If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Non–small cell lung cancer (NSCLC) in never-smokers (NS) is vastly different from those with a smoking history in terms of etiology, driver mutations, and immunotherapy responsiveness. This study compares the real-world overall survival (OS) of NSCLC patients by smoking history and mutation status.
The study included 30,310 advanced or metastatic NSCLC patients in the Flatiron Health EHR-derived database who received biomarker testing results (EGFR, ALK, ROS1, and BRAF), and initiated therapy between 2011 and 2017, with follow up through June 2018. OS by smoking and driver mutation groups was summarized via Kaplan-Meier survival estimates, and compared in the context of a multivariate Cox proportional hazard model.
OS differed by smoking and driver-mutation categories (adjusted and stratified P< .001). The median OS for wild-type (WT) smoking patients was 9.6 months, for mutated (MT) smokers was 19.4 months (adjusted and stratified hazard ratio [HR] relative to WT smokers 0.65; 95% CI 0.60-0.71), for WT NS was 15.1 months (HR 0.78; 95% CI 0.73-0.83 relative to WT smokers), and for MT NS was 25.5 months (HR 0.52; 95% CI 0.48-0.58 relative to WT smokers).
NS with NSCLC survived longer than those with smoking history, in both groups of WT and mutation-positive patients. Findings highlight that in NSCLC patients, a history of never smoking may have similar effect on hazard of death as the presence of an actionable mutation. Taken together, differences in heredity, mutations, and biologic history suggest that NS lung cancer is a distinct clinical entity and must be managed accordingly.
While adenocarcinoma is the dominant histology of lung cancer in never smokers, smoking related lung cancer demonstrates the full spectrum of histology including small cell, squamous (SCC), and nonsquamous cell (NSCC) types (Figure 1)
Over the last 2 decades, actionable molecular drivers of lung cancer have been defined including EGFR, ALK, and ROS1 and these occur almost exclusively in advanced non–small cell lung cancer (NSCLC) in never smokers. Here, we will distinguish between patients with an actionable mutation (MT) and those without documented a genetically defined, actionable mutation, or wild-type (WT) patients. Major racial and geographic differences exist in the prevalence of this genetically defined cancer. For instance, in East Asian women, 60% to 80% of lung cancers are diagnosed in never smokers
of all patients with advanced NSCLC, respectively. Inhibition of these driver mutations in the metastatic setting is more effective and less toxic than conventional chemotherapy and has become the standard of care.
Immunotherapy with PD-1/PD-L1 inhibition, alone or in combination with chemotherapy, has now replaced conventional chemotherapy as both the first and second-line treatment for patients with WT advanced NSCLC.
Data from an international retrospective registry study demonstrated objective response rates of less than 20% to single agent immunotherapy in patients with ROS1, MET, EGFR, HER2, RET, and ALK mutations with a median progression free survival of only 2.8 months.
Notably, 72% of the patients in this analysis were never smokers.
Lastly, heritability also appears to play a role in lung cancer in never smokers. Epidemiologic data using a familial database show statistical evidence of heritability in never-smokers lung cancer not seen in smokers.
These findings appear to be limited to those with nonsquamous NSCLC. Although possible germline mutations responsible for this observation have not been identified till date, the inheritance is observed even beyond seventh degree relatives and is not observed in the spouses of never smokers with lung cancer.
Treatment of lung cancer in a patient with MT never-smoking lung cancer is very different from the management of smoking associated lung cancer. Unfortunately, the majority of never smokers with lung cancer are WT, and thus do not benefit from molecular targeted therapy. Presently, those patients with WT never-smoking lung cancer are treated similarly to those with a smoking history, but their natural history has not been characterized. It is unclear if never smoking patients with WT lung cancer will behave more like smoking patients with WT lung cancer, never smoking patients with MT lung cancer, or somewhere in between. In the current manuscript we evaluated real world survival outcomes of patients with advanced NSCLC treated at both community and academic settings, and further stratified the outcomes by smoking status and presence or absence of electronic health record (EHR) evidence of EGFR, ALK, ROS1, or BRAF driver mutations.
