Lung cancer is the second most common type of. Lung Cancer in Women* Analysis of the National Surveillance, Epidemiology, and End Results Database

Transcription

1 Lung Cancer in Women* Analysis of the National Surveillance, Epidemiology, and End Results Database Jennifer B. Fu, MD; T. Ying Kau, PhD; Richard K. Severson, PhD; and Gregory P. Kalemkerian, MD Objectives: In order to further characterize the effect of gender on the clinicopathologic features and survival of patients with lung cancer, and to determine gender-associated differences in temporal trends, we analyzed data that had been entered into a population-based cancer database. Patients and methods: Data on demographics, stage at diagnosis, histology, initial therapy, and survival were obtained on all patients with primary bronchogenic carcinoma registered in the national Surveillance, Epidemiology, and End Results database from 1975 to Results: Of the 228,572 eligible patients, 35.8% were female. The median age at diagnosis was 66 years for both men and women. However, women accounted for 40.9% of patients who were < 50 years of age and for 35.4% of older patients. The incidence of lung cancer in men peaked at 72.5 per 100,000 person-years in 1984 and then declined to 47 per 100,000 person-years by In women, the incidence continued to rise to a peak of 33.1 per 100,000 person-years in 1991 before reaching a plateau at 30.2 to 32.3 per 100,000 person-years from 1992 to These changes have resulted in a marked narrowing of the male/female incidence ratio from 3.56 in 1975 to 1.56 in As initial treatment, women with local disease underwent surgery more frequently than did men. Stage-specific survival rates were better for women at all stages of disease (p < ). In a multivariate analysis, male gender was an independent negative prognostic factor (p < ). Conclusion: The incidence rate of lung cancer in women in the United States has reached a plateau. However, women are relatively overrepresented among younger patients, raising the question of gender-specific differences in the susceptibility to lung carcinogens. At each stage of the disease, the relative survival of women is better than that of men, with the largest difference noted in patients with local disease. (CHEST 2005; 127: ) Key words: epidemiology; gender; incidence; lung cancer; survival Abbreviations: ER estrogen receptor; NSCLC non-small cell lung cancer; RR relative risk; SCLC small cell lung cancer; SEER Surveillance, Epidemiology, and End Results *From the Department of Internal Medicine (Drs. Fu and Kalemkerian), University of Michigan, Ann Arbor, MI; and the Department of Family Medicine (Drs. Severson and Kau), Wayne State University, Detroit, MI. Presented in part at the Fourth International Lung Cancer Congress, Maui, HI, June 25 28, 2003, and the Tenth World Conference on Lung Cancer, Vancouver, BC, Canada, August 10 14, This research was supported in part by Surveillance, Epidemiology, and End Results contract No. N01-CN and NO1-PC from the National Cancer Institute, Bethesda, MD. Lung cancer is the second most common type of cancer diagnosed in women in the United States and surpassed breast cancer as the leading cause of cancer-related mortality in In 2003, it was estimated that 47% of the 171,900 new lung cancer cases and 44% of the 157,200 lung cancer deaths in the United States would occur in women. 1 Currently, lung cancer accounts for 25% of all cancerrelated deaths in women. Although the incidence of lung cancer in men has been declining for years, most studies have reported a continued increase in incidence in women. Gender-associated differences in the clinicopathologic characteristics and survival of patients with lung cancer appear to exist, but previous studies have yielded conflicting findings. Most studies 2 5 have Manuscript received May 6, 2004; revision accepted September 16, Reproduction of this article is prohibited without written permission from the American College of Chest Physicians ( permissions@chestnet.org). Correspondence to: Gregory P. Kalemkerian, MD, University of Michigan Medical Center, C350 MIB, 1500 E Medical Center Dr, Ann Arbor, MI ; kalemker@umich.edu 768 Clinical Investigations

