Abstract
In patients with metastatic non-small cell lung cancer (mNSCLC), the extent to which immunotherapy utilization rate varies by comorbidities is unclear. Using the National Cancer Database (NCDB) from 2015 to 2016, we assessed the association between levels of comorbidity and immunotherapy utilization among mNSCLC patients. Burden of comorbidities was ascertained based on the modified Charlson-Deyo score and categorized as an ordinal variable (0, 1 and ≥2). Immunotherapy utilization was determined based on registry data. Multivariable logistic regressions were employed to estimate adjusted odds ratios (aOR) and 95% confidence intervals (CI) for the comorbidity score while adjusting for sociodemographic factors, histopathological subtype, surgery, chemotherapy, radiotherapy, insurance, facility type, and other cancer history. Subgroup analyses were conducted by age and race/ethnicity. Overall, of the 89,030 patients with mNSCLC, 38.6% (N=34,382) had the comorbidity score of ≥1. Most patients were non-Hispanic White (82.3%, N=73,309) and aged ≥65 years (63.2%, N=56,300), with the mean age of 68.4 years (SD=10.6). Only 7.0% (N=6,220) of patients received immunotherapy during 2015–2106. Patients with a comorbidity score of ≥2 had a significantly lower rate of immunotherapy utilization versus those without comorbidities (aOR=0.85, 95% CI=0.78–0.93, p-trend<0.01). In subgroup analysis by age, association patterns were similar among patients younger than 65 and those aged 65–74 years. There were no significant differences in subgroup analysis by race/ethnicity, although statistical significance was only observed for White patients (comorbidity score ≥2 vs. 0: aOR=0.85, 95% CI=0.77–0.93, p-trend<0.01). In conclusion, mNSCLC patients with a high burden of comorbidities are less likely to receive immunotherapy.
Keywords: lung cancer, immunotherapy, comorbidity, epidemiology
Introduction
Lung cancer is the second most common malignancy in the United States and the leading cause of cancer-related death in the United States and around the globe. The Surveillance, Epidemiology, and End Results (SEER) Program estimates that 235,760 people in the United States will be diagnosed with lung cancer in 2021 and 131,880 lung cancer patients will die in the same year.1 Overall, approximately 85% of incident lung cancer cases are classified as non-small cell lung cancer (NSCLC),2 and about 40% of NSCLC patients have metastasis at the time of diagnosis.3 While conventional chemotherapy or targeted therapy may improve outcomes of patients with metastatic NSCLC (mNSCLC) to some extent, not all patients respond well to these therapies.4–6 However, with the development of immunotherapy, patients with mNSCLC have more options for treatment and a potential for improved survival.7 Since approval of the first immune checkpoint inhibitors (ICI) for NSCLC by the Food and Drug Administration (FDA) in 2015,8 immunotherapy has improved prognosis for many patients with mNSCLC.9
In addition to clinical treatment, several health-related factors can also influence survival of patients with mNSCLC. For example, comorbidities such as cardiovascular disease, diabetes, and liver disease can negatively impact prognosis of these patients;10 however, the extent to which such illnesses can impact immunotherapy utilization among patients with mNSCLC is not clear, suggesting there is a need to disentangle how immunotherapy utilization patterns vary by comorbidities. Evaluating associations between comorbidities and immunotherapy use may help identify gaps and inform treatment decisions that may improve outcomes for patients with mNSCLC.
In this study, we analyzed data from the National Cancer Database (NCDB) between 2015 and 2016 to examine how utilization rate of immunotherapy varies by comorbidities among patients with mNSCLC.
Methods
NCDB and study population
The National Cancer Database (NCDB) began as a joint project between the Commission on Cancer (CoC) of the American College of Surgeons and the American Cancer Society in 1988. In 2016, the database collected records from over 34 million cancer patients in the United States, making it the largest clinical cancer registry in the world. Relevant data in NCDB are extracted from patients’ medical records by certified tumor registrars. The reporting facilities are required to include information regarding cancer care taking place outside of their facilities even if they are not accredited by the CoC and all data are validated before being released.11 Specifically, the NCDB contains de-identified variables regarding patients’ demographic information, tumor characteristics, first-course treatment for cancer, and survival for approximately 70% of the US cancer population.12 A total of 101,555 patients with mNSCLC were identified in 2015–2016 NCDB; metastasis was defined by the seventh edition of the American Joint Committee of Cancer TNM staging system, and patients with “M=1” for NSCLC were treated as having metastasis. These mNSCLC cases were included for analysis if they had no missing values of comorbidities, immunotherapy use, and other relevant covariates. This yielded 89,030 (87.7%) cases for analysis. Compared with patients included for analysis, excluded patients had a slightly lower Charlson-Deyo score and younger age but they did not have a large difference in immunotherapy receipt rate (Supplementary Table 1).
