Recent Survival Gains in Stage IV NSCLC by Sociodemographic Strata

Abstract

Introduction

The management of stage IV NSCLC has been transformed by recent innovations. Nevertheless, access to medical innovations varies across sociodemographic groups in the United States, which may affect the rate of outcome improvements. Our objective was to evaluate the recent real-world gains in the survival of patients with stage IV NSCLC across sociodemographic groups.

Methods

The National Cancer Database was queried for treated patients diagnosed with stage IV NSCLC between 2010 and 2020. Data was analyzed in three eras (2010–2013, 2014–2017, and 2018–2020). Two-year survival was assessed using the Kaplan-Meier method. Adjusted mortality risk was calculated using stratified Cox analysis.

Results

A total of 393,586 patients with stage IV NSCLC received treatment. Chemotherapy administration decreased (from 64.8% to 25.1%), radiation therapy decreased (from 54.3% to 27.6%), and immunotherapy increased (from 2.0% to 51.8%). Between eras 1 and 3, median survival increased by 53.7% (6.7–10.3 mo); nevertheless, not all groups improved at the same pace. The median survival increased by 81% (from 8.3 to 15.0 mo) for Hispanic patients, by 54.7% (from 6.7 to 10.3 mo) for non-Hispanic Blacks, and by 46.7% (from 6.6 to 9.6 mo) for non-Hispanic Whites. The median survival of uninsured patients increased from 5.8 to 7.2 months (24.1%), whereas that of patients with private insurance increased from 8.6 to 14.7 months (70.9%).

Conclusions

The survival of patients with treated stage IV NSCLC has improved considerably over the past decade. Nevertheless, expected survival and the pace of improvement differed across sociodemographic groups. Further studies to understand this outcome variability may enhance the effectiveness and equity of NSCLC treatments.

Introduction

Lung cancer is the leading cause of cancer-related deaths both in the United States and worldwide.^1,2 Approximately 40% of patients diagnosed with NSCLC are diagnosed with stage IV disease, which has been associated with particularly poor outcomes, with an estimated 5-year survival rate of 5.8%.3, 4, 5

Platinum-based chemotherapy has been the cornerstone of first-line treatment for patients with stage IV NSCLC over the past 40 years.⁶ Over the past decade, the management of advanced NSCLC has changed in response to innovations such as immune checkpoint inhibitors, the development of increasingly effective targeted therapies such as tyrosine kinase inhibitors (TKIs), and new radiation protocols.6, 7, 8 These innovative treatments have been associated with improvements in survival for patients with stage IV NSCLC. For example, administration of immune checkpoint inhibitors improved progression-free survival from 6.0 months to 10.3 months compared with patients treated with chemotherapy.^8,9 For patients with EGFR mutations, osimertinib, a third-generation EGFR TKI, increased median progression-free survival from 10.2 months to 18.9 months versus standard TKIs.^10,11 Nevertheless, innovation in cancer care does not always fully translate into the survival benefits experienced in the clinical trial setting, as patients may not be as able or willing to comply with treatment regimens.^12,13 As a result, it is unclear how these innovative treatments have impacted the prognosis of U.S. patients with stage IV NSCLC in the real-world setting.

Historically, not all patients in the United States have had the same access to innovative treatments. Additionally, the rate of adoption of innovative treatments differs accross groups. For example, before Food and Drug Administration approval and in the 2 years after approval, Black patients with lung cancer were less likely to receive immunotherapy than White patients.¹⁴ A separate study found that in a group of patients with advanced-stage NSCLC, Black patients were less likely to receive immunotherapy even when stratified by insurance.¹⁵ In addition, certain demographic groups (African American, Hispanic, American Indian or Alaskan Native, people with lower socioeconomic status, and people who lived in remote or rural areas) have been shown to be less likely to receive comprehensive biomarker testing that could potentially make them eligible for innovative therapies such as targeted therapy or clinical trials.¹⁶ Given the potential for disparate access to innovative treatment, it is unclear how changing treatment paradigms have impacted survival across different sociodemographic groupings.