2 reported that women receive diagnoses at a younger median age, suggesting that they have an increased susceptibility to the development of lung cancer. However, the effect of gender on the lung cancer risk associated with tobacco use remains unclear. 3,6,7 Gender-related differences in the incidence of histologic subtypes and stage at presentation have also been variably reported. 2,4,8,9 Several studies have noted better overall survival rates in women with non-small cell lung cancer (NSCLC) following surgical resection. Female gender has also been noted to be a favorable prognostic factor in patients with small cell lung cancer (SCLC). 17,18 To clarify gender-associated differences and temporal trends in clinicopathologic characteristics and survival rates for patients with primary bronchogenic carcinoma, we analyzed data from patients who were registered in the large, population-based, national Surveillance, Epidemiology, and End Results (SEER) database over a 25-year period. Patients Materials and Methods The SEER program is a National Cancer Institute-funded initiative that has collected demographic and clinicopathologic data on newly diagnosed cancer patients in the United States since During the study period, the SEER system collected data from 9 to 11 selected geographic regions covering approximately 10 to 14% of the US population. The national SEER database collects information on all incident cases of cancer from hospitals, physicians offices, clinics, radiation therapy facilities, hospice facilities, nursing homes, and the state s office of vital statistics. This results in an estimated capture rate of 99% of all incident cancer cases. Data regarding demographics, primary tumor site, stage of disease, tumor histology, initial treatment (within 4 months of diagnosis), and survival were available throughout the study period. The national SEER data were screened to identify all patients with lung cancer who received diagnoses from 1975 to 1999 (337,036 patients). Patients were excluded based on the following (with values given as the percentage of patients excluded): diagnosis of cancer prior to lung cancer (13.7%); report by autopsy or death certificate only (2.1%); clinical diagnosis without biopsy (6.8%); in situ carcinoma (0.1%); and histologic identification other than small cell carcinoma, large cell carcinoma, squamous cell carcinoma, or adenocarcinoma (9.5%). The final study population consisted of 228,572 eligible patients. Tumor stage was defined by the extent of disease with the following three categories: (1) local disease, an invasive neoplasm confined entirely to the organ of origin; (2) regional disease, extension beyond the organ of origin directly into surrounding tissues and/or regional lymph nodes; and (3) distant disease, discontinuous metastases, excluding regional lymph nodes. Survival was defined as the time between the date of diagnosis and the date of death. Statistical Analysis Comparison between genders in the frequency distribution of cancer stage, age, race, tumor histology, and types of initial treatment was evaluated with a 2 heterogeneity test. All reported p values are two-sided. Age-adjusted incidence rates and relative survival rates were calculated using statistical software (SEER*Stat 4.2; National Cancer Institute SEER Program; Bethesda, MD). All incidence rates are per 100,000 person-years. The relative survival rate was calculated by the life-table method, in which the observed survival rate is adjusted for expected mortality from all causes. 19 All patients included in the study received diagnoses from 1975 to 1999, and follow-up data were available through December A comparable multivariate analysis using the Cox proportional hazards model was performed to evaluate the relative risk (RR) of death of male vs female patients while considering the potential confounding effects of, and effect modification by, selected covariates, including age at diagnosis, race, year of diagnosis, histology, stage, and initial therapy. 20 The model was adjusted for geographic area within the SEER registry. Survival curves were plotted (log-log plot) to test the proportional hazards assumption. A statistical software package (Statistical Analysis System; SAS Institute; Cary, NC) was used for 2 calculations and multivariate survival analyses. 21 Results Patient Characteristics The clinicopathologic characteristics of 228,572 patients with primary bronchogenic carcinoma who received diagnoses from 1975 to 1999 and were registered in the national SEER database are listed in Table 1. Overall, 81,843 patients (35.8%) were female. The median age at diagnosis was 66 years for both men and women. However, 8.6% of women Table 1 Characteristics of SEER Lung Cancer Patients, * Characteristic Male Patients Female p Value Patients, No. 146,729 (64.2) 81,843 (35.8) Age Median, yr yr 10,158 (6.9) 7,026 (8.6) yr 136,571 (93.1) 74,817 (91.4) Race White 122,721 (83.6) 70,918 (86.7) Black 16,498 (11.3) 7,235 (8.8) Other 7,335 (5.0) 3,560 (4.3) Unknown 175 (0.1) 130 (0.2) Histology Adenocarcinoma 48,772 (33.2) 36,561 (44.7) Squamous cell 53,198 (36.3) 17,493 (21.4) Large cell 17,786 (12.1) 9,261 (11.3) Small cell 26,973 (18.4) 18,528 (22.6) Stage Local 26,088 (17.8) 15,454 (18.9) Regional 43,517 (29.7) 22,724 (27.8) Distant 63,742 (43.4) 35,960 (43.9) Unknown 13,382 (9.1) 7,705 (9.4) *Values given as No. (%), unless otherwise indicated. p values pertain to the differences in the proportional incidence distributions between male and female patients in each category. CHEST / 127 / 3/ MARCH,