Patients and Clinical Data Extraction
This analysis uses data from the nationwide US Flatiron Health EHR-derived de-identified database comprising structured and unstructured data curated via technology-enabled abstraction and supplemented with third-party death information.
Institutional Review Board approval of the study protocol was obtained prior to study conduct, and included a waiver of informed consent. Data provided to third parties were de-identified according to the Health Insurance Portability and Accountability Act guidelines. This analysis consisted of observations on 30,310 patients receiving at least 1 line of therapy for advanced or metastatic NSCLC with advanced or metastatic diagnosis dates from January 1, 2011 to December 31, 2017, seen at 177 community practices and 2 academic medical centers in the United States.
This analysis was limited to patients who had a clinic visit or medication order within 90 days of advanced or metastatic disease diagnosis to minimize the potential impacts of patients who were not primarily engaged with the relevant practice; eg, patients seeking a second opinion. Patient data was collected through June 30, 2018, which provided at least 6 months of potential follow-up for all patients. Further, only documented driver mutation testing results (EGFR, BRAF, ALK, and ROS1) and PDL1 expression levels based on result data which were available by the date of initiation of first-line therapy were considered in this analysis. In particular, patients who were not tested for a particular mutation on or before first-line initiation were considered as WT for the relevant mutation. Smoking status was abstracted via EHR review and documented as a history of smoking, no history of smoking, or unknown. Mutation status was also abstracted from documentation of the presence or absence of mutation, type of testing performed (NGS, IHC, FISH, or PCR), source of sample (tissue or blood), and tissue type (primary tumor vs. metastatic site). Data on receipt of cancer directed oral therapy was abstracted and combined with structured intravenous therapy for the following variables: name of therapy, line of therapy, start and end dates of therapies, and the receipt of maintenance therapy or not.
The characteristics of the total study population were summarized as number and percent of patients within relevant categories, and compared via chi-squared tests with patients categorized by smoking status and presence of driver mutation. Overall survival, subject to right-censoring, was calculated as the time from date of initiation of first-line therapy to the date of death, censoring at the last visit date or end of most recent oral therapy. OS of patients within smoking status and driver mutation groups was summarized via Kaplan-Meier survival estimates, and analyzed using univariate and multivariate Cox proportional hazards models. The adjusted multivariate models included age, gender, histology, smoking status, race/ethnicity, PDL1 expression level and grouped first-line therapy, and were stratified by advanced diagnosis year within each practice, given anticipated practice-specific time trends. All hazard ratios (HR) reported in the text are from the multivariate, stratified model.
Our study population included 30,310 patients diagnosed with advanced NSCLC in the United States who received at least one line of therapy for advanced or metastatic disease. A patient inclusion flow diagram is provided in Appendix Figure AAA, which details exclusions to the full Flatiron Health advanced NSCLC patient cohort to arrive at the analytic cohort. The baseline characteristics of the total study population are summarized in Table 1. The median age was 68 years, 46.7% (n = 14,146) of patients were female, 76.8% (n = 20,814) were non-Hispanic white, 4.6% (n = 1387) had high (≥50%) PD-L1 expression and 12.8% (n = 3818) were never-smokers. PD-L1 status was not documented/unknown in 87.6% of the population. Overall, 15,931 (52.6%) patients had no EHR evidence of testing for targetable mutations (EGFR, ALK, ROS1, and BRAF) on or before first-line initiation, while 10,527 (34.7%) patients had no EHR evidence of testing for targetable mutations at any point in their disease course. Smoking status was unknown or not documented in 1.3% (n = 383) of the population. Actionable mutations were reported in 9.1% of the study population (n = 2763), differentially as 34.2% (n = 1305) in never-smokers and 5.5% (n = 1443) in those with a smoking history. Fifteen patients with unknown or not documented smoking status also had the presence of a mutation. Females were more likely to be never smokers, and were also more likely to have driver mutations regardless of smoking status. Asians consisted of only 2.6% (n = 709) of total patients with NSCLC in this US data set but represented 16% (n = 190) of the mutated never smokers. The most common histology was nonsquamous cell carcinoma (NSCC, 69.8%, n = 21,167), of which the vast majority would be expected to be adenocarcinomas. NSCC was substantially more common among never-smokers (86.8%, n = 3315 vs. 67.5%, n = 17,619 among patients with a smoking history) and patients with a targetable mutation (96.2%, n = 2643 vs. 67.3%, n = 18,291 among patients without an actionable mutation). Mutations were rare in patients with squamous cell histology, for both smokers and never smokers, 2.8% (n = 41) and 1.2% (n = 16), respectively.
