|
|
||
 |
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Journal of Clinical Oncology, Vol 24, No 15 (May 20), 2006: pp. 2245-2251 © 2006 American Society of Clinical Oncology. DOI: 10.1200/JCO.2005.04.8033 Never-Smokers With Lung Cancer: Epidemiologic Evidence of a Distinct Disease Entity
From the Department of Medical Oncology; Division of Clinical Trials and Epidemiological Sciences; Department of Therapeutic Radiology; Department of Surgical Oncology, National Cancer Centre; and the Department of Respiratory and Critical Care Medicine, Singapore General Hospital, Singapore, Singapore. Address reprint requests to Chee-Keong Toh, MD, Department of Medical Oncology, National Cancer Centre, 11 Hospital Dr, Singapore, Singapore 169610; e-mail: dmotck{at}nccs.com.sg
PURPOSE: Tobacco smoke is a definite causative agent for lung cancer. It is increasingly being recognized that never-smokers can be afflicted with non–small-cell lung cancer (NSCLC). We aim to assess survival differences between smokers and never-smokers with NSCLC.
PATIENTS AND METHODS: We analyzed 975 NSCLC patients who presented from January 1999 to December 2002. Clinical characteristics among current-, former- and never-smokers were tested using RESULTS: Of 975 patients, 59 had no smoking history and 33 had no quit time recorded. Of 883 patients analyzed, 286 patients (32.4%) were never-smokers. One hundred ninety-six never-smokers (68.5%) were females compared with 12% among current- and 13% among former-smokers (P < .001). There was a significant difference in histologic subtype between never-smokers and smokers: 69.9% with adenocarcinoma versus 39.9% (current-smokers) versus 47.3% (former-smokers); 5.9% with squamous cell carcinoma versus 35.7% (current-smokers) versus 28% (former-smokers; P < .001). Smokers had significantly poorer performance status (P = .002) and higher median age at diagnosis (P < .001) while more never-smokers presented with advanced disease (P = .002). Eight hundred and five patients (82.6%) died by May 30, 2005. The HR for smokers was significantly higher on both univariate and multivariate analysis (HR, 1.297; 95% CI, 1.040 to 1.619). CONCLUSION: Never-smokers comprised a high proportion of NSCLC patients in Singapore. Definite epidemiologic differences exist between never-smokers and smokers. Differences in survival outcome further suggest that the biology underlying the pathogenesis and behavior of the disease may be different for never-smokers.
The association of smoking as a risk factor for lung cancer has been well established from epidemiologic evidence since the 1950s.1,2 All of the earlier landmark studies were conducted in the Western countries where an estimate of 85% to 90% of lung cancer cases were attributed to smoking.3 As a result, never-smokers with lung cancer were under-represented and not well studied. Data from many Asian countries have emerged to establish the fact that never-smokers do constitute a proportion of non–small-cell lung cancer (NSCLC) patients that should not be ignored.4,5 These studies have demonstrated differences in epidemiologic characteristics and histologic subtypes between smokers and never-smokers, which led to the suggestion of existence of nontobacco related risk factors in the pathogenesis of NSCLC in a subgroup of patients. Possible risk factors included exposure to cooking fumes,6 hormones,7 menstrual cycle,8 and viral infection.9 Additional evidence that suggested differences in tumor biology between the never-smokers and smokers lay in the mutational frequencies and spectra observed in the tumor tissue itself.10,11 Lynch et al recently reported that a subgroup of patients with NSCLC, mainly adenocarcinoma, harbored a specific activating mutation in the epidermal growth factor receptor gene that correlated with clinical responsiveness to the tyrosine kinase inhibitor gefitinib.12 This was subsequently confirmed by other groups which showed consistently that mutations in the tyrosine kinase domain of epidermal growth factor receptor was more commonly found among women, never-smokers, Asians, and adenocarcinomas.13-18 All these data suggest that the pathogenesis of NSCLC in the never-smokers could possibly be different from the smokers. The treatment response and survival outcome could differ potentially between smokers and never-smokers with NSCLC given their possible differences in pathogenesis and tumor biology. Unfortunately, such differences (if they exist) were difficult to demonstrate as NSCLC has been treated as a single entity in a similar fashion, despite the well-recognized histologic heterogeneity. Several retrospective epidemiologic analyses have attempted to look at differences in survival outcome between smokers and never-smokers but were fraught with variables that might have affected the analysis.19-21 In this study, we hypothesize that never-smokers with NSCLC have different characteristics and survival outcome compared with smokers and should be regarded as a different disease entity. We previously reported our analysis on NSCLC patients who presented to the Department of Medical Oncology, National Cancer Centre.20 The study was limited by the fact that it mainly included patients with locally advanced and metastatic disease. In order to encompass all stages of the disease, we expanded our study population to include patients from the Departments of Respiratory and Critical Care Medicine, Radiation Oncology, and Surgical Oncology in two major referral centers in Singapore.