3 and only 6.9% of men were 50 years of age at the time of diagnosis (p ). Women accounted for 40.9% of patients who were 50 years of age, but for only 35.4% of those who were 50 years of age at the time of diagnosis. Analysis by race revealed that 8.8% of women and 11.3% of men were African- American (p ). The proportional occurrence of histologic subtypes also differed significantly between men and women with lung cancer (p ) [Table 1]. In women, adenocarcinoma (44.7%) was the most common histologic subtype, followed by small cell cancer (22.6%) and squamous cell (21.4%) carcinomas. In men, squamous cell carcinoma (36.3%) was the most common subtype, followed closely by adenocarcinoma (33.2%) and remotely by small cell carcinoma (18.4%). The distribution of disease stage showed little absolute difference between men and women, although these differences did reach statistical significance (p 0.001) [Table 1]. Distant disease was the most common stage at presentation for both men (43.4%) and women (43.9%). The greatest gender difference in stage occurrence was for regional disease, which was reported in 27.8% of women and 29.7% of men. Incidence Rates The age-adjusted incidence rate for lung cancer in men peaked at 72.5 per 100,000 person-years in 1984 and has declined steadily since then to 47 per 100,000 person-years in 1999 (Fig 1). However, the incidence rate in women continued to rise until 1991 when it peaked at 33.1 per 100,000 person-years before plateauing at 30.2 to 32.3 per 100,000 personyears from 1992 to 1999 (Fig 1). This has resulted in a dramatic narrowing of the male/female incidence ratio during the study period from 3.56 in 1975 to 1.56 in 1999 (Fig 1). Over the entire study period, the incidence rate of lung cancer was higher in men than in women regardless of age (Table 2). However, the male/ female incidence ratio for younger patients (ie, those 50 years of age) was substantially lower than that for older patients (ie, those 50 years) [1.49 vs 2.33, respectively]. A comparison of two study periods, from 1975 to 1987 and from 1988 to 1999, revealed that the incidence rate of lung cancer decreased over time in both men and women who were 50 years of age but that the decline was greater in men (40.9% vs 27.6%, respectively), resulting in a narrowing of the male/female incidence ratio from 1.64 to 1.34 (Table 2). A similar temporal comparison in patients 50 years of age showed that the incidence rate decreased by 13.5% in men but increased by 37.3% in women between the two time periods, resulting in a more profound narrowing of the male/female ratio in older patients from 3.00 to The incidence rates of lung cancer in both black and white women rose by equivalent degrees between 1975 to 1987 and 1988 to 1999 (29.8% vs 30.3%, respectively), with the incidence rate remaining slightly higher in black women throughout the Figure 1. Age-adjusted, gender-specific incidence rates of lung cancer, from 1975 to Clinical Investigations

4 Table 2 Temporal Comparison of Age-Adjusted Incidence Rates* Variable Men Women RR 95% CI Men Women RR 95% CI Men Women RR 95% CI Overall Age, yr Race White Black Histology Adenocarcinoma Squamous cell Large cell Small cell Stage Local Regional Distant *Incidence per 100,000 person-years. CI confidence interval. study period (31.3 vs 28.7 per 100,000 person-years) [Table 2]. Incidence rates for both black and white men fell during the study period. Although there was a slightly greater decrease noted in black men than in white men between 1975 to 1987 and 1988 to 1999 (17.5% vs 15.7%, respectively), the rate remained substantially higher in black men (85.7 vs 56.9 per 100,000 person-years) [Table 2]. The incidence rate of adenocarcinoma increased in both men and women between 1975 to 1987 and 1988 to 1999, with a greater relative increase noted in women (40.5% vs 9.3%, respectively) [Table 2]. In contrast, incidence rates for squamous cell and small cell carcinoma decreased in men, while increasing slightly in women (Table 2, Fig 2). Over the entire study period, the greatest disparity between men and women was noted for those with squamous cell carcinoma (male/female ratio, 3.90), and the least Figure 2. Age-adjusted, gender-specific incidence rates of histologic subtypes of lung cancer, from 1975 to CHEST / 127 / 3/ MARCH,