Table 1Patient Characteristics by Driver Mutation and Smoking Status
EHR evidence of EGFR mutations was identified in 7.1% (n = 2149) of the total advanced NSCLC population, in 27.4% of never smokers and in 4.2% of patients with a history of smoking. ALK rearrangement evidence was found in 1.4% (n = 437) of the total advanced NSCLC population, in 5.5% of never smokers, and in 0.9% of patients with a history of smoking. ROS1 mutation evidence was found in 0.2% (n = 64) of the total advanced NSCLC population, in 0.9% of never smokers, and in 0.1% of patients with a history of smoking. Of the 3002 patients with advanced NSCLC who initiated first-line treatment on or after July 1, 2017, EHR evidence of BRAF mutations was identified in 1.3% (n = 38) of these patients, in 1.0% of never smokers, and in 1.3% of patients with a history of smoking.
In the overall cohort, male gender, squamous or not-otherwise-specified (NOS) histology, and older age were independent predictors of a worse OS in both univariate and adjusted analysis. While there were significant survival differences between Black and Hispanic populations compared to non-Hispanic white patients, Asian patients had a substantially reduced risk of death in this analysis. Patients with lung cancer that had PD-L1 expression >50% had a modest improvement in survival. Both history of smoking and lack of targetable mutation were associated with worse survival outcomes, with WT smokers having the worst OS and MT never smokers having the best. Smokers with targetable mutations and never smokers without targetable mutations were intermediate, with smokers having a targetable mutation faring slightly better (Table 2).
Table 2Histology Specific Outcomes in Univariate and Multivariate Analysis
Univariate and multivariate models of survival outcomes based on smoking status (smokers vs. never smokers) are summarized in Table 3. Male patients were found to have a higher risk of death compared to females in both univariate and multivariate analysis among both smokers and never smokers. NSCLC NOS histology was also associated with a higher risk of death compared to the other histologies. Further, the negative impact of NOS histology appeared to be greater among never smokers than smokers. Never smokers with SCC also demonstrated a substantially increased risk of death compared to never smokers with NSCC. Age also substantially impacted risk of death, with never smokers demonstrating a greater negative impact of older age than smokers. Finally, PD-L1 expression >50% was associated with a lower risk of death among never smokers than low levels of PD-L1 expression. Supplemental eTable 2 details histology specific outcomes in univariate and multivariate analysis among patients who were tested for at least one targetable mutation before first-line initiation of therapy.
Table 3Smoking Specific Outcomes in Univariate and Multivariate Analysis
Notably, in the real world more than 75% of EGFR and ALK mutated patients received first-line TKI therapy, compared to 68.8% of ROS1 mutated patients. eTable 1 provides a summary of first-line therapies both overall and by smoking status.