Exposure and outcome of interest
Burden of comorbidities was measured using the modified Charlson-Deyo comorbidity score. The score is a weighted index based on 17 diseases and each of these comorbid conditions is assigned a weight from 1 to 613,14. Because only a small proportion of cases had a Charlson-Deyo comorbidity score higher than 2, this index was truncated to 0, 1, and 2+ in the dataset, with a score of 0 indicating no comorbid conditions present. Immunotherapy utilization for cancer treatment was the outcome of interest in our study and it was treated as a binary variable (used vs. not used). The NCDB had no detailed information regarding specific treatment agents.
Other covariates
Demographic variables included age at diagnosis (<65, 65–74, and ≥75 years), race/ethnicity (non-Hispanic White, non-Hispanic Black, American Indian, Asian/Pacific Islander, Hispanic, and other), and sex (female vs. male). Area-level data regarding education and income were inferred by linking patients’ zip codes to the US Census and US Department of Agriculture Economic Research Service datasets,11 and they were categorized as ordinal variables to approximate quartiles. Specifically, area-level educational background was measured by percent without high school degree, and income was reflected by median annual household income in that area. Individual-level health service-related factors included insurance (uninsured, private, Medicaid, Medicare, and other) and facility where patients received cancer diagnosis (non-academic vs. academic); these variables were included as confounders because they could affect likelihood of comorbidity diagnosis and immunotherapy receipt.15 Histopathological subtype (adenocarcinoma, squamous cell carcinoma, large cell/neuroendocrine carcinoma, and other) were included in our analysis and it was determined using International Classification of Diseases for Oncology (ICD-O)-3 morphology codes.16 Histories of surgery, chemotherapy, and radiation therapy were included. Number of cancers over the lifetime of patient was collected in NCDB. Selection of covariates was based on a priori knowledge and plausibility of association with our main exposure and outcome of interest.
Statistical analysis
First, we summarized distributions of study covariates in the overall sample and by Charlson-Deyo comorbidity score. Then, we reported the proportions of immunotherapy use by Charlson-Deyo comorbidity score and estimated the corresponding utilization rate and 95% confidence interval (CI). Three multivariable logistic regression models were used to estimate adjusted odds ratio (aOR) and 95% CI for Charlson-Deyo comorbidity score. The first model only adjusted for age and sex; the second further adjusted for race/ethnicity, insurance, facility type, history of surgery, chemotherapy, and radiotherapy, histopathological subtype, and history of more than one cancer. The third model additionally adjusted for area-level income and education because socioeconomic status is associated with burdens of comorbidities; in addition, previous studies have found that melanoma patients with lower education were less likely to receive immunotherapy, suggesting that these factors might be potential confounders for mNSCLC patients as well.15,17,18 To explore if complete-case analysis biased the estimates, we additionally adjusted for an indicator of exclusion in the first two models for patients without missing data of covariates in these models. In analysis, Charlson-Deyo comorbidity score was treated as an ordinal variable (0, 1, and ≥2) and patients with scores at 0 were the reference group. Tests for trend were conducted by treating the score as a continuous variable in the model.
Cancer patients of older age often have greater complexity in disease management which may defer treatment and racial/ethnic minorities have been shown to have a lower rate of treatment utilization,19,20 thus we hypothesized that there could be a synergistic effect between comorbidity and age and race/ethnicity. To explore this hypothesis, we conducted subgroup analysis for people with different ages (<65, 65–74, and ≥75 years) or race/ethnicity categories (non-Hispanic White, non-Hispanic Black, Hispanic, and other race/ethnicity group [American Indian, Asian/Pacific Islander, and other categories]). An interaction term between the Charlson-Deyo score and age or race/ethnicity was generated and included in the multivariable model, and the interaction terms were examined by the Wald test.
For current analysis, two-sided p values<0.05 were considered to be statistically significant. All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC).
Results
The distributions of study covariates are present in Table 1. Of the 89,030 patients included in our analysis, 38.6% had Charlson-Deyo comorbidity score≥1, the mean age was 68.4 years (SD=10.6), and 53.4% were male. Overall, 82.3% of the patients were non-Hispanic White, 12.7% were Black, 3.6% were Hispanic, and only 1.4% self-reported as American Indian, Asian/Pacific Islander, or other categories. Over half (50.5%) of our study population were living in areas where at least 10.9% of residents did not have high school degree and 45.4% of patients’ residential regions had a median annual household income lower than $50,354. Most patients were insured and Medicare was the most common (60.6%) type of insurance; in terms of facilities that provided medical services, 31.9% were associated with academic institutions or universities. About two-thirds (63.9%) of the cases were adenocarcinoma and 19.8% were squamous cell carcinoma. Only 2.2% of patients had surgery, 33.6% had chemotherapy, and 43.7% had radiotherapy. Most patients (79.6%) in our analysis did not have cancers other than mNSCLC. Patients with a higher Charlson-Deyo comorbidity score were more likely to be older, be male, live in areas with lower socio-economic status, be diagnosed from non-academic healthcare facilities, have histopathological subtypes other than adenocarcinoma, and have no chemotherapy or radiotherapy.