This study was conducted at the Singapore General Hospital and National Cancer Centre, Singapore. Patients had primary NSCLC diagnosed between January 1, 1999 and December 31, 2002. The study received the approval of the institutional review board of both centers. All patients diagnosed consecutively with available case records were reviewed. Some patients were excluded because of missing or irretrievable case notes. Individual case records were reviewed and selected epidemiologic characteristics, investigation results, American Joint Committee on Cancer (AJCC) stage, histologic subtypes, and Eastern Cooperative Oncology Group (ECOG) performance status were recorded. Selected epidemiologic characteristics included age, sex, and race of the patients. The presence of weight loss, comorbidities, type of treatment (surgery, radiotherapy, chemotherapy, targeted therapy, or combination of modalities) received as well as response to treatment were also reviewed. Weight loss was recorded in kilograms (kg) and significant weight loss was defined as more than 5 kg (taking an average weight of 50 kg, this will be equivalent to 10% of baseline body weight). The presence of comorbidities included one or more of the following conditions: diabetes mellitus, ischemic heart disease, hypertension, asthma, chronic obstructive lung disease, and pulmonary tuberculosis.
Smoking data included pack-years smoked and quit time for former smokers. Quit time was recorded as less than 2 years,
The smoking status was initially analyzed with three levels, namely current-, former- and never-smokers. The differences among current-, former-, and never-smokers were tested using The main outcome analyzed was all-cause mortality. Overall survival for each patient was measured from the date of diagnosis to the date of death or the date patient was last known to be alive for censored observation. Patients’ death data were checked with the Singapore Registry of Births and Deaths on 30 May 2005. The unadjusted hazard ratio (HR) and its 95% CI were calculated by univariate Cox regression. Additional multivariate analysis was done adjusting for those known prognostic factors and potential confounders that were significant in the univariate analysis (P < .10). Factors that were considered for inclusion included smoking status, sex, AJCC stage, ECOG status, presence of comorbidities, significant weight loss, and treatment received. Age at diagnosis was adjusted as a continuous variable in every Cox regression. Due to small numbers in the subgroups, ECOG status was categorized in two groups (0 to 1 v 2 to 4) and the histologic subtypes were similarly divided into two groups (squamous cell v nonsquamous cell). Patients with indeterminate weight loss or without weight loss information were excluded from analysis accordingly. Both current- and former-smokers were grouped together in the multivariate analysis as their HRs were similar on univariate analysis. Factors occurring on more than two levels were coded using dummy variables to estimate the HR between the different levels separately. P values were computed based on Wald tests. Although histologic subtype was not a significant factor in the univariate survival analysis, it was strongly associated with cigarette smoking. Therefore, the HRs of smokers compared with never-smokers were estimated and tested in the subgroup of histologic subtype. The smoking effect was also examined in subgroups of sex due to a major difference between the two groups as well as various categories of AJCC stage as stage is a well-known important prognostic factor. The overall survival curves and 5-year survival rates were calculated according to the Kaplan-Meier method. The differences in the curves were tested using the log-rank test.
Patient Characteristics A total of 975 NSCLC patients diagnosed between January 1999 and December 2002 were reviewed. Of the 975 patients, the majority 857 (87.9%) are ethnically Chinese. The Malays constitute 8.5%; the Indians 2.2% and 1.3% are of other ethnic groups (data not shown). Of the 975 patients, 59 had no smoking history and 33 did not have quit time recorded. There were 451 current- (51.1%), 146 former- (16.5%), and 286 never-smokers (32.4%; Table 1). There were significantly more males (88% of current- and 87% of former-smokers) and squamous cell carcinoma (35.7% of current- and 28% of former-smokers) among the smokers compared with the never-smokers (31.5% males and 5.9% squamous cell carcinoma). The proportion of females (68.5%) and adenocarcinoma (69.9%) was higher among the never-smokers.