5 gender disparity was for those with adenocarcinoma (male/female ratio, 1.65) [Table 2, Fig 2]. Gender differences in stage-specific incidence rates mirrored those of overall lung cancer incidence with male/female ratios of 2.1 to 2.4 over the entire study period (Table 2). However, there were substantial differences in stage-specific incidence trends over time. While the incidence rates for local and regional disease fell sharply in men from 1975 to 1987 and from 1988 to 1999, they remained relatively unchanged in women (Table 2). For distant disease, the incidence rate in men remained relatively unchanged, while it rose by 53% in women from 9.7 per 100,000 person-years from 1975 to 1987 to 14.9 per 100,000 person-years from 1988 to In 1988, the SEER coding guidelines underwent a change that resulted in a shift of patients from local or regional stage to distant stage of disease. However, this coding shift should have affected both men and women equally, suggesting that the unbalanced increase in distant disease in women may have been due to other factors. Initial Therapy The SEER database includes information on the first course of cancer-directed treatment received within 4 months of the initial diagnosis of cancer. For hospital/facility-based therapy, such as surgery and radiotherapy, the accuracy and completeness of these data are considered to be highly reliable, while for office-based therapy, such as chemotherapy, they are considered to be unreliable. Although the treatment information in the SEER database is not inclusive of all the therapy that patients may have received during the course of their disease, it is indicative of the level of aggressiveness with which the patient was initially managed. Surgery is the standard treatment for patients with localized, earlystage NSCLC who can tolerate lung resection, with radiotherapy reserved for those with limited pulmonary function. During the latter part of the study period, the standard treatment for most patients with regional NSCLC shifted from surgery and/or radiotherapy to chemotherapy plus radiotherapy. For patients with SCLC, local disease is rare, and the treatment of regional disease shifted from radiotherapy alone to radiotherapy plus chemotherapy during the study period. For patients with distant disease, either NSCLC or SCLC, the standard treatment is chemotherapy with radiotherapy reserved for the palliation of specific symptoms. For patients with local disease, more women underwent surgery than men (63.7% vs 56.3%, respectively; p ), while men were treated more frequently with radiotherapy (23.4% vs 18.2%, respectively; p ) [Table 3]. A similar trend, although much smaller in magnitude, was evident in patients with regional disease. The use of combinedmodality therapy is one estimate of the degree of aggressiveness with which patients are treated. For patients with regional disease, similar percentages of women (17.7%) and men (17.0%) were initially treated with both surgery and radiotherapy. In patients with distant disease, a slightly higher percentage of men underwent radiotherapy (50.3% vs 47.3%, respectively; p ). Survival Relative survival rates based on life-table analysis methods were calculated to estimate the effect of cancer on survival by compensating for the expected difference in baseline survival rates between male and female populations. However, in lung cancer patients, relative rates may not completely correct for age-specific differences in survival since the available life-tables are based on unselected populations and are not adjusted for increased smokingrelated, non-lung cancer mortality. Throughout the study period, the overall, 2-year, and 5-year relative survival rates were significantly higher in women than in men (p ) [Fig 3, Table 4]. During the course of the study period, overall survival rates increased in both men and women, with a slightly greater increase noted in Table 3 Treatment Modality by Stage and Gender, Stage Treatment Modality Patients, % Male Female p Value* Local Surgery alone Radiotherapy alone Surgery radiotherapy No surgery or radiotherapy Unknown Regional Surgery alone Radiotherapy alone Surgery radiotherapy No surgery or radiotherapy Unknown Distant Surgery alone Radiotherapy alone Surgery radiotherapy No surgery or radiotherapy Unknown All stages Surgery alone Radiotherapy alone Surgery radiotherapy No surgery or radiotherapy Unknown *p values pertain to the differences in the proportional incidence distributions between male and female patients in each category. 772 Clinical Investigations