Table 1.
Characteristics of patients with metastatic NSCLC from 2015 and 2016 NCDB (N= 89,030)
| Charlson-Deyo Score, N(%) |
||||
|---|---|---|---|---|
| Overall | 0 | 1 | ≥2 | |
| Characteristics | N (%) | (N=54,648) | (N=20,871) | (N=13,511) |
|
| ||||
| Age (years) | ||||
| <65 | 32,730 (36.8) | 21,875 (40.0) | 7,133 (34.2) | 3,722 (27.6) |
| 65–74 | 29,598 (33.2) | 17,440 (31.9) | 7,372 (35.3) | 4,786 (35.4) |
| ≥75 | 26,702 (30.0) | 15,333 (28.1) | 6,366 (30.5) | 5,003 (37.0) |
| Race/ethnicity | ||||
| Non-Hispanic White | 73,309 (82.3) | 44,828 (82.0) | 17,372 (83.2) | 11,109 (82.2) |
| Non-Hispanic Black | 11,265 (12.7) | 6,845 (12.5) | 2,554 (12.2) | 1,866 (13.8) |
| American Indian | 267 (0.3) | 166 (0.3) | 70 (0.3) | 31 (0.2) |
| Asian/Pacific Islander | 816 (0.9) | 616 (1.1) | 138 (0.7) | 62 (0.5) |
| Hispanic | 3,183 (3.6) | 2,066 (3.8) | 695 (3.3) | 422 (3.1) |
| Other | 190 (0.2) | 127 (0.2) | 42 (0.2) | 21 (0.2) |
| Sex | ||||
| Female | 41,471 (46.6) | 26,366 (48.3) | 9,534 (45.7) | 5,571 (41.2) |
| Male | 47,559 (53.4) | 28,282 (51.7) | 11,337 (54.3) | 7,940 (58.8) |
| Percent without high school degree | ||||
| ≥17,6% | 19,669 (22.1) | 11,626 (21.3) | 4,810 (23.0) | 3,233 (23.9) |
| 10.9–17.5% | 25,287 (28.4) | 15,196 (27.8) | 6,066 (29.1) | 4,025 (29.8) |
| 6.3–10.8% | 25,283 (28.4) | 15,598 (28.5) | 5,965 (28.6) | 3,720 (27.5) |
| <6.3% | 18,791 (21.1) | 12,228 (22.4) | 4,030 (19.3) | 2,533 (18.8) |
| Median annual household income | ||||
| <$40,227 | 19,233 (21.6) | 11,199 (20.5) | 4,768 (22.9) | 3,266 (24.2) |
| $40,227–50,353 | 21,209 (23.8) | 12,785 (23.4) | 5,136 (24.6) | 3,288 (24.3) |
| $50,354–63,332 | 21,095 (23.7) | 12,891 (23.6) | 4,970 (23.8) | 3,234 (23.9) |
| ≥$63,333 | 27,493 (30.9) | 17,773 (32.5) | 5,997 (28.7) | 3,723 (27.6) |
| Insurance type | ||||
| Uninsured | 2,642 (3.0) | 1,819 (3.3) | 552 (2.6) | 271 (2.0) |
| Private | 23,384 (26.3) | 16,362 (29.9) | 4,735 (22.7) | 2,287 (16.9) |
| Medicaid | 7,606 (8.5) | 4,799 (8.8) | 1,760 (8.4) | 1,047 (7.8) |
| Medicare | 53,912 (60.6) | 30,766 (56.3) | 13,449 (64.4) | 9,697 (71.8) |
| Other | 1,486 (1.7) | 902 (1.7) | 375 (1.8) | 209 (1.5) |
| Facility type | ||||
| Non-academic | 60,663 (68.1) | 36,036 (65.9) | 14,854 (71.2) | 9,773 (72.3) |
| Academic | 28,367 (31.9) | 18,612 (34.1) | 6,017 (28.8) | 3,738 (27.7) |
| Histopathological subtype | ||||
| Adenocarcinoma | 56,887 (63.9) | 36,084 (66.0) | 12,849 (61.6) | 7,954 (58.9) |
| Squamous cell carcinoma | 17,600 (19.8) | 9,690 (17.7) | 4,598 (22.0) | 3,312 (24.5) |
| Large cell/neuroendocrine carcinoma | 1,734 (2.0) | 1,023 (1.9) | 441 (2.1) | 270 (2.0) |
| Other | 12,809 (14.3) | 7,851 (14.4) | 2,983 (14.3) | 1,975 (14.6) |
| Surgery | ||||
| No | 87,085 (97.8) | 53,536 (98.0) | 20,327 (97.4) | 13,222 (97.9) |
| Yes | 1,945 (2.2) | 1,112 (2.0) | 544 (2.6) | 289 (2.1) |
| Chemotherapy | ||||