The National Cancer Database (NCDB) captures detailed information on approximately 70% of newly diagnosed cancers in the United States and serves as an effective resource for outcome-based research, including lung cancer research.^17,18 We evaluated trends in 2-year survival in patients diagnosed with stage IV NSCLC between 2010 and 2020 in the NCDB. Our objectives were to determine real-world outcomes during this dynamic time frame and identify trends in survival across different sociodemographic groups to potentially expose opportunities to address disparities.

Materials and Methods

Data Source

The NCDB is a database of oncology patients developed through a collaborative program of the Commission on Cancer of the American College of Surgeons and the American Cancer Society.¹⁸ The NCDB Participant User File for 2021 containing de-identified patient information was used for the purpose of this study. The study was performed in compliance with the institutional review board guidelines, and the requirement for patient consent was waived. Guidelines for Strengthening the Reporting of Observational Studies in Epidemiology guidelines were followed.

Study Sample

Patients diagnosed with stage IV NSCLC between 2010 and 2020 were included in this study. Patients who did not receive any treatment or had missing race/ethnicity or follow-up data were excluded (n = 25,506 excluded patients; the sensitivity analysis did not identify any obvious differences between this group and the study population). Recognizing that a series of progressive changes in lung cancer treatment occurred during this period, the overall sample was divided into three cohorts on the basis of the year era 1 (2010–2013), era 2 (2014–2017), and era 3 (2018–2020).

Variables

Independent variables

Race-ethnicity was considered in the following groups¹⁹: non-Hispanic White, Hispanic, non-Hispanic Black, Asian, American Indian or Alaskan Native, Pacific Islander or Native Hawaiian, and Other (including patients documented as “unknown”). Additional covariates included: age (18–65 y, 66–75 y, or ≥76 y), sex (male or female), region (Midwest, Northeast, South, West, or unknown), insurance (Medicaid, Medicare, Uninsured, Private, Other), median income (<$30,000, $30,000–$34,999, $35,000–$45,999, or ≥$46,000), facility type (Academic, Non-Academic, or unknown), Charlson score (0, 1, or ≥2), tumor histology (Adenocarcinoma, Squamous, or Other), treatment (Chemotherapy, Radiation, Chemoradiation, Immunotherapy, or Surgery). Some patients were documented to have been treated; nevertheless, the specific treatment was unknown. Additional covariate details are available online.²⁰

Dependent variables

The 2-year overall survival was determined in months from the time of diagnosis. Patients who were alive at the last follow-up or after 2 years were considered censored.

Missing Data Strategy

The cumulative percentage of missing data eligible for imputation was less than 2% across all variables and seemed to be missing at random. Therefore, a complete case study was conducted.

Statistical Analyses

Overall, the unadjusted 2-year survival was assessed using the Kaplan-Meier analysis. Stratified Cox model analysis controlling for age, sex, insurance status, Charlson score, region, facility type, and treatment was performed to assess the differences in survival by racial and ethnic group within each diagnosis period. We also assessed the proportion of changes in the median survival for each race across the diagnosis periods. The significance level was set at a two-sided p value of less than 0.05, and 95% confidence intervals (CIs) were used. Data were analyzed using SAS statistical software version 9.4.

Results

A total of 393,586 patients were identified, including 312,361 (79.4%) non-Hispanic White, 48,389 (12.3%) non-Hispanic Black, 13,540 (3.4%) Hispanic, and 13,329 (3.4%) Asian patients, as outlined in Table 1. The patients were predominantly male, 212,036 (53.9%), with a median age of 67 years (interquartile range: 60–75 y). There were trends in race, ethnicity, and economic factors. For example, non-Hispanic Blacks had the highest proportion of patients with the lowest income (<$30,000 per year) (30.8%), whereas Asians were less likely to have a low income (5.3%, p < 0.001). Hispanics comprised the highest percentage of uninsured patients (9.0%). Asians were more likely to have a lower Charlson score (0) (75.0%, p < 0.001), whereas non-Hispanic Blacks had the highest percentage of patients with a high Charlson score (≥2) (15.9%, p < 0.001) (Supplementary Table 1).