The median OS was 10.7 months (95% CI: 10.5-10.9), 9.9 months (95% CI: 9.7-10.1), and 22.5 months (95% CI: 21.2-23.7) for the overall advanced NSCLC patient population, WT patients with advanced NSCLC, and MT patients with advanced NSCLC respectively (multivariate stratified HR 0.65 (95% CI 0.60-0.70) for patients with EHR evidence of actionable mutations relative to those without). There was strong evidence that OS differed according to both smoking and driver mutation categories (multivariate stratified P< .001 for all comparisons). The median OS for smoking patients with WT mutation status was 9.6 months, while median OS for smoking patients with a targetable mutation was 19.4 months (HR 0.65 relative to WT smokers; 95% CI: 0.59-0.71). The median OS for WT never-smoking patients was 15.1 months (HR 0.78 relative to WT smokers; 95% CI: 0.74-0.83), while the median OS for never-smoking patients with a targetable mutation was 25.5 months (HR 0.52 relative to WT smokers; 95% CI: 0.47-0.57).
The median OS for all patients whose tumors had EGFR mutations was 22.7 (95% CI: 21.7-24.0) months, 25.1 (95% CI: 23.5-26.9) months for never-smoking patients with EGFR mutations and 20.1 (95% CI: 19.0-22.3) months for smoking patients with EGFR mutations. The median OS for all patients whose tumors had ALK mutations was 24.8 (95% CI: 20.5-31.0) months, 38.3 (95% CI: 27.2-57.1) months for never-smoking patients with ALK mutations and 16.6 (95% CI: 12.5-23.7) months for smoking patients with ALK mutations. The median OS for all patients whose tumors had ROS1 mutations was 18.7 (95% CI: 14.7-not reached) months, 21.3 (95% CI: 14.7 – not reached) months for never-smoking patients with ROS1 mutations and 17.8 (95% CI: 13.5 – not reached) months for smoking patients with ROS1 mutations. The median OS for all patients whose tumors had BRAF mutations was 11.1 (95% CI: 8.3-17.8) months, 15.7 (95% CI: 5.7 – not reached) months for never-smoking patients with BRAF mutations and 10.8 (95% CI: 8.0-19.3) months for smoking patients with BRAF mutations.
Lung cancer in never smokers is characterized by a different natural history, carcinogenesis, treatment approach, and response to immunotherapy
relative to lung cancer in smokers. In our real-world analysis of 30,310 patients with advanced NSCLC diagnosed from 2011 to 2017, 12.8% of cases were diagnosed in never smokers. Based on estimates from the American Cancer Society for 2019,
lung cancer in never smokers, as an entity separate from smokers’ lung cancer, would represent the seventh most common cause of cancer death in the United States. As the frequency of smoking continues to decline in the USA,
the proportion of never smokers among lung cancers will continue to increase, further highlighting the need to better characterize this disease.
While the dominant feature distinguishing the treatment and outcomes for lung cancer in never smokers from that in smokers is the presence of an actionable mutation, only one-third of lung cancers in never smokers have a targetable mutation. Our real-world analysis confirms that never-smoking status is common in lung cancer, but 65% of never smokers with lung cancer did not have an actionable mutation. At the population level of never-smoking patients with NSCLC, outcomes are mostly reflective of the majority subset of those without a known driver mutation. For these reasons, we sought to characterize the features and outcomes for lung cancer in smokers and never smokers by the presence or absence of actionable mutations with a goal to determine if never smokers without a mutation would behave more like their clinical history or somatic mutation.
The hazard ratios for smoking and the presence of an actionable mutation were similar in this analysis, suggesting that a history of smoking was associated with a similar effect on overall survival outcomes as does lack of an actionable mutation. We also found that survival was strongly affected by smoking status in both patients with and without mutations. It is well known that cigarette smoking strongly contributes to cardiovascular mortality,
which might shorten the survival of patients with lung cancer, but we suspect this influence to be small as cancer death is measured in weeks while cardiovascular death is measured in years. Accordingly, we hypothesize that smoking itself may exert an independent effect on cancer aggressiveness and the host. It is possible that smokers have a worse prognosis due to nontargetable mutations, or some other unidentified pathologic feature inherent to the tumor.