| No | 59,077 (66.4) | 34,727 (63.6) | 14,269 (68.4) | 10,081 (74.6) |
| Yes | 29,953 (33.6) | 19,921 (36.4) | 6,602 (31.6) | 3,430 (25.4) |
| Radiotherapy | ||||
| No | 50,123 (56.3) | 29,639 (54.2) | 12,104 (58.0) | 8,380 (62.0) |
| Yes | 38,907 (43.7) | 25,009 (45.8) | 8,767 (42.0) | 5,131 (38.0) |
| More than one cancer | ||||
| No | 70,888 (79.6) | 43,699 (80.0) | 16,612 (79.6) | 10,577 (78.3) |
| Yes | 18,142 (20.4) | 10,949 (20.0) | 4,259 (20.4) | 2,934 (21.7) |
Abbreviations: NCDB: National Cancer Database, NSCLC: non-small cell lung cancer
Overall, 7.0% (N=6,220) of patients received immunotherapy (Table 2); utilization rate decreased as the Charlson-Deyo comorbidity score increased (0: 7.7%, 95% CI=7.5%–7.9%; 1: 6.4%, 95% CI=6.0%–6.7%; ≥2: 5.0%, 95% CI=4.7%–5.4%). In the model adjusting for all covariates (Table 2), patients with a score≥2 had a significantly lower utilization rate compared to those with 0 (aOR=0.85, 95% CI=0.78–0.93, p-trend<0.01). Adjusting for indicator of exclusion did not change the effect measures in multivariable models (Supplementary Table 2).
Table 2.
Association between comorbidity and immunotherapy utilization among patients with metastatic NSCLC
| Charlson-Deyo Score | User/N | Utilization rate (%) and 95% CI | aOR and 95% CI* | aOR and 95% CI‡ | aOR and 95% CI† |
|---|---|---|---|---|---|
|
| |||||
| Overall sample (N=89,030) | |||||
| 0 | 4,212/54,648 | 7.7 (7.5, 7.9) | REF | REF | REF |
| 1 | 1,329/20,871 | 6.4 (6.0, 6.7) | 0.83 (0.78, 0.89) | 0.93 (0.87, 0.99) | 0.93 (0.87, 0.99) |
| ≥2 | 679/13,511 | 5.0 (4.7, 5.4) | 0.67 (0.62, 0.73) | 0.85 (0.78, 0.92) | 0.85 (0.78, 0.93) |
| p-trend<0.01 | p-trend<0.01 | p-trend<0.01 | |||
| By age | |||||
| <65 years (N=32,730) | |||||
| 0 | 2,035/21,875 | 9.3 (8.9, 9.7) | REF | REF | REF |
| 1 | 576/7,133 | 8.1 (7.4, 8.7) | 0.86 (0.78, 0.94) | 0.94 (0.85, 1.04) | 0.94 (0.85, 1.05) |
| ≥2 | 226/3,722 | 6.1 (5.3, 6.8) | 0.63 (0.55, 0.73) | 0.80 (0.69, 0.93) | 0.80 (0.69, 0.93) |
| p-trend<0.01 | p-trend<0.01 | p-trend<0.01 | |||
| 65–74 years (N=29,598) | |||||
| 0 | 1,392/17,440 | 8.0 (7.6, 8.4) | REF | REF | REF |
| 1 | 492/7,372 | 6.7 (6.1, 7.2) | 0.83 (0.74, 0.92) | 0.93 (0.83, 1.04) | 0.93 (0.83, 1.04) |
| ≥2 | 254/4,786 | 5.3 (4.7, 5.9) | 0.65 (0.56, 0.74) | 0.81 (0.70, 0.93) | 0.81 (0.70, 0.94) |
| p-trend<0.01 | p-trend<0.01 | p-trend<0.01 | |||
| ≥75 years (N=26,702) | |||||
| 0 | 785/15,333 | 5.1 (4.8, 5.5) | REF | REF | REF |
| 1 | 261/6,366 | 4.1 (3.6, 4.6) | 0.79 (0.68, 0.91) | 0.89 (0.77, 1.03) | 0.89 (0.77, 1.03) |
| ≥2 | 199/5,003 | 4.0 (3.4, 4.5) | 0.76 (0.65, 0.89) | 0.97 (0.82, 1.14) | 0.98 (0.83, 1.15) |
| p-trend<0.01 | p-trend=0.40 | p-trend=0.44 | |||
| p-interaction=0.24 | p-interaction=0.19 | p-interaction=0.19 | |||
| By race/ethnicity | |||||
| Non-Hispanic White (N=73,309) | |||||
| 0 | 3,497/44,828 | 7.8 (7.6, 8.1) | REF | REF | REF |