6 Figure 3. Gender-specific, relative 2-year and 5-year survival rates of patients with lung cancer, from 1975 to men, resulting in a modest narrowing of the gender gap. For the time period from 1975 to 1987, the 2-year and 5-year relative survival rates were 5.2% and 4.2% greater in women, respectively, while for the time period from 1988 to 1999, they were 4.5% and 3.5% greater in women. Stage-specific survival rates for women were also significantly higher than those for men for every stage of disease (Table 4). This gender-related survival difference was greatest in patients with local disease and declined as the extent of disease increased. However, the greatest degree of temporal narrowing of the gender gap was also noted in patients with local disease due to a greater degree of improvement in survival in men. For the time period from 1975 to 1987, the 2-year and 5-year relative survival rates for patients with local disease were 11.5% and 12.1% greater in women than in men, respectively, while for the timer period from 1988 to 1999, the female advantage was only 6.2% and 6.5% (Table 4). Women with regional and distant disease Table 4 Temporal Comparison of Relative 2-Year and 5-Year Stage-Specific Survival Stage Male Female p Value Male Female p Value Local 2-yr yr Regional 2-yr yr Distant 2-yr yr All stages 2-yr yr also fared significantly better than men throughout the study period, although the degree of difference in survival for these stages remained relatively stable over time, with an absolute variation of 1%. To determine whether differences in treatment could account for gender-related differences in survival, the survival rate for men and women who received the same initial treatment was evaluated. Among patients who underwent surgery as part of their initial treatment, survival in women was superior to that in men, suggesting that the lower frequency of surgical resection in men was not solely responsible for the gender-associated difference in survival. Female patients who underwent surgery alone had 2-year and 5-year survival rates of 74.3% and 56.8%, respectively, while the survival rates for male patients were 66.0% and 48.3%, respectively (p ). For patients who underwent radiotherapy, the relative 2-year and 5-year survival rates in men vs women were 15.5% vs 12.2% and 5.3% vs 3.4%, respectively (p ). The results of a multivariate analysis of the RR of death incorporating demographic, geographic, temporal, and clinicopathologic variables are shown in Table 5. During multivariate modeling, it became apparent that there was a significant interaction between gender and age for patients with local and regional disease, with the negative influence of male gender being greater in older patients. For example, relative to women who were 50 years of age, the risk of death in men who were 50 years of age with local disease was 1.08 (p 0.05), while the risk of death in men who were 50 years of age relative to women who were 50 years of age was 1.28 (1.92/ 1.50; p 0.05). Overall, male gender was an independent, negative prognostic indicator in all stages of disease, with the greatest negative impact on survival in patients with local disease. Advancing age also had CHEST / 127 / 3/ MARCH,

7 Table 5 Multivariate Analysis of Relative Risk of Death, * Variables Local Stage (n 39,868) Regional Stage (n 63,570) Distant Stage (n 94,911) Gender and age Male 50 yr 1.92 ( ) 1.46 ( ) 1.37 ( ) Male 50 yr 1.08 ( ) 1.07 ( ) 1.14 ( ) Female 50 yr 1.50 ( ) 1.26 ( ) 1.21 ( ) Female 50 yr Race Black 1.04 ( ) 1.03 ( ) 1.01 ( ) All Other 0.86 ( ) 0.92 ( ) 0.86 ( ) White Year (annual risk) 0.99 ( ) 0.98 ( ) 0.99 ( ) Histologic type Small cell 1.05 ( ) 0.81 ( ) 0.87 ( ) Squamous cell 1.15 ( ) 1.01 ( ) 1.02 ( ) Large cell 1.17 ( ) 1.06 ( ) 1.13 ( ) Adenocarcinoma Treatment Surgery, no radiation 0.19 ( ) 0.23 ( ) 0.48 ( ) Radiation, no surgery 0.75 ( ) 0.61 ( ) 0.74 ( ) Surgery radiation 0.34 ( ) 0.29 ( ) 0.52 ( ) No surgery, no radiation *Values are given as RR (95%) confidence interval. p 0.05 in comparison to baseline category with RR 1. p in comparison to baseline category with RR 1. a significant negative impact in all stages of disease. Black race had a small, but significant, negative influence on survival for patients with local and regional disease. For patients with local disease, any histologic diagnosis other than adenocarcinoma carried a significant negative impact. Discussion The majority of epidemiologic studies evaluating the effects of gender in lung cancer have been conducted in Europe, Japan, and Canada, raising questions as to the applicability of their results to women in the United States. Most prior studies in the United States were either geographically or temporally limited or were performed at academic centers with potentially biased referral