
      >7 cm2.48<0.00011.72<0.00011.64<0.00011.50<0.0001
      Clinical N0 (ref)1.001.001.001.00
      1N/AN/A0.980.38651.070.00341.20<0.0001
      2N/AN/AN/AN/A1.09<0.00011.35<0.0001
      3N/AN/AN/AN/A1.23<0.00011.39<0.0001
      X1.18<0.00011.200.00171.26<0.00011.28<0.0001
      TreatmentS (ref)1.001.001.001.00
      S-C0.78<0.00010.57<0.00010.75<0.00010.800.0399
      S-CR1.230.00750.770.00260.910.19571.160.5826
      C-S0.930.240.66<0.00010.53<0.00010.740.0429
      CR-S1.000.97660.64<0.00010.77<0.00010.960.8296
      C3.01<0.00011.96<0.00012.15<0.00011.69<0.0001
      R2.77<0.00012.32<0.00012.35<0.00012.07<0.0001
      CR2.11<0.00011.42<0.00011.46<0.00011.37<0.0001
      No treatment3.96<0.00013.15<0.00014.06<0.00013.60<0.0001
      HR, hazard ratio; ref, reference; NS, not statistically significant; NOS, not otherwise specified; N/A, not applicable; S, Surgery; C, chemotherapy; R, radiation; CR, chemoradiation.

      Discussion

      The findings of this report are consistent with those of other studies
      • Gadgeel S.M.
      • Ramalingam S.
      • Cummings G.
      • et al.
      Lung cancer in patients <50 years of age: the experience of an academic multidisciplinary program.
      • Maruyama R.
      • Yoshino I.
      • Yohena T.
      • et al.
      Lung cancer in patients younger than 40 years of age.
      • Mauri D.
      • Pentheroudakis G.
      • Bafaloukos D.
      • et al.
      Non-small cell lung cancer in the young: a retrospective analysis of diagnosis, management and outcome data.
      • Lara M.S.
      • Brunson A.
      • Wun T.
      • et al.
      Predictors of survival for younger patients less than 50 years of age with non-small cell lung cancer (NSCLC): a California Cancer Registry analysis.
      • Nugent M.D.
      • Edney M.T.
      • Hammerness P.G.
      • et al.
      Non–small cell lung cancer at the extremes of age: impact on diagnosis and treatment.
      • Subramanian J.
      • Morgensztern D.
      • Goodgame B.
      • et al.
      Distinctive characteristics of non-small cell lung cancer (NSCLC) in the young: a Surveillance, Epidemiology, and End Results (SEER) analysis.
      • Thomas A.
      • Chen Y.
      • Yu T.
      • et al.
      Trends and characteristics of young non-small cell lung cancer patients in the United States.
      • Kuo C.-W.
      • Chen Y.-M.
      • Chao J.-Y.
      • et al.
      Non-small cell lung cancer in very young and very old patients.
      and show that younger patients with NSCLC are more likely to be female, be nonwhite, have a lower comorbidity score, have adenocarcinoma, present with larger and more invasive tumors, and be more likely to be recommended for and undergo treatment. The data also show an increased overall and relative survival benefit with younger patients when adjusted for treatment received, comorbidity, tumor size, clinical nodal status, and patient demographic variables. This effect is larger at lower stages than at higher stages. These findings are consistent with those of studies done with registry data
      • Lara M.S.
      • Brunson A.
      • Wun T.
      • et al.
      Predictors of survival for younger patients less than 50 years of age with non-small cell lung cancer (NSCLC): a California Cancer Registry analysis.
      • Subramanian J.
      • Morgensztern D.
      • Goodgame B.
      • et al.
      Distinctive characteristics of non-small cell lung cancer (NSCLC) in the young: a Surveillance, Epidemiology, and End Results (SEER) analysis.
      • Thomas A.
      • Chen Y.
      • Yu T.
      • et al.
      Trends and characteristics of young non-small cell lung cancer patients in the United States.
      and smaller single-institution studies.
      • Nugent M.D.
      • Edney M.T.
      • Hammerness P.G.
      • et al.
      Non–small cell lung cancer at the extremes of age: impact on diagnosis and treatment.
      • Kuo C.-W.
      • Chen Y.-M.
      • Chao J.-Y.
      • et al.