| 1 | 1,099/17,372 | 6.3 (6.0, 6.7) | 0.81 (0.76, 0.87) | 0.91 (0.85, 0.98) | 0.92 (0.85, 0.98) |
| ≥2 | 560/11,109 | 5.0 (4.6, 5.5) | 0.67 (0.61, 0.73) | 0.84 (0.77, 0.93) | 0.85 (0.77, 0.93) |
| p-trend<0.01 | p-trend<0.01 | p-trend<0.01 | |||
| Non-Hispanic Black (N=11,265) | |||||
| 0 | 497/6,845 | 7.3 (6.7, 7.9) | REF | REF | REF |
| 1 | 159/2,554 | 6.2 (5.3, 7.2) | 0.87 (0.73, 1.05) | 0.94 (0.77, 1.13) | 0.94 (0.77, 1.13) |
| ≥2 | 94/1,866 | 5.0 (4.1, 6.0) | 0.72 (0.58, 0.91) | 0.86 (0.68, 1.09) | 0.86 (0.68, 1.09) |
| p-trend<0.01 | p-trend=0.18 | p-trend=0.18 | |||
| Hispanic (N=3,183) | |||||
| 0 | 133/2,066 | 6.4 (5.4, 7.5) | REF | REF | REF |
| 1 | 53/695 | 7.6 (5.7, 9.6) | 1.35 (0.97, 1.89) | 1.39 (0.98, 1.96) | 1.39 (0.98, 1.96) |
| ≥2 | 18/422 | 4.3 (2.3, 6.2) | 0.79 (0.47, 1.31) | 0.95 (0.56, 1.61) | 0.94 (0.55, 1.59) |
| p-trend=0.99 | p-trend=0.49 | p-trend=0.52 | |||
| Other (N=1,273)§ | |||||
| 0 | 85/909 | 9.4 (7.5, 11.2) | REF | REF | REF |
| 1 | 18/250 | 7.2 (4.0, 10.4) | 0.82 (0.48, 1.40) | 0.92 (0.53, 1.60) | 0.93 (0.53, 1.61) |
| ≥2 | 7/114 | 6.1 (1.7, 10.6) | 0.70 (0.31, 1.56) | 0.74 (0.32, 1.71) | 0.75 (0.33, 1.73) |
| p-trend=0.29 | p-trend=0.48 | p-trend=0.50 | |||
| p-interaction=0.31 | p-interaction=0.52 | p-interaction=0.51 | |||
Abbreviations: aOR: adjusted odds ratio, CI: confidence interval, cOR: crude odds ratio
This included 267 American Indians, 816 Asian/Pacific Islanders, and 190 patients belonging to other race/ethnicity groups.
The model adjusted for age and sex.
The model adjusted for age, sex, race/ethnicity, insurance, facility type, surgery, histopathological subtype, chemotherapy, radiotherapy, and history of more than one cancer.
The model adjusted for age, sex, race/ethnicity, insurance, facility type, surgery, histopathological subtype, chemotherapy, radiotherapy, history of more than one cancer, income, and education.
Association patterns of Charlson-Deyo comorbidity score were largely similar in patients younger than 65 years and those aged 65–74 years, suggesting a similar inverse association between a higher comorbidity burden and immunotherapy use across these age groups; different from younger subgroups, the association among patients aged≥75 years was almost null (comorbidity score≥2 vs. 0: aOR=0.98, 95% CI=0.83, 1.15, p-trend=0.44), although results of Wald tests did not suggest a significant interaction (p-interaction=0.19). Subgroup analysis by race/ethnicity yielded similar point estimates of aOR in non-Hispanic White (comorbidity score≥2 vs. 0: aOR=0.85, 95% CI=0.77–0.93, p-trend<0.01) and Black (comorbidity score≥2 vs. 0: aOR=0.86, 95% CI=0.68–1.09, p-trend=0.18) patients, whereas the association in Hispanic patients was close to null (comorbidity score≥2 vs. 0: aOR=0.94, 95% CI=0.55–1.59, p-trend=0.52). No significant interaction was observed for subgroup analysis by race/ethnicity.