populations. Wingo et al 22 previously published data from the SEER database on gender-specific and race-specific lung cancer incidence and mortality rates within an analysis on the association of lung cancer and tobacco use. The population-based study presented here, the broadest analysis to date of the effects of gender on lung cancer patients in the United States, was performed in an attempt to resolve several epidemiologic conflicts that arose in previous studies. The large population source and long study period allowed the evaluation not only of overall differences in the incidence and clinicopathologic features of lung cancer between men and women, but also of temporal trends. Most, but not all, prior studies have reported that women are diagnosed with lung cancer at an earlier age than men. 2,4,10,12 14,16 In our study, the median age at diagnosis in both men and women was 66 years, which was substantially older than that reported for women in many previous studies. However, we noted that women were proportionately overrepresented in the group of patients who were 50 years of age. Although many investigators have suggested that female smokers have an increased susceptibility to the development of lung cancer, prior epidemiologic studies have come to no uniform conclusion on this issue. While some prospective studies 3,6,9 have reported that female smokers have a greater risk of developing lung cancer than male smokers, others have demonstrated comparable risks regardless of gender when controlling for tobacco smoke exposure. Our finding that women were overrepresented among younger patients also suggests, but clearly does not prove, that women may have an increased susceptibility to tobacco carcinogens. Although the link between smoking and lung cancer is indisputable, the biological basis for a possible gender-specific difference in susceptibility remains unclear. 23,24 Zang and Wynder 7 estimated that women had a 1.5-fold higher RR of lung cancer 774 Clinical Investigations

8 than men, despite the findings that women were more likely to be never-smokers and that men started smoking earlier, inhaled more deeply, and smoked more cigarettes per day. This study also reported that the dose-response odds ratios for the cumulative exposure to cigarette smoke were 1.2- fold to 1.7-fold higher in women for all histologic types. Risch et al 25 found that the association between cigarette consumption and lung cancer risk was stronger for women than for men, regardless of histologic subtype. For example, the odds ratios for developing lung cancer for a 40-pack-year smoker relative to a lifelong nonsmoker were 27.9 in women and 9.6 in men. These findings and those of other studies 26 reporting similar results suggest that the greater risk of lung cancer in female smokers relative to male smokers could be due to an increased susceptibility to the carcinogenic effects of tobacco smoke in women. The difference in incidence rates for lung cancer between men and women has been narrowing over the past 20 years. From 1975 to 1999 the male/ female ratio fell from 3.56 to If the current trends continue, the rates of lung cancer in men and women in the United States will be equal within 10 to 15 years. Despite the good news that the incidence rate for lung cancer in women is no longer rising, the prolonged plateau that has persisted throughout the 1990s without an apparent downward trend is rather disturbing, especially viewed in light of the relatively brief plateau in incidence rates in men that presaged the current decline. The existence of differences in the proportional incidence of the various histologic subtypes of lung cancer between men and women has been welldocumented. Our study confirms that the major histologic subtype in women over the last 20 years has been adenocarcinoma, while this subtype has only risen to prominence in men over the past 10 years. The greatest gender differential continues to be seen for squamous cell carcinoma, a finding that appears to be due to the relatively low risk of this subtype in women. The reasons behind these histologic differences remain unclear but may be related to gender-specific differences in smoking habits or the type of cigarettes smoked. Alternatively, some reports of estrogen receptor (ER) expression by human lung cancer cells have suggested that exogenous or endogenous estrogens may play a role in the development of adenocarcinoma of the lung in women. Prior studies that have addressed the effect of gender on the outcome of lung cancer have suggested that female gender carries some survival advantage. The present study confirms that women had statistically better outcomes than men at all stages of disease, with the greatest survival difference noted in local-stage disease. Several studies have reported gender-specific differences in survival in surgically treated patients with NSCLC, with women uniformly having better outcomes. 