Table 1.

Overall Sample Characteristics

Variable	Category	Hispanic (%) n = 13,540		Non-Hispanic White (%) n = 312,361		Non-Hispanic Black (%) n = 48,389		American Indian or Alaskan Native (%) n = 1216		Asian (%) N = 13,329		Native Hawaiian or Pacific Islander (%) n = 909		Other (%) N = 3842		p value
Age	Median (IQR)	65 (57–64)		68 (60–75)		64 (57–72)		65 (57–72)		66 (58–75)		64 (57–72)		66 (58–74)
	≤65 y	6790	50.1	129899	41.6	26621	55.0	643	52.9	6301	47.3	495	54.5	1871	48.7	<0.001
	66–75 y	4034	29.8	105432	33.8	14256	29.5	392	32.2	3966	29.8	267	29.4	1204	31.3
	>75 y	2716	20.1	77030	24.7	7512	15.5	181	14.9	3062	23.0	147	16.2	767	20.0
Sex	Male	7686	56.8	167470	53.6	26809	55.4	623	51.2	6836	51.3	509	56.0	2103	54.7	<0.001
	Female	5854	43.2	144891	46.4	21580	44.6	593	48.8	6493	48.7	400	44.0	1739	45.3
Insurance	Uninsured	1224	9.0	9525	3.0	2862	5.9	44	3.6	558	4.2	24	2.6	203	5.3	<0.001
	Private	3759	27.8	89980	28.8	12387	25.6	281	23.1	5030	37.7	327	36.0	1292	33.6
	Medicaid	2383	17.6	21936	7.0	8243	17.0	151	12.4	1810	13.6	153	16.8	444	11.6
	Medicare	6052	44.7	185454	59.4	23892	49.4	615	50.6	5816	43.6	389	42.8	1789	46.6
	Other	122	0.9	5466	1.7	1005	2.1	125	10.3	115	0.9	16	1.8	114	3.0
Charlson score	0	9098	67.2	193868	62.1	29520	61.0	724	59.5	9997	75.0	583	64.1	2695	70.1	<0.001
	1	2892	21.4	74935	24.0	11197	23.1	312	25.7	2323	17.4	191	21.0	739	19.2
	≥2	1550	11.4	43558	13.9	7672	15.9	180	14.8	1009	7.6	135	14.9	408	10.6
Facility	Non-academic	7614	56.2	221812	71.0	27791	57.4	961	79.0	7148	53.6	565	62.2	2095	54.5	<0.001
	Academic	5528	40.8	88708	28.4	20113	41.6	238	19.6	5862	44.0	331	36.4	1698	44.2
	Unknown	398	2.9	1841	0.6	485	1.0	17	1.4	319	2.4	13	1.4	49	1.3
Histology	Adenocarcinoma	9737	71.9	208305	66.7	32486	67.1	761	62.6	10852	81.4	686	75.5	2691	70.0	<0.001
	Squamous	2260	16.7	66459	21.3	9970	20.6	309	25.4	1388	10.4	142	15.6	677	17.6
	Other	1543	11.4	37597	12.0	5933	12.3	146	12.0	1089	8.2	81	8.9	474	12.3

IQR, interquartile range.

Changes in Stage IV NSCLC Treatment

The use of chemotherapy as a single-modality therapy decreased during the three diagnosis eras (31.8% in era 1, 25.6% in era 2, and 12.7% in era 3), a trend observed in all socioeconomic groups (Fig. 1). The use of radiation, alone or in combination with chemotherapy, decreased from 54.3% in era 1 to 27.6% in era 3. The use of surgery also decreased in all racial and ethnic groups between eras 1 and 3. In contrast, the use of immunotherapy increased for all racial and ethnic groups across the diagnosis eras, with the greatest increase occurring between eras 2 and 3.