Discussion
Our study found that only a small fraction of patients with mNSCLC received immunotherapy during 2015–2016. Although the multivariable model suggested that there was no substantial difference in utilization rate between patients with Charlson-Deyo Score at 0 and 1, patients with Charlson-Deyo Score≥2 had an approximately 15% relative reduction in likelihood of immunotherapy utilization compared to those without major comorbidities. Compared to younger age groups, having Charlson-Deyo score≥2 in older subgroup (75+ years) appeared to have minimal impact, although there was no significant interaction by age; because short life expectancy is strongly associated with a lower rate of treatment utilization,21 we hypothesize that older age may obscure the impacts of comorbidity on immunotherapy use. We observed similar magnitude of associations between comorbidities and immunotherapy use among non-Hispanic White and Black patients; however, comorbidities did not appear to impact immunotherapy utilization among Hispanic patients.
Our findings are consistent with and extend results from previous research of similar topics. For example, by analyzing 4,014 NSCLC patients in the SEER Kentucky Cancer Registry (2007–2011), researchers found that NSCLC patients with higher comorbidity score were less likely to receive surgery, chemotherapy, or radiation therapy.22 In addition, Freeman et al.15 analyzed 16,906 stage III melanoma patients diagnosed between 2006–2016 from the NCDB and explored association between Charlson-Deyo Score and receipt of immunotherapy. Their analysis suggested that patients with Charlson-Deyo Score≥2 had a 30% relative decrease in odds of immunotherapy use compared to patients without comorbidities, whereas the utilization rate did not differ between patients with the Charlson-Deyo Score at 1 and 0. Although that study analyzed a different type of cancer, effect size of Charlson-Deyo Score is similar to our observed inverse association between comorbidities and immunotherapy use in patients with mNSCLC.
Several underlying mechanisms may help us explain the inverse association between burden of comorbidities and immunotherapy use in patients with mNSCLC. First, co-existing illnesses may limit cancer patients’ or physicians’ consideration of newer therapies and result in underutilization of appropriate anti-tumor treatment.23,24 Because of the complexity of care needed for multimorbidity, a variety of medical expertise will be required for management of cancer and other illnesses simultaneously, which can cause barriers to timely immunotherapy receipt.25,26 In addition, delivery of therapy to lung cancer patients with multiple health problems requires significant care coordination, resulting in delayed or unstandardized treatment.26 Second, although there is no consistent evidence suggesting that comorbidities can increase risk of immune-related adverse events (irAEs) among cancer patients receiving immunotherapy,27 physicians may still be reluctant to recommend this treatment to patients with mNSCLC who have a higher burden of comorbid conditions; the reason is that comorbidities are associated with a worse prognosis in lung cancer patients,10 which may greatly offset any benefit gained by immunotherapy. For example, a study of microsimulation models to investigate the cost-effectiveness of pembrolizumab for stage IV NSCLC patients found that treatment with pembrolizumab did not represent a cost-effective strategy compared to chemotherapy in patients with varying comorbidity burden.28 Third, ICI was initially approved for second line use to treat mNSCLC in 2015,29 suggesting that patients with mNSCLC had to start with chemotherapy between 2015–2016; this indicates that patients with unfavorable prognostic features (e.g. higher burden of comorbidities) might have done poorly with chemotherapy and therefore were not in a condition to receive immunotherapy. Furthermore, survival of mNSCLC is not favorable and it can be further reduced by comorbidities, suggesting a great proportion of these patients with comorbidities may seek palliative care rather than curative treatment like immunotherapy.
Our study has some merits in design and analysis. The NCDB is the largest clinical cancer registry in the world,11 and the large sample size and robust measurement of cancer-related characteristics ensure a good statistical power and reduce measurement error. The stratified analysis by race/ethnicity helped us disentangle if there were racial disparities regarding the impact of comorbidities on immunotherapy utilization. However, several limitations should be noted when interpreting our results. First, The NCDB does not include information on the specific types of treatment agent, which makes it impossible to explore associations with a specific treatment modality. Second, although the Charlson-Deyo comorbidity score reflects the burden of co-existing illnesses, it does not represent the severity of these diseases which can also strongly influence receipt of immunotherapy, suggesting that some residual confounding may bias our estimates to some extent; in addition, lack of information on specific types of illnesses makes us unable to investigate impact of a certain disease on immunotherapy use in NCDB. Lastly, several types of immunotherapy modalities have been approved by the FDA for NSCLC patients after 201630–32 and clinicians’ awareness of these new options have likely substantially improved recently, suggesting utilization patterns of immunotherapy may be different among patients with mNSCLC in the current era.