10,12 16 Other investigators have found 17,18 similar results favoring women in patients with SCLC. One retrospective, single institution study 15 from Japan evaluated 1,123 consecutive patients with NSCLC and found that women with adenocarcinoma exhibited the most favorable prognosis, with multivariate analysis confirming that female gender was an independent, favorable prognostic factor. Alexiou et al 13 reported that women had significantly better survival with pathologic stage I lung cancer, irrespective of histologic subtype. They also noted a nonsignificant trend toward higher survival rates in women with stage II and III disease. It seems clear that women with lung cancer have higher overall survival rates than their male counterparts. The reasons for this survival advantage have not been identified, but are likely due to a variety of factors. In our study, surgery was performed more frequently in women than in men with local-stage disease. While this may have been due to a somewhat older age distribution and possibly to more severe comorbidity in men, the exact reasons for this gender-specific difference in treatment are not discernable from our data. Since surgery offers patients with local-stage disease the best chance for long-term survival, this difference in treatment may partially explain the significantly superior survival noted for women with local-stage disease. However, our finding that gender remained an independent prognostic factor in a stage-specific multivariate analysis suggests that gender-specific variations in stage, initial therapy, and histologic subtype are not entirely responsible for the female survival advantage. Hormonal influences may play an important role in the pathogenesis of lung cancer and the survival of lung cancer patients. One study 27 demonstrated the gender-specific expression of ER, with ER expression in lung tumors and normal lung tissue in 85% and 31% of women, respectively, and in 15% and 0% of men, respectively. Other studies 28 have suggested that the expression of ER maybeinvolved in the development of lung adenocarcinoma. However, some reports 29,30 have offered conflicting data regarding the expression and potential influence of hormonal factors in the development of NSCLC. The interaction between gender and age that was noted during our multivariate modeling demonstrated that the relative gender-specific survival advantage was greater in the older patient population. This interaction may be due in part to potentially greater differences in comorbidity between the genwww.chestjournal.org CHEST / 127 / 3/ MARCH,

9 ders in older patients than in younger patients. However, it also might argue against a hormonal explanation for gender-specific differences in survival, since younger women are more hormonally active. Similarly, a Japanese study found that women who were 60 years of age with lung cancer had a higher overall survival rate than men, while those who were 60 years of age did not. 12 The present study has several limitations, including the lack of data on exposure to tobacco smoke, on the use of chemotherapy, and on treatments throughout the course of the disease. Like many epidemiologic studies, our review was retrospective, thereby restricting us to the use of the data at hand and limiting the interpretation of our findings. The SEER database also lacks information on comorbid conditions, some of which, such as cardiovascular disease, may also have gender-specific differences in incidence. It is possible that some of the gender-specific survival differences noted in this study may be related to comorbidities. The size of the population analyzed in the present study led to the identification of many statistically significant differences despite the presence of small, and possibly clinically irrelevant, absolute differences. In the reporting of our results, we have tried to focus on gender-specific variations that were at a level that would suggest a clinically or epidemiologically important difference. Despite these limitations, the present analysis of a large, population-based cancer registry offers significant insights into the effects of gender on the clinicopathologic features and prognosis of lung cancer, and on the temporal trends that have occurred over the past 25 years. Several of our findings, especially those pertaining to incidence and survival, deserve further study to evaluate the biological basis of these gender-specific