      Non-small cell lung cancer in very young and very old patients.
      Other studies that have shown an absence of a difference in survival between younger and older patients may have lacked the power to detect smaller stage-specific differences.
      • Gadgeel S.M.
      • Ramalingam S.
      • Cummings G.
      • et al.
      Lung cancer in patients <50 years of age: the experience of an academic multidisciplinary program.
      • Maruyama R.
      • Yoshino I.
      • Yohena T.
      • et al.
      Lung cancer in patients younger than 40 years of age.
      • Mauri D.
      • Pentheroudakis G.
      • Bafaloukos D.
      • et al.
      Non-small cell lung cancer in the young: a retrospective analysis of diagnosis, management and outcome data.
      • Skarin A.T.
      • Herbst R.S.
      • Leong T.L.
      • et al.
      Lung cancer in patients under age 40.
      Youth plays an independent role in overall survival at lower stages. In this analysis, there was a clinically and statistically significant difference between 5-year overall survival for stages I and II benefiting younger patients, whereas in stages III and IV there was little appreciable clinical difference despite achievement of statistical significance. This discrepancy is due to the increased lethality of the disease at more advanced stages, leading to lower overall survival in both groups. For example, the median survival of older patients at stage IV is 35% less than that of younger patients, but this relative difference amounts to only an extra 3.1 months of life for younger patients (8.8 months [95% CI: 8.3–9.3] versus 5.7 months [95% CI: 5.6–5.8]).
      Youth also has an impact on relative survival at lower stages. A difference between overall and relative survival would suggest that patients die of causes other than their lung cancer. A lack of difference between overall and relative survival would imply that all observed mortality is due to lung cancer. A difference was observed in stages I and II in the older patients only, meaning that older patients died of lung cancer as well as of other causes. One explanation for this is the increased burden of comorbidity in the older patients.
      • Howlader N.
      • Mariotto A.B.
      • Woloshin S.
      • et al.
      Providing clinicians and patients with actual prognosis: cancer in the context of competing causes of death.
      In younger patients at all stages and in older patients at stages III and IV, this difference was not observed, suggesting that these patients die almost exclusively of their lung cancer and that comorbidity does not contribute to mortality at more advanced stages.
      Relative survival data were derived from life table data published by the Centers for Disease Control and Prevention.
      • Arias E.
      United States life tables, 2011.
      One of the limitations of using relative survival is that the study group must be similar to the population at large. In general, patients with lung cancer typically use tobacco and have respiratory disease more frequently than do their counterparts with no cancer.
      • Mannino D.M.
      • Aguayo S.M.
      • Petty T.L.
      • et al.
      Low lung function and incident lung cancer in the United States: data from the first national health and nutrition examination survey follow-up.
      Because tobacco use and respiratory disease are not measured in the NCDB, this trend cannot be validated in the data set. However, relative survival in lung cancer was validated in the SEER registry, as cancer-specific survival is a variable that can be derived in that registry.
      • Howlader N.
      • Ries L.A.G.
      • Mariotto A.B.
      • et al.
      Improved estimates of cancer-specific survival rates from population-based data.
      In that analysis, there was a difference of less than 2% between relative survival and cancer-specific survival, indicating that relative survival was a reasonable proxy for cancer-specific survival.
      • Howlader N.
      • Ries L.A.G.
      • Mariotto A.B.
      • et al.
      Improved estimates of cancer-specific survival rates from population-based data.
      There is no reason to believe that the prevalence of smoking, respiratory disease, and other unmeasured confounders is different in the SEER registry and the NCDB, so it is reasonable to assume that the population in the current study is not significantly different from the population at large for the purposes of relative survival analysis. Therefore, relative survival is a reasonable proxy for cancer-specific survival in the NCDB.
      Although there are stage-specific treatment guidelines for NSCLC,
      • Detterbeck F.C.
      • Lewis S.Z.
      • Diekemper R.
      • et al.