Our results are concordant with the previously described outcomes of the Cancer Care Outcomes Research and Surveillance (CanCORS) large observational cohort. In CanCORS, it was shown that compared with never-smoking patients, the adjusted hazard of death from any cause was 29% higher among former smoking patients (HR 1.29; 95% CI: 1.08-1.55), and 39% higher among current/recent smoking patients (HR 1.39; 95% CI: 1.16-1.67).
This study has several limitations. A fundamental limitation of an analysis using real-world data collected and abstracted from the EHR is the completeness of the data. Real-world EHR based data is also subject to shifting practice, testing, EHR usage, patient populations, and time trends, in addition to variation in each of these by site and provider. An additional limitation is due to the definition of actionable mutation based on EHR documentation of the relevant mutations. Further, test results on some targetable mutations may have emerged after initiation of first-line therapy (eg, in patients with visceral crisis requiring urgent treatment while awaiting mutation results), which could bias the survival association. We expect that some portion of patients without EHR documentation of an actionable mutation did in fact have such a mutation. Additionally, only 75% of those with actionable mutations were documented to have matched targeted therapy, with a potential for missingness. This may in part be explained by patient on study drugs as part of a clinical trial, which are blinded to prevent re-identification.
The difference in survival outcomes related to smoking may have been greater with a more precise smoking history. In the current study, never smokers were defined as those without a smoking history, whereas smokers were defined as those individuals who currently smoke or have ever smoked tobacco, cigarettes, cigars, and pipes. Thus, the smoking group included a heterogeneous group of patients: current smokers, former smokers (irrespective of the duration of quitting smoking) and even light smokers. The inclusion of light smokers in the smoking group might have increased the number of patients with actionable mutations in this group, and could have subsequently skewed the results. This differs from accepted definitions of smoking status by the National Center of Health Statistics, where never smokers are defined as adults who never have smoked, or who have smoked less than 100 cigarettes in their lifetime, former smokers are adults who have smoked greater than 100 cigarettes in their lifetime but have quit smoking, and current smokers are adults who have smoked greater than 100 cigarettes/lifetime and currently smoke. Lastly, smoking status was not documented in 1.3% of the advanced NSCLC population and may have been subject to provider interpretation, both of which may have affected our results.
NSCLC in WT never smoking patients is a biologically distinct disease with different risk factors and improved survival outcomes compared to lung cancer in WT smoking patients. These findings highlight that in patients with NSCLC, a history of smoking was associated with a similar magnitude (and opposite) of effect on survival as documentation of the presence of an actionable mutation.
Clinical Practice Points
Lung cancer in never smokers is a distinct clinical entity with unique epidemiologic, biologic, and molecular features that distinguish it from lung cancer developing in those with a smoking history. For those never-smoking NSCLC patients with driver mutations, first line targeted therapy can provide long progression free and overall survivals with good quality of life. Unfortunately, a large proportion of never-smokers will not have targetable mutations, and their outcomes with standard therapy are less well characterized.
In this analysis of more than 30,000 American patients with NSCLC, we found that the hazard ratios for smoking and the presence of an actionable mutation were similar. This suggests that in the real-world setting, a history of smoking is associated with a similar effect on overall survival outcomes as does lack of an actionable mutation. We also found that overall survival among real-world patients with NSCLC was strongly affected by smoking status irrespective of mutation status. This study reinforces the importance of obtaining an accurate smoking history in clinical practice and promoting smoking cessation among patients with NSCLC undergoing treatment. Additionally, this information can be used to prognosticate in the clinical setting.
Supported by the National Cancer Institute of the National Institutes of Health under award number P30CA042014.
Dr. Kerrigan and Dr. Haaland report travel assistance from Flatiron Health. Dr. Haaland has served as a consultant for Astra Zeneca and Prometric Life Sciences. Dr. Adamson is an employee of Flatiron Health, a subsidiary of Roche, and owns stock in Roche. Drs. Puri and Akerley report no relevant financial COI. Ms. Xuechen Wang reports no COI. Dr. Patel reports institutional funding from Takeda.