As expected, the majority of the patients presented with advanced stage III-IV disease. Interestingly, never-smokers tend to present with more advanced disease than the current- and former-smokers (P < .001) and were diagnosed earlier at a median age of 62 years old (P < .001). The never-smokers had better ECOG status and a higher proportion of them received some form of treatment for their disease. There were no statistically significant differences between the groups with regards to the presence of significant weight loss (P = .329) or comorbidities (P = .126). As a substantial number of women were never-smokers, additional analyses were done within the never-smoking group to examine if the observed differences mentioned above had a sex differential (Table 2). Both sexes were similar in age, ECOG status, comorbidities, and weight loss. However, more females had adenocarcinoma (P = .009)
Univariate Survival Analysis At the time of analysis, a total of 805 deaths (82.6%) had occurred. One patient was excluded from this analysis, as his survival status was not clear after his diagnosis. The 5-year overall survival rates for the smokers and never-smokers were 7.7% and 10.8%, respectively (P = .0003; Fig 1). Accounting for sex differences, the 5-year overall survival rates of smokers and never-smokers among males were 7.7% and 10% while the 4-year survival rates were 12.4% and 17.6% among females respectively (results not shown). The unadjusted HR for current-smokers (HR, 1.364; 95% CI 1.156 to 1.609) and former-smokers (HR, 1.274; 95% CI 1.022 to 1.587) were significantly different from never-smokers (Table 3). Female patients had slightly lower risk of dying compared with males (HR, 0.877; 95% CI 0.753 to 1.021).
The HR of dying increased with poorer ECOG status and advanced AJCC stage. Never-smokers fared better compared with smokers at all stages, a result which was more obvious in patients with early stages. Patients with significant weight loss or who did not receive any form of treatment had an increased risk of dying. There was a slightly lower risk of dying for patients with nonsquamous cell carcinoma compared with those with squamous cell carcinoma. The presence of comorbidities increases the HR of dying (P = .073).
Multivariate Survival Analysis
Separate subset analyses were further conducted for sex, histologic subtype, and AJCC stage (Table 5.) Among the males, there was a higher risk of dying for smokers, but this was not statistically significant (HR, 1.293; 95% CI 0.971 to 1.723). Similarly, the HR for female smokers was higher but statistically nonsignificant (HR, 1.370; 95% CI, 0.956 to 1.963.) However, one must bear in mind that such subset analyses would result in loss of statistical power to detect a significant impact of smoking status on survival outcome when analyzed for sex. Among the nonsquamous cell carcinoma subtype, the smokers had a higher risk of dying compared with the never-smokers (HR, 1.257; 95% CI, 0.989 to 1.598). The small sample size of never-smokers with squamous cell histology may account for the failure to detect an impact of smoking status in this subset. The smokers have a higher HR of dying at all stages but the risk is much higher among patients with early stages (stage I-II) compared with the later stages (stage III and IV; Fig 2).
With a significant proportion of never-smokers among NSCLC patients in our population, we were able to compare differences in the characteristics and survival outcome between the current-, former- and never-smokers. It has been shown that lung cancer risk decreases with increasing duration of smoking abstinence.22 In the Nurses Health Study, former smokers had an age-adjusted risk reduction of 40% compared with current smokers after 2 to 5 years of quitting smoking.23 This was similarly observed in the study by Ebbert et al.24 Thus, we analyzed our data classifying former smokers as those who had quit smoking for more than 2 years, with the understanding that 2 years is arbitrarily defined as there is no consensus on an appropriate cut off time period. The current and former-smokers were similar in many characteristics and both have higher HR of dying than the never-smokers. This suggests that former smokers still retain many characteristics of current smokers, which is concordant with our hypothesis that never-smokers with NSCLC may be biologically different from those who had ever smoked before. Never-smokers constituted 32.4% of NSCLC patients in our population, with 14.6% among the males and 72.8% among the females. This proportion is among the highest in the world,25 except among the Japanese who reported almost similar proportions of never-smokers.26 We were unable to get accurate documentation of passive exposure to environmental tobacco smoke, which is a limitation of our study. However, we found differences between the smokers and never-smokers despite grouping all never-smokers as a group regardless of passive exposure. It is an interesting and significant observation that never-smokers were diagnosed earlier than their smoking counterparts (5 years and 10 years earlier than current- and former-smokers, respectively). This was similarly observed in a study by Koo et al which found that younger lung cancer patients were more likely to have been never-smokers in Asian populations while it was the opposite in the Western population.25 It has been shown that there is a clear dose-response relation between smoking and lung cancer where the risk increases with the number of cigarettes smoked, years of smoking, and earlier age at onset of smoking.27 Thus, our findings could be further strengthened if the age of onset of smoking and