Obesity as a Predictive Factor Among Patients With Lung Cancer and Cutaneous Melanoma Treated With Immune Checkpoint Inhibitors and EGFR Inhibitors

This was a 2011-2019 SEER-Medicare-based retrospective cohort study of patients with 1) non-small-cell lung cancer (NSCLC) receiving ICI; 2) cutaneous melanoma receiving ICI; and 3) NSCLC receiving EGFRi. The SEER-Medicare database represents a linkage of SEER and Medicare claims data, which are two large population-based sources, respectively, providing cancer-related and health services information on Medicare beneficiaries with cancer. SEER data sources include the SEER cancer registries, which cover 48% of the U.S. population, track the incidence of persons diagnosed with cancer and collect follow-up data until death. Medicare is a Federal health insurance program and covers older adults 65 years or older and people with disabilities. The 2011-2019 SEER-Medicare data includes Medicare beneficiaries in the SEER data who were diagnosed with cancer between 2011 and 2017 as well as their Medicare claims through 2019. Cancer diagnosis and characteristics were identified from the SEER data components while non-cancer diagnoses were identified from the Outpatient (OUTPAT), Carrier Claims (NCH), and Medical Provider Analysis and Review (MedPAR) datasets of the Medicare claims files while procedures were identified from OUTPAT and NCH datasets of the Medicare claims files. The study was approved by West Virginia Institutional Review Board.

Patients with histologically confirmed primary non-small-cell lung cancer (primary site codes: C340, C341, C342, C343, C348, C349: histology codes: 8010, 8012, 8013, 8020, 8046, 8050, 8051, 8052, 8070, 8071, 8072, 8073, 8074, 8075, 8076, 8077, 8078, 8140, 8141, 8143, 8147, 8250, 8251, 8252, 8253, 8254, 8255, 8260, 8310, 8430, 8480, 8481, 8490, 8560, 8570, 8571, 8572, 8573, 8574, 8575) and cutaneous melanoma (primary site codes: C440, C441, C442, C443, C444, C445, C446, C447, C448, C449; histology codes: 8720-8790) were included if they were diagnosed between 2012-2017, were 66 years or older at cancer diagnosis, and initiated ICI or EGFRi within one year of diagnosis. NSCLC and melanoma patients initiating ICI had continuous Medicare Parts A and B enrolment within 12 months pre- and post-diagnosis with no HMO enrolment during the same period. NSCLC patients initiating EGFRi were also required to have the same enrolment criteria in addition to continuous Medicare Part D enrolment within 12 months post-diagnosis.

ICI were identified using Healthcare Common Procedure Coding System codes (HCPCS procedure codes, Table S1) while EGFRi were identified from Part D Event (PDE) file using National Drug Codes (NDC). For ICI administrations, duration of each cycle was determination based on dose administered and recommended dose for each ICI. For example, administration of 1200mg of atezolizumab was taken as a 21-day cycle for melanoma while administration of 840mg was taken as a 14-day cycle – recommended dosage is 1200mg every 3 weeks. For EGFRi prescription fills, dates of prescription fill were adjusted to account for early refills. Patients commencing more than one ICI or EGFRi on the same date were excluded. Date of treatment initiation and termination were ascertained. Treatment initiation date was defined as the date of prescription fill for the first ICI or EGFRi. Termination date was defined as end date for the last prescription fill, given that the gap between consecutive prescription fills was less than 90 days. Treatment duration was defined as the number of months between treatment initiation and treatment termination.

Patients were followed from treatment initiation until the earliest of recurrence or death, end of follow-up, or end of study period. The main outcomes of interest were recurrence-free survival (RFS) and overall survival (OS). Recurrence was defined treatment with chemotherapy, radiotherapy, or surgery occurring more than 90 days after initial treatment termination. Time to recurrence was defined as the number of days between treatment initiation and recurrence, death or end of follow-up, whichever occurred first. Overall survival was defined as the number of days between treatment initiation and death from any cause or end of follow-up, whichever occurred first. Due to differences in data collection and quality between SEER and Medicare data, patients whose follow-up ended on or before their date of diagnosis were excluded to fulfil the criteria for temporality in longitudinal cohort studies.

Patients with obesity were identified using validated ICD-9-CM and ICD-10-CM diagnostic codes (specificity >95%)¹⁴ from OUTPAT, NCH, and MedPAR datasets. Obesity is defined as a body mass index of 30 or more. The presence of obesity was ascertained as a diagnosis for the condition within 365 days before cancer diagnosis.

Covariates included age, sex, cancer stage at diagnosis, race/ethnicity, year of diagnosis, comorbidity score, type of ICI or EGFRi initiated, and time to treatment. Age, sex, cancer, race/ethnicity, and year of diagnosis were ascertained on the date of cancer diagnosis. Comorbidity was based on the Charlson comorbidity index (CCI) and was ascertained within the one-year pre-diagnosis period.¹⁵ ICI/EGFRi exposure and time to treatment were ascertained in the year following cancer diagnosis. Time to treatment was defined as the number of months between dates of diagnosis and treatment initiation.