      Executive summary: diagnosis and management of lung cancer, 3rd ed: American College of Chest Physicians evidence-based clinical practice guidelines.
      in reality each stage represents a heterogeneous collection of tumor types, and no single treatment can be applied across an entire stage. The model showed that tumor size, clinical nodal status, and treatment received were all significant predictors of mortality at each stage. When treatment received was broken down by tumor stage (see Fig. 4), there were statistically significant differences in how each age group was treated at each stage, which may impact survival. For example, older patients with stage II disease are more likely than younger patients to receive radiation therapy only (p < 0.0001), which has an HR for mortality of 2.32 (p < 0.0001), instead of surgery with adjuvant chemotherapy (HR = 0.57, p < 0.0001), which is the standard of care. Nadpara et al.
      • Nadpara P.A.
      • Madhavan S.S.
      • Tworek C.
      • et al.
      Guideline-concordant lung cancer care and associated health outcomes among elderly patients in the United States.
      showed that receipt of guideline-concordant care was significantly associated with a survival benefit in the elderly. The analysis from this report suggests that this principle applies to younger patients as well. The reason for patients not receiving guideline-concordant care is likely multifactorial and is not captured well within the NCDB. It is an interesting thought that younger patients were more likely to be recommended treatment and less likely to refuse treatment (see Table 1). The clinical, cultural, and personal factors behind these decisions remain unknown.
      Comorbidity has both direct and indirect effects on survival, and increasing age brings with it increasing comorbidities (76% of younger patients have no comorbidities versus 58% of older patients, p < 0.0001). Comorbidity is known to be an independent predictor of mortality in hospital-based cancer registries, and the HR for comorbidity decreases with increasing cancer stage, both of which are consistent with this analysis.
      • Piccirillo J.F.
      • Tierney R.M.
      • Costas I.
      • et al.
      Prognostic importance of comorbidity in a hospital-based cancer registry.
      • Read W.L.
      • Tierney R.M.
      • Page N.C.
      • et al.
      Differential prognostic impact of comorbidity.
      Comorbidity indirectly affects survival by influencing treatment decisions. It has also been shown that patients with more comorbidities are less likely to receive treatment according to published guidelines independent of patient age, and this effect is more pronounced at lower stages.
      • Nadpara P.A.
      • Madhavan S.S.
      • Tworek C.
      • et al.
      Guideline-concordant lung cancer care and associated health outcomes among elderly patients in the United States.
      • de Rijke J.M.
      • Schouten L.J.
      • Velde GPMt
      • et al.
      Influence of age, comorbidity and performance status on the choice of treatment for patients with non-small cell lung cancer; results of a population-based study.
      Therefore, it may be concluded that comorbidity has direct effects on survival as shown in the Cox model and indirect effects on survival by influencing treatment choices.
      Lung cancer in younger patients may serve as an emotional trigger that can introduce bias toward overtreatment. NSCLC is a heterogeneous disease with many variations in presentation. Younger patients with lung cancer typically present later, with more symptoms, and with more advanced tumors on average.
      • Skarin A.T.
      • Herbst R.S.
      • Leong T.L.
      • et al.
      Lung cancer in patients under age 40.
      Although the NCDB does not provide information on patient presentation or imaging, other studies have shown that lung cancer diagnosed on the basis of chest radiography or computed tomography scan alone in asymptomatic patients has a significantly higher volume doubling time compared with in patients who were symptomatic.
      • Detterbeck F.C.
      Turning gray: the natural history of lung cancer over time.
      • Usuda K.
      • Saito Y.
      • Sagawa M.
      • et al.
      Tumor doubling time and prognostic assessment of patients with primary lung cancer.
      It is also known that patients who are asymptomatic at diagnosis have better cancer-specific survival when controlling for age, histologic diagnosis, tumor size, and nodal status.
      • Hanagiri T.
      • Sugio K.
      • Mizukami M.
      • et al.
      Postoperative prognosis in patients with non-small cell lung cancer according to the method of initial detection.
      Therefore, one would expect that this phenomenon would disproportionately benefit older patients and narrow the survival gap between the younger and older patients.
      In this study clinical staging was used instead of pathologic staging. This approach was necessary to capture a representative sample of patients managed nonoperatively. The discordance between clinical and pathologic staging in patients who were eventually pathologically staged in this study was approximately 15%, with 10% being up-staged and 5% being down-staged, and this was similar across both age groups (data not shown). We acknowledge that this observation introduces an inherent bias into the survival data because survival is correlated more strongly with pathologic than clinical stage; however, using pathologic staging in some patients and not in others would also introduce bias. Because upstaging is more common than downstaging, patients who undergo surgery and therefore have pathologic staging information would be more likely to have higher stages than those undergoing nonsurgical therapy. Furthermore, pathologic staging is typically not available when initial treatment decisions are being made.