the pack-years smoked could be compared between the current and former-smokers; unfortunately, some of the details could not be obtained in our retrospective analysis. Never-smokers exposed to passive smoke could have a lower amount of carcinogen compared with the smokers and should, by inference, develop lung cancer later. However, this is not the case in our study and the study by Koo et al.25 Hence it is possible that factors exist other than passive exposure to tobacco smoke as well as a subset among the Asians (particularly the Chinese ethnic group) that are constitutionally ill-disposed to handle the putative causative factor(s), with the consequent earlier onset of cancer. Moreover, a meta-analysis in nonsmoking Chinese women found an absence of correlation between smoke exposure and adenocarcinoma in this group of patients.28 The retrospective nature of our study precluded accurate data collection of family history of lung cancer. Such data would have been useful in the interpretation and additional analysis of the reasons for an earlier onset of disease among the never-smokers. A worrisome observation from our study is that never-smokers were diagnosed at a more advanced stage compared with the smokers. This may either be due to late presentation on the part of the patients or late diagnosis on the part of the clinicians as lung cancer in never-smokers is comparatively less common and not at the top of a diagnostic list. Thus, the public and clinicians should be aware that lung cancer is not purely a disease of the smokers. To further support our belief that never-smokers should be viewed as a separate entity from the smokers, we found that the survival outcome between the groups were significantly different. After adjusting for known prognostic factors, multivariate analysis showed that smokers have an increased HR of dying as compared with the never-smokers. Nordquist et al also found an improved survival in never-smokers, however, they did not include performance status and comorbidities in their study and therefore the smokers could have fared poorer as they were presumed to have poorer ECOG status and had more comorbidities.21 Tammemagi et al also found smoking to be an independent predictor of poorer survival, but only 8.3% of their patients were never-smokers.19 In our previous study, we did not find a significant difference in survival between the two groups and this was likely due to the smaller sample size involved and inclusion of mainly patients with advanced disease.20 There is a lower than expected incidence of comorbidities among smokers, which could be due to an underestimate given the limitations of a retrospective study. Moreover, our study showed that comorbidities did not have any impact on survival outcome on univariate analysis. This could be due to the definition of comorbidities and our inability to grade its severity in our study. Complicating the issue is the intricate relationship between sex and smoking status, as there are significantly more never-smokers among the females. Some groups have advocated that women who smoke are more susceptible to lung cancer and those with lung cancer behave differently from men. However, this is still controversial as other studies have failed to demonstrate this result.29 In our study, we did not find sex to be a significant prognostic factor. Except for histologic subtypes, we did not find other differences in presentations between the sexes within the never-smokers. As presented in Table 2, the similarities between the female and male never-smokers are striking. The significantly higher proportion of adenocarcinoma among the female never-smokers can be explained by the higher proportion of tumors it was not possible to subtype in the male subgroup. This is in contrast to the study by de Perrot et al, which found that women with NSCLC had a better survival but their cohort included mainly patients with early disease who had undergone surgery.30 We also found that among patients with nonsquamous cell carcinoma histology, the never-smokers have a lower risk of dying compared with the smokers whereas there is no difference in HR between the two groups among patients with squamous cell carcinoma. However, this result must be interpreted with caution given the small number of never-smokers with squamous cell histology. This difference in outcome within histologic subtypes needs to be confirmed with additional studies as most large retrospective studies do not find histologic subtype an important prognostic factor.31,32 Using gene expression arrays, several studies have identified distinct differences between the two groups.10,33,34 The findings that the molecular profiles of NSCLC are different between smokers and never-smokers support a causative agent(s) other than tobacco smoke, as a single etiologic agent should result in similar, if not comparable, molecular aberrations. Future efforts should focus on conducting studies to further delineate underlying biologic differences, identify potential non–tobacco-related risk factor(s), and further refine treatment strategies for these two groups of patients.
The authors indicated no potential conflicts of interest.
Presented in part at the 11th World Conference on Lung Cancer, Barcelona, Spain, July 3-6, 2005. Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
Related Correspondence
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 or Kruskal-Wallis test. The hazard ratio (HR) for death and its 95% CI were calculated by Cox regression. 
2 to 5 years, 
T transversions in lung cancers reflect the primary mutagenic signature of DNA-damage by tobacco smoke. Carcinogenesis 22:367-374, 2001