Baseline characteristics were summarized using descriptive statistics – frequency (percentages) for categorical variables and mean (standard deviation) for continuous variables. Comparisons were made between patients with and without obesity using chi-square tests for categorical variables, t-tests for continuous variables, and Wilcoxon's rank sum test for discrete variables. To determine the association between obesity status and RFS and OS, univariate and multivariate Cox regression were modeled. Cancer stage was excluded from the regression models due to high levels of missingness (>50%). Sensitivity analysis was conducted for the three cohorts, excluding patients who developed cachexia post-cancer diagnosis. Analysis was conducted using SAS 9.4.

Table 1 shows the characteristics of patients with NSCLC receiving ICIs, stratified by obesity status. A total of 2,592 patients were included. Average age was 74.3 years (±5.9 years) and the majority of patients were male (52.2%), White (84.1%), diagnosed in 2017 (46.1%), and received Nivolumab (49.6%). Also, median CCI score was 1 (Interquartile range=0-2, range=0-17) while median time to treatment was 6 months (interquartile range (IQR)=3-9 months). Compared to non-obese patients, a higher proportion of NSCLC patients with obesity were diagnosed in 2017 (54.8% vs 45.1%, p<0.01) with obese patients being significantly younger (73.4 years±5.2 years vs 74.4 years±5.9) and having higher CCI scores (2 (IQR=1-4) vs 1 (IQR=0-2). But there was no significant difference by sex, race/ethnicity, type of ICI initiated and time to treatment. Overall median times to recurrence was 476 days (95% confidence interval [95% CI]=446-516 days) and median time to death from treatment initiation was 479 days (95% CI=458-518 days). However, time to recurrence was significantly shorter among those who were obese (392 days (95% CI=330-468 days) vs 493 days (95% CI=464-525 days), p<0.01) with no statistically significant difference in time to death (443 days (95% CI=404-488 days) vs 491 days (95% CI=462-524 days), p=0.35)

Table 2 shows the characteristics of patients with cutaneous melanoma who received ICIs. A total of 880 patients were included, of whom the majority were male (65.2%), White (95.3%), diagnosed in 2017 (33.5%) and received ipilimumab (49.0%). The mean age was 76.1 years±7.2 years with a median CCI score of 0 (IQR=0-2, range=0-11 months) and median time to treatment from diagnosis of 4 months (IQR=2-8 months). Compared to non-obese patient with cutaneous melanoma, a higher proportion of obese patients were diagnosed in 2017 (47.3% vs 31.9%, p=0.02) with no significant difference by sex, race/ethnicity, type of ICI initiated and time to treatment. Those who were obese were also significantly younger (74.7 years±6.1 years) vs 76.3 years±7.3 years, p=0.02) but had higher CCI scores (2 (IQR=0-4) vs 0 (IQR=0-2, p<0.0001)

Overall median time to treated recurrence from ICI commencement was 343 days (95% CI=320-400 days) with no statistically significant difference by obesity status (378 days (95% CI=281-507 days) vs 341 (95% CI=319-401 days), p=0.35). Overall median time to death was 1,409 days (95% CI=1,107-2,136 days) and the difference was not statistically different by obesity status (984 days (95% CI=637 days-upper limit not reached) vs 1,428 days (95% CI=1,168 days-upper limit not reached), p=0.28)

Table 3 shows the characteristics of patients with NSCLC initiating EGFR inhibitors, stratified by obesity status (N=2,524). Average age was 75.5 years (±6.5 years) and the majority of patients were male (65.3%), White (62.1%), and received erlotinib with decreasing number of patients initiating an EGFR inhibitor from 2012 to 2017. Median CCI score was 0 (Interquartile range=0-1, range=0-11) while median time to treatment was 2 months (IQR=1-6 months). Compared to non-obese patients, a higher proportion of patients with obesity were White (76.2% vs 61.3%, p<0.001) and diagnosed in 2017 (22.2% vs 14.0%, p=0.04) with obese patients being significantly younger (73.7 years±6.1 years vs 75.6 years±6.5) and having higher CCI scores (2 (IQR=1-3 vs 0 (IQR=0-1) and later time to treatment (3 months (IQR=1-7 months vs 2 months (IQR=1-6 months, p=0.02). But there was no significant difference by sex and type of EGFRi initiated. Median times to treated recurrence was 1,187 days (95% CI=1,100-1,357 days) and median time to death from treatment initiation was 671 days (95% CI=636-704 days). However, time to treated recurrence was significantly faster among those who were obese (664 days (95% CI=496-778 days vs 1,231 days (95% CI=1,127-1,379 days, p<0.0001) with no statistically significant difference in time to death (559 days (95% CI=440-763 days) vs 673 days (95% CI=639-707 days), p=0.35)