Figure 1.

Distribution of treatment across race-ethnic groups for the three diagnosis periods. Percentage of patients that underwent each specific treatment across the three diagnosis eras. Immunotherapy combinations included immunotherapy in addition to other treatments, other combinations (no immuno) included other combinations of treatments not previously represented when immunotherapy was not a part of the treatment regimen. AI/AN, American Indian or Alaskan Native; Chemo, chemotherapy; Chemorad, chemotherapy and radiation; Immuno, immunotherapy; NH Black, non-Hispanic Black; NH/PI, Native Hawaiian or Pacific Islander; NH White, non-Hispanic White.

Treatment patterns varied across racial groups. Non-Hispanic White and American Indian or Alaskan Native (AI or AN) patients had the highest rates of immunotherapy use in the third era (53.7% and 51.6%, respectively), increasing from 1.6% and 2.1%, respectively, in the first era for each group. In contrast, Asians had the lowest rate of immunotherapy use in the third era (34.2 %).

Unadjusted Survival

All sociodemographic groups reported considerable improvements in median survival across the three diagnostic eras (Fig. 2). The median survival rates were highest across all three eras for Asian patients (12.9 mo, 17.0 mo, and 22.5 mo) and Hispanic patients (8.3 mo, 10.1 mo, and 15.0 mo) and lowest among AI or AN patients (6.9 mo, 6.9 mo, and 9.0 mo), non-Hispanic White patients (6.6 mo, 7.4 mo, and 9.6 mo), and non-Hispanic Black patients (6.7 mo, 7.8 mo, and 10.3 mo).

Figure 2.

Median survival in months by race-ethnicity for the three diagnosis periods. Median survival (mo) by race-ethnicity for the three diagnosis eras. Era 1: 2010 to 2013, Era 2: 2014 to 2017, Era 3: 2018 to 2020.

To assess the rate of improvement over time, the proportionate change between eras was assessed across different socioeconomic groups. There was considerable variability in the rate of increase in the median survival across different races and ethnicities (Fig. 3). There seemed to be an acceleration in survival gains among Hispanic patients during the study period, with a 21.4% increase in overall survival between eras 1 and 2 and a 48.8% increase between eras 2 and 3. For Asian patients, the relative increase in the median survival between eras 1 and 2 and eras 2 and 3 was stable at 32%. The lowest relative change across the study interval was seen in AI or AN patients, with no change in survival between eras 1 and 2 and a 30.6% change between eras 2 and 3.

Figure 3.

The proportion of change in median survival between diagnosis periods. Era 1 (2010–2013), Era 2 (2014–2017), Era 3 (2018–2020). Proportional change is described as a percentage change. For example, the difference between era 1 and era 2 for Asian patients was a proportional change of 1.32 and a percentage change of 32%. AI/AN, American Indian or Alaskan Native; NH Black, non-Hispanic Black; NH White, non-Hispanic White; NH/PI, Native Hawaiian or Pacific Islander

Changes in survival were evaluated across the other sociodemographic groups (Supplementary Fig. 1). There were considerable differences noted in insurance status with the highest median survival across all three diagnosis eras for patients with private insurance (8.6 mo, 10.3 mo, and 14.7 mo) and the lowest for uninsured patients (5.8 mo, 5.9 mo, and 7.2 mo). Patients treated at academic centers had better median survival across all three diagnosis periods (7.8 mo, 9.1 mo, 12.3 mo) than patients treated at non-academic centers (6.2 mo, 7.1 mo, and 9.3 mo).