In conclusion, mNSCLC patients with a higher burden of comorbidities are less likely to use immunotherapy for treatment. This pattern remains among patients younger than 75 years and is consistent in non-Hispanic White and Black patients. Previous research suggests that a higher burden of co-existing illnesses is associated with a higher risk of non-cancer deaths, particularly in patients with less severe cancers;33 however, there is no real-world evidence indicating that such burden can worsen survival in mNSCLC patients receiving immunotherapy, and currently available trials usually enrolled healthier patients that could not represent burden of comorbidities in real-world patients with mNSCLC.34 Thus, in the future, real-world studies with more detailed measures of comorbidities, therapeutic agents, and data on survival outcomes will help shed light on these associations and investigate whether higher burdens of comorbidities are empirically associated with poor immunotherapy outcomes. This information might provide physicians with critical evidence needed to evaluate the suitability of patients for immunotherapy.
Supplementary Material
Funding source:
Georgetown University CCSG Pilot Developmental Award
Footnotes
Conflict of interests: None of the authors had conflict of interest
References
- 1.Siegel RL, Miller KD, Fuchs HE, Jemal A. Cancer Statistics, 2021. CA Cancer J Clin. 2021;71(1):7–33. [DOI] [PubMed] [Google Scholar]
- 2.Jemal A, Ward E, Thun MJ. Contemporary lung cancer trends among U.S. women. Cancer Epidemiol Biomarkers Prev. 2005;14(3):582–585. [DOI] [PubMed] [Google Scholar]
- 3.Morgensztern D, Ng SH, Gao F, Govindan R. Trends in stage distribution for patients with non-small cell lung cancer: a National Cancer Database survey. J Thorac Oncol. 2010;5(1):29–33. [DOI] [PubMed] [Google Scholar]
- 4.Desai S, Kim C, Veytsman I. Role of Anti-EGFR Targeted Therapies in Stage III Locally Advanced Non-small Cell Lung Cancer: Give or Not to Give? Curr Oncol Rep. 2019;21(9):84. [DOI] [PubMed] [Google Scholar]
- 5.Artal Cortes A, Calera Urquizu L, Hernando Cubero J. Adjuvant chemotherapy in non-small cell lung cancer: state-of-the-art. Transl Lung Cancer Res. 2015;4(2):191–197. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Scheff RJ, Schneider BJ. Non-small-cell lung cancer: treatment of late stage disease: chemotherapeutics and new frontiers. Semin Intervent Radiol. 2013;30(2):191–198. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Dobosz P, Dzieciatkowski T. The Intriguing History of Cancer Immunotherapy. Front Immunol. 2019;10:2965. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Atkins MB, Clark JI, Quinn DI. Immune checkpoint inhibitors in advanced renal cell carcinoma: experience to date and future directions. Ann Oncol. 2017;28(7):1484–1494. [DOI] [PubMed] [Google Scholar]
- 9.Horn L, Spigel DR, Vokes EE, et al. Nivolumab Versus Docetaxel in Previously Treated Patients With Advanced Non-Small-Cell Lung Cancer: Two-Year Outcomes From Two Randomized, Open-Label, Phase III Trials (CheckMate 017 and CheckMate 057). J Clin Oncol. 2017;35(35):3924–3933. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Islam KM, Jiang X, Anggondowati T, Lin G, Ganti AK. Comorbidity and Survival in Lung Cancer Patients. Cancer Epidemiol Biomarkers Prev. 2015;24(7):1079–1085. [DOI] [PubMed] [Google Scholar]
- 11.Boffa DJ, Rosen JE, Mallin K, et al. Using the National Cancer Database for Outcomes Research: A Review. JAMA Oncol. 2017;3(12):1722–1728. [DOI] [PubMed] [Google Scholar]
- 12.Bilimoria KY, Stewart AK, Winchester DP, Ko CY. The National Cancer Data Base: a powerful initiative to improve cancer care in the United States. Ann Surg Oncol. 2008;15(3):683–690. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.de Groot V, Beckerman H, Lankhorst GJ, Bouter LM. How to measure comorbidity. a critical review of available methods. J Clin Epidemiol. 2003;56(3):221–229. [DOI] [PubMed] [Google Scholar]
- 14.Charlson/Deyo Score. http://ncdbpuf2013.facs.org/content/charlsondeyo-comorbidity-index [Last accessed Aug 20th 2020].