Table 4 shows the results of the Cox regression for the association between RFS and obesity status as well as OS and obesity status for both patients with NSCLC and cutaneous melanoma who received ICI. Among patients with NSCLC, obesity was significantly associated with worse RFS, even after controlling for other covariates. The hazard of treated recurrence was 36% higher among patients with obesity (Hazard ratio (HR)=1.36 95% confidence interval (95% CI)=1.12-1.65, p<0.01). However, there was no significant association between obesity and OS. Other covariates significantly associated with RFS among NSCLC patients included age, year of diagnosis, CCI, time to treatment, and ICI initiated. Respectively, the hazard for treated recurrence declined by 1% with increasing age (HR=0.99, 95% CI=0.99-1.00, p<0.01) and by 7% with increasing CCI scores (HR=0.93, 95% CI=0.90-0.98, p<0.01) but increased 32% with each increasing calendar year of diagnosis (HR=1.32, 95% CI=1.06-1.19-1.46, p<0.0001) and by 4% with each month between diagnosis and treatment (HR=1.04, 95% CI=1.02-1.06, p<0.01). Compared to Nivolumab, atezolizumab (HR=2.02, 95% CI=1.49-2.73, p<0.0001) and durvalumab (HR=1.64, 95% CI=1.23-2.18, p<0.001) were associated with higher hazard for treated recurrence while pembrolizumab (HR=0.83, 95% CI=0.70-0.97) was associated with lower hazard for treated recurrence. For overall survival, being female (HR=0.84, 95% CI=0.76-0.95, p<0.01) was associated with 16% lower hazard of death while time to treatment (HR=1.15, 95% CI=1.13-1.18, p<0.0001) was associated with 15% higher hazard for death. Also, durvalumab (HR=0.29, 95% CI=0.17-0.50, p<0.0001) and pembrolizumab (HR=0.65, 95% CI=0.56-0.76, p<0.0001) were associated with lower hazard of death while atezolizumab (HR=1.49, 95% CI=1.15-1.93, p<0.01) was associated with hazard of death.

Among patients with cutaneous melanoma, there was no significant association between obesity and RFS or between obesity and OS. Covariates associated with RFS included year of diagnosis, time to treatment, and type of ICI initiated. Every increasing year of diagnosis (HR=1.14, 95% CI=1.07-1.22, p<0.001) and month between diagnosis and treatment (HR=1.05, 95% CI=1.02-1.07, p<0.001) were, respectively, associated with 14% and 5% higher hazard of treated recurrence. Compared to Nivolumab, ipilimumab (HR=2.17, 95% CI=1.68-2.81, p<0.0001) was associated with higher hazard of treated recurrence. Additionally, covariates associated with OS included age, race/ethnicity, year of diagnosis, CCI, and time to treatment. Each increasing calendar year of diagnosis (HR=0.74, 95% CI=0.67-0.81, p<0.001) was associated with lower hazard of death while age (HR=1.02, 95% CI=1.00-1.04, p<0.05), being Hispanic (HR=2.81, 95% CI=1.15-6.88, p<0.05), CCI (HR=1.14, 95% CI=1.07-1.21, p<0.001) and time to treatment (HR=1.11, 95% CI=1.08-1.15, p<0.0001) were associated with higher hazard of death.

Table 5 shows the results of the Cox regression for the association between RFS and obesity status as well as OS and obesity status for patients with NSCLC who received EGFRi. Similar to patients who received ICI, obesity was significantly associated with worse RFS, independent of the covariates while there was no significant association with OS. Patients who were obese had 48% higher hazard of treated recurrence, compared to non-obese patients (HR=1.48, 95% CI=1.11-1.97). Covariates associated with RFS included race/ethnicity, year of diagnosis, CCI, time to treatment, and type of EGFRi initiated. Compared to being White, being Black (HR=0.64, 95% CI=0.47-0.87, p<0.01), Hispanic (HR=0.44, 95% CI=0.21-0.92, p<0.05), or of other race/ethnicity (HR=0.79, 95% CI=0.67-0.93, p<0.01) were each associated with lower hazard of treated recurrence. Compared to gefitinib, use of afatinib (HR=2.30, 95% CI=1.12-4.73, p<0.05) was associated with 130% higher hazard of treated recurrence. Each increasing year of diagnosis was associated with 29% increase in hazard of treated recurrence (HR=1.29, 95% CI=1.23-1.37, p<0.0001). Additionally, each unit increase in CCI scores was associated with 9% increase in hazard of treated recurrence (HR=1.09, 95% CI=1.03-1.14, p<0.01), while each month increase in time to treatment was associated with 6% increase in hazard of treated recurrence (HR=1.06, 95% CI=1.03-1.08, p<0.0001). For overall survival, age (HR=1.02, 95% CI=1.01-1.03, p<0.0001) and time to treatment (HR=1.10, 95% CI=1.09-1.12, p<0.0001) were associated with 2% and 10% increased hazard of death while year of diagnosis was associated with 7% decline in hazard of death (HR=0.93, 95% CI=0.90-0.97, p<0.001).

Obesity with associated with RFS but not OS among patients with NSCLC with no association observed among patients with CM. Given that the link between obesity and NSCLC risk remains controversial in the literature, the observed association with RFS may be linked to the absence of the paradoxical effect of improve outcomes that may be seen with cancers caused by obesity such as CM. We propose that the underlying biology maybe different in obese NSCLC patients which needs to be further explored to understand the mechanistic underpinning of this finding which may provide for new targets based on BMI of a NSCLC patient.