      As for additional limitations, the NCDB is a retrospective database that does not capture all variables of interest with respect to lung cancer. Most notably, performance status is not available. Performance status has been shown to be a predictor of mortality and is known to correlate with increasing age.
      • de Rijke J.M.
      • Schouten L.J.
      • Velde GPMt
      • et al.
      Influence of age, comorbidity and performance status on the choice of treatment for patients with non-small cell lung cancer; results of a population-based study.
      • Hsu C.L.
      • Chen K.Y.
      • Shih J.Y.
      • et al.
      Advanced non-small cell lung cancer in patients aged 45 years or younger: outcomes and prognostic factors.
      The NCDB does not capture tobacco use. Tobacco use has an association with mortality, and younger people with lung cancer do not smoke as much as their older counterparts.
      • Bryant A.S.
      • Cerfolio R.J.
      Differences in outcomes between younger and older patients with non–small cell lung cancer.
      We assume that our data are similar in that younger people smoke less. This variable could certainly be a latent one in our model that is captured within the age variable to some extent. Although it has been shown that younger patients are treated more aggressively, the NCDB supplies only limited information as to the rationale behind treatment decisions. Many other factors that are not captured by the NCDB may play a role in these decisions in addition to age; they include pulmonary function testing, patient preference, cultural considerations, and provider judgment.
      In this analysis, overall and relative survival in younger patients with NSCLC is better than in older patients, with most of the benefit derived from stages I and II. Younger patients are typically healthier and are offered more aggressive initial therapy at all stages of disease, yet they see only a marginal benefit in survival in more advanced stages. Aggressive treatment of younger patients may not confer a meaningful survival benefit, especially at more advanced stages. Treatment decisions should not be based solely on age but rather should be individualized to the patient and tumor characteristics. Further investigation is needed to understand tumor- and treatment-specific factors that affect survival among younger patients with NSCLC.

      Acknowledgments

      The American College of Surgeons and the Commission on Cancer have not verified and are not responsible for the analytic or statistical methodology used, or the conclusions drawn from these data by the investigator.

      References

        • Siegel R.L.
        • Miller K.D.
        • Jemal A.
        Cancer statistics, 2015.
        CA: Cancer J Clin. 2015; 65: 5-29
        • Gadgeel S.M.
        • Ramalingam S.
        • Cummings G.
        • et al.
        Lung cancer in patients <50 years of age: the experience of an academic multidisciplinary program.
        Chest. 1999; 115: 1232-1236
        • Maruyama R.
        • Yoshino I.
        • Yohena T.
        • et al.
        Lung cancer in patients younger than 40 years of age.
        J Surg Oncol. 2001; 77: 208-212
        • Mauri D.
        • Pentheroudakis G.
        • Bafaloukos D.
        • et al.
        Non-small cell lung cancer in the young: a retrospective analysis of diagnosis, management and outcome data.
        Anticancer Res. 2006; 26: 3175-3181
        • Skarin A.T.
        • Herbst R.S.
        • Leong T.L.
        • et al.
        Lung cancer in patients under age 40.
        Lung Cancer. 2001; 32: 255-264
        • Lara M.S.
        • Brunson A.
        • Wun T.
        • et al.
        Predictors of survival for younger patients less than 50 years of age with non-small cell lung cancer (NSCLC): a California Cancer Registry analysis.
        Lung Cancer. 2014; 85: 264-269
        • Nugent M.D.
        • Edney M.T.
        • Hammerness P.G.
        • et al.
        Non–small cell lung cancer at the extremes of age: impact on diagnosis and treatment.
        Ann Thorac Surg. 1997; 63: 193-197
        • Subramanian J.
        • Morgensztern D.
        • Goodgame B.
        • et al.
        Distinctive characteristics of non-small cell lung cancer (NSCLC) in the young: a Surveillance, Epidemiology, and End Results (SEER) analysis.
        J Thorac Oncol. 2010; 5: 23-28
        • Thomas A.
        • Chen Y.
        • Yu T.
        • et al.
        Trends and characteristics of young non-small cell lung cancer patients in the United States.