- 15.Freeman SC, Satish M, Walters RW. Comorbidity burden on receipt of adjuvant immunotherapy and survival in patients with stage III melanoma: an analysis of the National Cancer Database. Int J Dermatol. 2020. [DOI] [PubMed] [Google Scholar]
- 16.Lewis DR, Check DP, Caporaso NE, Travis WD, Devesa SS. US lung cancer trends by histologic type. Cancer. 2014;120(18):2883–2892. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Zhang D, Advani S, Zhu Z, Dang L, Walter LC, Braithwaite D. Mammography use in relation to comorbidities and functional limitations among older breast cancer survivors. J Cancer Surviv. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Demb J, Abraham L, Miglioretti DL, et al. Screening Mammography Outcomes: Risk of Breast Cancer and Mortality by Comorbidity Score and Age. J Natl Cancer Inst. 2020;112(6):599–606. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Shugarman LR, Mack K, Sorbero ME, et al. Race and sex differences in the receipt of timely and appropriate lung cancer treatment. Med Care. 2009;47(7):774–781. [DOI] [PubMed] [Google Scholar]
- 20.Neulander EZ, Duncan RC, Tiguert R, Posey JT, Soloway MS. Deferred treatment of localized prostate cancer in the elderly: the impact of the age and stage at the time of diagnosis on the treatment decision. BJU Int. 2000;85(6):699–704. [DOI] [PubMed] [Google Scholar]
- 21.Given B, Given CW. Older adults and cancer treatment. Cancer. 2008;113(12 Suppl):3505–3511. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Rios J, Gosain R, Goulart BH, et al. Treatment and outcomes of non-small-cell lung cancer patients with high comorbidity. Cancer Manag Res. 2018;10:167–175. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Ding R, Zhu D, He P, Ma Y, Chen Z, Shi X. Comorbidity in lung cancer patients and its association with medical service cost and treatment choice in China. BMC Cancer. 2020;20(1):250. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 24.Geraci JM, Escalante CP, Freeman JL, Goodwin JS. Comorbid disease and cancer: the need for more relevant conceptual models in health services research. J Clin Oncol. 2005;23(30):7399–7404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Corkum M, Urquhart R, Kendell C, Burge F, Porter G, Johnston G. Impact of comorbidity and healthcare utilization on colorectal cancer stage at diagnosis: literature review. Cancer Causes Control. 2012;23(2):213–220. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Sarfati D, Koczwara B, Jackson C. The impact of comorbidity on cancer and its treatment. CA Cancer J Clin. 2016;66(4):337–350. [DOI] [PubMed] [Google Scholar]
- 27.van Holstein Y, Kapiteijn E, Bastiaannet E, van den Bos F, Portielje J, de Glas NA. Efficacy and Adverse Events of Immunotherapy with Checkpoint Inhibitors in Older Patients with Cancer. Drugs Aging. 2019;36(10):927–938. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Criss SD, Palazzo L, Watson TR, et al. Cost-effectiveness of pembrolizumab for advanced non-small cell lung cancer patients with varying comorbidity burden. PLoS One. 2020;15(1):e0228288. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Pacheco JM, Camidge DR, Doebele RC, Schenk E. A Changing of the Guard: Immune Checkpoint Inhibitors With and Without Chemotherapy as First Line Treatment for Metastatic Non-small Cell Lung Cancer. Front Oncol. 2019;9:195. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Wang L, Ma Q, Yao R, Liu J. Current status and development of anti-PD-1/PD-L1 immunotherapy for lung cancer. Int Immunopharmacol. 2020;79:106088. [DOI] [PubMed] [Google Scholar]
- 31.Ribeiro TB, Ribeiro A, Rodrigues LO, Harada G, Nobre MRC. U.S. Food and Drug Administration anticancer drug approval trends from 2016 to 2018 for lung, colorectal, breast, and prostate cancer. Int J Technol Assess Health Care. 2020;36(1):20–28. [DOI] [PubMed] [Google Scholar]
- 32.Vaddepally RK, Kharel P, Pandey R, Garje R, Chandra AB. Review of Indications of FDA-Approved Immune Checkpoint Inhibitors per NCCN Guidelines with the Level of Evidence. Cancers (Basel). 2020;12(3). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Edwards BK, Noone AM, Mariotto AB, et al. Annual Report to the Nation on the status of cancer, 1975–2010, featuring prevalence of comorbidity and impact on survival among persons with lung, colorectal, breast, or prostate cancer. Cancer. 2014;120(9):1290–1314. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sedrak MS, Freedman RA, Cohen HJ, et al. Older adult participation in cancer clinical trials: A systematic review of barriers and interventions. CA Cancer J Clin. 2020. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.