differences. ACKNOWLEDGMENT: We are grateful to Dr. Mary Varterasian for critical review of the manuscript. References 1 Jemal A, Murray T, Samuels A, et al. Cancer statistics, CA Cancer J Clin 2003; 53: Ferguson MF, Skosey C, Hoffman PC, et al. Sex-associated differences in presentation and survival in patients with lung cancer. J Clin Oncol 1990; 8: Agudo A, Ahrens W, Benhamou E, et al. Lung cancer and cigarette smoking in women: a multi-center case control study in Europe. Int J Cancer 2000; 88: Radzikowska E, Glaz P, Roszkowski K. Lung cancer in women: age, smoking, histology, performance status, stage, initial treatment and survival: population-based study of 20,561 cases. Ann Oncol 2002; 13: McDuffie H, Klaassen D, Dosman J, et al. Female-male differences in patients with primary lung cancer. Cancer 1987; 59: Kreuzer M, Boffetta P, Whitley E, et al. Gender differences in lung cancer risk by smoking: a multicentre case-control study in Germany and Italy. Br J Cancer 2000; 82: Zang E, Wynder E. Differences in lung cancer risk between men and women: examination of the evidence. J Natl Cancer Inst 1996; 88: Lienert T, Serke M, Loddenkemper R, et al. Lung cancer in young females. Eur Respir J 2000; 16: Koyi H, Hillerdal G, Branden E. A prospective study of a total material of lung cancer from a county in Sweden : gender, symptoms, type, stage, and smoking habits. Lung Cancer 2002; 36: Perrot M, Licker M, Spiliopoulous A, et al. Sex differences in presentation, management and prognosis of patients with non-small cell lung carcinoma. J Thorac Cardiovasc Surg 2000; 119: Baldini E, Strauss GM. Women and lung cancer: waiting to exhale. Chest 1997; 112(suppl):S229 S Minami H, Yoshimura M, Tsubota N, et al. Lung cancer in women: sex-associated differences in survival of patients undergoing resection for lung cancer. Chest 2000; 118: Alexiou C, Onyeaka CV, Morgan W, et al. Do women live longer following lung resection for carcinoma? Eur J Cardiothorac Surg 2002; 21: Ferguson MK, Wang J, Vokes E, et al. Sex-associated differences in survival of patients undergoing resection for lung cancer. Ann Thorac Surg 2000; 69: Yoshino I, Baba H, Suemitsu R, et al. A time trend of profile and surgical results in 1123 patients with non-small cell lung cancer. Surgery 2002; 131:S242 S Ouellette D, Desbiens G, Emond C, et al. Lung cancer in women compared with men: stage, treatment, and survival. Ann Thorac Surg 1998; 66: Spiegelman D, Maurer LH, Ware JH, et al. Prognostic factors in small-cell carcinoma of the lung: an analysis of 1,521 patients. J Clin Oncol 1989; 7: Osterlind K, PK Andersen. Prognostic factors in small cell lung cancer: multivariate model based on 778 patients treated with chemotherapy with or without irradiation. Cancer Res 1986; 46: Ederer F, Axtell LM, Cutler SJ. The relative survival rate: a statistical methodology. Natl Cancer Inst Monogr 1961; 6: Cox DR. Regression models and life tables. J R Stat Soc B 1972; 34: SAS Institute. SAS user s guide: statistics, version 5 edition. Cary, NC: SAS Institute, 1985; Wingo PA, Ries LAG, Giovino GA, et al. Annual report to the nation on the status of Cancer, , with a special section on lung cancer and tobacco smoking. J Natl Cancer Inst 1999; 91: Osann K, Lowery J, Schell MJ. Small cell lung cancer in women: risk associated with smoking, prior respiratory disease, and occupation. Lung Cancer 2000; 28: Williams RR, Horm JW. Association of cancer sites with tobacco and alcohol consumption and socioeconomic status of patients: interview study from the Third National Cancer Survey. J Natl Cancer Inst 1977; 58: Risch HA, Howe GR, Jain M, et al. Are female smokers at higher risk for lung cancer than male smokers? A casecontrol analysis of histology type. Am J Epidemiol 1993; 138: Harris RE, Zang EA, Anderson JI, et al. Race and sex 776 Clinical Investigations

10 differences in lung cancer risk associated with cigarette smoking. Int J Epidemiol 1993; 22: Fasco M, Hurteau G, Spivack S. Gender-dependent expression of alpha and beta estrogen receptors in human nontumor and tumor lung tissue. Mol Cell Endocrinol 2002; 188: Omoto Y, Kobayashi Y, Nishida K, et al. Expression, function, and clinical implications of estrogen receptor in human lung cancers. Biochem Biophys Res Commun 2001; 285: Stabile LP, Davis A, Gubish CT, et al. Human non-small cell lung tumors and cells derived from normal lung express both estrogen receptor and and show biological responses to estrogen. Cancer Res 2002; 62: Di Nunno L, Larsson L, Rinehart JJ, et al. Estrogen and progesterone receptors in non-small cell lung cancer in 248 consecutive patients who underwent surgical resection. Arch Pathol Lab Med 2000; 124: CHEST / 127 / 3/ MARCH,