        Front Oncol. 2015; 5: 113
        • Kuo C.-W.
        • Chen Y.-M.
        • Chao J.-Y.
        • et al.
        Non-small cell lung cancer in very young and very old patients.
        Chest. 2000; 117: 354-357
        • Lerro C.C.
        • Robbins A.S.
        • Phillips J.
        • et al.
        Comparison of cases captured in the National Cancer Data Base with those in population-based central cancer registries.
        Ann Surg Oncol. 2013; 20: 1759-1765
        • Greene F.L.
        • Page D.L.
        • Fleming I.D.
        • et al.
        AJCC Cancer Staging Manual.
        6th ed. Springer, New York, NY2002
        • Edge S.
        • Byrd D.
        • Compton C.
        • et al.
        AJCC Cancer Staging Manual.
        7th ed. Springer, New York, NY2010
        • Goldstraw P.
        • Chansky K.
        • Crowley J.
        • et al.
        The IASLC Lung Cancer Staging Project: proposals for revision of the TNM stage groupings in the forthcoming (eighth) edition of the TNM classification for lung cancer.
        J Thorac Oncol. 2016; 11: 39-51
        • Arias E.
        United States life tables, 2011.
        Natl Vital Stat Rep. 2015; : 64-163
        • Howlader N.
        • Ries L.A.G.
        • Mariotto A.B.
        • et al.
        Improved estimates of cancer-specific survival rates from population-based data.
        J Natl Cancer Inst. 2010; 102: 1584-1598
        • Howlader N.
        • Mariotto A.B.
        • Woloshin S.
        • et al.
        Providing clinicians and patients with actual prognosis: cancer in the context of competing causes of death.
        J Natl Cancer Inst Monogr. 2014; : 255-264
        • Mannino D.M.
        • Aguayo S.M.
        • Petty T.L.
        • et al.
        Low lung function and incident lung cancer in the United States: data from the first national health and nutrition examination survey follow-up.
        Arch Intern Med. 2003; 163: 1475-1480
        • Detterbeck F.C.
        • Lewis S.Z.
        • Diekemper R.
        • et al.
        Executive summary: diagnosis and management of lung cancer, 3rd ed: American College of Chest Physicians evidence-based clinical practice guidelines.
        Chest. 2013; 143: 7S-37S
        • Nadpara P.A.
        • Madhavan S.S.
        • Tworek C.
        • et al.
        Guideline-concordant lung cancer care and associated health outcomes among elderly patients in the United States.
        J Geriatr Oncol. 2015; 6: 101-110
        • Piccirillo J.F.
        • Tierney R.M.
        • Costas I.
        • et al.
        Prognostic importance of comorbidity in a hospital-based cancer registry.
        JAMA. 2004; 291: 2441-2447
        • Read W.L.
        • Tierney R.M.
        • Page N.C.
        • et al.
        Differential prognostic impact of comorbidity.
        J Clin Oncol. 2004; 22: 3099-3103
        • de Rijke J.M.
        • Schouten L.J.
        • Velde GPMt
        • et al.
        Influence of age, comorbidity and performance status on the choice of treatment for patients with non-small cell lung cancer; results of a population-based study.
        Lung Cancer. 2004; 46: 233-245
        • Detterbeck F.C.
        Turning gray: the natural history of lung cancer over time.
        J Thorac Oncol. 2008; 3: 781-792
        • Usuda K.
        • Saito Y.
        • Sagawa M.
        • et al.
        Tumor doubling time and prognostic assessment of patients with primary lung cancer.
        Cancer. 1994; : 2239-2244
        • Hanagiri T.
        • Sugio K.
        • Mizukami M.
        • et al.
        Postoperative prognosis in patients with non-small cell lung cancer according to the method of initial detection.
        J Thorac Oncol. 2007; 2: 907-911
        • Hsu C.L.
        • Chen K.Y.
        • Shih J.Y.
        • et al.
        Advanced non-small cell lung cancer in patients aged 45 years or younger: outcomes and prognostic factors.
        BMC Cancer. 2012; 12: 241
        • Bryant A.S.
        • Cerfolio R.J.
        Differences in outcomes between younger and older patients with non–small cell lung cancer.
        Ann Thorac Surg. 2008; 85: 1735-1739