Adjusted Survival

Adjusted Cox analysis was performed using non-Hispanic White patients as the reference group. Almost all groups fared better than the non-Hispanic White patients (Fig. 4). Asian patients had the lowest mortality hazard across all three diagnostic eras (0.70, 95% CI: 0.67–0.73, p < 0.0001 in era 1; 0.74, 95% CI: 0.71–0.76, p < 0.0001 in era 2; and 0.75, 95% CI: 0.72–0.79, p < 0.0001 in era 3). AI or AN patients had the highest mortality risk, although the difference was not statistically significant (Supplementary Table 2).

Figure 4.

HRs per race-ethnicity for the three diagnosis periods. Forest Plot of HRs obtained through Cox analysis. AI/AN, American Indian or Alaskan Native; 95% CI, confidence interval; HR, hazard ratio; NH Black, non-Hispanic Black; NH/PI, Native Hawaiian or Pacific Islander; NH White, non-Hispanic White.

Sites of Metastasis

Owing to the heterogeneity of stage IV disease, we evaluated the different sites of metastases to assess the differences across racial and ethnic groups. (Supplementary Table 3). Non-Hispanic Black patients had the highest percentage of patients with multiple-site metastases, whereas Asian patients had the lowest number of multiple-site metastases (73.4% versus 64.5%). The non-Hispanic White group had the highest percentage of patients with brain metastases (68.9%).

Discussion

The survival of patients with stage IV NSCLC in real-world settings has improved considerably over the past decade in all sociodemographic groups. These findings are consistent with a trend in declining mortality for NSCLC in the United States.²¹ This current study focused on stage IV NSCLC, which in theory should be more directly related to the behavior of the cancer and the effectiveness of treatment, as most deaths in patients with stage IV lung cancer are related to their malignancy.²² To some extent, improved survival would be attributable to the increased use of innovative treatments that have been linked to improved survival.^15,21 There have been a number of studies that have raised questions about the adoption and impact of innovative treatment in the real-world setting.^13,23 Nevertheless, the observed dramatic changes in treatment approaches and impressive gains in survival are reassuring that trial discoveries are impacting real-world practice and outcomes in the United States.

This study identified differences in survival across sociodemographic groups. Survival is highly variable across races and ethnicities. Hispanic and Asian patients consistently had the most favorable survival, whereas AI or AN, non-Hispanic Blacks, and non-Hispanic Whites consistently had the worst survival. This is consistent with several other reports of NSCLC survival differences across racial and ethnic groups.24, 25, 26 In terms of contributing factors to these observed differences, Asian patients likely had a higher proportion of EGFR mutations, which can have a better prognosis and respond to third-generation TKIs.^10,11,27 Meanwhile, the impact of immunotherapy seems to be less in the setting of the EGFR mutation group, which could explain the lower rate of immunotherapy in this cohort.²⁸ Asian patients may also benefit from advantages attributable to diet, weight, and less frequent smoking before cancer diagnosis.²⁹ Unfortunately, some of these granular details, including smoking status, are not available in the NCDB. For Hispanic patients, better survival has been attributed to better overall health in some settings, although it remains unlikely that this fully explains the differences noted.²⁴ Differences in mutation profiles such as high frequency of EGFR mutations and low frequency of KRAS mutations in Hispanic populations have also been attributed to survival differences.³⁰ In terms of the lower survival in non-Hispanic Black patients, non-Hispanic White patients, and American Indian/Alaskan Native patients, this is consistent with prior reports that found lower survival in these groups.^25,31

Several additional attributes such as insurance status are associated with survival. In the third era, the survival rate of patients with private insurance was twice that of those without insurance. This is consistent with previous studies that have found that insurance status continues to be an important factor in obtaining access to care in advanced-stage NSCLC.³² One study conducted on patients with Stage III NSCLC found that compared with uninsured patients and patients with Medicare or Medicaid, patients with private insurance were more likely to be adequately diagnosed and treated, resulting in better overall survival.^33,34 One of the proposed contributing factors for this difference in survival is that patients with no insurance or Medicaid are less likely to undergo curative procedures.³²

In addition, we found that patients treated at academic centers consistently had better survival rates than those treated at non-academic centers. This aligns with prior studies, including an earlier study examining the NCDB (before this period of paradigm change), which found that the type of initial treatment center significantly influences survival, favoring academic centers.²¹ Academic centers are at the forefront of the development of novel therapies with access to clinical trials and are likely to be associated with the improved survival noted in patients treated at these centers.²¹

Limitations

This study has several limitations in addition to those usually associated with observational studies. Although the NCDB captures many treatment- and patient-related variables, it does not include potentially important details. For example, tumor profiling and smoking status, which could explain both treatment decisions and potential differences in outcomes, were not included. Specific agents used for systemic therapy were not included, which could be important, as there have been increases in efficacy. Nevertheless, not all patients with stage IV disease are the same in terms of their disease status. We attempted to evaluate this using a query of the sites involved at the time of diagnosis. Most patients in each group presented with multiple-site involvement (approximately 70%). Nevertheless, we did note differences in the specific sites involved, and it is possible that the efficacy of treatments could vary by the site of treatment involvement (e.g., bioavailability may differ by the organ involved). In addition, the staging modalities employed to determine the stage IV status, such as positron emission tomography scanning, brain magnetic resonance imaging, and the extent to which suspected systemic metastases were biopsy-confirmed, were not captured in the NCDB.

Another limitation of this study is that some racial and ethnic groups are represented in very small numbers in the NCDB. For example, only 1216 of our sample (0.3%) were American Indian or Alaskan Natives, 13,329 (3.4%) were Asian, 13,540 (3.4%) were Hispanic, 909 (0.2%) were native Hawaiian or Pacific Islanders, and 3842 (1.0%) were identified as other races. This may have affected our ability to detect statistically significant differences in some areas where they would otherwise have been noted with a larger sample size. That being said, previous work into the representativeness of different groups in the NCDB suggested that coverage was similar for Whites (73%) and Blacks (73%), intermediate for Asians and Pacific Islanders (68%), and lowest for Hispanics (55%) and American Indian or Alaskan Natives (AI or AN) (41%).³⁵ The NCDB only captures overall survival. Given that this was a study of stage IV NSCLC, it is likely that most deaths were attributable to cancer²²; nevertheless, it is possible that our observed differences were influenced by other sources of mortality.

Finally, relative differences in overall survival are an imperfect reflection of equity in cancer care, without understanding the potential survival of each group if ideally managed and supported. Ideally, we would want to consider whether every group had the same opportunity to reach their optimal health outcome, which likely differs from group to group and patient to patient.³⁶ That being said, overall survival is an important perspective that can fuel deeper examinations into equity in this setting.³⁷

Conclusion

In this study, the overall survival of stage IV NSCLC increased broadly across the population during a time when immunotherapy use was increasing in the United States and targeted therapies were becoming more potent. Nevertheless, the median survival and rate of improvement were quite variable among different sociodemographic groups. Further studies are warranted to understand the variability in survival to potentially expose opportunities to improve the effectiveness and equity of NSCLC care.

CRedit Authorship Contribution Statement

Oluwaseun F. Ayoade: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Validation, Visualization, Writing - original draft, Writing - review & editing.

Maureen E. Canavan: Data curation, Formal analysis, Methodology, Visualization, Validation, Writing - review & editing.

Emily J. Zolfaghari: Visualization, Writing - review & editing.

Giorgio Caturegli: Visualization, Writing - review & editing.

So Yeon Kim: Conceptualization, Investigation, Validation, Visualization, Writing - review & editing.

Daniel J. Boffa: Conceptualization, Investigation, Methodology, Project administration, Validation, Visualization, Writing - review & editing.

Disclosure

The authors declare no conflict of interest.